5. all published paper

© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimAdv. Mater. 2010, 22, 3729–3734 3729
www.advmat.de
www.MaterialsViews.com
COMMUNICATION
By Halei Zhai, Wenge Jiang, Jinhui Tao, Siyi Lin, Xiaobin Chu, Xurong Xu,
and Ruikang Tang*
Self-Assembled Organic–Inorganic Hybrid Elastic Crystal
via Biomimetic Mineralization
[*] H. Zhai, W. Jiang, J. Tao, S. Lin, X. Chu, Dr. X. Xu, Prof. R. Tang
Center for Biomaterials and Biopathways
and Department of Chemistry
Zhejiang University
Hangzhou, 310027 (P.R. China)
E-mail: rtang@zju.edu.cn
Dr. X. Xu, Prof. R. Tang
State Key Laboratory of Silicon Materials
Zhejiang University
Hangzhou, 310027 (P.R. China)
DOI: 10.1002/adma.201000941
It is generally accepted that biomaterials have unique physi-
cochemical properties.[1] Inspired by biological systems, sci-
entists have been studying biomimetic methods to fabricate
functional materials.[2] Almost all biomaterials possess a
common multi-component feature.[1,3] These composites fre-
quently have ordered organic–inorganic hybrid structures and
their properties are distinct from the individual components.
For example, in a multilayered complex of inorganic aragonite
tablets and organic substrate, the fracture toughness of nacre is
significantly improved to three thousand times greater than that
of synthetic aragonite.[4] Another striking example is biological
bone. In bone, the hydroxyapatite (HAP) phase crystallizes in
the nanoscaled channels formed by the staggered alignment
of the protein matrix. The typical HAP crystals in bone are
plate-shaped with extremely thin thickness (1.5–2 nm), which
is the smallest known dimension of the biologically formed
crystals.[5] In nature the organic and inorganic components inti-
mately associate into well-organized hybrid structures to ensure
optimal strength and flexural stress.[6,7] Therefore, in biomi-
metic designs and fabrications the formation of such ordered
nanostructures is a key challenge.
The formation of inorganic crystals in living organisms is
regulated by the organic matrix. Generally, different organic
templates and additives lead to variety in the morphology, size,
orientation, and assembly of the inorganic crystal by medi-
ating its nucleation and growth.[8,9] Although many organic–
inorganic nanocomposites have been reported,[10] the self-
formation of ultrathin organic–inorganic substructures is still
difficult to achieve by using a simple bottom-up approach. But
the self-formed ordered and intimate combination of organic
additives and inorganic crystals at the nanoscale is a crucial
requirement for bioactive composites.[11] Here we prepare an
organic–inorganic hybrid crystal by the self-assembly of cal-
cium phosphate, surfactant, and protein. This hybrid crystal is
composed of uniform and alternate organic–inorganic layers
at the nanoscale. Both the inorganic crystalline phase and
organic phases in the hybrid crystals have an ultrathin thick-
ness of 1–2 nm. The two ordered components form simultane-
ously during the crystal generation so that they integrate well
with each other to form a superstructure. It is of great impor-
tance that such biomimetic crystals are considerably flexible
and elastic.
It is believed that functional organic molecules can interact
with calcium species at the organic–inorganic interfaces to
modulate the growth and assembly process of the inorganic
crystals. The globular protein bovine serum albumin (BSA),
which comprises a single chain of 583 amino acid residues,
is one of the most studied proteins. It is widely used as a
model protein in many fields including biomimetic miner-
alization.[12] Surfactants are widely applied as the crystalliza-
tion templates in many biomimetic studies.[13] However, the
cooperation of different organic additives has been frequently
overlooked in previous works because of the complicated inter-
actions in the system.[14] Actually, the interactions of a sur-
factant molecule and protein are widely found in biological
systems, for example, the interaction of protein with cell mem-
brane surfactants. The selected two compounds can represent
the protein matrix and special small functional molecules in
biomimetic mineralization studies. Usually, proteins and sur-
factants can form complexes in solution, which are frequently
described by a “necklace bead model”. The micelle-like clus-
ters of surfactants scatter along the polypeptide chains like the
pearls in a necklace.[15] The hydrophilic groups of micelles
are exposed to aqueous solutions and their configuration can
be adjusted. In such protein–surfactant complexes, the protein
is functionalized by the surfactant; meanwhile the aggrega-
tion behavior of the surfactant is also affected by the protein
structure. Here we find that the complex of BSA and an ani-
onic surfactant (sodium bis(2-ethylhexyl) sulfosuccinate, AOT)
could self-assemble into regular rhombus plates with a spe-
cific organic–inorganic substructure in a calcium phosphate
solution.
Scanning electron microscopy (SEM) shows the uniform
rhombic plates formed by the collaboration of calcium phos-
phate, BSA, and AOT (Figure 1a).The typical rhombs are
300–400 nm in the long axis and 200–300 nm in the short axis.
Their typical thickness is 80–100 nm. These rhombs are stable
and their structures can endure in solution or in air for months.
The energy-dispersive X-ray spectroscopy (EDS) reveals the pres-
ence of calcium and phosphate ions in the rhombs; the atomic
ratio of Ca:P is around 1.5. In addition to the elements of C
and O, S was also detected (Figure S1 and Table S1 of the Sup-
porting Information), indicating the presence of AOT (–SO3
2−).
The organic–inorganic hybrid composite was also confirmed by

3730
www.advmat.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 3729–3734
COMMUNICATION
characteristic substructure: two independent sets of diffrac-
tion peaks were detected by using wide-angle X-ray diffraction
(WAXD) and small-angle X-ray diffraction (SAXD) (Figure 1e).
In the small-angle region, a typical reflection characteristic
of lamellar structures is observed. The interspacing distance,
d = 3.12 nm, was calculated by using the reflection peak at
2θ = 2.83° ((001) reflection of the rhomb crystals). The (002)
and (003) reflections were detected at 2θ = 5.71° (d = 1.55 nm)
in SAXD and 2θ = 8.45° (d = 1.05 nm) in WAXD, respectively.
These sharp peaks show the rhombs had a highly ordered
lamellar structure. The other WAXD peaks in the normal range
(2θ > 10°) indicate that the crystallized mineral phase is a HAP-
like phase. These examinations clearly demonstrate that there
are two independent lattice structures within a rhomb crystal. It
is important that the organic and inorganic phases are orderly
arranged to form the hybrid materials rather than the simple
and disordered mixture. By using a side view of the ultrathin
Fourier transform infrared spectroscopy (FTIR). The peaks at
1737, 1459, and 1419 cm−1 are the characteristic peaks of AOT,
while the bands at 1655 (amide I) and 1553 cm−1 (amide II) indi-
cate the presence of BSA. In addition, the broad peaks at 1023 and
567cm−1 areduetothepresenceoftheinorganicphosphategroup
(Figure 1b). Thermogravimetric analysis (TGA) and differential
scanning calorimetry (DSC) showed the presence of 21.4 %
organic component (the organics decomposed at temperatures
of 200–500 °C) and 62.1 % inorganic composite (the residue at
temperatures above 500 °C, Figure 1c). From these results, we
can conclude that the rhombs are the hybrid materials of inor-
ganic (calcium phosphate) and organic phases (BSA and AOT).
The regular rhombs were examined by means of trans-
mission electron microscopy (TEM, Figure 1d). The selected
area electron diffraction (SAED) pattern shows the inor-
ganic phase is in a crystalline form and the pattern is similar
to that of HAP tiny crystallites. Abnormally, the crystal has a
Figure 1. a) SEM image of the rhombs. Inset: enlargement of the rhomb in the white circle. b) FTIR curves of the rhombs (bottom) and AOT (top).
The characteristic peaks for BSA, AOT, and phosphate, are marked as circles, triangles, squares, respectively. c) TGA and DSC analysis under a nitrogen
atmosphere. The weight percentages of water and organic component are labeled. d) Transmission electron microscopy (TEM) image of the rhombs.
Inset: selected area electron diffraction (SAED) pattern corresponding to the white circled area. e) Wide-angle and small-angle (inset) X-ray diffraction
(WAXD and SAXD, respectively) patterns of the rhombs. f) TEM side view of an ultrathin sectioned rhomb.

3731
www.advmat.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimAdv. Mater. 2010, 22, 3729–3734
COMMUNICATION
a strong binding effect with calcium ions as
a result of the highly charged –SO3
2−
groups
(Figure S2, Supporting Information). But the
interaction between calcium and BSA was rel-
atively poor. Since AOT molecules aggregated
onto the BSA chains according to the “neck-
lace bead model”, the local concentrations of
calcium around the BSA–AOT complex were
greater than that in the bulk solution so that
the AOT aggregates on BSA provided the het-
erogeneous nucleation sites for calcium phos-
phate. Moreover, the AOT molecules were
organized by the BSA structure so that the
complexes could induce the ordered assembly
of calcium phosphate. We suggest that the
mineral surfaces also act as the stable solid
substrates for the self-assembly of the BSA–
AOT complex. Thus, the lamellar organic–
inorganic structures could be bottom-up
assembled in the solutions spontaneously.
Accordingly, the substructure of the hybrid
rhombs is the alternate combination of the
ultrathin nanocrystal layer and the BSA–
AOT monolayer (Figure 2b), which is analo-
gous to the nanoscale characteristics of many
natural hybrid composites.[1,3,6,7]
Structured materials are usually asso-
ciated with unique physicochemical and
biological properties.[16]
Both advantages
of inorganic and organic phases can be
present in one hybrid material if these two
components can be well-integrated at the
nanoscale.[17]
Although the main compo-
nent of the rhombs is the crystallized cal-
cium phosphate, a rigid inorganic phase,
flexile and elastic behavior of the hybrid
crystal was obtained. Figure 3a illustrates a side view of a
rhomb: the whole crystal and its organic–inorganic layers are
bent to some extent. Interestingly, a similar bent wave shape
can be seen in the typical organic–inorganic hybrid reinforced
materials such as some polymer–clay nanocoposites.[18]
In
order to confirm the mechanical features of the material, a
force curve examination using atomic force microscopy (AFM,
Figure 3b) was applied. The cantilever was very sensitive to
the tip force and its deflection curve could qualitatively repre-
sent the hardness of the examined surface. In contrast to the
typical sudden and straight force–deflection lines for the rigid
silicon substrate (modulus of 130 GPa, which is similar to that
of pure HAP crystals: 112 GPa[19]
), the loading force increased
smoothly with an increase of the deflection degree of the
AFM cantilever. The buffer effect in the AFM force examina-
tion indicates that the rhombs are not rigid. This characteristic
was similar to that of a typical soft material, polystyrene (PS,
modulus of about 3 GPa). It is interesting that no obvious per-
manent damage or indention point was detected on the rhomb
surface after the loading–unloading cycles (inset of Figure 3b)
in the AFM examination. In order to quantitatively understand
the mechanical properties of the hybrid, a nanoindentation
measurement with a diamond indenter tip was additonally
section of the rhombs under TEM, the lamellar structure is
shown in Figure 1f: the dark region corresponds to the inor-
ganic phase (crystallized calcium phosphate) and the light one is
the organic phase. The individual organic and inorganic phases
are alternately stacked. Each layer structure could be identified
readily at the nanoscale in the hybrid crystal. These two distinct
units are well integrated so that the complete hybrid crystals
can be finally produced at the nanoscale. The thickness of each
organic–inorganic unit is about 3.2 ± 0.2 nm, which is in good
agreement with the calculated d value from the SAXD study.
It is noted that the thickenss of the mineral layer is only about
2 nm; this dimension is close to that of biological ultrathin
HAP crystallites formed between the collagen fibers of bone.
In order to understand the substructure of the rhombs, the
organic component was partially degraded by a 5 % NaOCl solu-
tion. Thus, the mineral layer in the complex could be observed
directly by TEM (Figure 2a). Small crystalline platelets, tens of
nanometers in dimension (length and width), were frequently
observed. In a rhomb crystal, the locations of inorganic crystal-
line platelets are restricted by the adjacent protein–BSA organic
frames. Thus, the continuous inorganic ultrathin layers might
be formed between the frames by using the nanocrystallites.
The conductivity investigations showed that AOT molecules had
Figure 2. a) TEM image of the rhombs etched by 5 % NaOCl; The inset is its fast Fourier trans-
form (FFT) image. b) Substructures of the organic–inorganic rhombs. AOT: small molecules
with round head; BSA: long dark chains; mineral phase: rectangles.

3732
www.advmat.de
COMMUNICATION
could even partially recover during the unloading processes.
In contrast, the unloading curves should be vertical if the solid
phase was rigid.[20] Since the indentation depth was greater
than 20 % of the sample thickness, the Bec model[21]
for a thin
soft material on a hard substrate was applied in the estimation
of the modulus (see details in Supporting Information). By
using the loading–unloading curves, the calculated modulus of
the organic–inorganic rhombs was 6.64 ± 1.41 GPa. This value
was even lower than the modulus of elastic-featured human
vertebral trabeculae, 13.5 ± 2.0 GPa.[22] Similar to biological
bone, both the elastic and hardness features were successfully
integrated by the nanostructured assembly of organic and inor-
ganic ultrathin phases, implying that the hybrid rhombs resolve
the brittleness shortage of inorganic crystals and improve the
material’s toughness. Actually, this is a smart and important
strategy of living organisms to generate functional biomaterials
by means of hybrid nanostructures.
Many research efforts often focus on the controlling effect
of the organic matrix on inorganic mineralization processes,
which mediates the size, morphology, and orientation of inor-
ganic crystals. Such an understanding implies a one-way con-
trol of inorganic phase formation by organic additives. Thus,
the organic templates are often required prior to the controlled
crystallization in order to obtain hybrid materials. However,
this understanding is not suitable in the current case. It was
noted that the BSA–AOT complexes could not form the rhomb
structure spontaneously in calcium solutions. Neither our
experiment nor the published literature detected the BSA–AOT
rhomb in the absence of any mineral phase. Only poorly crystal-
line calcium phosphate spherical particles were obtained if only
BSA was added into the calcium phosphate solution. Besides,
AOT alone resulted in the conventional rod-like HAP crystals
(Figure S3, Supporting Information) without any substructure.
Clearly, the formation of the hybrid rhombs is attributed to the
coexistence of BSA, AOT, and calcium phosphate, which is an
emergent process. As mentioned above, the presence of the
inorganic part also induces the assembly and structure of the
organic components during mineralization.[23] Additionally,
the changes of BSA and AOT concentrations within a certain
range only affects the size and morphology of the resultant
rhombs (Figure S4, Supporting Information). However, their
internal substructure was not altered at all (Figure S5). This
phenomenon could be explained by the regulation effect of
surfactant on the complex assembly, which has been demon-
strated by previous work.[15]
We noted that the assembly process rather than conven-
tional crystal growth occurred in the rhomb formation. No
obvious signal between 50 and 100 nm was observed during
the whole reaction process by dynamic light scattering (DLS,
Figure 4). At the initial stage of crystallization, two individual
distribution peaks existed in the DLS pattern. The small one
(∼20 nm) represented the BSA–AOT building block in the reac-
tion solution (Figure S6, Supporting Information), while the
large one (∼300 nm) belonged to the final product. The frac-
tion of the building block decreased gradually with the reaction,
while the intensity of the product increased. Eventually, only
the final product could be found at the end of the experiment.
The product size did not increase during the reaction. Accord-
ingly, the ex situ electron microscopy studies also demonstrated
performed on the rhombs so that the modulus of the material
could be calculated.[20] The solid and dashed lines represent the
loading and unloading processes, respectively (Figure 3c). The
relatively great indentation depth with different loading forces
from 25 to 40 μN were used to demonstrate the elastic charac-
teristic of the whole nanoplate well. Under such great external
forces, the deformation of the plates was significant. However,
the thin crystals did not collapsed and the depressed surfaces
Figure 3. a) TEM image of ultrathin section of the rhomb. b) Atomic force
microscopy (AFM) force curves of silicon substrate, rhombs (Rh) and poly-
styrene (PS). Cantilever deflection represents the deformation distance
of the sensitive AFM cantilever. Inset: AFM image of the plate after the
loading–unloading cycle. c) The nanoindentation curves of rhombs. The
displacement here means the indentation distance from the surface.

3733
www.advmat.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimAdv. Mater. 2010, 22, 3729–3734
COMMUNICATION
and nanoindentation were prepared by spin-coating 100 μL of slurry on
silicon wafers (3000 rpm). For ultrathin-sectioned TEM examination,
rhombs were embedded in 0.5 mL of epoxy. The mixture was solidified
at 80 °C for 12 h and then carefully microtomed by a Reichert–Jung
Ultracut E using a diamond knife. The typical thickness of the ultrathin
sections was ∼80 nm.
Characterization: SEM was performed by using a HITACHI S-4800
microscope at an accelerating voltage of 5 kV. FTIR spectroscopy (Nicolet
Nexus 670) was used to determine the composition of the products.
Thermogravimetric analysis was carried out by a TA Instrument SDT
Q600. The experiment was measured in a temperature range of 22–600 °C
at a heating rate of 10 °C min−1 under nitrogen atmosphere. TEM
observations were performed by a Philips CM200UT microscope at a
typical accelerating voltage of 160 kV. WAXD and SAXD were carried out
by means of a Rigaku D/max-2550pc instrument with monochromatized
Cu Kα radiation and a scanning step of 0.02°. AFM images were
collected by a Veeco multimode scanning probe microscope with Nano
IVa controller. The measurements were performed using an E head and
a silica tip (Veeco) on a cantilever with a spring constant of 40 N m−1
in tapping mode with filters off, with a scanning rate of 20−60 Hz. The
qualitative measurement of the mechanical properties was performed
by the cantilever deflection in the AFM force curve. The data was
collected for 200 individual force curves on 10 different rhombs. The
nanoindentation measurements were performed by a Tribo-Indenter
In-Situ Nanomechanical Test System with a Berkovich diamond indenter
(tip radius of about 50 nm). The system was calibrated by using fused
quartz before indentation. The data was collected using TestWorks 4
(MTS Systems). The modulus was calculated using the Oliver and Pharr
method and the substrate effect was corrected by the Bec model. The
DLS measurements were taken by using a Brookhaven Instruments 90
Plus particle size analyzer. Conductivity measurements were carried out
by Conducometer DDS-11A at 30 °C. The conductivity electrode was
calibrated using 0.01 M KCl solution prior to use.
Supporting Information
Supporting Information is available online from Wiley InterScience or
from the author.
Acknowledgements
We thank Haihua Pan and Yuan Su for their helpful discussions, Yuewen
Wang, Jieru Wang, Yin Xu, and Xiaoming Tang for assistance in material
characterization techniques. This work was supported by the National
Natural Science Foundation of China (20601023 and 20871102),
Zhejiang Provincial Natural Science Foundation (R407087), the
Fundamental Research Funds for the Central Universities and Daming
Biomineralization Foundation.
Received: March 16, 2010
Revised: April 5, 2010
Published online: July 21, 2010
the absence of intermediate solid or phase during the growth.
The DLS result reveals an abnormal pathway in the organic–
inorganic hybrid material assembly. We suppose that the BSA–
AOT complexes induce the mineral crystallization firstly and
then they are restructured by the mineral phase to form the
alternative layer structure by a cooperative effect. However, the
detailed mechanism needs further investigation.
In this Communication we demonstrate that organic–inor-
ganic hybrid rhombs with a lamellar superstructure can be
self-generated by protein, surfactant molecules, and mineral
phases. Each crystal contains two basic nanoscaled subunits: the
ultrathin inorganic mineral and organic ultrathin layers. These
layers are formed simultaneously and integrate well by self-
assembly to generate the hybrid crystals. During this process
the cooperative effect between the organic and inorganic phases
is key. The ordered organic–inorganic nanostructure confers the
optimum mechanical properties on the resultant hybrid mate-
rial. The current study provides further evidence of the biomi-
metic fabrication of functional materials.
Experimental Section
Materials: Triply distilled CO2-free water was used in the experiment.
Ca(NO3)2 and (NH4)2HPO4 were of analytical grade and their solutions
were filtered twice through Millipore films (0.22 μm) prior to use. BSA
(Albumin Bovine fraction V, BR, purity >98 %) and AOT (Aldrich) were
used directly without further purification.
Preparation: An aqueous solution (100 mL) containing AOT (4 mM)
and BSA (1 mg mL−1
) was prepared. The solution pH was adjusted to
10.0 ± 0.5 at room temperature by ammonia solution (3 M). Ca(NO3)2
solution (50 mL, 5mM, pH = 10.0 ± 0.5) was added to the mixed solution
at a rate of 10 mL min−1
and the solution was stired for 30 min. After that,
(NH4)2HPO4 solution (50 mL, 5mM, pH = 10.0 ± 0.5) was added dropwise
at a rate of 1.5 mL min−1. The slurry was examined by DLS periodically
and the formed solids were collected by high-speed centrifugation at
10 000 rpm. All the solid samples were washed by water three times
and were vacuum-dried at 35 ± 1 °C. Freshly prepared rhombs were
dispersed in ethanol (∼0.5 mg mL−1
) and collected on carbon-coated
copper grids for TEM examination. Samples for AFM measurements
Figure 4. Dynamic light scattering (DLS) size distribution curves at dif-
ferent stages during the emergent formation of rhombs. The percentage
values are calculated by using the statistics of the particle amounts.
[1] S. Mann, Biomineralization: Principles and Concepts in Bioinorganic
Materials Chemistry, Oxford University Press, Oxford 2001.
[2] a) L. P. Lee, R. Szema, Science 2005, 310, 1148; b) C. Sanchez, H.
Arribart, M. M. G. Guille, Nat. Mater. 2005, 4, 277; c) T. Kato, A.
Sugawara, N. Hosoda, Adv. Mater. 2002, 14, 869; d) T. Sun, L. Feng,
X. Gao, L. Jiang, Acc. Chem. Res. 2005, 38, 644.
[3] a) N. Watabe, J. Ultrastruct. Res. 1965, 12, 351; b) L. C. Palmer, C. J.
Newcomb, S. R. Kaltz, E. D. Spoerke, S. I. Stupp, Chem. Rev. 2008,
108, 4754; c) H. O. Fabritius, C. Sachs, P. R. Triguero, D. Roobe,
Adv. Mater. 2009, 21, 391.

3734
www.advmat.de
COMMUNICATION
2005, 285, 562; d) S. Chodankar, V. K. Aswal, P. A. Hassan, A. G.
Wagh, Phys. B 2007, 398, 112; e) T. Chakraborty, I. Chakraborty, S. P.
Moulik, S. Ghosh, Langmuir 2009, 25, 3062.
[16] a) J. Aizenberg, A. Tkachenko, S. Weiner, L. Addadi, G. Hendler,
Nature 2001, 412, 819; b) B. Pokroy, V. Demensky, E. Zolotoyabko,
Adv. Funct. Mater. 2009, 19, 1054.
[17] a) Z. Tang, N. A. Kotov, S. Magonov, B. Ozturk, Nat. Mater. 2003,
2, 413; b) L. J. Bonderer, A. R. Studart, L. J. Gauckler, Science 2008,
319, 1069.
[18] a) P. Podsiadlo, A. K. Kaushik, E. M. Arruda, A. M. Waas, B. S. Shim,
J. Xu, H. Nandivada, B. G. Pumplin, J. Lahann, A. Ramamoorthy, N.
A. Kotov, Science 2007, 318, 80; b) M. A. Priolo, D. Gamboa, J. C.
Grunlan, ACS Appl. Mater. Interfaces 2009, 2, 312.
[19] G. Dewith, H. J. A. Vandijk, N. Hattu, K. Prijs, J. Mater. Sci. 1981, 16,
1592.
[20] W. C. Oliver, G. M. Pharr, J. Mater. Res. 1992, 7, 1564.
[21] a) S. Bec, A. Tonck, J. M. Georges, E. Georges, J. L. Loubet, Philos.
Mag. A-Phys. Condens. Matter Struct. Defect Mech. Prop. 1996, 74,
1061; b) S. Roche, S. Bec, J. L. Loubet, in Mechanical Properties
Derived from Nanostructuring Materials, Vol. 778 (Eds: D. F. Bahr,
H. Kung, N. R. Moody, K. J. Wahl), Materials Research Society,
Warrendale PA 2003, p. 117; c) G. Hochstetter, A. Jimenez, J. P.
Cano, E. Felder, Tribol. Int. 2003, 36, 973.
[22] J. Y. Rho, T. Y. Tsui, G. M. Pharr, Biomaterials 1997, 18, 1325.
[23] M. Antonietti, M. Breulmann, C. G. Göltners, H. Cölfen, K. K. W.
Wong, D. Walsh, S. Mann, Chem. Eur. J. 1998, 4, 2493.
[4] a) J. D. Currey, Proc. R. Soc. London, Ser. B 1977, 196, 443; b) A.
P. Jackson, J. F. V. Vincent, R. M. Turner, J. Mater. Sci. 1990, 25,
3173.
[5] C. Burger, H. W. Zhou, H. Wang, I. Sics, B. S. Hsiao, B. Chu,
L. Graham, M. J. Glimcher, Biophys. J. 2008, 95, 1985.
[6] S. Weiner, H. D. Wagner, Annu. Rev. Mater. Sci. 1998, 28, 271.
[7] D. Liu, H. D. Wagner, S. Weiner, J. Mater. Sci. Mater. Med. 2000, 11,
49.
[8] F. C. Meldrum, H. Cölfen, Chem. Rev. 2008, 108, 4332.
[9] a) G. Falini, S. Albeck, S. Weiner, L. Addadi, Science 1996, 271,
67; b) R. Kniep, S. Busch, Angew. Chem. Int. Ed. 1996, 35, 2624;
c) J. Aizenberg, A. J. Black, G. M. Whitesides, Nature 1999, 398, 495;
d) S. Sadasivan, D. Khushalani, S. Mann, Chem. Mater. 2005, 17,
2765.
[10] a) S. Mann, Nat. Mater. 2009, 8, 781; b) R.-Q. Song, H. Cölfen, Adv.
Mater. 2010, 22, 1301.
[11] H. Gao, B. Ji, I. L. Jäger, E. Arzt, P. Fratzl, Proc. Natl. Acad. Sci. U. S. A.
2003, 100, 5597.
[12] J. Xie, J. Y. Lee, D. I. C. Wang, J. Phy. Chem. C 2007, 111, 10226.
[13] C. E. Fowler, M. Li, S. Mann, H. C. Margolis, J. Mater. Chem. 2005,
15, 3317.
[14] a) L. Qi, J. Li, J. Ma, Adv. Mater. 2002, 14, 300; b) A. Kotachi,
T. Miura, H. Imai, Chem. Mater. 2004, 16, 3191.
[15] a) N. J. Turro, X. Lei, K. P. Ananthapadmanabhan, M. Aronson,
Langmuir 1995, 11, 2525; b) C. K. Ober, G. Wegner, Adv. Mater.
1997, 9, 17; c) S. De, A. Girigoswami, S. Das, J. Colloid Interface Sci.

Controlled formation of calcium-phosphate-based hybrid mesocrystals by
organic–inorganic co-assembly
Halei Zhai,a
Xiaobin Chu,a
Li Li,a
Xurong Xuab
and Ruikang Tang*ab
Received 28th July 2010, Accepted 27th August 2010
DOI: 10.1039/c0nr00542h
An understanding of controlled formation of biomimetic mesocrystals is of great importance in
materials chemistry and engineering. Here we report that organic–inorganic hybrid plates and even
mesocrystals can be conveniently synthesized using a one-pot reaction in a mixed system of protein
(bovine serum albumin (BSA)), surfactant (sodium bis(2-ethylhexyl) sulfosuccinate (AOT)) and
supersaturated calcium phosphate solution. The morphologies of calcium-phosphate-based products
are analogous to the general inorganic crystals but they have abnormal and interesting substructures.
The hybrids are constructed by the alternate stacking of organic layer (thickness of 1.31 nm) and
well-crystallized inorganic mineral layer (thickness of 2.13 nm) at the nanoscale. Their morphologies
(spindle, rhomboid and round) and sizes (200 nm–2 mm) can be tuned gradually by changing BSA,
AOT and calcium phosphate concentrations. This modulation effect can be explained by a competition
between the anisotropic and isotropic assembly of the ultrathin plate-like units. The anisotropic
assembly confers mesocrystal characteristics on the hybrids while the round ones are the results of
isotropic assembly. However, the basic lamellar organic–inorganic substructure remains unchanged
during the hybrid formation, which is a key factor to ensure the self-assembly from molecule to
micrometre scale. A morphological ternary diagram of BSA–AOT–calcium phosphate is used to
describe this controlled formation process, providing a feasible strategy to prepare the required
materials. This study highlights the cooperative effect of macromolecule (frame structure), small
biomolecule (binding sites) and mineral phase (main component) on the generation and regulation of
biomimetic hybrid mesocrystals.
Introduction
Scientists are eager to mimic nature’s ability to design functional
materials whose properties are often superior to the synthetic
ones. In nature, biominerals are widely produced by bacteria,
protists, plants, invertebrates and vertebrates, including
humankind.1
These biological materials are featured by a smart
combination of multi-components especially in the form of
integrated organic–inorganic hybrid materials, in which the
organic parts are often proteins and low-molecular-mass mole-
cules.2
They are constructed by using organic components to
control the nucleation, growth, organization and transformation
of inorganic phases. Interactions between organic and inorganic
phases at the molecular level, although complex, are common
occurrences to determine the size, shape, and properties of the
resulting products.1,3
Different from the synthesized ones, the
functions of biominerals depend to a large extent on the ordered
association of biomolecules with mineral phases. The organized
hybrid materials, unlike the single components, can be tailored
into different compositions and morphologies, e.g. bone,4
tooth5
and mollusc shells6
etc., to ensure the optimal mechanical and
physicochemical characteristics.
The controls that determine the sizes, shapes, and properties of
crystals are a key to addressing numerous challenges in material
designs and applications. It has been revealed that organic
molecules can influence the shape and properties of inorganic
crystals.7
However, it is difficult for the two distinct organic and
inorganic phases to spontaneously assemble into highly ordered
structures. In living organisms, biological mineralization is able
to combine particular building blocks or entities into functional
hybrid composites. An understanding of these biochemical
controls is essential and important, not only to study
biomineralization mechanisms further, but also to design novel
hybrid materials and processing technologies. Despite the
complicated hierarchical structures of biominerals, their basic
building blocks are frequently the nano-sized organic–inorganic
composites.8
Therefore, an ordered and periodic assembly of
organic and inorganic nanophases at the nanoscale is crucial to
biomimetically synthesize hybrid materials. But, how can we
design ordered hybrid composites and how can we conveniently
control their structures, sizes and morphologies under mild
conditions?
Although organic–inorganic hybrid materials have been
approached by various methods such as layer-by-layer (LbL)9
and template-directed crystallization,10
the bottom-up fabrica-
tion from ions or molecules is still a great challenge in the
laboratory since the control of periodic deposition is difficult to
achieve at hierarchical scales. In conventional biomimetic crys-
tallization studies, organic molecules, which act as structure-
directing agents, modulate the crystal morphology by their
a
Centre for Biomaterials and Biopathways, Zhejiang University,
Hangzhou, Zhejiang, 310027, China. E-mail: rtang@zju.edu.cn; Fax:
+86-571-87953736; Tel: +86-571-87953736
b
State Key Laboratory of Silicon Materials, Zhejiang University,
Hangzhou, Zhejiang, 310027, China
2456 | Nanoscale, 2010, 2, 2456–2462 This journal is ª The Royal Society of Chemistry 2010
PAPER www.rsc.org/nanoscale | Nanoscale
Publishedon13October2010.Downloadedon24/01/201605:58:33. View Article Online / Journal Homepage / Table of Contents for this issue

selective absorption onto crystal faces, altering crystal facet
stability and growth kinetics.7,11
Recently, a non-classical crystal
growth pathway based upon nano assembly has received
considerable attention.12
The nanoparticles, which are directed
by specific organic additives, can act as the basic building units to
assemble into superstructures or mesocrystals. During such
a process, the organic molecules (especially macromolecules)
selectively absorb and interact with primary nanocrystals. The
assembly process follows programmed arrangement into high
order hybrid structures.13
The morphology can be tuned by
varying the interactions between different organic and inorganic
phases. However, the one step bottom-up process, which starts
from the molecular level rather than from preformed nano-
particle precursors, may be readily able to control the orientation
and order of assembly processes to form integrated hybrid
nanocomposite. But this strategy requires a precise and sponta-
neous co-assembly of both organic and inorganic phases
alternately at both the molecular level and the nanoscale.14
In this paper, we reported an easy but effective method for
direct synthesis of organic–inorganic hybrid mesocrystals by
a emergent co-assembly process of protein (bovine serum
albumin (BSA)) and surfactant (sodium bis(2-ethylhexyl) sulfo-
succinate (AOT)) in a supersaturated calcium phosphate solu-
tion. The calcium-phosphate-based hybrid crystals with lamellar
structure have different properties from conventional ones. Here
we emphasize that the size and morphology of the resulting
hybrids could be regulated readily by varying BSA, AOT and
calcium phosphate concentrations according to a suggested
morphological ternary diagram. This study provided a novel
pathway to one-pot preparation of functional hybrid crystal
materials with tuneable size and morphologies by organic–inor-
ganic co-assembly.
Results and discussion
It is believed that functional organic molecules can interact with
calcium species at the organic–inorganic interfaces to modulate
the growth and assemble of inorganic crystals. BSA is one of the
most studied proteins but this biological macromolecule is not an
effective modifier in calcium phosphate crystallization.15
It has
been previously confirmed that the interaction between BSA and
calcium or phosphate ions in aqueous solutions is poor.16
BSA
itself is inert in mineral deposition. In contrast, many surfactant
molecules are widely used as effective promoters and templates in
biomimetic calcium mineralization since their hydrophilic groups
(especially the sulfonate and carboxylate groups) provide active
binding sites to calcium ions. AOT is one among typical agents
that can modulate calcium phosphate precipitation significantly.
AOT molecules have a strong binding effect with calcium ions
due to their highly charged -SO3
2À
groups.16,17
However, hierar-
chical or complicated biomineral-like structures cannot be
achieved by using this small molecule due to the lack of higher-
order structures. In our control experiments, only poor crystal-
line HAP was obtained if BSA was added into the supersaturated
calcium phosphate solutions; AOT alone produced the conven-
tional rod-like HAP crystals without any organized hybrid
structure. These results matched the previous studies and
understandings well. However, the cooperative effect of BSA and
AOT in the calcium phosphate solution could lead to the
formation of unique hybrids in a one-pot reaction.
Under an experimental condition of 2 mM AOT, 1 mg mlÀ1
BSA and 1.25 mM calcium ions (the molar ratio of calcium to
phosphate was fixed at 1.67 in all experiments), the uniform
rhombic plates precipitated spontaneously as shown by scanning
electron microscopy (SEM, Fig. 1(A)). Their size distribution
was homogeneous. The typical rhombic plates were 1.23 Æ 0.21
and 0.91 Æ 0.18 mm along their long and short axes, respectively
(statistical results from $100 plates); the aspect ratio was about
1.4. The thickness of the plates was 130 Æ 20 nm. These rhombic
plates had exactly same morphology (Fig. 1(B)) and this char-
acteristic was similar to the general inorganic crystals. However,
the chemical compositions of the obtained plates were relatively
complicated. Besides the elements of calcium and phosphorus,
the element of sulfur was detected in the solids by using energy-
dispersive X-ray spectroscopy (EDS). This result indicated the
presence of AOT (-SO3
2À
) in the hybrid plates. It was also
revealed that inorganic part in the plates was a kind of calcium
phosphate minerals with Ca : P molar ratio of 1.5–1.6. The
coexistence of organic–inorganic components was also
confirmed by Fourier transform infrared spectroscopy (FT-IR,
Fig. 1(C)). The peaks at 1737, 1459 and 1419 cmÀ1
were the
characteristic signals of AOT, while the bands at 1656 (amide I)
and 1555 cmÀ1
(amide II) showed the involvement of BSA in the
solids.18
The broad peaks at 1022 and 564 cmÀ1
were assigned to
the inorganic phosphate groups.19
Thermogravimetric analysis
(TGA) showed that the mineral phase was the main composition
in the solids. The weight loss of 38% between 100 and 500
C was
corresponded predominantly to removal of the organic phase,
while the weight contents of the inorganic phases were 62%. In
addition, the plates became ‘crimped-paper’-like after calcina-
tions at 500
C in air for 2 h. Without the organic frame, the
solids became brittle and the structures were collapsed readily
into small pieces under an ultrasonic condition. Many previous
studies suggested that the organic compounds play a regulation
role in inorganic mineralization rather than being involved in
Fig. 1 (A) SEM image of the rhombic plates. (B) Enlarged image of the
rhombic plate in the white circle; the double-headed arrow shows the
extended orientation. (C) FT-IR pattern of the products. (D) The
rhombic plates after calcination at 500
C in air.
This journal is ª The Royal Society of Chemistry 2010 Nanoscale, 2010, 2, 2456–2462 | 2457
Publishedon13October2010.Downloadedon24/01/201605:58:33. View Article Online

structural recombination. However, the current results implied
that BSA and AOT were the key components in the hybrid
construction. Thus, these solids were different from the other
precipitated inorganic crystals in the presence of organic addi-
tives.
The resulting rhombic plates shared the same size and aniso-
tropic morphology similar as general inorganic crystals.
However, in-depth examination revealed that they were distinct
from the conventional calcium phosphate crystals.20
The
rhombic plates were examined by wide angle X-ray diffraction
(WAXD, Fig. 2(A)) and as expectated, the crystalline HAP-like
calcium phosphate phase was detected. The WAXD pattern was
very similar to that of pure HAP but small peak shifts were also
observed. We suggested that the binding effect between the
organic component and calcium ions would cause the lattice
distortion. The lattice structure of the inorganic phase could be
revealed at the atomic scale by using high resolution transmission
electron microscopy from a top view of the plates (HRTEM,
Fig. 2(B)). This image represents a typical ultrathin inorganic
crystal layer embedded in the rhombic plates. However, another
independent set of diffraction peaks was found in the X-ray
diffraction (XRD) pattern, which revealed that a superstructure
was present in the hybrids. The characteristic peaks of lamellar
structure (interspacing distance, d ¼ 3.43 nm) could be found
from both small angle X-ray diffraction (SAXD) and WAXD
(Fig. 2(A)), indicating an ordered arrangement of subunits along
a crystallographic direction rather than a simple mixture of the
organic and inorganic phases. A side view of the ultra-thin
sectioned samples under transmission electron microscopy
(TEM, Fig. 2(C)) confirmed the internal structure: the organic
layers (light, 1.31 nm) and the inorganic layers (dark, 2.13 nm)
alternately stacked at the nanoscale to form the compact hybrid
structure. Thus, the organic molecules (BSA and AOT) were well
organized to form the layered organic phase. Each organic–
inorganic ultra-thin unit had a thickness of 3.44 nm, which
agreed with the XRD data, 3.43 nm, and the individual inorganic
layer was a calcium phosphate crystal plate with a thickness of
only 2.13 nm. These nanoplates acted as the building blocks that
could self-assemble together with the organic layers to generate
the lamellar complex. Additionally, a wave-like superficial
texture of the hybrids could be observed (Fig. 1(B)) and the
profiles were similar to the hybrid crystal morphology. This
phenomenon indicated that the assembly might be an anisotropic
process.
In order to understand the orientation of each inorganic layer,
selected area electron diffraction study (SAED, Fig. 2(D)) was
applied. It was noted that the anisotropic diffraction dots rather
than the isotropic diffraction rings were obtained during the
examination of a whole rhombic crystal, which represented
a similar characteristic of single crystal. It was interesting that the
orientation reflected by these dots (arrow in the insert image) was
exactly same as the long axis of the examined rhombic crystal.
Such a coincidence implied that all the ultrathin inorganic crystal
layers within the hybrid plates should share the same crystallo-
graphic orientation. Additionally, the experimental diffractions
dots of the whole crystal were almost same as the fast Fourier
transform (FFT) result (Fig. 2(B)) of an individual crystal layer.
Therefore, the formed hybrid crystal exhibited similar features to
a single crystal; however, it had additional superlattice structure.
Since the rhombic plates had a specific morphology while they
were not constructed as the conventional single crystals, these
hybrids could be considered as a kind of artificial meso-
crystal.12,21
However, the imperfect dots on Fig. 2 (D) might
indicate that the misaligned orientation still occurred during
nano assembly. Since the material was constructed by ultrathin
calcium phosphate units, it was interesting that flexible and
elastic features were conferred onto the mesocrystal along the
lamellar packing direction in spite of that; its main composition
was a brittle ceramic phase. These mechanical properties of the
hybrids had been characterized by our previous study,16
demonstrating the advantages of organized assembly for
formation of mesocrystals in material functionalization.
The convenient control of the size and morphology of the
organic–inorganic hybrids and mesocrystals is a challenge,
although those for single hybrid crystals are nowadays sophis-
ticated. In our experiments, the calcium phosphate–BSA–AOT
hybrid mesocrystals with different size and morphology could be
feasibly regulated within a simple reaction system by changing
the reactant concentrations. We fixed BSA and calcium
concentration at 0.5 mg mlÀ1
and 1.25 mM, respectively. When
the AOT concentration was 1.00 mM, the obtained hybrid plates
were not rhombic plates any more. Their shapes became spindle-
like. The hybrid plates changed into a round shape when the
AOT concentration was increased to 4.00 mM. However, the
further decreasing or increasing of AOT concentration result into
the disappearance of the co-assembly or hybrid in the system. In
this experiment, their morphologies were gradually adjustable
from spindle, to rhombus to round by increasing the AOT
concentration from 1.00 to 4.00 mM (Fig. 3). During the
evolution process, the length along the short axis of the formed
Fig. 2 (A) WAXD and SAXD (insert) patterns of the rhombi;
(B) HRTEM of a rhombus (top view). Insert: FFT simulation result;
(C) TEM image of ultra-thin sectioned rhomb from side view. The values
of 2.13, 1.31 and 3.44 nm corresponded to the thicknesses of inorganic
(dark), organic (light) and organic–inorganic complex layers, respec-
tively. Insert: TEM image of the side view of the ultra-thin sections of the
plates, bar is 0.5mm. (C) is the enlargement of the region within the white
circle; (D) TEM image of the hybrids. Insert was the SEAD pattern
(white circle area). The HRTEM image in (B) was also obtained on the
same area by the in situ technique. Arrows showed that each individual
inorganic plate in the hybrid shared the same crystallographic orienta-
tion, which was the long axis of the rhombus.

hybrid plates did not change significantly, it was maintained at
300–400 nm. However, the long axis kept on decreasing from
1.50 mm to 300–400 nm with increasing AOT concentration.
Accordingly, the hybrid morphology became isotropic. This
phenomenon implied that AOT component was an important
factor to control the a degree of anisotropic co-assembly of the
hybrids.
Although the morphologies and sizes of the resulted hybrid
plates were influenced remarkably by the changing of AOT
concentrations in the reaction solutions, the internal organic–
inorganic subunit remained. The WAXD and SAXD patterns of
the spindles, rhombi and rounds were exactly same without any
change. But the misalignments of each individual inorganic layer
in the hybrid increased with the increasing of AOT concentra-
tion. The crystallographic mismatch of the inorganic layers could
be examined by using SEAD. During the evolution from the
regular rhombi to round shapes, the diffraction dots disappeared
gradually while the diffraction rings existed (Fig. 4). This
tendency indicated that the preferred orientation of the thin
calcium phosphate planes in the hybrid was weakened. Although
AOT itself could result in aggregates in solution to induce
calcium mineralization, the aggregation was simple and isotropic
due to the lack of complicated configuration. Therefore, it was
reasonable that the excessive AOT could destroy the anisotropic
assembly of the ultrathin mineral plates in the hybrid rhombi.
Although the inorganic and organic layers were still packed
layer-by-layer strictly along the thickness direction, the crystal-
lographic directions of the inorganic crystal planes in the hybrids
became disordered. The anisotropic assembly transformed the
orientation of the long axes into the isotropic mode with
increasing AOT concentration; thus, the round plates were
finally yielded at 4.00 mM AOT and the hybrid was not meso-
crystalline any more. Besides, it should be mentioned that the
percentages of organic and inorganic contents in the hybrid
solids was not changed significantly during the morphology
modulation; in which the inorganic content was kept within
a range of 69–72% from the spindles to the rounds.
Besides the AOT concentrations, the formation of hybrid
crystals could be also adjustable by BSA concentration. In this
examination, the concentrations of AOT and calcium were
maintained at 2 mM and 1.25 mM, respectively, and the BSA
concentrations were increased from 0.25 to 2.00 mg mlÀ1
. It was
noted that the morphologies of hybrid plates underwent another
gradual evolution from the irregular quadrilaterals to rhombi
and then to plump spindles (Fig. 5). The sizes and aspect ratios of
the hybrids increased from 200 nm to 2 mm and 1.1 to 2.0,
respectively, during the modulation. Although the hybrid width
increased along the short axis, the more extended length along
the long axis indicated that the anisotropy assembly process was
affected significantly by the protein concentration. It was noticed
that in biomineralization, the complicated hierarchical building
structures of biominerals are frequently contributed by the
ordered aggregates of proteins. Again, the basic organic–inor-
ganic units and their ordered packing behaviours were not
changed during the morphology and size regulations. It was
mentioned that, when the BSA concentration increased, the role
of AOT in the synthesis decreased. Therefore, the ratio or the
Fig. 3 SEM images of the hybrids synthesised at AOT concentrations of
1.00 (A), 2.00 (B) and 4.00 mM (C). (D)–(F) are the corresponding XRD
patterns of (A)–(C), respectively.
Fig. 4 During the morphology change from rhombus (A) to round (B),
anisotropic diffraction dots became isotropic rings in the corresponding
SEAD pattern.
Fig. 5 SEM images of the hybrids at BSA concentration of 0.25 (A),
1.13 (B) and 2.00 mg mlÀ1
(C).

cooperative effect of BSA and AOT was another key factor in
mesocrystal formation and regulation.
It was known that the co-assembly could not occur in the
absence of the inorganic phase. Thereby, it was reasonable that
the concentrations of calcium and phosphate could control the
mesocrystal formation too (Fig. 6). Under BSA and AOT
concentrations of 1.00 mg mlÀ1
and 2.00 mM, respectively, the
resulting rhombi shared the same intermediate state with
increasing calcium and phosphate concentration in the reaction
solution. If calcium concentration was decreased to 0.63 mM, the
poly-dispersed quadrilaterals-like plates (size of 400–800 nm)
formed with the small aspect ratio of 1.1. If the concentration
was increased to 2.50 mM, the slender spindle-like plates were
obtained and their size distribution was 1.8–2.3 mm with an
aspect ratio was 2.5. From the evolution from quadrilaterals,
rhombi to slender spindles, it could be seen that the anisotropic
co-assembly process was enhanced.
The previous studies of biomimetic fabrication of hybrid
materials with artificial molecules such as peptide-amphiphile,22
block copolymer,23
and amphiphilic dendro-calixarene,24
sug-
gested that the specific sites and sterically constrained effect may
control the assembly of the organic template and then the size
and morphology of the final hybrid materials. Different from the
above-mentioned understanding, under our experimental
conditions, the change of BSA and AOT concentrations were
directly related to the different modification state of BSA. The
BSA protein, which was constituted by a single chain of
583 amino acid residues, acted as a stable and relatively rigid
fragment connected with the special motif (AOT aggregates).25
The hydrophilic groups of aggregates exposed to aqueous solu-
tions and their configuration can be adjusted. The highly charged
group (–SO3
2À
) in AOT could greatly interact with calcium ions
and then modulated calcium phosphate precipitation
significantly, which had been demonstrated experimentally in
many works and in our previous paper.16,26
However, the binding
ability of BSA with calcium ions is weak and the controlling
effect on the mineral formation is relatively poor. As a result,
BSA acted as structural frame while the AOT aggregates
provided the nucleation sites of mineral during the co-assembly
process. In the current study, BSA macromolcules combined
with smaller AOT molecules to form a BSA–AOT complex and
such a modified protein could effective control the crysallization
and assembly of the calcium phosphate mineral. To some extent,
this method provides an efficient way to turn a non-mineraliza-
tion protein into a mineralization protein by using surfactants.
The conformation of the macromolecules restricts the assembly
only along certain specific directions. However, the larger
concentration of AOT is accompanied by an increase in the
amount and size of AOT aggregates, offering more sites for
the assembly process.27
As a result, the controlling effect from the
protein was counteracted and the assembly process could happen
at more directions to form the isotropic rounds. Furthermore,
increasing the amount or the relative amount of BSA concen-
trations partly restricted the assembly process in specific prefer-
ential orientations by spatial configuration to form the
anisotropic hybrids or mesocrystals.28
Thus, the co-assembly
process preferred to occur in certain directions, especially along
the long axis of the hybrid plates rather than the short axis.
Although the short axis partly extended under some experi-
mental cases, the greatly increase along the long axis resulted into
the spindles-like mesocrystal formation. The competitive
controlling effect of BSA and AOT led to the transformation of
an isotropic and anisotropic assembly process during hybrid
crystal construction. Thus, the formation of different hybrids
and mesocrystals with tuneable size and morphologies could be
achieved.
An anisotropic co-assembly process could also be promoted by
increasing the mineral ion concentrations. In the formation
process of mesocrystals, the inorganic precursor controlled the
size and morphology of the final product by tuning the amounts,
size and shapes of the nano-sized building blocks.29
Under our
experimental conditions, the controlling role of mesocrystal
growth became dominant in greater saturation to decide the
product size and structure. As the preferred orientation of
the calcium phosphate crystal plates is parallel to the long axis of
the rhombic plates, the fast growth of the calcium phosphate
plate crystals along this preferred orientation promoted the
formation of the slender spindle-like plates with larger aspect
ratios during the co-assembly process. However, the interaction
between BSA-AOT complex and calcium phosphate crystal was
also responsible for the co-assembly of the organic and inorganic
phase to form highly ordered hybrid materials and maintain their
internal structure.
Actually, the generation of hybrid material via the cooperative
effect of macromolecules (mainly proteins), small biomolecules
and the mineral phase is a common strategy in natural bio-
mineralization.30
In the biological construction, high-molecular-
weight macromolecules, such as collagen, act as support matrix
to provide a structural frame for the mineralization, the
biomineralization proteins themselves have nucleation sites
but most matrices receive mineralization function by binding
and stabilizing functional motifs that are carboxylate- or
Fig. 6 SEM images of the hybrids at calcium concentration of 0.63 (A),
1.56 (B) and 2.50 mM. In all experiments, the ratio of calcium to phos-
phate in the reaction solution was maintained at 1.67.

sulfonate-rich. Thus, the combination of organic–inorganic
mineralization interfaces and the organized organic matrices can
concentrate the mineral ions to induce the deposition as well as to
regulate the size, morphology and orientation of the inorganic
building blocks to form integrated organic–inorganic hybrid
composites with complicated structure. We suggest that in this
system, BSA is the structural frame to control the anisotropic
assembly; the adsorption of AOT onto BSA enhances the
mineralization ability of the protein; and the mineral acts as an
inorganic conjunction phase to solidify the organic–inorganic
hybrid structure. In the experiments, the increase of BSA
promoted the formation of larger hybrid plates with increased
aspect ratio, while AOT exhibited the opposite controlling effect.
The increasing of inorganic concentrations preferred the
formation of slender hybrid plates with a larger size. In order to
show the controlling effect of the reactant concentrations, the
simplified morphological maps in the form of solution ternary
diagrams was proposed (Fig. 7). The biomimetic formation
hybrid and mesocrystals could be yielded in the grey region. In
the specific regions, the formed hybrid plates had a similar size
and morphology. From points A to B, the increase of aspect ratio
was preferred as the hybrid rounds transformed into the spindle
ones. Since the anisotropic assembly behaviour was enhanced,
this evolution implied that the resulting mesocrystals became
more organized and the mismatch degrees of the inorganic layers
in the hybrids could be reduced. From points A to C, both the
size and aspect ratio of the resulted hybrids were increased and
their morphologies were changed from rounds to spindles. From
points B to C, the hybrids turned from wide spindles to slender
spindles with increased size and aspect ratio too. By using this
morphological ternary diagram, we could design readily hybrids
and mesocrystals with the required size and morphology.
Conclusions
We demonstrate that the ordered and uniform hybrids or mes-
ocrystals can be biomimetically synthesized by the co-assembly
of proteins, small functional molecules and minerals using
a simple one-pot reaction. Their size distributions and
morphologies can be adjusted by varying the component
concentration in reaction solutions. The anisotropic co-assembly
of the BSA–AOT complex and ultrathin calcium phosphate
crystal plates is a key to the control of mesocrystal formation.
A morphological ternary diagram can be used to design different
hybrid materials as requireed. This work may give another
inspiration to the assembly of multi components into one inte-
grated hybrid material with a highly ordered structure.
Furthermore, the bottom-up pathway of controlled fabrication
may be developed as a simple and effective strategy to prepare
feasibly functional hybrid and mesocrystal materials.
Experimental
Materials
Triply distilled water was used in all the experiments. Ca(NO3)2
and (NH4)2HPO4 were of analytical and their solution were
filtered twice using 0.22mm Millipore films prior to use. BSA
(Albumin Bovine fraction V, BR, purity 98%, LABMAX) and
AOT (Aldrich) were used without any further purification.
Hybrid plate preparation
Using a typical experiment as an example, 100 ml aqueous
solution containing 4 mM AOT and 0.20 g BSA was mixed with
50 ml Ca(NO3)2 solution (5mM). The solution pH was adjusted
to 10.0 Æ 0.5 at room temperature by 3 M ammonia solution.
Then 50 ml (NH4)2HPO4 solution (3 mM, pH ¼ 10.0 Æ 0.5) was
added dropwise at a rate of 1.5 ml minÀ1
. The reaction solution
contained 2.00 mM AOT, 1.00 mg mlÀ1
BSA, 1.25 mM Ca(NO3)2
and 0.75 mM (NH4)2HPO4. The mixture was gently stirred at
30 Æ 1
C for 24 h. The precipitated solids were collected by
centrifugation at 6000 rpm. The solid were washed by water for
three times and were vacuum-dried at 35 Æ 1
C. In order to
examine the controlling effect of reactant concentrations on
hybrid formation, different concentrations of AOT, BSA and
calcium phosphate ions were used and all the experimental
processes were the same.
Fig. 7 Controlled synthesis of hybrids by a morphological ternary diagram. The co-assembly occurred within the grey area and the formation of
mesocrystals was preferred in its left and bottom sections. The typical morphology of the final products were also demonstrated. Bar ¼ 1mm.

Characterizations
SEM was performed by using a HITACHI S-4800 at a typical
acceleration voltage of 5 kV. FT-IR spectra (Nicolet Nexus 670)
were applied to analysis the hybrid compositions. WAXD and
SAXD were characterized by a Rigaku D/max-2550pc with
monochromatized Cu-Ka radiation; the scanning step was 0.02
.
TGA was performed by a TA Instrument SDT Q600. The
experiment was measured in a temperature range from room
temperature to 1000
C under nitrogen atmosphere. TEM
observations were performed by a CM200UT TEM (Philips) at
an acceleration voltage of 160 kV. During the ultra-thin
sectioned TEM examination, rhombi were embedded in epoxy.
The mixture was solidiﬁed at 80
C for 12 h and then carefully
microtomed by a Reichert-Jung Ultracut E using a diamond
knife.
Acknowledgements
We thank Jieru Wang, Xinting Cong, Xiaomin Tang, Yin Xu
and Linshen Chen for their help with characterization, Haihua
Pan and Yuan Su for discussions. This work was supported by
the Fundamental Research Funds for the Central Universities,
National Natural Science Foundation of China (20871102),
Zhejiang Provincial Natural Science Foundation (R407087) and
Daming Biomineralization Foundation.
Notes and references
1 S. Mann, Biomineralization: Principles and Concepts in Bioinorganic
Materials Chemistry, Oxford University Press, 2001.
2 L. Bedouet, F. Rusconi, M. Rousseau, D. Duplat, A. Marie,
L. Dubost, K. Le Ny, S. Berland, J. Peduzzi and E. Lopez, Comp.
Biochem. Physiol., Part B: Biochem. Mol. Biol., 2006, 144, 532–543;
J. L. Arias and M. a. S. Fernacndez, Chem. Rev., 2008, 108,
4475–4482.
3 C. E. Killian and F. H. Wilt, Chem. Rev., 2008, 108, 4463–4474;
J. S. Evans, Chem. Rev., 2008, 108, 4455–4462.
4 S. Weiner and H. D. Wagner, Annu. Rev. Mater. Sci., 1998, 28,
271–298.
5 S. Busch, U. Schwarz and R. Kniep, Chem. Mater., 2001, 13,
3260–3271.
6 N. Watabe, J. Ultrastruct. Res., 1965, 12, 351–370.
7 F. C. Meldrum and H. C€olfen, Chem. Rev., 2008, 108, 4332–4432.
8 R. Z. Wang, Z. Suo, A. G. Evans, N. Yao and I. A. Aksay, J. Mater.
Res., 2001, 16, 2485–2493; H. J. Gao, B. H. Ji, I. L. Jager, E. Arzt and
P. Fratzl, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 5597–5600.
9 Z. Tang, N. A. Kotov, S. Magonov and B. Ozturk, Nat. Mater., 2003,
2, 413–418; P. Podsiadlo, A. K. Kaushik, E. M. Arruda, A. M. Waas,
B. S. Shim, J. Xu, H. Nandivada, B. G. Pumplin, J. Lahann,
A. Ramamoorthy and N. A. Kotov, Science, 2007, 318, 80–83.
10 N. Gehrke, N. Nassif, N. Pinna, M. Antonietti, H. S. Gupta and
H. C€olfen, Chem. Mater., 2005, 17, 6514–6516; P. H. Kithva,
L. Grondahl, R. Kumar, D. Martin and M. Trau, Nanoscale, 2009,
1, 229–232.
11 N. A. J. M. Sommerdijk and G. d. With, Chem. Rev., 2008, 108,
4499–4550.
12 R. Q. Song and H. C€olfen, Adv. Mater., 2010, 22, 1301–1330.
13 M. Li, H. C€olfen and S. Mann, J. Mater. Chem., 2004, 14, 2269–2276.
14 S. Mann, Nat. Mater., 2009, 8, 781–792.
15 R. I. Martin and P. W. Brown, J. Mater. Sci.: Mater. Med., 1994, 5,
96–102; K. L. Yadav and P. W. Brown, J. Biomed. Mater. Res., 2003,
65a, 158–163.
16 H. Zhai, W. Jiang, J. Tao, S. Lin, X. Chu, X. Xu and R. Tang, Adv.
Mater., 2010, 22, 3729–3734.
17 C. E. Fowler, M. Li, S. Mann and H. C. Margolis, J. Mater. Chem.,
2005, 15, 3317–3325.
18 G. Falini, S. Weiner and L. Addadi, Calcif. Tissue Int., 2003, 72,
548–554.
19 S. J. Gadaleta, E. P. Paschalis, F. Betts, R. Mendelsohn and
A. L. Boskey, Calcif. Tissue Int., 1996, 58, 9–16.
20 J. Song, V. Malathong and C. R. Bertozzi, J. Am. Chem. Soc., 2005,
127, 3366–3372; A. Ethirajan, U. Ziener, A. Chuvilin, U. Kaiser,
H. C€olfen and K. Landfester, Adv. Funct. Mater., 2008, 18,
2221–2227; Y. Zhang and J. Lu, Cryst. Growth Des., 2008, 8,
2101–2107.
21 A.-W. Xu, M. Antonietti, S.-H. Yu and H. C€olfen, Adv. Mater., 2008,
20, 1333–1338.
22 J. D. Hartgerink, E. Beniash and S. I. Stupp, Science, 2001, 294,
1684–1688; V. M. Yuwono and J. D. Hartgerink, Langmuir, 2007,
23, 5033–5038.
23 Z. H. Nie, D. Fava, E. Kumacheva, S. Zou, G. C. Walker and
M. Rubinstein, Nat. Mater., 2007, 6, 609–614; H. Wang, A. J. Patil,
K. Liu, S. Petrov, S. Mann, M. A. Winnik and I. Manners, Adv.
Mater., 2009, 21, 1805–1808.
24 M. Kellermann, W. Bauer, A. Hirsch, B. Schade, K. Ludwig and
C. B€ottcher, Angew. Chem., Int. Ed., 2004, 43, 2959–2962.
25 N. J. Turro, X. G. Lei, K. P. Ananthapadmanabhan and M. Aronson,
Langmuir, 1995, 11, 2525–2533; S. De, A. Girigoswami and S. Das,
J. Colloid Interface Sci., 2005, 285, 562–573.
26 S. Sarda, M. Heughebaert and A. Lebugle, Chem. Mater., 1999, 11,
2722–2727.
27 C. K. Ober and G. Wegner, Adv. Mater., 1997, 9, 17–31.
28 H.-A. Klok, J. F. Langenwalter and S. Lecommandoux,
Macromolecules, 2000, 33, 7819–7826; X. Kong and S. A. Jenekhe,
Macromolecules, 2004, 37, 8180–8183; L. Rubatat, X. Kong,
S. A. Jenekhe, J. Ruokolainen, M. Hojeij and R. Mezzenga,
Macromolecules, 2008, 41, 1846–1852; A. Sacnchez-Ferrer and
R. Mezzenga, Macromolecules, 2010, 43, 1093–1100.
29 H. C€olfen and M. Antonietti, Angew. Chem., Int. Ed., 2005, 44,
5576–5591.
30 N. Kroger, R. Deutzmann, C. Bergsdorf and M. Sumper, Proc. Natl.
Acad. Sci. U. S. A., 2000, 97, 14133–14138; L. C. Palmer,
C. J. Newcomb, S. R. Kaltz, E. D. Spoerke and S. I. Stupp, Chem.
Rev., 2008, 108, 4754–4783.

Spontaneously amplified homochiral organic–inorganic
nano-helix complexes via self-proliferation†
Halei Zhai,a
Yan Quan,a
Li Li,a
Xiang-Yang Liu,b
Xurong Xuc
and Ruikang Tang*ac
Most spiral coiled biomaterials in nature, such as gastropod shells, are homochiral, and the favoured chiral
feature can be precisely inherited. This inspired us that selected material structures, including chirality,
could be specifically replicated into the self-similar populations; however, a physicochemical
understanding of the material-based heritage is unknown. We study the homochirality by using calcium
phosphate mineralization in the presence of racemic amphiphilic molecules and biological protein. The
organic–inorganic hybrid materials with spiral coiling characteristics are produced at the nanoscale. The
resulted helixes are chiral with the left- and right-handed characteristics, which are agglomerated
hierarchically to from clusters and networks. It is interesting that each cluster or network is homochiral
so that the enantiomorphs can be separated readily. Actually, each homochiral architecture is evolved
from an original chiral helix, demonstrating the heritage of the matrix chirality during the material
proliferation under a racemic condition. By using the Ginzburg–Landaue expression we find that the
chiral recognition in the organic–inorganic hybrid formation may be determined by a spontaneous
chiral separation and immobilization of asymmetric amphiphilic molecules on the mineral surface, which
transferred the structural information from the mother matrix to the descendants by an energetic
control. This study shows how biomolecules guide the selective amplification of chiral materials via
spontaneous self-replication. Such a strategy can be applied generally in the design and production of
artificial materials with self-similar structure characteristics.
1 Introduction
Through long time periods of evolution, most spiral coiled bio-
materials in nature, like gastropod shells, adopt a specic
homochirality.1,2
For example, the majority of current gastropod
shells have a right-handed (R-) coiling pattern (Fig. 1a).3
The
chiral minority was eliminated eventually by a frequency-
dependent selection and the dominant one proliferated.4,5
This
inspires us that materials with specic structural properties,
like chirality, can spontaneously develop into a large self-similar
community.6
Biologically, it is accepted that regularly expressed
biomolecules, together with inorganic minerals, constitute the
physical chirality of gastropod offsprings under the guidance of
a controlling gene.7
For instance, at the growth front of shells,
the tiny chitin nanocrystals behave as the amphiphilic mole-
cules and self-assemble into the liquid crystal layers (Fig. 1b).8
Fig. 1 The chirality of gastropod shells and a schematic drawing of the shell
mineralization front. (a) General gastropod species have right-handed shells. (b)
During the natural generation of shell structure b-chitin molecules assemble into
supermolecules (chitin crystallites) and their liquid-crystal layers induce the spiral
mineralization of calcium carbonate (this scheme is prepared based upon a
mechanism proposed by Cartwright et al.).8,9
a
Centre for Biomaterials and Biopathways and Department of Chemistry, Zhejiang
University, Hangzhou, 310027, China. E-mail: rtang@zju.edu.cn; Fax: +86 571-
8795-3736
b
Department of Physics and Department of Chemistry, National University of
Singapore, Singapore 117542, Singapore
c
Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou, Zhejiang
310027, China
† Electronic supplementary information (ESI) available: Supporting gures and
tables. See DOI: 10.1039/c3nr33782k
Cite this: Nanoscale, 2013, 5, 3006
Received 23rd November 2012
Accepted 29th January 2013
DOI: 10.1039/c3nr33782k
www.rsc.org/nanoscale
Nanoscale
PAPER
Publishedon06February2013.DownloadedbyUniversityofCalifornia-SanFranciscoon24/01/201605:50:31.
View Article Online
View Journal | View Issue

These chitin layers provide growing sites for the inorganic
phase and modulate the mineralization together with related
proteins. Thus, it follows that the spiral micro-pattern consti-
tuted by a chitin–calcium carbonate lamellar structure is grad-
ually constructed (Fig. 1b).9
In this sense, an understanding on
the physicochemical regulations of the organic–inorganic bio-
inspired materials with selective chirality will advance our
knowledge in chemistry and materials sciences. A challenging
question to be addressed is whether we can mimic the self-
evolution (symmetry breaking) process of shells in our labora-
tories so that the chiral materials can be separated and propa-
gated to generate self-similar articial production.
It has been demonstrated that some organic molecules can
control the morphology of biominerals, like calcium phosphate
and calcium carbonate crystals.10
The chiral organic molecules
usually act as templates to control the crystal morphology,
rather than incorporated organic composition to constitute the
chiral hybrid materials.11,12
In the articial design of chiral
nanomaterials, a variety of dispersed chiral superstructures,
such as nano-helixes and nano-tubes, can be generated with the
twisted assembly of chiral molecules, or even nano-sized crys-
talline units.13
However, each nano-helix or nano-tube is con-
structed by independent assembly, rather than a successive
proliferation procedure to pass down the chirality and nal
formation of the homochiral complex. As a result, the archi-
tecture of a homochiral material complex is rarely achieved.14
Herein, by employing a racemic mixture of a chiral amphiphile
(bis-(2-ethylhexyl) sulfosuccinate sodium salt, AOT) and bovine
serum albumin (BSA) in supersaturated calcium phosphate
solution, two kinds of chiral organic–inorganic hybrid nano-
helixes (L- and R-enantiomers) can spontaneously form and
each kind of chiral helix eventually proliferates into a larger
homochiral helix complex. We feel that such an experimental
phenomenon may be relevant to the proliferation of chiral
materials.
2 Experimental section
2.1 Materials
Triply distilled CO2-free water was used in the experiment.
Ca(NO3)2 and (NH4)2HPO4 were of analytical grade and their
solutions were ltered twice through 0.22 mm Millipore lms
prior to use. BSA (Albumin Bovine fraction V, BR, purity 98%)
and AOT (Aldrich, racemic mixture) were directly used without
further purication.
2.2 Preparation of the homochiral nano-helix complex
The temperature during all the synthesis processes was main-
tained at 30 Æ 1
C. Briey, a 100 ml aqueous solution con-
taining 1 mM AOT and 1 mg mlÀ1
BSA was prepared. The
solution pH was adjusted to 10.0 Æ 0.5 by 3 M ammonia solu-
tion. 50 ml Ca(NO3)2 solution (5 mM, pH ¼ 10.0 Æ 0.5) was
added. Then 50 ml (NH4)2HPO4 solution (3 mM, pH ¼ 10.0 Æ
0.5) was added dropwise at a rate of 1.5 ml minÀ1
. The solution
was gently stirred for 10 h and the formed solids were
collected by centrifugation at 3600 rpm. All the solid samples
were washed by water three times and were vacuum-dried at
35 Æ 1
C. Freshly prepared samples were dispersed in ethanol
($0.5 mg mlÀ1
) and collected on carbon-coated copper grids for
TEM examinations. In the seed growth experiment, 1/20 percent
of the obtained product underwent intense ultrasonic treat-
ment (KUDOS, 35 kHz, 20 min) and the helix clusters or
networks were collapsed into dispersed helixes. Then the
dispersed helixes were added as seeds into freshly prepared
reaction solutions and the reaction solutions were collected by
centrifugation and observed with Transmission Electron
Microscopy (TEM). For ultrathin sectioned TEM examination,
dried samples were embedded in 0.5 ml epoxy. The mixture was
solidied at 80
C for 12 h and then carefully microtomed by a
Reichert-Jung Ultracut E ultramicrotome using a diamond
knife.
2.3 Au-labelled BSA absorptions
BSA-Au nanoparticles were synthesized according to the work by
J. Xie et al.15
Briey, 5 ml 10 mM HAuCl4 solution was added into
5 ml 50 mg mlÀ1
BSA solutions and stirred for 5 min. Then,
0.5 ml 1 M NaOH was added and the solution was kept at 37
C
for 24 h. The product was dialyzed with 1000 ml distilled water
for 24 h. The BSA-Au was used instead of pure BSA in order to
probe the location of BSA on the surface of the helix.
2.4 Examination of calcium concentrations
The concentration of free calcium ions in BSA, AOT, BSA + AOT
solutions were measured by a PCa-1 calcium ion selective
electrode with a saturated calomel electrode as the reference
electrode. The electrode was calibrated according to the
instructions before use.
2.5 Characterizations
Scanning electron microscopy (SEM) was performed by using a
HITACHI S-4800 eld-emission scanning electron microscope
at an acceleration voltage of 5 kV. Fourier-transform infrared
spectroscopy (FT-IR, Nicolet Nexus 670) was used to determine
the composition of the products. Thermogravimetric analysis
(TGA) was carried out by a TA Instrument SDT Q600. The
experiments were measured over a temperature range of 22–
800
C at a rate of 10
C minÀ1
under air atmosphere. TEM
observations were performed by a JEM-1200EX at a typical
acceleration voltage of 80 kV. Small angle X-ray diﬀraction
(SAXRD) and Wide angle X-ray diﬀraction (WAXRD) were char-
acterized by a Rigaku D/max-2550pc with monochromatized Cu
Ka radiation and the scanning step was 0.02
. Solid state
nuclear magnetic resonance (ssNMR) was kindly performed by
Prof. Jarry Chan's group at the National Taiwan University on a
Bruker DSX300 NMR spectrometer.
3 Results and discussion
3.1 Structure and composition of the nano-helix
In our biomimetic case, AOT and BSA were adopted as the
models for biological amphiphilic and proteins, respectively.
AOT is of asymmetric double-chain amphiphile (Fig. S1†). It can
Paper Nanoscale
View Article Online

assemble into various mesomorphous phases, which have been
widely used in biomimetic crystallization.16
BSA is one of the
common proteins in biomineralization studies.17
The syner-
gistic effect of AOT and BSA on calcium phosphate minerali-
zation gave rise to the formation of nano-helix (Fig. 2 and S2†).
In the control experiments, the use of AOT and BSA alone only
generated the calcium phosphate nanorods and nanospheres,
respectively (Fig. S3†). Clearly, the helix formation was attrib-
uted to the coexistence of AOT, BSA and calcium phosphate.
It followed that the individual helixes could further develop
into micron-sized aggregated clusters (Fig. 2a and b). In an
individual cluster, the nano helixes extended radially outward
from a dense core, indicating the successive proliferation
procedure. Furthermore, some clusters connected with each
other to form a larger network (Fig. 2 and S4†). As the basic
building blocks of the clusters, the helixes were chiral and they
had two kinds of spiral enantiomers, L- and R-forms. Although
the overall amounts of the L- and R-helixes in the reaction
system were equal (Fig. S5†), only one helix enantiomer could be
identied within a cluster or connected network (Fig. 2a–c, see
more in ESI†). This suggested that the spontaneous chiral
recognition and chiral separation occurred during the cluster
and network generation. Concerning the composition and
structure of the helixes, they were constituted by organic and
mineral phases, which accounted for 20.5 wt% and 58.8 wt%,
respectively (the rest 20.7% was attributed to absorbed and
crystal water, Fig. S6†). FT-IR (Fig. 2d) and energy-dispersive
X-ray spectroscopy (EDS, Fig. S7†) revealed that the main
components in the helixes were AOT and a calcium phosphate
phase. X-ray diffraction (Fig. 2e) showed that the mineral phase
was close to brushite. Moreover, the mineral phase in the helix
was conrmed by Multiple Pulse Sequence Nuclear Magnetic
Resonance spectroscopy (CRAMPS-NMR) and a Heteronuclear
Correlation (HETCOR) spectrum between the 31
P and 1
H nuclei
31
P{1
H} combined rotation, indicating that the phosphate
groups were protonated (HPO4
2À
) in the calcium mineral
(Fig. 2f, S8 and Table S1†). The NMR data indicated the absence
of PO4
3À
in the complex. As a result, the calcium phosphate
species containing PO4
3À
groups such as hydroxyapatite, octa-
calcium phosphate and tri-calcium phosphate could be
excluded in the phase analysis. Additionally, Ca(H2PO4)2 could
not be precipitated under our experimental conditions due to
its high solubility. Both examinations shows that the signals of
the helix were close to those of brushite. Therefore, the brush-
ite-like mineral was considered as the primary inorganic
component in the helixes.
The internal structure of the nano-helixes could be consid-
ered as the alternative and spiral stacking of thin calcium
phosphate phase and AOT bilayers. The cross-section images of
the nano-helix showed that the thickness of the wall of the
nano-helix was about 2.1–2.2 nm (Fig. 3a–c). Furthermore,
SAXRD and WAXRD results also showed the alternative lamellar
superstructure in the helixes with a constant interspacing
distance (d ¼ 3.34 nm). The lamellar structure was also
demonstrated by TEM (Fig. 2e): the dark lines (1.7 nm) and light
lines (1.6 nm) correspond to the inorganic calcium phosphate
and organic AOT ultrathin layers, respectively. The thickness of
each organic–inorganic hybrid unit was about 3.3 Æ 0.2 nm,
which is in agreement with the d value calculated from the
SAXRD data. In the spiral helix, there existed a pitch angle of
about 43
between the strip edge and the long axis. It was noted
that the AOT molecules preferred to assemble into a bilayer
structure. Concisely, the organic bilayer could have a thickness
of about 1.6 nm if the molecules tilted by 43
. There were two
mirror forms for both the helix pitch angle and the AOT tilt
angle, +43
or À43
, as the denitions (Fig. 3d and e). The
mirror packing of AOT corresponded to the formation of R- and
L-enantiomers of the helixes. Apart from AOT, a small amount
of BSA was detected in the helix by FT-IR and 13
C{1
H} NMR
(Fig. S9†). Using nano Au particle labelled BSA as the imaging
agent, we found that the protein did not incorporate into the
hybrid inner structure, but absorbed onto the helix wall
surfaces (Fig. S10†), which might be due to its relatively large
dimension.19
We suggested that BSA served as a surface or
Fig. 2 Characterizations of nanohelixes. (a and b) Homochiral clusters consisting
of R-helixes and L-helixes, respectively. (c) A homochiral helix network; circles
indicate the cluster centres; inset is a magnification of the rectangular region. (d)
FT-IR of the helix and pure AOT. The typical and undisturbed peaks of AOT
(1750 cmÀ1
), BSA (1540 cmÀ1
, amino) and phosphate ions could be noted. The
peak located at 2342 cmÀ1
was generally attributed to CO2 from the air during the
FT-IR determination. (e) SAXRD and WAXRD patterns of the helixes. d ¼ 3.34 nm
and d ¼ 1.65 nm represent the first and the second diffractions of the lamellar
structure in the helixes, respectively. WAXRD also showed that the mineral phase
was similar to brushite. The XRD peaks of 11.3
and 31.0
were close to those of
brushite (020) and (121), respectively. The case of the small left-shift of charac-
teristic peaks could be found in small nanocrystals.18
The inset TEM image shows
each organic (light line)–inorganic (dark line) unit in the helix. (f) 31
P{1
H}HETCOR
spectra between the 31
P and 1
H nuclei measured in the helixes. The spectra was
acquired at a spinning frequency of 10 kHz and the contact time was set to 2.5 ms.
A total of 64 transients with an increment of 100 ms was accumulated.
Nanoscale Paper
View Article Online

structure stabilizer for the hybrid spiral strips. An experimental
fact was that no chiral product formed by using BSA alone.
During biomineralization, some biomacromolecules can adopt
an extended conformation when they interact with the inor-
ganic phase surface.20
Therefore, there was a possibility that the
BSA molecules might incorporate into the helix using their
extended forms. However, we could conrm that the AOT
molecules with highly charged sulphuric groups, rather than
BSA, were the primary organic composition in the nano-helix.
For example, the FT-IR study showed a very weak amino peak
(Fig. 2d) in the composites, implying the ignorable contents of
BSA in comparison with the strong AOT bands. Therefore, it was
suggested that the hybrid helixes were formed by the ordered
assembly of the calcium phosphate mineralized layer and the
AOT bilayer, and were stabilized by BSA absorption on the
hybrid surfaces. In this architecture, the assembly behaviour of
chiral AOT molecules in the hybrid helixes determined the
material’s chirality.
3.2 Proliferation of the nano-helixes
Originally, the homochiral clusters and networks evolved from a
single nano-helix (mother matrix). The preformed chiral nano-
helix spontaneously passed down the structure information
(chirality) from one generation to the next and then generated
the homochiral complexes (Fig. 4). Firstly, tiny hybrid buds
sprouted from the surface of the matrix (Fig. 4a). Both organic
and inorganic parts of the buds directly integrated with the
corresponding parts of mother matrix. This could be considered
as a kind of matrix outgrowth. Secondly, these hybrid buds grew
longer and twisted into the helical ribbons. At this stage, the
organic and inorganic parts at the growing front of hybrid buds
did not integrate with mother matrix any more. At this time,
there should be a choice in the twist direction (L- or R-).
Nevertheless, we noted that the newly formed helical ribbons
replicated precisely the twist direction (chirality) of the mother
matrix. This meant that the mother matrix induced the later
AOT molecules to assemble into a coherent packing direction,
even the AOT bilayers at the budding region and growth front
were separated by calcium phosphate layers. Thus, the original
structure was inherited through the budding and proliferation
process (Fig. 4b). Thirdly, the homochiral proliferation process
of the helixes continued by generating more “daughters” and
“grand-daughters” based upon the matrix. Due to the space
limitation, the newly formed helixes tended to stretch outward,
which generated radial homochiral clusters (Fig. 4c). Finally, a
few of the helixes at the cluster edge acted as “bridges” to
provide additional growing sites for new buds and initiated
another proliferation process (Fig. 4e). This new proliferation
Fig. 3 (a and b) Cross-section images of nano-helixes under TEM. (c) Schematic
structure of the nano helixes (dark grey: inorganic phase; light grey: organic
phase). (d and e) TEM and schemes of the R- and L-helix. The width of the AOT
bilayer is 1.6 nm from TEM observation. As AOT molecules have a length of 1.1
nm, AOT molecules in a bilayer should arrange with a tilt angle of about 43
.
Note: the AOT molecules in the same bilayer are simply treated as direct contact
and this small variation of tilt angle doesn't aﬀect our qualitative analysis.
Fig. 4 TEM images of the evolution from a single helix to a homochiral cluster
and then a homochiral network (community). (a) A sprouting bud from R-helix
matrix for the new “daughter” helix generation. (b) Growth and twist of the
“daughter” helix, which duplicated the chiral feature to be R-form; insets show
the details of the growth front on the matrix. (c) More buds formed and they
replicated the structure of matrix precisely. (d) Rudiment of the homochiral helix
cluster; insets: magniﬁcation of the branching sites. (e) Homochiral helix cluster
(R-form); arrows indicate the proliferation directions of the cluster; inset shows
the new buds formed at an extended helix. (f) Homochiral helix networks (R-
form); arrows show the proliferation directions.
Paper Nanoscale
View Article Online

could happen at multiple directions (Fig. 4f). Through the self-
repeating processes, a single nano-sized helix eventually evolved
into a large homochiral complex (network) at the micrometer
scale (Fig. 4f). In each network, the chirality of the newly born
helixes was precisely “inherited” from the original mother
matrix from generation to generation, which can be considered
as a spontaneous process of material-based self-proliferation.
We note that it is impossible for the dispersed helixes aer
intense ultrasonic treatment to aggregate into homochiral
clusters again. However, aer the dispersed helixes are re-
dispersed into the freshly prepared reaction solutions as seeds,
the time to induce the formation of helix clusters can be rela-
tively reduced according to the seed amount, indicating that the
mother helix acts as the seed to induce the proliferation of new
helixes to form helix clusters (Fig. S11†). As a result, in each
network, the chirality of the newly born helixes was precisely
“inherited” from the original mother matrix from generation to
generation, which can be considered as a spontaneous process
of material-based self-proliferation.
3.3 Model for the homochiral material
Analogous to the shell formation, the co-assembly of organic and
inorganic phases is restricted at the local domains of the growth
front (Fig. 4a and b). AOT molecules are more able of binding
calcium ions than BSA (Table S2†) and these amphiphilic
molecules greatly modulate the growth of calcium phosphate
species.21
In this case, the helix formation and replication are
controlled dominantly by the assembly behaviours of AOT at the
growth front. The AOT molecules can tightly absorb onto the
surface of calcium phosphate species with a strong binding
effect between calcium ions and sulphuric groups, which facili-
tates the assembly of AOT bilayers.22
Unlike the free AOT mole-
cules in aqueous solution, the relatively rigid CaP crystal, rather
than the mobile water layer, can imobilize the adjacent AOT
bilayers. Thus, the hybrid structure could be ‘solidied’ and
stabilized with a decreasing disordered uctuation of AOT
molecules comparing the free state, which then induces the next
layer mineralization.23
This alternative and cooperated deposi-
tion of the AOT bilayer and calcium phosphate phase layer
gradually constitutes a thin AOT–calcium phosphate hybrid
strip, which is similar to the associated assembly of lipids and
inorganic phase reported by Seddon, et.al.13b
In our work, the chiral AOT molecules are responsible for the
twist of the hybrid strip to form the L- and R-nano-helixes.
Although the BSA used here is constituted with chiral L-amino
acids, the chirality of nano-helixes are unlikely to be controlled
by BSA. Due to its large size, it is difficult to incorporate into the
ordered structure with ultra-small units of 1.7 nm (calcium
phosphate layer) and 1.6 nm (organic layer), while the twisted
arrangement of these units forms the chirality at the nanoscale.
In addition, the equal number of L- and R-nano-helixes also
indicates that the BSA with single chiral units (L-amino acids)
has little contribution to the chirality of the nano-helixes. Many
works have been reported that strips constituted with chiral
molecules tend to twist into nano-helixes to reduce the elastic
energy.24
Similarly, the chirality of the nano-helixes in our
system is determined by the assembly behaviour of chiral AOT
molecules.
However, the racemic mixture generally dilutes the chiral
interaction between the chiral molecules, so that the chiral
superstructure might fail to form.25
Nevertheless, some studies
have shown that both R- and L-enantiomers can emerge in
racemic systems if an energy favoured chiral phase separation
occurs, especially for the lipids with chiral headgroups and
inexible double chains structures.26
Interestingly, AOT owns a
similar structure and phase behaviour to these lipids.27
More-
over, chiral molecules can also undergo a phase separation
when they are restricted at interfaces.28
Therefore, in our
system, it follows that a spontaneous chiral phase separation of
amphiphilic AOT may occur on the calcium phosphate mineral
substrate, resulting in the bilayers with exactly the same
molecular packing behaviour.
Due to the complicated structure, the conformation infor-
mation (chirality) of AOT in each nano-sized helix is difficult to
identify. Besides, methods of the synthesis or separation of AOT
diastereoisomers is rarely reported.29
Based upon the mirror
arrangement of AOT in L- and R-helixes, we divide the AOT
molecules into two types with different tilt directions of +43
or À43
. Aer this simplication, only the tilt angle needs to be
taken into account in the qualitative analysis of the energy
during the formation process. The AOT molecules in the bila-
yers have two different tilt angles, +43
or À43
, which can be
considered as the enantiomers to induce R- and L-chiral helix
formations, respectively (Fig. 3d and e). The favoured tilt angle
should maintain the same value during the alternative dispo-
sition. Thus, the energy favoured recognition is a key to main-
tain the molecular assembly according to the chiral breaking
model supposed by Selinger et al.30
The model suggests that the
elastic energy of the strip can be reduced by a chiral separation
even under racemic conditions. An order parameter, j, is
introduced, which is treated as the local net amount of right-
handed minus le-handed molecular packing here. The elastic
free energy, F, of the thin chiral bilayer strip can be written as
eqn (1)
F ¼
ð
dS

1
2
k

1
r
2
þ
1
2
k0

1
r
2
cos2
f À lHPj

1
r

sin f cos f
þ
1
2
KðVjÞ2
þ
1
2
tj2
þ
1
4
uj4

þ Eedge (1)
where, S is the area, the rst term is the standard Helfrich
bending energy of the hybrid membrane and the coefficient k is
the isotropic rigidity. In the right side of eqn (1), the second
term represents the anisotropy of the rigidity and the coefficient
k0
is the anisotropic term and f is the title angle of the chiral
molecules (Fig. 5a); the third term is a chiral term that favours
twisting in a tilt angle f; the coefficient lHP, is the chirality
parameter, which exists only in chiral membranes and depends
on the chiral order. The sign of lHP can be changed when the
membrane transforms into its mirror image. lHPj increases
with the greater chiral phase separation degree of j. The last
three terms in the bracket are the Ginzburg–Landau expres-
sions in powers of j, which represent the free energy change
Nanoscale Paper
View Article Online

during the ordering transition. The values of K and u are
temperature independent constants while the coefficient of t
relates to temperature and t 0 for chiral phase separation.30,31
Because AOT molecules in the helix have the xed tilt angle of
43
, the domain wall energy on the edge is a constant.
When simultaneously minimizing the free energy over tilt
angle and radius, r, in eqn (1), the following is obtained,
f0 ¼ arctan

k þ k0
k
1=
4

(2)
r0 ¼
k
1=
4
ðk þ k0
Þ
1=
4
h
k
1=
2
þ ðk þ k0
Þ
1=
2
i
lHPj
(3)
in which, (k + k0
)/k represents the energy cost for the ratio of the
bend parallel to the tilt direction to bend perpendicular to
the tilt direction. In our system, its value is about 0.76 because
fAOT ¼ 43
, which indicates that the hybrid strip favours
twisting parallel to the AOT tilt direction. Besides, the radius of
the nano-helix equals 1.33k/lHPj. Usually, the lipid amphi-
philes form helical tubes or a helix with a larger diameter of
hundreds nanometres or even a few micrometers. Since the
organic–inorganic hybrid structure exists in our helixes, it is
reasonable that the rigidity coefficient k should be greater than
single component chiral lipid membranes. As a result, lHPj
must be signicantly greater to produce such a slender helix
with a very small radius ($10 to 25 nm).
We note that the favoured energy barrier DF plays an
essential role to control the twist direction and chiral prolifer-
ation of nano-helixes (eqn (1)). Here, the qualitative description
energy barrier DF between the racemic state (j ¼ 0) and sepa-
ration state (jmÆ) can be described by eqn (4),30
DF ¼
ð
dS

lHPjmÆ

1
r

sin f cos f þ
1
2
KðVjmÆÞ2
þ
1
2
tjmÆ
2
þ
1
4
ujmÆ
4

(4)
As the radius of the nano-helix is r f 1/lHPj, the relatively
small radius of the helix (10–25 nm) implies that lHPj is of great
value in our case, which facilitates the chiral separation. The
equation shows that the free energy of the strip has two local
minima representing the two types of energy favoured AOT
packing with the mirror symmetry (Fig. 5b) if chiral phase
separation occurs. Without chiral phase separation, the strip
cannot twist into a helix because the radius becomes innite
when j ¼ 0 (r / N).
Fig. 5b shows the constant arrangement of AOT molecules
(constant tilt angle of +43
or À43
) within a strip is energeti-
cally preferred due to an energy barrier. First, the energy barrier
DF can promote the formation of energy favoured and stable
nano-helixes, rather than unstable hybrid strips. If the different
arrangements of AOT molecules (L- and R-) coexist in the same
strip, the elastic energy increases so that the resulting strip
becomes unstable (Fig. 5b, middle). Therefore, the elastic
energy cannot be reduced and generate twisted nano-helixes. By
contrast, the same arrangements of AOT molecules (L- or R-) can
successfully reduce the unfavoured elastic energy and twist to
form L- or R-nano-helixes, respectively (Fig. 5b, le and right).
Second, the energy barrier of DF is also responsible for the
homochiral proliferation. In our system, the preformed helix
matrix has an inductive effect on the sequent proliferation
because the emerging organic and inorganic parts in the new
buds directly extend from their mother matrix. Thus, new buds
share the same AOT packing form with the mother matrix. The
same AOT packing can be replicated under the guidance of the
mother matrix due to the favoured energy reduction, which
means that L- to L- or R- to R-proliferation is a preferential way.
Subsequently, the buds grow following the determined AOT
packing to form a new chiral helix with the same chirality. For
example, the new buds generated from the R-nano-helixes in
Fig. 4 faithfully adopt the R-twist direction and keep the selected
form during the growth process. The mutated proliferation of
L- to R- or R- to L- also require extra energy to overcome DF in
comparison with the matched L- to L- or R- to R-. Accordingly,
the chiral structure proliferation always initiates at the pre-
formed helixes and amplies the chiral structure from the
mother matrix to subsequent generations. Finally, large
homochiral complexes (helix clusters and networks) can be
generated under the guidance of the energy controlled recog-
nition of AOT packing.
4 Conclusions
This study reveals that the homochiral complex of the organic–
inorganic hybrid helix can form via a self-proliferation process.
The energy controlled chiral recognitions and separations of
asymmetric chiral AOT molecules are essential in both helix
formation and homochiral proliferation. The nding is of
importance to approach homochiral biomimetic materials in
the laboratory. We expect this strategy of bio-inspired chiral
structure proliferation can be developed into a convenient
pathway for the articial synthesis of self-similar functional
materials.
Acknowledgements
We thank Prof. Jerry Chen for the ssNMR studies, Dr Jinhui Tao,
Dr Haihua Pan and Yuan Su for discussions, Hua Wang, Jieru
Fig. 5 (a) The geometry of AOT molecules in the helix discussed in eqn (1) for helix
formation. (b) Two local minima of the elastic free energy (F) with symmetry
packing (jm+ or jmÀ) lead to an energy barrier of DF, which ensures the oriented
packing vector of AOT bilayers to produce chiral helix and homochiral proliferation.
Paper Nanoscale
View Article Online

Wang, Yin Xu and Xiaoming Tang, Xinting Cong, Yalin Li for
characterizations. This work was supported by the National
Natural Science Foundation of China (91127003) and the
Fundamental Research Funds for the Central Universities.
1 (a) M. J. French, Invention and evolution: design in nature and
engineering, Cambridge Univ Pr, 1994; (b) S. Mann,
Biomineralization: Principles and Concepts in Bioinorganic
Materials Chemistry, Oxford University Press, 2001.
2 (a) R. Ueshima and T. Asami, Nature, 2003, 425, 679–679; (b)
C. Grande and N. H. Patel, Nature, 2008, 457, 1007–1011.
3 M. Schilthuizen and A. Davison, Naturwissenschaen, 2005,
92, 504–515.
4 (a) G. J. Vermeij, A natural history of shells, Princeton Univ Pr,
1995; (b) P. Y. Parkhaev, Paleontological Journal, 2007, 41,
233–240.
5 (a) O. Lipton, Malacologia, 1979, 19, 129–146; M. S. Johnson,
Heredity, 1982, 49, 145–151; (b) T. Asami, R. H. Cowie and
K. Ohbayashi, Am. Nat., 1998, 152, 225–236.
6 (a) S. Mann and G. A. Ozin, Nature, 1996, 382, 313–318; (b)
H. Cölfen and S. Mann, Angew. Chem., Int. Ed., 2003, 42,
2350–2365; (c) S. Zhang, Nat. Biotechnol., 2003, 21, 1171–
1178.
7 R. Kuroda, B. Endo, M. Abe and M. Shimizu, Nature, 2009,
462, 790–794.
8 J. H. E. Cartwright and A. G. Checa, J. R. Soc. Interface, 2007,
4, 491–504.
9 J. H. E. Cartwright, A. G. Checa, B. Escribano and C. I. Sainz-
D´ıaz, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 10499–
10504.
10 (a) M. Antonietti, M. Breulmann, C. G. Göltner, H. Cölfen,
K. K. W. Wong, D. Walsh and S. Mann, Chem.–Eur. J.,
1999, 4, 2493–2500; (b) A. Bigi, B. Bracci and S. Panzavolta,
Biomaterials, 2004, 25, 2893–2899; (c) D. Gebauer,
H. Cölfen, A. Verch and M. Antonietti, Adv. Mater., 2008,
21, 435–439; (d) L. Wang and G. H. Nancollas, Chem. Rev.,
2008, 108, 4628.
11 (a) G. Falini, M. Gazzano and A. Ripamonti, J. Mater. Chem.,
2000, 10, 535–538; (b) C. Orme, A. Noy, A. Wierzbicki,
M. McBride, M. Grantham, H. Teng, P. Dove and
J. DeYoreo, Nature, 2001, 411, 775–779; (c) M. Olszta,
E. Douglas and L. Gower, Calcif. Tissue Int., 2003, 72, 583–
591; (d) O. Casse, K. Kita-Tokarczyk, A. H. E. Müller,
W. Meier, A. Taubert, O. Casse, K. Kita-Tokarczyk,
A. H. E. Müller, W. Meier and A. Taubert, Faraday Discuss.,
2008, 139, 179–197.
12 (a) C. Göbel, P. Simon, J. Buder, H. Tlatlik and R. Kniep,
J. Mater. Chem., 2004, 14, 2225–2230; (b) T. Tsuji, K. Onuma,
A. Yamamoto, M. Iijima and K. Shiba, Proc. Natl. Acad. Sci.
U. S. A., 2008, 105, 16866–16870; (c) Y. J. Wu, T. W. T. Tsai
and J. C. C. Chan, Cryst. Growth Des., 2012, 12, 547–549.
13 (a) T. Kunitake and N. Yamada, J. Chem. Soc., Chem.
Commun., 1986, 655–656; (b) A. M. Seddon, H. M. Patel,
S. L. Burkett and S. Mann, Angew. Chem., Int. Ed., 2002, 41,
2988–2991; (c) H. Imai and Y. Oaki, Angew. Chem., 2004,
116, 1387–1392; (d) T. Shimizu, M. Masuda and
H. Minamikawa, Chem. Rev., 2005, 105, 1401–1444.
14 (a) M. Fischlechner and E. Donath, Angew. Chem., Int. Ed.,
2007, 46, 3184–3193; (b) H. Robson Marsden and A. Kros,
Angew. Chem., Int. Ed., 2010, 49, 2988–3005.
15 J. Xie, Y. Zheng and J. Y. Ying, J. Am. Chem. Soc., 2009, 131,
888–889.
16 (a) C. E. Fowler, M. Li, S. Mann and H. C. Margolis, J. Mater.
Chem., 2005, 15, 3317–3325; (b) H. Chen, Z. Tang, J. Liu,
K. Sun, S. R. Chang, M. C. Peters, J. F. Manseld,
A. Czajka-Jakubowska and B. H. Clarkson, Adv. Mater.,
2006, 18, 1846–1851.
17 (a) G. Yin, Z. Liu, J. Zhan, F. Ding and N. Yuan, Chem. Eng. J.,
2002, 87, 181–186; (b) J. Xie, J. Y. Lee and D. I. C. Wang,
J. Phys. Chem. C, 2007, 111, 10226–10232.
18 R. Yogamalar, R. Srinivasan, A. Vinu, K. Ariga and A. C. Bose,
Solid State Commun., 2009, 149, 1919–1923.
19 A. Wright and M. Thompson, Biophys. J., 1975, 15, 137–
141.
20 Y. Y. Kim, L. Ribeiro, F. Maillot, O. Ward, S. J. Eichhorn and
F. C. Meldrum, Adv. Mater., 2010, 22, 2082–2086.
21 (a) S. Sarda, M. Heughebaert and A. Lebugle, Chem. Mater.,
1999, 11, 2722–2727; (b) M. Bujan, M. Sikiric, N. Filipović-
Vinceković, N. Vdović, N. Garti and H. Füredi-Milhofer,
Langmuir, 2001, 17, 6461–6470.
22 (a) Z. Li, A. Weller, R. Thomas, A. Rennie, J. Webster,
J. Penfold, R. Heenan and R. Cubitt, J. Phys. Chem. B, 1999,
103, 10800–10806; (b) M. S. Hellsing, A. R. Rennie and
A. V. Hughes, Langmuir, 2011, 27, 4669–4678.
23 Z. X. Li, J. R. Lu, R. K. Thomas, A. Weller, J. Penfold,
J. R. P. Webster, D. S. Sivia and A. R. Rennie, Langmuir,
2001, 17, 5858–5864.
24 J. Selinger, F. MacKintosh and J. Schnur, Phys. Rev. E: Stat.
Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 1996, 53,
3804–3838.
25 P. Nelson and T. Powers, J. Phys. II, 1993, 3, 1535–1569.
26 (a) A. Singh, T. G. Burke, J. M. Calvert, J. H. Georger,
B. Herendeen, R. R. Price, P. E. Schoen and P. Yager,
Chem. Phys. Lipids, 1988, 47, 135–148; (b) M. S. Spector,
J. V. Selinger, A. Singh, J. M. Rodriguez, R. R. Price and
J. M. Schnur, Langmuir, 1998, 14, 3493–3500.
27 U. Olsson, T. C. Wong and O. Soederman, J. Phys. Chem.,
1990, 94, 5356–5361.
28 (a) C. J. Eckhardt, N. M. Peachey, D. R. Swanson,
J. M. Takacs, M. A. Khan, X. Gong, J. H. Kim, J. Wang and
R. A. Uphaus, Nature, 1993, 362, 614–616; (b) H. Fang,
L. C. Giancarlo and G. W. Flynn, J. Phys. Chem. B, 1998,
102, 7311–7315; (c) S. De Feyter, A. Gesquière,
P. C. M. Grim, F. C. De Schryver, S. Valiyaveettil,
C. Meiners, M. Sieffert and K. Müllen, Langmuir, 1999, 15,
2817–2822.
29 C. Larpent and X. Chasseray, Tetrahedron, 1992, 48, 3903–
3914.
30 J. V. Selinger, M. S. Spector and J. M. Schnur, J. Phys. Chem. B,
2001, 105, 7157–7169.
31 L. M. Blinov, Structure and Properties of Liquid Crystals,
Springer Verlag, 2010, vol. 123.
Nanoscale Paper
View Article Online

Lamellar organic–inorganic architecture via classical screw growth
Yan Quan,a
Halei Zhai,a
Zhisen Zhang,a
Xurong Xu*b
and Ruikang Tang*ab
Received 22nd May 2012, Accepted 6th July 2012
DOI: 10.1039/c2ce25805f
The fabrication of organic–inorganic composites with well-defined lamellar internal structure is of
great interest in current materials society. Inspired by the biomineralization of nacre, we found that
an organic–inorganic lamellar hybrid can be achieved spontaneously and readily using classical screw
growth, which is well-described by Burton–Cabrera–Frank (BCF) theory for solution crystal growth.
Herein, we demonstrate that a combination of calcium phosphate and sodium bis(2-ethylhexyl)
sulfosuccinate in the presence of bovine serum albumin leads to hybrid crystals with nacre-like
structure via the conventional crystallization strategy. Accordingly, solution techniques for
crystallization regulation can be used readily to control product habits. This study demonstrates how
the BCF mechanism is of relevance in biomimetic composition generation. Such a biomimetic
approach may aid in creating novel organic–inorganic composites through classical pathways.
1 Introduction
In biological systems, organic and inorganic components
intimately associate into well-organized nanocomposite materi-
als with optimized performances.1,2
Nacre, the inner shell layer
of mother of pearl, provides a fascinating example of the power
of nature, which can assemble structures with remarkable
mechanical strength and toughness (Fig. 1a).3–5
To ensure
optimal mechanical characteristics, the calcium carbonate
crystals in nacre arrange into parallel laminas and these
inorganic layers are separated by sheets of organic matrix
composed of biological macromolecules, such as chitin and
proteins.6–8
This wonderful arrangement of organic–inorganic
lamellar structure can improve the material’s toughness sig-
nificantly. For example, the toughness of nacre is 3000 times
higher than that of pure calcium carbonate crystals.9,10
Inspired
by such a remarkable characteristic, scientists have endeavoured
to design hybrid composites with nacre-like architecture.11
Various methods such as layer-by-layer (LbL) assembly,12,13
freeze casting14,15
and colloidal-based synthesis16
etc. have been
used. In laboratories, LbL may be the most common technique,
which, as suggested by its name, consists of a layer-by-layer
assembly by dipping the material into first one component then
another to make multilayered composites like nacre.17
However, nature is more sophisticated in using self-assembly
strategies to construct structurally well-defined arrays,18
provid-
ing the basis of a wide variety of complex structures. Cartwright
et al. and Wada revealed that nacre is generated by simulta-
neously integrating the growth of the inorganic and organic
phases via a conventional crystal growth process rather than by
the artificial LbL deposition.19,20
At the growth fronts of nacre,
chitin crystallites act as amphiphilic molecules and self-assemble
into liquid crystal layers. Then, chitin layers along with protein
serve as templates and modulate the mineralization process. In
the ultimate section, chitin–calcium carbonate lamellar is
gradually constructed at the growing surface.21
An experimental
proof is that spiral patterns (Fig. 1b) are frequently found on the
growing surface of nacre under a scanning electron microscope
(SEM), which suggests a classical mechanism of screw growth.21
It is well known that either inorganic ions or biomolecules can be
organized into highly ordered structures at the atom scale by
forming corresponding crystals via screw growth.22,23
This screw
growth mechanism, as suggested by BCF theory,24
can give rise
to crystals universally under various conditions including
biomineralization.25
The natural formation of nacre inspires us
to design functional materials using more efficient pathways.26
An attempt at biomimetic lamellar hybrid fabrication through
conventional crystal growth will have many important techno-
logical applications in materials science and will provide an in-
depth understanding of the physicochemical mechanisms about
a
Centre for Biomaterials and Biopathways, Department of Chemistry and
State Key Laboratory of Silicon Materials, Zhejiang University,
Hangzhou, Zhejiang, 310027, China. E-mail: rtang@zju.edu.cn;
Fax: +86 571 87953736; Tel: +86 571 87953736
b
Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou,
Zhejiang, 310027, China. E-mail: xxu@zju.edu.cn
Fig. 1 (a) Nacre is the inner shell layer with optimized mechanical
strength. (b) The spiral growth pattern on nacre’s growing surface
indicates a BCF mechanism in this biomineralization process; this image
was prepared based upon Cartwright et al.’s observations.21
CrystEngComm Dynamic Article Links
Cite this: CrystEngComm, 2012, 14, 7184–7188
www.rsc.org/crystengcomm PAPER
7184 | CrystEngComm, 2012, 14, 7184–7188 This journal is ß The Royal Society of Chemistry 2012
Publishedon06July2012.DownloadedbyUniversityofCalifornia-SanFranciscoon24/01/201605:54:24. View Article Online / Journal Homepage / Table of Contents for this issue

bio-constructions of composite materials with complicated
structures.
We have found that a combination of sodium bis(2-ethylhexyl)
sulfosuccinate (AOT, an anionic amphiphilic molecule) and
bovine serum albumin (BSA, one of the common proteins in
biomineralization studies) on calcium phosphate biomimetic
mineralization gives rise to a spontaneous assembly of hybrid
crystals with regular rhombus morphology.27
These hybrid
crystals have a lamellar structure with the alternate stacking of
inorganic phase layers and organic phase layers, which has been
demonstrated in our previous work.27
Although the mechanical
properties of the material have been studied, the formation
mechanism is still a mystery.27
Herein, we reveal that the nacre-
like lamellar structure can be constructed artificially via a
classical screw growth mechanism, which is similar to the
biological pathway for nacre formation. Furthermore, the
morphology of the hybrid crystals can be tuned using crystal
growth techniques, exhibiting a significant advantage over the
other synthesis strategies such as LbL. Coincidentally, biominer-
alization systems are also sophisticated in such crystallization
regulation to produce various composite materials in nature.
2 Experimental section
2.1 Materials and preparation
BSA and AOT were purchased from LABMAX and Sigma,
respectively. Ca(NO3)2?4H2O and (NH4)2HPO4 were from
Aladdin. All chemicals were used without any further purifica-
tion and all solutions were filtered through 0.22 mm Millipore
membranes prior to use.
In a typical synthesis experiment, 50 ml Ca(NO3)2 (5 mM) was
added to 100 ml solution containing 4 mM AOT and 2 mg ml21
BSA and the pH was adjusted to 10.0 ¡ 0.5 using 5 M
NH3?H2O. Then, 50 ml (NH4)2HPO4 (3 mM, pH of 10.0 ¡ 0.5)
was added dropwise at a rate of 1.5 ml min21
to initiate the
precipitation. The typical reaction period was 24 h and the solids
were collected by centrifugation (6000 rpm) at the end of the
experiment.
2.2 Re-growth and dissolution
The above-prepared solids were used as seed crystals. About 1 mg
seeds were immersed into 50 mL freshly prepared aqueous solutions
with different compositions: (i) 2.0 mM AOT, 1.0 mg ml21
BSA,
1.25 mM Ca(NO3)2, 0.75 mM (NH4)2HPO4; (ii) 2.0 mM AOT,
1 mg ml21
BSA, 2.50 mM Ca(NO3)2, 1.50 mM (NH4)2HPO4. The
different solutions could result in an alteration of crystal habit. The
re-growth period was 12 h.
In a dissolution experiment, about 1 mg crystals were
dispersed in 10 ml 10 mM tris(hydroxymethyl)aminomethane
buffer solution with a pH of 8.8. During the experiment, solids
were withdrawn periodically from the slurry for examination.
2.3 Characterization
In transmission electron microscopy (TEM) studies, the reaction
suspensions were dropped on carbon-coated copper grids and
dried in air. The observations were performed using a Philips CM
200UT at a typical accelerating voltage of 160 kV. For ultrathin-
sectioned TEM examination, the dried rhombic crystals were
embedded in 0.5 ml of epoxy. The mixture was solidified at 80 uC
for 12 h and sliced using a Reichert-Jung Ultracut. The typical
thickness of an ultrathin section was 80 nm.
SEM was performed by using a HITACHI S-4800 at an
accelerating voltage of 5 kV. Wide angle X-ray diffraction
(WAXD) and small angle X-ray diffraction (SAXD) were carried
out with a Rigaku D/max-2550pc with monochromatized Cu-Ka
radiation. AFM was performed with a Nanoscope IVa (Veeco,
USA) on the seed crystals. All images were acquired in contact
mode. The tip force exerted on the surface was optimized to
reduce the imaging artefacts.
The obtained hybrids had a rhombic crystal-like morphology
and contained two basic nanoscale subunits: the organic layer
and the ultrathin calcium phosphate (CaP) inorganic layer. A
sectioned TEM study revealed that both organic and inorganic
components were orderly integrated to generate the lamellar
hybrid structure (Fig. 2a) within the hybrid: the dark lines and
bright lines represent mineral and organic phases, respectively
(under TEM, the inorganic phase result in higher contrast due to
the relatively greater electron density, e.g. the electron densities
of Ca and P in the mineral phase are much greater than those of
C and H in the organic phase). The inorganic layers (2.13 nm),
together with organic layers (1.31 nm), constituted a basic unit
(3.44 nm) for the complex. This ordered structure was also
demonstrated by WAXD and SAXD (Fig. 2b), showing an
alternate structure (d = 3.44 nm) and brushite phase in the CaP
layer. Such an organic–inorganic nanostructure conferred
optimized mechanical properties on this artificial material,
especially elastic properties. For example, its modulus, 6.64 ¡
1.41 GPa,25
is even lower than that of elastic-featured human
vertebral trabeculae, 13.5 ¡ 2.0 GPa.27
In contrast, typical
moduli of pure CaP compounds are always 90 GPa.28
This
elastic property shows that the hybrid can be an excellent
candidate for a mechanical substitute in tissue engineering and
also directs to a bone-like structural design.
With the magnification under SEM, we could note that the
hybrid crystal surfaces were made up of a striking arrangement
Fig. 2 (a) SEM image of the spontaneously formed AOT–calcium
phosphate organic–inorganic hybrid architecture; insert shows the
lamellar structure inside the hybrid from a side view of an ultrathin
section: dark and light lines represent the inorganic calcium phosphate
layers and the organic AOT bilayers, respectively. (b) WAXD and SAXD
patterns showing the lamellar structure. Note: the WAXD pattern could
be fitted using brushite (JCPDS 09-0077); however, the peaks of the
standard brushite, 30.51u and 29.26u, were shifted to 30.14u and 28.59u,
respectively. These shifts could be explained by the nano size effect of the
ultrathin mineral layer.
This journal is ß The Royal Society of Chemistry 2012 CrystEngComm, 2012, 14, 7184–7188 | 7185
Publishedon06July2012.DownloadedbyUniversityofCalifornia-SanFranciscoon24/01/201605:54:24. View Article Online

of spiral and labyrinthine patterns. Different types of spirals, such
as left-handed-coiled (Fig. 3a), right-handed-coiled (Fig. 3b) and
paired (Fig. 3c), spread over the hybrid surfaces. These typical
patterns were exactly similar to those observed in nacre
formation.21
Generally, such microscopic spirals often appear
on crystal surfaces when crystals grow slowly in solution. About
sixty years ago, Frank et al. proposed an explanation, suggesting
that dislocations lead to crystal growth.24
Because the crystal
plane around a screw dislocation is a helicoid, the steps advance in
a spiral staircase fashion without any need of nucleating new
layers. The emerged steps gradually wave from centre to edges and
then continuously form new surface. Analogously, the alternate
layered structure was constructed by the spontaneous generation
and movement of the hybrid step outward in this case. If the
movements of spiral steps were blocked or failed to reach the edge
of the crystal, the growth of crystal would be restricted at the local
region but keep on promoting protuberant spirals. This step
termination could be demonstrated by some ‘‘incomplete’’ crystal
(Fig. 3d); the layered structure inside the hybrid could be
identified readily. The observed spiral stair was solid experimental
proof to confirm the screw mechanism for the hybrid formation.
The classical crystal growth theory suggests that the tangential
growth mechanism leads to growth hillocks formed of piles of
the original dislocation resource of the hillocks.29,30
Fig. 4a
shows details about the original dislocation resource of a hillock
on the surface of a hybrid. In our experiments, the sizes of
original steps were located within a range of 12–20 nm. This
dimension could be roughly considered as the critical diameter
size of the growth steps. A simplified spiral model, the
Archimedean spiral, was used to describe the screw around the
hillock. If the rates of advance of the steps in every direction
were the same, the terrace width, W, in the screw should equal to
4prc, in which rc was the radius of curvature of the step at the
emergence point of dislocation.31
The averaged values of W were
70–125 nm and thereby, the calculated 2rc from W values were
about 12–20 nm, which agreed with the direct measurements.
We found that the deviations for measurements of W and 2rc
were relatively great in this case, which should be attributed to
the anisotropic characteristics of the screws. Accordingly, we
introduced Wx and Wy to represent the terrace widths along the
different directions (Fig. 4b). Wx mostly lay in a range of 70 ¡
10 nm, while Wy, 110 ¡ 15 nm (Fig. 4c). The ratio of Wy to Wx
was about 1.5. This value was close to the ratio of the hybrid
crystal dimensions along the same directions. This consistence
indicated that the morphology of the crystals was dominated by
the geometrical feature of steps, which was also proposed by the
single screw growth model in BCF theory.
AFM characterizations revealed more details about the spiral
steps (Fig. 4d), the height between each terrace always raised
approximately 6.88 ¡ 0.30 nm (Fig. 4e). Occasionally, the
dislocation step at the growth hillock was developed into an out
extended layer (Fig. 4f) when the newly formed step terrace failed
to extend on the presented surface. The ‘‘unexpected born’’ layer
had a thickness of approximately 7 nm. This phenomenon could
be used to represent the step height independently. Our study has
showed that the inorganic and organic layers had thicknesses of
2.13 nm and 1.31 nm, respectively (Fig. 2a). The dimension of each
integrated organic–inorganic layered structure was 3.44 nm
(Fig. 2b) and therefore, each step composed two composite
layers. Why the hybrid material selected two units of the mineral–
AOT complexes to establish a step has not been resolved yet.
Nevertheless, this hybrid step was relevant to the spiral growth
fronts in nacre formation. In nature, the step front for nacre
growth typically involves three components: mineral, protein and
chitin acting as amphiphilic molecules. Coincidentally, the three
Fig. 3 SEM images of hybrid crystals. (a) Left-handed spiral pattern.
(b) Right-handed spiral pattern. (c) Paired spiral pattern. (d) Spiral stairs
in an ‘‘incomplete’’ crystal.
Fig. 4 Measurement of the growth step on hybrid surfaces. (a) SEM
image of a hillock source, 2rc represents the critical diameter of the spiral
step. (b) Scheme of the anisotropic screw, x is the short axis, y is the long
axis. (c) Statistical histogram of the Wx and Wy measurements; the curves
are produced based on the Gaussian fits. (d–e) AFM height image and
section analysis of the hybrid surface revealing that the step height may
correspond to the dimension of two organic–inorganic composite layers.
(f) SEM of an independent grown layer with a thickness of y7 nm.

biological components were represented by calcium phosphate,
BSA and AOT, respectively, in our biomimetic case.
We emphasized that the lamellar architectures could not be
produced without BSA or AOT. Our previous work showed that
if only BSA or AOT existed in the calcium phosphate solution,
CaP nanoparticles or nanorods could be induced, respectively,27
which were not organic–inorganic structured. It has been
concluded that the cooperative effect of BSA and AOT on
CaP mineralization was essential in the hybrid formation.27
In
the hybrid, the AOT bilayers and the CaP ultrathin layers were
the primary component but BSA played an adsorption role to
stabilize the structure.27
The relatively large size of BSA made it
impossible for the molecule to be present inside the hybrid.
Actually, we also labelled BSA using Au nanoparticles and
examinations showed that BSA only adsorbed onto the material
surfaces. Another proof is that the hybrids could also be
produced if BSA was replaced by silk fibroin, implying that BSA
did not participate in the inner structure but was an important
additive to ensure the hybrid formation. During the solution
growth, the anisotropic adsorption of additives on the steps
always results in a change of step morphology at the microscopic
level,32
leading to an alternation of crystal habit at the
macroscopic level. Accordingly, we have found that different
BSA concentrations could lead to expected changes of hybrid
morphology.33
This phenomenon implied that the hybrid crystal
could be tunable by conventional crystallization techniques to
obtain the required dimensions and morphologies.
A hybrid crystal with size 736 nm 6 558 nm 6 135 nm
(Fig. 5a) was used as a seed crystal. By incubation in a freshly
prepared reaction solution (see experimental section for
details), the crystal could become large with new dimensions of
1260 nm 6 949 nm 6 176 nm (Fig. 5b), demonstrating an ideal
3-dimensional growth behaviour. In contrast, the LbL deposi-
tion resulted in only 1-dimensional (thickness) increase in the
products. It was noticed that the screw steps remained, which
were similar to the original ones on the seed. Generally, crystal
habit or morphology is relevant to the steps.34,35
The above-
mentioned anisotropic growth spiral (Wy to Wx was about 1.5)
led to rhombic crystal formation. Under conditions in which
the concentrations of calcium and phosphate were doubled in
the reaction solution, the anisotropic feature of the spiral
was enhanced; the spiral elongated in a spindle-like fashion.
Accordingly, the grown crystal evolved from a rhomb into a
spindle (Fig. 5c) with significantly increased dimensions along
the y axis. This was a specific example of the morphology and
dimensions of the hybrid crystals being adjusted by solution
composition in conventional crystallization methods. Such
solution-based regulation is sophisticated in natural biominer-
alization but was unavailable in other artificial techniques for
lamellar fabrication.
Dissolution is not a simple reversed process of crystal growth.
Actually, dissolution was initiated from etched pits, which were
always produced at the point characterized by the highest stress
on crystal surfaces.36
It is well known that the presence of screw
dislocations causes stress and that the stress is also radiating
outward from the screw centre and decreasing with radial
distance. It follows that the greatest stress on the crystal surface
is at the screw centre. Fig. 5d–f show a typical dissolution
process of the nacre-like hybrids. At the initial stage, the
dissolution pits were always born at the screw centres, which
were also the centres of crystal surfaces (Fig. 5d). During the
dissolutions, the pits become deeper and larger to provide
dissolution contributions and they shared similar anisotropic
features with the growth ones. The layered structure inside the
pits could also be observed under TEM (Fig. 5e). Interestingly,
the pits shared similar anisotropic features with the growth steps.
Analogous to a single screw growth mechanism, a single pit was
frequently observed in the dissolution and this feature resulted in
an eye-like structure (Fig. 5f). Again, the observed dissolution
phenomenon supported the BCF model for the hybrid forma-
tion. On the contrary, if the hybrid had formed by the LbL
deposition, the hybrid would be peeled layer by layer rather than
by dissolution from the centre.
Fig. 5 Growth of rhombic crystals in different conditions and with
different periods of dissolution in solution. (a) SEM of seed crystal. (b)
SEM of the re-grown crystal. (c) SEM of re-grown crystal under different
solution conditions; the morphologies of crystals are changing with the
step morphology change. (d) Initially formed pit on the dissolving hybrid
surfaces. (e) TEM of an intermediate dissolution state. (f) SEM of an
‘‘eye-like’’ crystal resulting from dissolution.
Fig. 6 Scheme of the classical growth model. The spiral hillock
represents the steps on the surface of a crystal. The magnification shows
the step consists of two hybrid layers. The grey part in the right corner
represents calcium phosphate; the molecule with two tails represents
AOT, BSA proteins adsorbed on the step stabilize the structure.
This journal is ß The Royal Society of Chemistry 2012 CrystEngComm, 2012, 14, 7184–7188 | 7187

Based on above experimental results, we proposed a model for
the hybrid formation (Fig. 6). A screw hillock emerging on the
growing surface dominated the crystallization. The complete step
contained two organic–inorganic complex units and the lamellar
organic–inorganic crystals were generated by the continued
generation and expansion of steps from the surface centre. The
anisotropic movement and the morphological characteristics of
the hillock step controlled the habit of hybrid crystals, which
could be adjusted by either solution composition or BSA
adsorption.
4 Conclusions
In summary, we have shown an alternative understanding of
complicated lamellar composite generation using the classical
screw crystal growth mechanism. The study brings inspiration to
biomimetic materials preparation using conventional pathways.
Such a simple attempt at lamellar hybrid crystal fabrication will
have many important technological applications in materials
science and will also provide an in-depth understanding about
biomimetic constructions of composite materials.
References
1 H. Co¨lfen and S. Mann, Angew. Chem., Int. Ed., 2003, 42, 2350.
2 G. K. Xu, W. Lu, X. Q. Feng and S. W. Yu, Soft Matter, 2011, 7,
4828.
3 L. Xie, F. Zhu, Y. Zhou, C. Yang and R. Zhang, Prog. Mol. Subcell.
Biol., 2011, 52, 331.
4 X. Li, W. C. Chang, Y. J. Chao, R. Wang and M. Chang, Nano Lett.,
2004, 4, 613.
5 E. Munch, M. E. Launey, D. H. Alsem, E. Saiz, A. P. Tomsia and
R. O. Ritchie, Science, 2008, 322, 1516.
6 M. Suzuki, K. Saruwatari, T. Kogure, Y. Yamamoto, T. Nishimura,
T. Kato and H. Nagasawa, Science, 2009, 325, 1388.
7 S. Blank, M. Arnoldi, S. Khoshnavaz, L. Treccani, M. Kuntz, K.
Mann, G. Grathwohl and M. Fritz, J. Microsc., 2003, 212, 280.
8 F. Nudelman, B. A. Gotliv, L. Addadi and S. Weiner, J. Struct. Biol.,
2006, 153, 176.
9 K. S. Katti, D. R. Katti, S. M. Pradhan and A. Bhosle, J. Mater.
Res., 2005, 20, 1097.
10 L. Be´douet, M. Jose´ Schuller, F. Marin, C. Milet, E. Lopez and M.
Giraud, Comp. Biochem. Physiol., Part B: Biochem. Mol. Biol., 2001,
128, 389.
11 L. Wang and M. C. Boyce, Adv. Funct. Mater., 2010, 20, 3025.
12 Z. Tang, N. A. Kotov, S. Magonov and B. Ozturk, Nat. Mater.,
2003, 2, 413.
13 P. Podsiadlo, Z. Liu, D. Paterson, P. B. Messersmith and N. A.
Kotov, Adv. Mater., 2007, 19, 949.
14 E. Munch, M. E. Launey, D. H. Alsem, E. Saiz, A. P. Tomsia and
R. O. Ritchie, Science, 2008, 322, 1516.
15 S. Deville, E. Saiz, R. K. Nalla and A. P. Tomsia, Science, 2006, 311,
515.
16 L. J. Bonderer, A. R. Studart and L. J. Gauckler, Science, 2008, 319,
1069.
17 C. Jiang and V. V. Tsukruk, Adv. Mater., 2006, 18, 829.
18 Y. Oaki and H. Imai, Angew. Chem., Int. Ed., 2005, 44, 6571.
19 J. H. E. Cartwright, A. G. Checa, B. Escribano and C. I. Sainz-Dı´az,
Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 10499.
20 K. Wada, Nature, 1966, 211, 1427.
21 J. H. E. Cartwright and A. G. Checa, J. R. Soc. Interface, 2007, 4,
491.
22 J. Christoffersen and M. R. Christoffersen, J. Cryst. Growth, 1990,
100, 203.
23 T. A. Land, A. J. Malkin, Y. G. Kuznetsov, A. McPherson and J. J.
De Yoreo, Phys. Rev. Lett., 1995, 75, 2774.
24 W. K. Burton, N. Carbrera and F.C. Frank, Philos. Trans. R. Soc.
London, Ser. A, 1951, 243, 299.
25 J. De Yoreo, L. Zepeda-Ruiz, R. Friddle, S. Qiu, L. Wasylenki, A.
Chernov, G. Gilmer and P. Dove, Cryst. Growth Des., 2009, 9, 5135.
26 A. Sellinger, P. M. Weiss, A. Nguyen, Y. Lu, R. A. Assink, W. Gong
and C. J. Brinker, Nature, 1998, 394, 256.
27 H. Zhai, W. Jiang, J. Tao, S. Lin, X. Chu, X. Xu and R. Tang, Adv.
Mater., 2010, 22, 3729.
28 M. Milosevski, J. Bossert, D. Milosevski and N. Gruevska, Ceram.
Int., 1999, 25, 693.
29 A. Chernov, Sov. Phys. Usp., 1961, 4, 116.
30 C. M. Pina Martı´nez, U. Becker, P. Risthaus, D. Bosbach and A.
Putnis, Nature, 1998, 395, 483.
31 I. V. Markov, Crystal Growth for Beginners: Fundamentals of
Nucleation, Crystal Growth and Epitaxy, World Scientific Pub. Co.
Inc., Singapore, 2003.
32 G. Fu, S. R. Qiu, C. A. Orme, D. E. Morse and J. J. De Yoreo, Adv.
Mater., 2005, 17, 2678.
33 H. Zhai, X. Chu, L. Li, X. Xu and R. Tang, Nanoscale, 2010, 2, 2456.
34 J. J. De Yoreo and P. M. Dove, Science, 2004, 306, 1301.
35 H. H. Teng, P. M. Dove, C. A. Orme and J. J. De Yoreo, Science,
1998, 282, 724.
36 A. C. Lasaga and A. Luttge, Science, 2001, 291, 2400.

Biomimetic graphene oxide–hydroxyapatite
composites via in situ mineralization and
hierarchical assembly†
Yaling Li,‡a
Cuilian Liu,‡a
Halei Zhai,a
Genxing Zhu,a
Haihua Pan,ab
Xurong Xuab
and Ruikang Tang*ab
A graphene oxide–hydroxyapatite hybrid is synthesized via in situ
mineralization. The integrated HAP nanoplates share similar size,
morphology and orientations with those of natural bones. With their
excellent mechanical properties and biocompatibility, the composites
offer potential applications in load-bearing bone repair, scaffold
materials and as an alternative model for biomimetic research.
Nature has created various excellent materials during the
process of evolution. The huge diversity of elaborate hierar-
chical structures existing in biological systems is increasingly
becoming a source of inspiration for scientists to design
advanced materials.1
These biominerals are usually integrated
organic–inorganic hybrids with distinguished mechanical
properties, which are quite distinct from individual compo-
nents. Despite of the highly controlled hierarchical structures,
another common feature is that the biominerals oen involve
nanocrystals as building block arranged in high order with
organic molecule as the supporting matrix.2
For example, bone
is mainly composed of ultrathin plate-like hydroxyapatite (HAP)
nanocrystals (2–5 nm in thickness) and collagen, in which HAP
crystals are parallel aligned and tightly interact with the
collagen bers.3
As one of the most remarkable materials in
nature, bone usually serves as an elastic structural frame and
internal organ protection in body (with modulus about 10 to 20
GPa (ref. 4 and 5)). The nanoscale feature of bone minerals can
confer the optimum strength and the maximum tolerance of
aws on the tissues.2
Although there have been a mass of
researches on fabrication of bone-like composites,6,7
the arti-
cial design of materials mimicking bone both in structure and
mechanical property still remains a great challenge.
Biominerals in tissues are usually formed under the control
of macromolecular templates of proteins, peptides, and poly-
saccharides.8–10
During the mineralization process, the organic
matrixes are required to provide not only mechanical support
but also effective control over minerals crystal nucleation and
growth to obtain highly ordered deposition and integration.11
Accordingly, the organic templates are desired to possess the
advantages of localized nucleation and ordered assembly at
nanoscale. Graphene, a single layer of carbon atoms tightly
packed into a honeycomb lattice, has attracted tremendous
attention for its remarkable physical properties.12
Graphene
oxide (GO), one of the most important derivatives of graphene,
can be considered as consisting of graphene sheets decorated
with hydrophilic oxygen functional groups (hydroxyl, epoxide,
and carboxyl group).13
Accordingly, it can act as a useful
building block for versatile functional materials synthesis.
Various GO-based composites with specic functions have been
reported,14
especially for medical and biological applications,
such as tissue engineering,15
drug delivery,16
cellular imaging,17
biosensor,18
and antibacterial materials.19
However, the
previous in vitro and in vivo studies show that GO might become
a health hazard.20
GO can be internalized by cells, and then
escape from subcellular compartments, travel within the cyto-
plasm, and translocate into the nucleuses.21
To adjust the
cytotoxicity, biomacromolecules such as chitosan,16
gelatin,22
Tween,23
have been used to modify GO sheets so as to alleviate
the potential risks. Biominerals, like HAP, exhibiting excellent
biocompatibility, have also been suggested to composite with
GO to improve weak mechanical properties of the pure HAP as
well as reducing the toxicity of GO.24,25
However, we note that the
reported fabrication methods are relatively complicated or time
consuming, and specic macromolecules are usually required
to pre-modify the GO sheets. More importantly, the uncon-
trolled precipitation process of calcium phosphate on GO
surface usually leads to random and weak combination between
HAP and GO sheets.
In this work, we directly used GO as a mineralization
substrate and reinforce component to produce the biomimetic
a
Center for Biomaterials and Biopathways, Department of Chemistry, Zhejiang
University, Hangzhou, China. E-mail: rtang@zju.edu.cn; Fax: +86 571 87953736;
Tel: +86 571 87953736
b
Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou, China
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c4ra02821j
‡ These authors contributed equally to the work.
Cite this: RSC Adv., 2014, 4, 25398
Received 31st March 2014
Accepted 30th May 2014
DOI: 10.1039/c4ra02821j
www.rsc.org/advances
25398 | RSC Adv., 2014, 4, 25398–25403 This journal is © The Royal Society of Chemistry 2014
RSC Advances
COMMUNICATION
Publishedon30May2014.DownloadedbyUniversityofCalifornia-SanFranciscoon24/01/201605:54:02.
View Article Online
View Journal | View Issue

GO–HAP composites via a facile one-step in situ crystallization
method. GO could be considered as a two-dimensional (2-D)
hydrophilic macromolecule26
with abundant mineralization
related groups (hydroxyl, and carboxyl group). The unique 2-D
geometry of GO could regulate the inorganic phase deposition
onto the surface of GO.27
With precise control of mineralization,
plate-like HAP could nucleate and growth on GO surface. These
HAP plates tightly bind with GO with their (100) face. Thus, it
was rather readily to obtain parallel arrangement at 3-D scale as
the layered stack and assembly of the GO sheets.28
Via a vacuum
assisted self-assembly process, a GO–HAP paper would be easily
obtained, in which the plate-like nanocrystals were in parallel
arrangement on GO. Accordingly, the elastic modulus of the
resulted paper could be comparable to modules of natural bone
and the resulted composite material exhibited excellent
biocompatibility.
GO was prepared from pristine graphite by a modied
Hummers and Offema method.29
GO and calcium chloride were
dispersed in ethylene glycol–water mixture solvent (170 ml
ethylene glycol and 30 ml water), followed by ultrasonication for
30 min. Aerwards, disodium hydrogen phosphate aqueous
solution was added to initiate the in situ mineralization of
calcium phosphate on GO sheets and the reaction was kept at
85 Æ 1
C for 12 h to accelerate the mineralization process.
Under transmission electron microscopy (TEM), the GO sheets
were multilayers with size of a few micrometers (Fig. 1A). Aer
the mineralization process (Fig. 1B), the GO surfaces were
covered by the newly formed nanoplates, which were typically
tens of nanometers in length and width (Fig. 1C). The thickness
of HAP plate was several nanometers by measuring the standing
ones (Fig. 1C, white arrows), which might be induced by the
wrinkle or the fold of GO (Fig. 1C, black arrows). A direct
measurement by atomic force microscopy (AFM, Fig. S1†) also
conrmed that the thickness of the plates was 4.12 Æ 0.52 nm.
The strong diffraction ring in selected area electron diffraction
(SAED, Fig. 1B) could be assigned to the (002), (211) and (222)
planes of HAP. The Ca/P ratios determined by energy dispersive
spectroscopy analysis (EDS, Fig. S2†) were about 1.676 and the
value was consistent with the stoichiometric ratio of Ca/P in
HAP. HRTEM image (Fig. 1D) showed that the exposed surface
of HAP nanoplates was (100) planes, indicating that the HAP
nanoplates bind with GO by the (100) planes. X-Ray Diffraction
patterns (XRD, Fig. 1E) further demonstrated the formation of
HAP. The XRD peaks at 25.9
, 31.8
and 39.8
were indexed to
the (002), (211) and (310) of HAP (JCPDF card # 09-0432),
respectively. The strong and sharp peak of GO at 2q ¼ 10.44
indicate the (001) interlayer spacing of 0.85 nm and AFM
examination showed that GO sheets had a thickness of 0.97 Æ
0.39 nm. This value was much larger than that of pristine
graphite (0.34 nm) due to the introduction of oxygen-containing
functional groups on the graphite sheets.30
However, aer the
mineralization, the (001) reection peak of layered GO almost
disappeared, which was consistent with previous studies that
the diffraction peaks became weakened or even disappear
whenever the regular stacks of GO sheets were exfoliated.31
Further, the small differences between GO and GO–HAP in the
Raman study (Fig. S3†) indicated that the GO was not thor-
oughly reduced to graphene during the mineralization.32
The
weight ratio of HAP–GO in composites was 2.12 from TGA
results (Fig. 1F, the inuence of adsorbed water was elimi-
nated). The initial weight loss around 100
C in the samples was
due to the evaporation of absorbed water. Around 250
C, there
was an obvious weight loss in GO and GO–HAP, which was
attributed to the decomposition of the residual oxygen-
Fig. 1 TEM images of GO (A) and GO–HAP (B and C) composites, inset in B (right, up) is selected area electron diffraction (SAED) pattern. Both
GO and GO–HAP showed good dispersity in water. (D) HRTEM image of HAP nanoplates on GO sheets, inset was the FFT image of crystal lattice.
(E) XRD patterns of GO and GO–HAP powder samples. (F) TGA profiles of GO–HAP, GO and HAP. (G and H) XPS analysis of the C1s region in GO
and GO–HAP. A large loss of oxygen-functional groups after a one-step synthesis procedure is evident.
This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 25398–25403 | 25399
Communication RSC Advances
View Article Online

containing groups. The sharp weight loss above 450
C was
caused by the thermal decomposition of GO.25
Notably, the
weight percents of inorganic and organic components in GO–
HAP composites (HAP, 67.9%, GO, 32.1%) were quite similar to
that of bone, in which the mineral part contributes with 65–70%
to the tissue and the organic part, 25–30%.3
X-ray photoelectron
spectroscopy (XPS) was conducted to further investigate the
chemical compositions of samples. High-resolution C1s spectra
of GO and GO–HAP (Fig. 1G and H) showed that four different
types of carbon components were existed: C–C (284.5 eV), C–O
(C–OH) (286.5 eV), C]O (287.8 eV) and O–C]O (289.1 eV).
Although some oxygen-containing groups remained in GO–
HAP, the peak intensities were much weaker in comparison
with pure GO. These phenomena indicated that GO was
partially deoxygenated during the mineralization process,
which was mainly caused by the reduction process with
ethylene glycol.31
To further investigate the formation process of the GO–HAP
composites, various samples were separated from the reaction
mixture at different time intervals, and then were observed
under TEM (Fig. S4†). The samples, with a short reaction time
for 2 min, were GO sheets with disordered precursors (small
pieces of several nanometres, Fig. S4A†) on their surfaces,
which were conrmed as poorly crystallized minerals (Fig. S5†).
With the reaction proceeding from 1 to 4 h, the nanoplates on
GO sheets gradually grew up and spread on GO surface. Aer 8
h, the crystal growth was completed and the surfaces of GO
sheets were covered with HAP nanoplates. The increasing of
crystallinity of the deposited HAP minerals with the reaction
time could be revealed by XRD (Fig. S5†). Scheme 1 demon-
strates a possible formation mechanism of the as-obtained GO–
HAP composites. It was known that GO sheets were decorated
with abundant oxygen-containing groups, especially hydroxyl
and carboxyl groups.13
These functional groups acted as anchor
sites and enabled in situ formation of HAP mineral phase on the
surfaces of GO sheets. In the initial stage, calcium ions, formed
by the dissolution of CaCl2 in ethylene glycol and water,
favourably bounded with these oxygen-containing groups. With
the addition of Na2HPO4 aqueous solution, a large number of
nuclei formed on GO sheets to induce HAP crystallization. The
morphology of HAP crystals was related to the specic EG–water
mixed solvent. In this system, EG provided a medium for the
controlled release of free calcium and phosphate ions from
their electrolyte solids, which would possibly reduce the driving
force of homogeneous nucleation and promote HAP growing on
GO substrates.54
It was found that similar plate-like HAP crystals
also formed without GO (Fig. S6†). We noted that the water
content determined the mineralization process of HAP on GO
surface. The less water content would slow down the deposition
process and obtained HAP with less crystallinity (Fig. S7†).
Accordingly, with simple control of mixture solvent, HAP
nanoplates could precisely form on GO surface with heteroge-
neous crystallization. In this process, the in situ mineralization
was a key to achieve the structured GO–HAP complex. In order
to exclude free HAP nanoplates attached on GO sheets, as-
synthesized HAP nanoplates were added into reaction solution
instead of ions precursors (Ca2+
and HPO4
2À
). Aer 12 h, TEM
images (Fig. S8A†) showed that there were some HAP crystals
sparsely covering on GO sheets, but aer ultrasonication
(40 kHz, 180 W, 25
C) for 2 h (Fig. S8B†), the crystals became
visibly less. In contrast, aer the same ultrasonic treatment,
GO–HAP composites underwent almost no obvious change
and there were nearly no scattered HAP nanoplates found
(Fig. S8C†). It followed that the HAP crystals were rooted on the
GO sheets, which could be understood as the integration of HAP
and GO phases by the hydrophilic groups on sheets during the
in situ mineralization process.
It has been demonstrated that the apatite nanocrystals can
provide the organic–inorganic nanocomposite in biological
bone with the favorable mechanical properties.2,33
We noted
that the dimensions of the resulted HAP nanoplates on GO
sheets were fairly similar to those in bone tissues.3
The GO–HAP
composite could be constructed into a well-ordered macro-
scopic structure with the bone-like features for a mechanical
examination. The resulted GO–HAP sheets were well-dispersed
in water (inset in Fig. 1B) and could be self-assembled into a
paper-like material under a directional ow.28
In the present
work, we got a free-standing paper via vacuum ltration of
colloidal dispersions of the GO–HAP sheets (Fig. 2A). Fig. 2B
and C showed that the obtained paper was uniform, complete
and exible. SEM image (inset in Fig. 2D) of the fracture surface
of the GO–HAP paper revealed the lamellar structure within the
Scheme 1 The proposed in suit mineralization mechanism of HAP on
GO sheets. CaP: crystal nucleus formed on GO sheets, HAP: HAP
nanoplates.
Fig. 2 (A) Self-assembly process of GO–HAP sheets during vacuum
filtration. (B and C) Digital photograph of GO–HAP paper. (D) XRD
pattern of the GO–HAP paper. In comparison with Fig. 1D, the (002)
reflection of HAP disappears in the paper-like assembly. Inset is SEM
image of fracture section of GO–HAP paper, revealing the lamellar
structure.
RSC Advances Communication
View Article Online

bulk material. Notably, XRD pattern of the GO–HAP paper
(Fig. 2D) displayed that (002) plane reection (25.9
) of HAP
disappeared, while the reection of (100) and (300) planes
(10.8
, and 32.9
, respectively) got evident enhancements. By
using the soware of PeakFit v4.12, the peaks of (211), (112) and
(300) in XRD patterns of GO–HAP powder and paper samples
(Fig. S9†) were separated to calculate the peak area ratio of (300)
to (002) planes (shown as I(300)/(002)). I(300)/(002) of powder sample
was only 1.02, while the value for paper sample was 29.41. Such
a signicant difference between GO–HAP powder and paper
samples (Fig. 1E and 2D) was attributed to the unique orienta-
tion of HAP nanoplates on GO sheets and the subsequent
orderly assembly. It was indicated by HRTEM (Fig. 1D) that the
plate-like HAP crystals were integrated with GO sheets via (100)
face, and packed into high ordered lamellar structure in our
GO–HAP paper. Moreover, due to the 2-D geometry of GO, not
only the HAP nanoplates, but also all their (100) planes were
approximately parallel to each other. The unique structure
resulted in the obvious enhancement of (300) reection and
disappearance of (002) reection.55
In biological bone, the
ultrathin HAP nanoplates are oriented along the long axes of the
collagen brils with their (100) planes parallel to each other.34
Therefore, the GO–HAP paper shared the similar hybrid struc-
ture with that of natural bone.
This unique structure of highly ordered nanoplates
embedded in the relatively so GO matrix would lead to an
optimal mechanical performance. Typical stress–strain curves
of GO–HAP and GO papers were shown in Fig. 3. Three regimes
of deformation were observed: straightening, almost linear
(“elastic”), and plastic.28
The initial modulus (EI) of GO–HAP
paper was 13.6 GPa, which was 223% higher than that of
unmodied GO paper (4.2 GPa), indicating that GO–HAP paper
was signicantly stiffer than the pure GO one against the initial
loading (Table 1). It was proposed that the initial tensile load
can lead to structural sliding of GO sheets to overcome physical
wrinkling or “waviness” that resulted from the fabrication
process and thus to achieve the best interlocking geometry.35
Correspondingly, the modulus continued to increase as the
samples straightened and entered the linear region. The
modulus (EII) of GO–HAP paper during the linear part was 16.9
GPa, 231% higher than the GO paper (5.1 GPa). The modulus of
our GO–HAP paper were higher than reported modulus values
for bucky paper (10 GPa),36
graphite foil ($5 GPa),37
and paper-
like materials ($5–15 GPa).22,38,39
The tensile strength (s) of GO–
HAP paper was 75.6 MPa, 78% higher than the GO paper.
However, the GO–HAP paper underwent a reduction of tough-
ness due to the integration of the rigid HAP crystals. The ulti-
mate strain (3) and fracture toughness (U) of GO–HAP paper
were 0.53%, 214.9 kJ mÀ3
, while these values for GO paper were
1.28%, 313.6 kJ mÀ3
, respectively. Nevertheless, compared with
the current used cross-linking agents to fabricate GO-based
composites such as polyallylamine (GO–PAA, 0.32%, 180 kJ
mÀ3
),40
poly(vinyl alcohol) (GO–PVA, 0.27%, 100 kJ mÀ3
),41
glutaraldehyde (GO–GA, 0.4%, 200 kJ mÀ3
),42
or borate (GO–
borate, 0.15%, 140 kJ mÀ3
),43
GO–HAP paper here was more
tougher. These results indicated that the HAP nanoplates
played a pivotal role in retaining toughness as the stiffness
increased, which were both equally important in load-bearing
materials design.
The improvement of mechanical strength was originated
from the ordered GO–HAP layered structure at nanoscale.
Under tensile stress, the deformation mechanism was similar
as a staggered model of load transfer in bone matrix.33
It was
shown that as soon as the structural size reaches the critical
length (the size of fracture process zone), materials become
insensitive to aws.2
Thus, the nanometer size of the mineral
crystals in biocomposites became important to ensure the
optimum fracture strength and maximum tolerance of aws.
More importantly, the effective load transfer between minerals
and so matrix also played a key role in damage shielding.44
As
previously mentioned, the binding force between GO sheets and
HAP nanoplates were strong, mainly resulted from the high
specic surface areas and in suit crystallization process. In the
composite, the HAP crystal orientations were induced and
controlled by the GO substrates during the in situ mineraliza-
tion. And the resulted GO–HAP sheets could be further self-
assembled to form the free-standing paper with the lamellar
structure. When the GO–HAP paper was exposed to an applied
tensile stress, the load could be transferred by so GO sheets via
shear between rigid HAP plates. Since HAP crystals could bear
most of stress, the strength of the hybrid material was signi-
cantly improved.
Recently, repair of load-bearing defects resulting from
disease or trauma becomes a critical problem for bone tissue
engineering.45
HAP, for its excellent biocompatibility, has been
extensively studied for this application. However, the conven-
tionally synthesized HAP crystallites cannot have sufficient
mechanical strength to repair these defects directly, therefore,
have been limited to the non-load-bearing applications.41,45
Aer composite with GO, the mechanical properties of the
resulted GO–HAP composites (Table 1) were greatly improved.
Compared with some bone tissues, the elastic modulus of GO–
HAP paper was higher than that of the mineralized collagen
bers (3–7 GPa),46
rat vertebra (11–13 GPa),47
bovine distal
femora (9–12 GPa),4
red deer anthler (7–8 GPa),48
andFig. 3 Stress–strain curves of GO–HAP and GO papers.
View Article Online

comparable to human femur bone (13–15 GPa),49
and human
tibia bone (13–16 GPa).50
Accordingly, with the comparable
stiffness to that of bones, the GO–HAP composite showed
promising application in bone tissue engineering.
Different from GO, the GO–HAP composites were also
featured by their excellent biocompatibility. In vitro cytotoxicity
test (MTT assay) was conducted to evaluate the GO–HAP mate-
rial for its potential application in biomedicine and bioengi-
neering. Human osteosarcoma cells (MG63), a representative of
human osteoblast-like cell, were used in this biological assess-
ment. As shown in Fig. 4A, GO had an unneglectable toxicity on
MG63 cells, presenting a dose-dependent cytotoxic effect. At the
highest concentration (200 mg LÀ1
), only 52% of the cells
remain viable. However, aer modied by HAP crystals, the
toxicity of GO was reduced remarkably. Even aer 24 h exposure
to the GO–HAP hybrid materials, the relative cell viability could
keep at a high level of 81–88% and these values were almost
unaffected by the material concentrations. In the parallel
experiment, we selected the conventional HAP to repeat the
experiment. As expected, HAP had relatively high cell viability
(88–94%) under all applied concentrations. These results
revealed that aer modication with HAP, the biocompatibility
of GO had been signicantly enhanced, which could be
comparable to the HAP biomineral. The cell attachment and
morphology on these different substrates had also been exam-
ined and the glass was used as blank control. The uorescent
staining images (Fig. 4B–E) showed that the cell density
increased from GO lm, to GO–HAP lm, to HAP lm, which
was consistent with MTT results. Most cells on the lms and
glass were well-spread and exhibited an elongated and highly
branched morphology, revealing that cells were well adhered on
substrates. However, compared to GO–HAP lm, cells on GO
lm were rather less, revealing the poor biocompatibility of GO.
Aer the mineralization modication, the cell toxicity of GO
could be markedly reduced by the high coverage percentages
and well-ordered orientation of HAP crystals. Thus, the devel-
oped GO–HAP composite shared the similar structure,
mechanical strength and bioactivity with the natural bone,
which could be specically suitable for the load-bearing
substitution.
Conclusions
In summary, we synthesized GO–HAP composites via biomi-
metic in situ mineralization and they can assembly into the
highly ordered bone-like structure. The tensile strength and
Young's modulus of the GO–HAP paper can achieve the optimal
level of the biological bone and the material also possesses the
excellent biocompatibility. Since the GO–HAP composites
mimic natural bone in both structure and function, we suggest
that the GO–HAP composites may offer a potential in bone
tissue repair and an alternative research model for biomimetic
bone.
Acknowledgements
We thank Jieru Wang, Xinting cong, Yiting Xu and Xiaoming
Tang for assistance in material characterizations. This work was
supported by the Fundamental Research Funds for the Central
Universities and the National Natural Science Foundation of
China (No. 91127003).
1 S. Mann, Biomineralization: principles and concepts in
bioinorganic materials chemistry, Oxford University Press,
Oxford, 2001.
2 H. Gao, B. Ji, I. L. Jager, E. Arzt and P. Fratzl, Proc. Natl. Acad.
Sci. U. S. A., 2003, 100, 5597–5600.
3 L. C. Palmer, C. J. Newcomb, S. R. Kaltz, E. D. Spoerke and
S. I. Stupp, Chem. Rev., 2008, 108, 4754–4783.
Table 1 Mechanical properties of bone, HAP, GO and GO–HAP papers. Note: for HAP powder, it is very difficult to obtain the tension–stress
curve to calculate the values of tensile strength, strain and work of fracture for bulk HAP powdersa
EI [GPa] EII [GPa] s [MPa] 3 [%] U [kJ mÀ3
]
Bone 10–20 (ref. 4 and 5) 89–114 (ref. 51) 1.1–2.5 (ref. 52) 120–875 (ref. 53)
HAP 5.18–5.92 (ref. 25) — — —
GO paper 4.2 5.1 42.3 1.28 313.6
GO–HAP paper 13.6 16.9 75.6 0.53 214.9
a
EI ¼ modulus in the initial region; EII ¼ modulus during the “linear” part; s ¼ ultimate strength; 3 ¼ ultimate strain; U ¼ work of fracture.
Fig. 4 (A) Relative cell viability of human osteosarcoma cells (MG-63)
treated with GO, GO–HAP and conventional HAP at various concen-
trations, and fluorescent images of MG63 cells cultured on (B) glass,
(C) GO, (D) conventional HAP and (E) GO–HAP films for 24 h.
RSC Advances Communication
View Article Online

4 R. B. Ashman and R. Jae Young, J. Biomech., 1988, 21, 177–
181.
5 J.-Y. Rho, L. Kuhn-Spearing and P. Zioupos, Med. Eng. Phys.,
1998, 20, 92–102.
6 E. D. Spoerke, S. G. Anthony and S. I. Stupp, Adv. Mater.,
2009, 21, 425–430.
7 A. M. Collins, N. J. V. Skaer, T. Gheysens, D. Knight,
C. Bertram, H. I. Roach, R. O. C. Oreffo, S. Von-Aulock,
T. Baris, J. Skinner and S. Mann, Adv. Mater., 2009, 21, 75–78.
8 R. Lakshminarayanan, R. M. Kini and S. Valiyaveettil, Proc.
Natl. Acad. Sci. U. S. A., 2002, 99, 5155–5159.
9 Y.-Y. Hu, A. Rawal and K. Schmidt-Rohr, Proc. Natl. Acad. Sci.
U. S. A., 2010, 107, 22425–22429.
10 J. Aizenberg, G. Lambert, S. Weiner and L. Addadi, J. Am.
Chem. Soc., 2001, 124, 32–39.
11 S. Weiner, W. Traub and S. B. Parker, Philos. Trans. R. Soc., B,
1984, 304, 425–434.
12 A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183–191.
13 D. R. Dreyer, S. Park, C. W. Bielawski and R. S. Ruoff, Chem.
Soc. Rev., 2010, 39, 228–240.
14 X. Huang, X. Qi, F. Boey and H. Zhang, Chem. Soc. Rev., 2012,
41, 666–686.
15 S. H. Ku, M. Lee and C. B. Park, Adv. Healthcare Mater., 2013,
2, 244–260.
16 H. Bao, Y. Pan, Y. Ping, N. G. Sahoo, T. Wu, L. Li, J. Li and
L. H. Gan, Small, 2011, 7, 1569–1578.
17 Y. Wang, W. C. Lee, K. K. Manga, P. K. Ang, J. Lu, Y. P. Liu,
C. T. Lim and K. P. Loh, Adv. Mater., 2012, 24, 4285–4290.
18 Y. Liu, D. Yu, C. Zeng, Z. Miao and L. Dai, Langmuir, 2010, 26,
6158–6160.
19 W. Hu, C. Peng, W. Luo, M. Lv, X. Li, D. Li, Q. Huang and
C. Fan, ACS Nano, 2010, 4, 4317–4323.
20 K. H. Liao, Y. S. Lin, C. W. Macosko and C. L. Haynes, ACS
Appl. Mater. Interfaces, 2011, 3, 2607–2615.
21 A. Bianco, Angew. Chem., Int. Ed., 2013, 52, 4986–4997.
22 C. Wan, M. Frydrych and B. Chen, So Matter, 2011, 7, 6159.
23 S. Park, N. Mohanty, J. W. Suk, A. Nagaraja, J. An, R. D. Piner,
W. Cai, D. R. Dreyer, V. Berry and R. S. Ruoff, Adv. Mater.,
2010, 22, 1736–1740.
24 S. Kim, S. H. Ku, S. Y. Lim, J. H. Kim and C. B. Park, Adv.
Mater., 2011, 23, 2009–2014.
25 M. Li, Y. Wang, Q. Liu, Q. Li, Y. Cheng, Y. Zheng, T. Xi and
S. Wei, J. Mater. Chem. B, 2013, 1, 475.
26 L. J. Cote, F. Kim and J. Huang, J. Am. Chem. Soc., 2008, 131,
1043–1049.
27 S. Chen, J. Zhu, X. Wu, Q. Han and X. Wang, ACS Nano, 2010,
4, 2822–2830.
28 D. A. Dikin, S. Stankovich, E. J. Zimney, R. D. Piner,
G. H. B. Dommett, G. Evmenenko, S. T. Nguyen and
R. S. Ruoff, Nature, 2007, 448, 457–460.
29 W. Chen and L. Yan, Adv. Mater., 2012, 24, 6229–6233.
30 C. Xu, X. Wu, J. Zhu and X. Wang, Carbon, 2008, 46, 386–389.
31 C. Xu, X. Wang and J. Zhu, J. Phys. Chem. C, 2008, 112,
19841–19845.
32 K. N. Kudin, B. Ozbas, H. C. Schniepp, R. K. Prud'Homme,
I. A. Aksay and R. Car, Nano Lett., 2008, 8, 36–41.
33 H. S. Gupta, J. Seto, W. Wagermaier, P. Zaslansky,
P. Boesecke and P. Fratzl, Proc. Natl. Acad. Sci. U. S. A.,
2006, 103, 17741–17746.
34 M. J. Olszta, X. Cheng, S. S. Jee, R. Kumar, Y.-Y. Kim,
M. J. Kaufman, E. P. Douglas and L. B. Gower, Mater. Sci.
Eng., R, 2007, 58, 77–116.
35 S. Park, K.-S. Lee, G. Bozoklu, W. Cai, S. T. Nguyen and
R. S. Ruoff, ACS Nano, 2008, 2, 572–578.
36 X. Zhang, T. Sreekumar, T. Liu and S. Kumar, J. Phys. Chem.
B, 2004, 108, 16435–16440.
37 M. Dowell and R. Howard, Carbon, 1986, 24, 311–323.
38 D. Zhong, Q. Yang, L. Guo, S. Dou, K. Liu and L. Jiang,
Nanoscale, 2013, 5, 5758–5764.
39 Y. Xu, W. Hong, H. Bai, C. Li and G. Shi, Carbon, 2009, 47,
3538–3543.
40 S. Park, D. A. Dikin, S. T. Nguyen and R. S. Ruoff, J. Phys.
Chem. C, 2009, 113, 15801–15804.
41 K. W. Putz, O. C. Compton, M. J. Palmeri, S. T. Nguyen and
L. C. Brinson, Adv. Funct. Mater., 2010, 20, 3322–3329.
42 Y. Gao, L.-Q. Liu, S.-Z. Zu, K. Peng, D. Zhou, B.-H. Han and
Z. Zhang, ACS Nano, 2011, 5, 2134–2141.
43 Z. An, O. C. Compton, K. W. Putz, L. C. Brinson and
S. T. Nguyen, Adv. Mater., 2011, 23, 3842–3846.
44 B. Ji and H. Gao, J. Mech. Phys. Solids, 2004, 52, 1963–1990.
45 A. J. Wagoner Johnson and B. A. Herschler, Acta Biomater.,
2011, 7, 16–30.
46 F. Yuan, S. R. Stock, D. R. Haeffner, J. D. Almer,
D. C. Dunand and L. C. Brinson, Biomech. Model.
Mechanobiol., 2011, 10, 147–160.
47 E. Hamed, I. Jasiuk, A. Yoo, Y. Lee and T. Liszka, J. R. Soc.,
Interface, 2012, 9, 1654–1673.
48 J. Currey, T. Landete-Castillejos, J. Estevez, F. Ceacero,
A. Olguin, A. Garcia and L. Gallego, J. Exp. Biol., 2009, 212,
3985–3993.
49 P. K. Zysset, X. Edward Guo, C. Edward Hoffler, K. E. Moore
and S. A. Goldstein, J. Biomech., 1999, 32, 1005–1012.
50 J. Y. Rho, R. B. Ashman and C. H. Turner, J. Biomech., 1993,
26, 111–119.
51 M. Akao, H. Aoki and K. Kato, J. Mater. Sci., 1981, 16, 809–
812.
52 D. L. Kopperdahl and T. M. Keaveny, J. Biomech., 1998, 31,
601–608.
53 P. Zioupos and J. D. Currey, Bone, 1998, 22, 57–66.
54 J. Tao, W. Jiang, H. Zhai, H. Pan, X. Xu and R. Tang, Cryst.
Growth Des., 2008, 8, 2227–2234.
55 Z. Zhuang and M. Aizawa, J. Mater. Sci.: Mater. Med., 2013,
24, 1211–1216.
View Article Online

Controls of Tricalcium Phosphate Single-Crystal Formation from Its
Amorphous Precursor by Interfacial Energy
Jinhui Tao,†
Haihua Pan,†
Halei Zhai,†
Jieru Wang,‡
Li Li,†
Jia Wu,†
Wenge Jiang,†
Xurong Xu,†
and Ruikang Tang*
Department of Chemistry and Center for Biomaterials and Biopathways and Centers of Analysis and
Measurement, Zhejiang UniVersity, Hangzhou 310027, China
ReceiVed October 10, 2008; ReVised Manuscript ReceiVed February 3, 2009
ABSTRACT: Different from the conventional solution precipitation, amorphous precursor involves widely in biomineralizations.
It is believed that the development of crystalline structures with a well-defined shape in biological systems is essentially facilitated
by the occurrence of these transient amorphous phases. However, the previous studies have not elucidated the physicochemical
factors influencing the transformation from the transient phase into the stable phase. In this study, the evolutions from the amorphous
calcium phosphate to the different-shaped (hexagon and octahedron; octahedron is an unexpected morphology of the crystal with
space group of R3jc) single crystals of β-tricalcium phosphate (β-TCP) were examined. The hexagonal β-TCP crystals were formed
via the phase transformation of amorphous precursor in CaCl2-Na2HPO4-ethylene glycol solution; however, the octahedral β-TCP
crystals were formed in Ca(OH)2-(NH4)2HPO4-ethylene glycol solution. Because the interfacial energies between amorphous phase
and crystals were much smaller than those between solutions and crystals, the crystallization of the β-TCP phase occurred directly
in the amorphous substrate rather than from the solution. It was interesting that the final morphology of product was also determined
by the interfacial energy between the transformed crystal and solution. The current work demonstrated that the amorphous precursor
epitaxial nucleation process and morphology selection of crystals in the amorphous phase could also be understood by an interfacial
energy control. This result might provide an in-depth understanding of the biomimetic synthesis of crystals via a pathway of amorphous
precursors.
Introduction
The ability to synthetically tune sizes, structures, and mor-
phologies of inorganic crystals is an important objective in
current materials science and device fabrications. The same
crystal may have different applications as its properties change
with size or shape.1
For example, the catalytic property of
platinum has been found to be highly dependent on which facets
terminate the surface.2
In our previous study, the mineralization
and demineralization behaviors of biomaterials such as β-tri-
calcium phosphate (β-TCP, Ca3(PO4)2) and hydroxyapatite
(HAP) are highly dependent on their exposed surfaces to
biological milieus,3
which are also related to the protein
adsorptions and cell attachments.4
Size- and shape-controlled
synthesis of many inorganic compounds such as noble metals,5
semiconductors,6
and magnetites7
have been achieved to
modulate their electrical, optical, magnetic, and catalytic proper-
ties. In contrast, the challenge of controlling crystal shape of
biominerals has been met with a limited success. But crystal
polymorph is an important feature of natural biominerals.
Different from the conventional solution precipitation, it has
been observed that amorphous precursor involves widely in
biological crystallizations. Living organisms usually use amor-
phous phases as the building materials, stabilizing them over
their lifetime, or depositing them as transient phases that
transform in a controlled manner into the specific crystalline
structure and morphology. For example, during the formation
of calcitic sea urchin spine and larval spicules, the amorphous
calcium carbonate is first formed before the final crystal
generation.8,9
Amorphous materials are also identified during
the formations of mollusk and skeletal minerals.9-11
It is
believed that the development of crystalline structures with a
well-defined shape in biological systems is essentially facilitated
by the occurrence of these transient amorphous phases.8-11
However, the previous studies have not elucidated the physi-
cochemical factors influencing the transformation from the
transient phase into the stable phase. Biological control over
the selection of mineral form and morphology indicates complex
interactions between the organism and the amorphous precursor,
which are not fully discovered. In this study, we examine the
evolution from the amorphous precursor to the different-shaped
(hexagon and octahedron, octahedron is an unexpected mor-
phology of the crystal with space group R3jc) single crystals. It
is revealed experimentally that crystal nucleated directly from
the amorphous precursor. The epitaxial nucleation process and
shape selection of crystals in the amorphous phase can be
addressed by an interfacial energetic control. This result provides
an in-depth understanding of the biomimetic crystallizations via
a pathway of amorphous precursors.
Calcium phosphates have excellent biocompatible properties
since they are main component of biological bone and tooth.12
In particular, β-TCP, an important resorbable calcium phosphate
biomaterials, is an intermediate phase of calcium phosphate.
β-TCP has been used as an ideal candidate for bone substitute,13
inorganic filling of biodegradable composites,14
substrate for
evaluation of cell seeding efficacy, proliferation, osteogenic
differentiation,15
and carrier for bone growth factors to stimulate
bone healing and formation, because of its excellent osteocon-
ductive and biodegenerative characteristics.16
Besides, it can
also find other applications of this compound, which involve
drug carrier, luminescence materials, and catalyst.17
It has been
reported that the protein adsorption property of β-TCP is
dependent upon its size and terminal facets.4
The synthesis
method with size and shape control ability may provide an
effective way for the biological modulation. There are several
synthesis methods to produce β-TCP but none of them can form
* Corresponding author. Tel/Fax: 86-571-87953736. E-mail: rtang@
zju.edu.cn.
†
Department of Chemistry and Center for Biomaterials and Biopathways,
Zhejiang University.
‡
Centers of Analysis and Measurement, Zhejiang University.
CRYSTAL
GROWTH
DESIGN
2009
VOL. 9, NO. 7
3154–3160
10.1021/cg801130w CCC: $40.75  2009 American Chemical Society
Published on Web 05/12/2009
DownloadedbyZHEJIANGUNIVonJuly27,2009
PublishedonMay12,2009onhttp://pubs.acs.org|doi:10.1021/cg801130w

the uniform and faceted crystals. These conventional methods
include solid-state reactions between CaHPO4 and CaCO3,
(NH4)2HPO4 and CaCO3, NH4H2PO4 and CaCO3, Ca2P2O7 and
CaCO3,18
or wet-chemical methods.19
The solid-state reactions
usually take place at high temperatures (∼1000 °C) and the
formed products are usually agglomerated without any defined
shapes. The wet-chemical route results in calcium deficient
apatite (CDHA), which is then transformed into β-TCP by
calcination at 700-800 °C. Although some other methods have
been tried to fabricate nano β-TCP, the shape and structure-
property relationship for this material can hardly be controlled.19e
All these methods cannot be used to understand the biological
formations of calcium phosphate. Herein, we propose a bioin-
spired pathway for a large scale synthesis of β-TCP using
amorphous calcium phosphate (ACP) as the precursor. Hexagon
and octahedron of the well-crystallized β-TCP can be achieved
from the identical ACP precursor under the different solvent
conditions.
Amorphous Precursor. One-tenth of a gram of CaCl2 ·2H2O was
added into 50 mL of ethylene glycol (EG), and the mixture was heated
to 150 °C under vigorous stirring. Next, 1.36 mL of 0.3 M Na2HPO4
(aqueous solution) and 120 µL of 1.3 M NaOH (aqueous solution) were
mixed with 20 mL of EG at a temperature of 105 °C. The phosphate-
containing EG solution was poured into the calcium-containing ethylene
glycol solution within 10 s. The precipitation sustained for 5 s and the
slurry was then poured into a vial immersed in ice-acetone bath (-16
°C) to quench reaction. The solids were collected by centrifugation
(10 000 g) and -4 °C. They were washed with ethanol for 3 times.
Synthesis of β-TCP via Amorphous Precursor. For hexagons, 20
mg of amorphous precursor was dispersed in 70 mL of EG containing
CaCl2 (7.6 mM) and Na2HPO4 (3.7 mM), and the slurry was heated to
150 °C. For octahedron, 20 mg of precursor was dispersed in 70 mL
of ethylene glycol containing Ca(OH)2 (7.6 mM) and (NH4)2HPO4 (3.7
mM), and the slurry was heated to 150 °C. In the size-controlled
synthesis, the precursor amounts were altered accordingly. For the study
of evolution process, the samples were withdrawn from the reaction
milieu periodically using a glass pipet. The extractions were injected
into vials immersed in ice-acetone bath (-16 °C) to quench the
reaction. All the above solids were harvested by centrifuging at 4000 g
and -4 °C. The tablets were washed with ethanol and water repeatedly
3 times to remove the residual solvent and other impurities. The crystals
were dried under a vacuum condition at room temperature.
Interfacial Energy Determinations. Solid samples were dispersed
in chloroform-ethanol mixed solvent (v:v ) 2:1) with a weight ratio
of 0.6%. A 100 µL of this dispersion was carefully dipped onto the
silicon substrates. The solvent was evaporated in air at room temperature
and the films could be formed on the substrates. The growth solutions
of hexagon and octahedron were filtered through membrane with pore
diameter of 220 nm before use. The surface tensions of these solutions
were measured by pendant method at 20 °C and relative humidity was
70%. To measure the interfacial energy of solid films in air, we used
four probing liquids: water, EG, n-octane, and DMSO. The contact
angles were measured by sessile drop and thin layer wicking at 20 °C
and relative humidity of 70%. At least five independent values were
measured for each solid film and liquid.
Characterizations. Transmission electron microscopy (TEM) ob-
servations were performed by using JEM-200CX TEM (JEOL, Japan)
at an acceleration voltage of 160 kV and JEM-2010HR HRTEM (JEOL,
Japan) at an acceleration voltage of 200 kV. Scanning electron
microscopy (SEM) characterization was performed on S-4800 field-
emission scanning electron microscope (HITACHI, Japan) at an
acceleration voltage of 5 kV. The phase of the solids was examined
by X-ray diffraction (XRD, D/max-2550pc Rigaku, Japan) with
monochromatized Cu KR radiation. The FT-IR spectra were collected
form 4000 to 400 cm-1
in transmission mode by a Nexus-670
spectrometer (Nicolet, USA). The contact angle data were measured
on an OCA15+ optical contact-measuring device (Data Physics
Instruments GmbH, Germany).
Simulations. Computer simulations were performed using the
morphology modules of Material Studio 3.1 packages. The initial
configuration of β-TCP crystal was taken from the X-ray crystal
structure. Initial face list was generated by Bravais-Friedel Donnay-
Harker (BFDH) method, which used the crystal lattice and symmetry
to generate a list of possible growth faces. The minimum d-spacing
was set to be 1.3 Å. The maximum of indices along a, b, c was chosen
to be 5, 5, 10 respectively. Finally, A face list consist of 1942 unique
crystal facets was generated. This face list was used as input for further
calculation of attachment energy. In the part of energy calculation
consistent-valence force field (CVFF) was used. Ewald summation
method was adopted for treatment of electrostatic terms with accuracy
of 0.001 kcal/mol. An atom-based summation method was applied for
van der Waals terms with the cutoff distance of 1.25 nm.
Results and Discussion
ACP is the least stable of the calcium phosphate phases and
it is identified at the early stage of the biological formations of
apatite.11
Amorphous mineral is moldable; this characteristic
results in the diverse crystal structures of bioinorganic crystals.
In the current study, the precursor ACP is synthesized and
stabilized in the laboratory by mixing of CaCl2 and Na2HPO4
in EG. TEM and SEM images of the resulting ACP precipitates
are shown in Figure 1. Energy-dispersive spectroscopy (EDS)
and chemical analysis (atomic adsorption for calcium and UV
for phosphate) shows the solids mainly contained calcium and
phosphorus and their molar ratio is 1.47 ( 0.05. The chemical
composition of the resulted ACP is similar to Ca3(PO4)2. The
selected area electron diffraction (SAED) pattern is weak and
dispersive, indicating the poor crystallinity of the phase (insert
of Figure 1A). FT-IR result shows the broad and featureless
phosphate absorption bands (Figure 1D). The triply degenerated
asymmetric stretching (1087, 1046, and 1032 cm-1
) and bending
vibrations of PO4
3-
(602, 574, and 561 cm-1
) in crystallized
solids are not detected.20
These results confirm that the
precipitate in EG is the amorphous phase. The peaks of CO3
2-
(1419 and 874 cm-1
) in FT-IR implies that some carbonate ions
incorporated into the ACP.20
The incorporation of carbonate is
a common phenomenon during the formation of biological
calcium phosphate. The presence of HPO4
2-
in the amorphous
precursor may also contributes to the absorption at 874 cm-1
.20
It is previously revealed that the short-range order is always
present in the bulk of amorphous phase including ACP.21
The
similar result is also observed in our samples. The high-
resolution TEM (HRTEM) study shows a few of nano ordered
domains in the amorphous phase for their different contrasts
compared with the surrounding disordered regions (Figure 1B).
Such an order-related contrast has also been reported in some
amorphous binary alloys.22
However, this short-range order
cannot be detected by conventional XRD and the amorphous
nature of the precipitates is clarified by the featureless humps
in its pattern (Figure 1E).
The formed ACP solids can be stabilized up to several months
under vacuum conditions at room temperature. We study the
phase transformation at temperature of 150 °C in EG in the
presence of calcium and phosphate ions (these ions are used to
prevent the dissolution of ACP in the solvent). ACP solids are
redispersed in a CaCl2-Na2HPO4-EG solution. The hexagon
can be eventually formed from ACP (Figure 2A, 2B). The
typical diameter of the hexagonal face can be tuned from 550
nm to 1 µm by changing precursor concentration from 10 mg/
70 mL to 40 mg/70 mL (powder to solution). The thickness of
the hexagon, ∼220-250 nm, keeps almost unchanged under
the different experimental conditions (Figure 3). The XRD
pattern collected on the hexagons can be indexed to β-TCP (R3jc,
a ) b ) 10.42 Å, c ) 37.38 Å; R ) β ) 90°, γ ) 120°, JCPDS
Calcium Phosphate Phase Transformation Crystal Growth Design, Vol. 9, No. 7, 2009 3155

09-0169). Ca, P, and O elements are detected by EDS, and
the measured calcium to phosphate molar ratio is 1.51 ( 0.02.3a
HRTEM and SAED show that the hexagon is terminated by
{100} and {001} planes (Figures 2C, 2D, and 3). The size of
product increases in proportional to precursor concentration in
the range from 10 mg/70 mL to 40 mg/70 mL (Figure 3). Further
increase in precursor concentration (40 mg/70 mL) has no
obvious influence on the product size any more. However, it is
also noted that the different concentrations of calcium (5-8
mM) and phosphate ions (2.5-4 mM) cannot result in a
significant change of crystal sizes. This result implies the
essential role of ACP precursor in the formation of β-TCP. The
reaction temperature can influence the phase and shape of the
products. β-TCP hexagons with rough {100} planes are formed
at a temperature of 115 °C. At 100 °C, the hexagons with
smooth apex, surface cracks, and holes (see Figure S1 in the
Supporting Information) can be resulted and they become
roundlike. Thus, the crystal perfection and crystallinity are
temperature-dependent.
The morphology of transformed β-TCP is dependent on the
solution conditions. It is abnormal but interesting that octahe-
drons with dimensions of 300∼400 nm can be observed in
Ca(OH)2-(NH4)2HPO4-EG solution by using ACP as the
starting material (Figure 2E). Even though β-TCP indexed in
the space group R3jc is not expected to grow with this
exceptional morphology.23
XRD experiments of samples con-
firm that the resulted material is β-TCP (see Figure S2 in the
Supporting Information). It is found that the surface of
octahedron is not atomic flat under HRTEM and SEM (Figure
2E-H). Some atomic steps can be observed on the surfaces.
The lattice planes parallel to the surfaces are uniquely indexed
as (006) and (101j) according to the lattice spacings, 0.622 and
Figure 1. (A) TEM image and the corresponding SAED pattern of ACP, the precursor, extracted from the reaction at 15 s. (B) HRTEM image of
the amorphous precursor in and no crystal lattice fringe is observed. (C) SEM image of the amorphous precursor. (D, E) FT-IR spectrum and XRD
pattern of the precursor.
Figure 2. SEM and HRTEM images of final β-TCP hexagon and octahedron. (A) SEM of a typical hexagon. (B) TEM of β-TCP hexagon recorded
along [001] zone axis. (C, D) HRTEM images of the right and left side marked in B. (E) SEM of a typical octahedron. (F) TEM image of β-TCP
octahedron recorded along the [010] direction; the angle between the adjacent surfaces is 76°. (G) HRTEM image recorded from the left side
surface marked in F; the lattice fringe parallel with the outer surface is corresponding to (006) lattice plane. (H) HRTEM image recorded from the
right side surface marked in F. The lattice fringe parallel with the outer surface is corresponding to (101j) lattice plane.
3156 Crystal Growth Design, Vol. 9, No. 7, 2009 Tao et al.

0.877 nm, respectively. As a confirmation of these indices, the
angle between the planes is 77° by calculation, which is
consistent with the measured value, 76°. It is important to
mention that biominerals usually have shapes that defy the strict
geometric restrictions of 230 classical space groups. The
symmetry breaking during the phase transformation from
isotropic amorphous to anisotropic crystal is an interesting
phenomenon. Its reason is unclear and further efforts will be
paid for an explanation.
It is well-reported that β-TCP cannot be formed in aqueous
solution as the involvement of proton and hydroxyl ions.19
If
the same amorphous phase is dispersed in water or calcium
phosphate aqueous solution, the resulted products are the rodlike
hydroxyapatite nanocrystals (Figure 4). However, the pure
β-TCP phase can be synthesized in nonaqueous solvent such
as EG or methanol.3a,19e
EG is a solvent with relatively high
dielectric constant that can dissolve many salts. Another property
of EG is its high boiling points (∼196 °C), which is suitable
for synthesis of highly crystallized materials.5d
Xia and co-
workers have successfully controlled the shape of noble metals
nanocrystals using EG.5a,c,d
It is found that EG has a strong
effect on mechanism of ion solvation and dissociation.3a
The
molar conductivity of ions in aqueous environment is ap-
proximately one order larger than that in EG, which indicates
that greater activities of ions in water than in EG. It is suggested
that the relatively low driving force for precipitation in EG may
be beneficial to the formations of uniform crystals.6c,24
To some extent, our results have phenomenological similari-
ties to the “gel-sol” mechanism proposed by Sugimoto.25
This
mechanism is first proposed on the basis of a metal hydroxide
gel to be transformed into uniform metal oxide sol through a
dissolution-recrystallization process. During this process, a
highly viscous metal hydroxide gel network is used as a matrix
for holding the nuclei and growing particles to protect them
from aggregation even in strong ionic strength conditions, and
also as a reservoir of metal ions or hydroxide ions to compensate
a drastically reduced supersaturation during the growth of
crystal. The dissolution-recrystallization model is frequently
discussed in the phase transformation of calcium minerals.
However, our system shows a different pathway that the crystal
nucleates directly at the precursor and its shape can be controlled
just by changing the growth environment. The precursor need
not to dissolve to provide nutrient for crystallization, and it can
be understood by an energetic controls of the interfaces. To
investigate the evolution process of the ACP precursor in EG,
we withdrew the samples from the same reaction system
periodically (Figure 5). The extracted mixture is quickly
Figure 3. TEM and SEM images and SAED pattern of samples synthesized at 150 °C using different precursor amounts. The total volume of EG
is 70 mL (CaCl2 and Na2HPO4 concentrations of 7.6 and 3.7 mM, respectively). (A, C) 10 mg precursor. (B) SAED pattern of the hexagon marked
in A. (D, E) 15 mg precursor. (F, G) 26 mg precursor. (H, I) 40 mg precursor. This study shows that the size of hexagonal plates can be adjustable
by the amorphous precursor amounts.
Figure 4. Hydroxyapatite nanorods formed after phase transformation
by amorphous precursor in water. (A) TEM image of the sample. (B)
Corresponding SAED pattern of this sample in A clarified the phase is
hydroxyapatite.

quenched to -16 °C to terminate the reaction, and the product
is collected by centrifugation (4000 g) at -4 °C. After
transformation for 10s, the amorphous solids are still continuous
without significant change. However, the clusters with ∼2-5
nm among the amorphous precursor are detected inside the ACP.
These initially crystallized clusters are considered to provide
the nucleation and growth sites for the crystal phase. The lattice
structure in these clusters can be detected and the interplanar
spacing, 0.208 nm, is consistent with the crystallographic data
of the (00,18) plane of β-TCP (Figure 5A). The clusters are
randomly distributed in the amorphous phase as the ringlike
diffraction patterns are obtained. The density and size of these
clusters increase with the reaction process. At a time of 20 s,
the spherical aggregates begin to form within the amorphous
phase. A covering film of lower contrast on their surfaces acts
as a buffer between original precursor and aggregates (Figure
5B). This buffer layer is an indication of nucleation inside the
amorphous precursor. Furthermore, the spherical aggregate is
remolded into crystallites with hexagonal shapes at 32 s (Figure
5C). The creation of well-faceted crystallites distorts the
precursor film for the generation of stress between these two
phases (inset SEM image in Figure 5C). It also shows that the
crystallites and the precursors are actually integrated without
any obvious boundary. Another important experimental phe-
nomenon is that the increasing the crystallized phase is
proportional to the decreasing of the amorphous one. It is also
noted that in the current phase transformation system, the solid
precursor, ACP, shares a similar chemical composition (Ca/P
) 1.47 ( 0.05) with β-TCP (Ca/P ) 1.51 ( 0.02) crystallites
and no additional ions are required during the evolution. Thus,
we conclude that the crystallite may directly nucleated by
solid-solid phase transformation from the precursor (Figure
5D). The amorphous precursor epitaxial nucleation rather than
so-called dissolution-recrystallization25
can explain the revealed
evolution process indicated by TEM observation as well as in
the viewpoint of interfacial energy control. Figure 6A shows
the initial state of the phase transformation in quite a short time
(within 1 min). An extended reaction time (∼1-2 min) leads
to a significant decrease in precursor amount and increase in
the crystal sizes (Figure 6B). When the reaction time is
prolonged to 3 min, the amorphous precursor disappears
completely and only the crystals with smooth outer surfaces
can be observed (Figure 6C). The sharp edge forms within 7
min. Further extension of reaction time shows no obvious
improvement in crystallinity and size.
According to Ostwald’s phase rule,26
the first formed phase
in polymorphism is normally the one that is closest in free
energy to the mother solution; that is, the least stable phase,
followed by phases with increasing stability. Amorphous
precursor mediated crystallization is a specific example of
Ostwald’s rule that has attracted great attention. Experimentally,
this mechanism is observed during the crystal growth of proteins
and colloids.27
As revealed by the previous literature,28
the
nucleation rate, Γ, can be represented as eq 1
where ∆G* is the height of the free energy barrier separating
the metastable phase from the crystal phase. The kinetic factor,
ν, is a measure of the rate at which critical nuclei, once formed,
transform into larger crystallites. The variation of the nucleation
rate is dominated by the variation in the barrier height. The
form of ∆G* can be given by the classical nucleation theory
where γ is the interfacial energy per unit area of the phase
interface, F is the number density of the solid phase, and ∆µ is
the difference in chemical potential between the metastable
phase and the crystal phase.
After the addition of amorphous precursor in our system, the
new equilibrium between the amorphous precursor and solution
is reached. The free energy barrier, ∆G*, is directly determined
by the interfacial energy, γ. The surface tension components
of amorphous precursor, hexagon and octahedron are determined
by wicking techniques with probing liquids of water, n-octane,
ethylene glycol, and DMSO. The solid surface tension compo-
nents, Lifshitz-van der Waals (γLW
) and Lewis acid-base (γAB
) 2(γ+
γ-
)1/2
)29
are obtained when Young eq 3 is solved
where the subscripts S and L represent the solid surface and
test liquids, respectively. γ+
is the Lewis acid (electron-acceptor)
and γ-
is the Lewis base (electron-donor) parameters. θ is the
contact angle between the test liquid and solid surface. The
observed contact angles of the various liquids on the ACP,
hexagon, and octahedron β-TCP are listed in Table 1. The
standard parameters of liquids and the calculated results of
amorphous precursor, hexagon, and octahedron are given in
Table 2.
The total interfacial tension between two different condensed
phases can be estimated from eq 4
Figure 5. HRTEM images of the phase transformation within 1 min.
(A) At 10s, different contrasts indicate that the clusters generated among
the amorphous matrix. (B) Spherical aggregates are formed in the
precursor with a low contrast buffer layer at about 20 s. (C) Spherical
aggregate remolded into hexagonal crystallite at about 32 s. The inset
SEM image indicates that the crystallite stems from the amorphous
precursor as the continuous connection between the precursor and the
crystallite. (D) Sample extracted at 50 s. The hexagon grows at the
expense of precursor. The inset SEM image indicates that the crystallite
has an improved shape.
Γ ) νexp(-∆G∗
/kBT) (1)
∆G∗
) 16πγ3
/(3F2
∆µ2
) (2)
(1 + cos θ)γL ) 2(√γS
LW
γL
LW
+ √γS
+
γL
-
+ √γS
-
γL
+
)
(3)
γij ) (√γi
LW
- √γj
LW
)2
+ 2(√γi
+
γi
-
+ √γj
+
γj
-
- √γi
+
γj
-
-
√γi
-
γj
+
) (4)

The interfacial energies between the amorphous precursor and
β-TCP hexagon (γAm-Hex), amorphous precursor and β-TCP
octahedron (γAm-Oct) are 0.026 and 0.006 mJ/m2
, respectively,
which are calculated from eq 4 using the data in Table 2. In
contrast, the interfacial energy between β-TCP hexagon and
CaCl2-Na2HPO-EG solution (γSH-Hex), β-TCP octahedron and
Ca(OH)2-(NH4)2HPO4-EG solution (γSO-Oct) are 1.65 and 1.50
mJ/m2
, respectively, by using eq 5 and data in Tables 1 and 2
The interfacial energy between crystal and amorphous precursor
(γAm-Hex or γAm-Oct) is lower compared to that between crystal
and solution (γSH-Hex or γSO-Oct) by about two magnitudes. Hence,
the free energy barrier between amorphous precursor and β-TCP
is far lower than that between β-TCP crystals and solutions
according to eq 2. Thus, the nucleation of β-TCP in the
amorphous precursor is thermodynamically preferred as shown
in eq 1. The amorphous precursor epitaxial nucleation process
occurs during the formation of both hexagon and octahedron.
Interestingly, the life spans of amorphous precursor in these
two solutions are quite different. The amorphous precursor
disappeared at 2.3 min in the formation of hexagon (Figure 6C).
In the case of octahedron formation, the precursor still exists widely
at the reaction time of 2.5 min (Figure 6F). Besides, the interfacial
energies between the growth solutions and crystals with different
shapes are quite different. The interfacial energies between
octahedron and CaCl2-Na2HPO4-EG solution (γSH-Oct), hexagon,
and Ca(OH)2-(NH4)2HPO4-EG solution (γSO-Hex) are 2.24 and
1.80 mJ/m2
, respectively. It is mentioned that γSO-Hex is larger
than γSH-Hex and γSH-Oct is larger than γSO-Oct. Therefore, both
formations of octahedron in CaCl2-Na2HPO4-EG solution and
hexagon in Ca(OH)2-(NH4)2HPO4-EG solution are thermo-
dynamically unfavorable for their relatively large interfacial
energies. This also indicates that the final morphology of crystal
is determined by the crystal-solution interfacial energies,
because the amorphous precursor disappears eventually. Only
the crystal-solution interfaces are present at the end of phase
transformation. Therefore, the amorphous precursor alters the
kinetic evolution pathway instead of changes the thermodynami-
cally stable shape of product. Furthermore, our attachment
energy calculation of β-TCP without any additives have selected
out the lattice planes with the lowest attachment energy, that
is, the planes with the lowest growth rate that determine the
final morphology.30
The lattice planes that enclose the hexagon
and octahedron have the lowest attachment energies in the plane
list of β-TCP phase, as shown in Table 3.
Conclusion
In summary, β-TCP crystals with different morphologies and
sizes are synthesized in an organic solvent using ACP as the
starting material. In this method, the resulted nano octahedrons
can be even against the classical crystal symmetry of β-TCP. It
Figure 6. SEM images of the sample evolutions. (A) Initial hexagon sample in the amorphous phase at 30 s, the crystallites are indicated by the
white arrows. (B) Samples at 1.2 min. The crystallites increase in both density and size; at this stage, the precursor coexists with the crystals. (C)
Samples at 2.3 min. The precursor has completely disappeared and the hexagons result. (D) The 3D atomic model of β-TCP hexagon. (E) Initial
octahedrons at 30 s; their morphology is spherelike. (F) Octahedron sample at 2.5 min, The white arrow indicates the crystallites, but they are not
fully developed; octahedral shape can be observed at this stage but the amorphous precursor still exists. (G) Sample at 90 min; the uniform octahedral
crystals formed. The inset is a high-magnification image of the β-TCP octahedron. (H) The 3D atomic model of β-TCP octahedron.
Table 1. Contact Angles of Probing Liquids on Amorphous
Precursor, β-TCP Hexagon, and Octahedron Surfaces at 20 °C and
Relative Humidity of 70%
ACP hexagon octahedron
DMSO 21.9 ( 3.0 19.4 ( 4.1 19.0 ( 0.5
EG 22.6 ( 2.8 16.3 ( 0.7 22.8 ( 2.0
n-octane 0 0 0
water 21.6 ( 3.6 ≈0 18.8 ( 2.7
CaCl2-Na2HPO4-EG 19.3 ( 0.9 25.5 ( 0.6
Ca(OH)2-(NH4)2HPO4-EG 18.7 ( 0.6 22.4 ( 3.5
Table 2. Surface Tension Components of Different Solvents and
Parameters of Amorphous Precursor, β-TCP Hexagon, and β-TCP
Octahedron Determined by Wicking Method at 20 °C (mJ/m2
)
γ γLW
γAB
γ+
γ-
DMSO 44.00 36.00 8.00 0.50 32.00
EG 48.00 29.00 19.00 1.92 47.00
n-octane 21.62 21.62 0 0 0
water 72.80 21.80 51.00 25.50 25.50
CaCl2-Na2HPO4-EG 48.45
Ca(OH)2-(NH4)2HPO4-EG 48.12
ACP 45.63 21.86 23.77 2.14 65.87
hexagon 47.38 21.79 25.59 2.21 74.11
octahedron 45.98 22.03 23.95 2.12 67.62
γsolid-solution ) γsolid - γsolutioncos θ (5)

is found that the crystallization of crystalline phases can occur
and develop directly within the ACP phases because of the lower
interfacial energies between the solids. However, the final shape
of crystals is controlled by alteration of crystal-solution
interfacial energy. The crystallized phase can also be controlled
by intervention the ACP precursor epitaxial crystallization by
different temperatures and precursor concentrations, etc. This
study suggests that a combination of amorphous precursors and
energetic controls can provide a novel strategy of material
manufacture and its mechanism may be applied in the studies
of biomineralization.
Acknowledgment. We thank Dr. Dexi Zhu and Prof. Hui
Ye for their assistance in the examinations. This work was
supported by National Natural Science Foundation of China
(20571064 and 20601023) and Cheung Kong Scholars Program
(RT).
Supporting Information Available: Samples synthesized at dif-
ferent temperatures (Figure S1), XRD patterns of hexagon and
octahedron β-TCP crystals (Figure S2) (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
References
(1) (a) Burda, C.; Chen, X.; Narayanan, R.; EI-Sayed, M. A. Chem. ReV.
2005, 105, 1025. (b) Sugimoto, T. Chem. Eng. Technol. 2003, 26,
313.
(2) (a) Ren, J.; Tilley, R. D. J. Am. Chem. Soc. 2007, 129, 3287. (b) Tian,
N.; Zhou, Z.; Sun, S.; Ding, Y.; Wang, Z. L. Science 2007, 316, 732.
(3) (a) Tao, J.; Jiang, W.; Zhai, H.; Pan, H.; Xu, X.; Tang, R. Cryst.
Growth Des. 2008, 8, 2227. (b) Pan, H.; Tao, J.; Yu, X.; Fu, L.; Zhang,
J.; Zeng, X.; Xu, G.; Tang, R. J. Phys. Chem. B 2008, 112, 7162.
(4) (a) Dos Santos, E. A.; Farina, M.; Soares, G. A.; Anselme, K. J. Mater.
Sci: Mater. Med. 2008, 19, 2307. (b) Yin, X.; Stott, M. J. J. Chem.
Phys. 2006, 124, 124701.
(5) (a) Sun, Y.; Xia, Y. Science 2002, 298, 2176. (b) Habas, S. E.; Lee,
H.; Radmilovic, V.; Somorjai, G. A.; Yang, P. Nat. Mater. 2007, 6,
692. (c) Xiong, Y.; Cai, H.; Wiley, B. J.; Wang, J.; Kim, M. J.; Xia,
Y. J. Am. Chem. Soc. 2007, 129, 3665. (d) Wiley, B. J.; Sun, Y.;
Mayers, B.; Xia, Y. Chem.sEur. J. 2005, 11, 454. (e) Jin, R.; Cao,
Y.; Mirkin, C. A.; Kelly, K. L.; Schatz, G. C.; Zheng, J. G. Science
2001, 294, 1901.
(6) (a) Peng, X. AdV. Mater. 2003, 15, 459. (b) Manna, L.; Scher, E. C.;
Alivisatos, A. P. J. Am. Chem. Soc. 2000, 122, 12700. (c) Yin, Y.;
Alivisatos, A. P. Nature 2005, 437, 664.
(7) (a) Hinotsu, T.; Jeyadevan, B.; Chinnasamy, C. N.; Shinoda, K.; Tohji,
K. J. Appl. Phys. 2004, 95, 7477. (b) Sun, S.; Murray, C. B.; Weller,
D.; Folks, L.; Moser, A. Science 2000, 287, 1989. (c) Chen, M.; Kim,
J.; Liu, J. P.; Fan, H.; Sun, S. J. Am. Chem. Soc. 2006, 128, 7132.
(8) (a) Beniash, E.; Aizenberg, J.; Addadi, L.; Weiner, S. Proc. R. Soc.
London, Ser. B 1997, 264, 461. (b) Politi, Y.; Arad, T.; Klein, E.;
Weiner, S.; Addadi, L. Science 2004, 306, 1161.
(9) Addadi, L.; Raz, S.; Weiner, S. AdV. Mater. 2003, 15, 959.
(10) Weiss, I. M.; Tuross, N.; Addadi, L.; Weiner, S. J. Exp. Zool. 2002,
293, 478.
(11) (a) Lowenstam, H. A.; Weiner, S. Science 1985, 227, 51. (b) Mahamid,
J.; Sharir, A.; Addadi, L.; Weiner, S. Proc. Natl. Acad. Sci. U. S. A.
2008, 105, 12748. (c) Tao, J.; Pan, H.; Wang, J.; Wu, J.; Wang, B.;
Xu, X.; Tang, R. J. Phys. Chem. C 2008, 112, 14929. (d) Le´ve´que, I.;
Cusack, M.; Davis, S. A.; Mann, S. Angew. Chem., Int. Ed. 2004, 43,
885.
(12) Dorozhkin, S. V.; Epple, M. Angew. Chem., Int. Ed. 2002, 41, 3130.
(13) (a) van Haaren, E. H.; Smit, T. H.; Phipps, K.; Wuisman, P. I. J. M.;
Blunn, G.; Heyligers, I. C. J. Bone Joint Surg. 2005, 87-B, 267. (b)
Ogose, A.; Hotta, T.; Kawashima, H.; Kondo, N.; Gu, W.; Kamura,
T.; Endo, N. J. Biomed. Mater. Res. 2004, 72B, 94. (c) Fujita, R.;
Yokoyama, A.; Nodasaka, Y.; Kohgo, T.; Kawasaki, T. Tissue Cell
2003, 35, 427. (d) Guo, X.; Wang, C.; Zhang, Y.; Xia, R.; Hu, M.;
Duan, C.; Zhao, Q.; Dong, L.; Lu, J.; Song, Y. Tissue Eng. 2004, 10,
1818.
(14) Lee, Y. M.; Park, Y. J.; Lee, S. J.; Ku, Y.; Han, S. B.; Choi, S. M.;
Klokkevold, P. R.; Chung, C. P. J. Periodontol 2000, 71, 410.
(15) (a) Sous, M.; Bareille, R.; Rouais, F.; Cle´ment, D.; Ame´deé, J.; Dupuy,
B.; Baquey, C. Biomaterials 1998, 19, 2147. (b) Takahashi, Y.;
Yamamoto, M.; Tabata, Y. Biomaterials 2005, 26, 3587.
(16) Laffargue, P.; Fialdes, P.; Frayssinet, P.; Rtaimate, M.; Hildebrand,
H. F.; Marchandise, X. J. Biomed. Mater. Res. 2000, 49, 415.
(17) (a) Gineba, M. P.; Traykova, T.; Planell, J. A. J. Controlled Release
2006, 113, 102. (b) Paul, W.; Sharma, C. P. J. Biomater. Appl. 2003,
17, 253. (c) Donker, H.; Smit, W. M. A.; Blasse, G. J. Electrochem.
Soc. 1989, 136, 3130. (d) Legrouri, A.; Lenzi, J.; Lenzi, M. React.
Kinet. Catal. Lett. 1998, 65, 227.
(18) (a) Yashima, M.; Sakai, A.; Kamiyama, T.; Hoshikawa, A. J. Solid
State Chem. 2003, 175, 272. (b) Bigi, A.; Foresti, E.; Gandolfi, M.;
Gazzano, M.; Roveri, N. J. Inorg. Biochem. 1997, 66, 259. (c) Pan,
Y.; Huang, J.; Shao, C. Y. J. Mater. Sci. 2003, 38, 1049. (d) Wei, X.;
Akinc, M. J. Am. Ceram. Soc. 2007, 90, 2709. (e) Belik, A. A.; Izumi,
F.; Stefanovich, S. Y.; Malakho, A. P.; Lazoryak, B. I.; Leonidov,
I. A.; Leonidova, O. N.; Davydov, S. A. Chem. Mater. 2002, 14, 3197.
(19) (a) O¨ zgu¨r Engin, N.; Cu¨neyt Tas, A. J. Am. Ceram. Soc. 2000, 83,
1581. (b) Gibson, I. R.; Rehman, I.; Best, S. M.; Bonfield, W. J. Mater.
Sci.:Mater. Med. 2000, 11, 533. (c) Kannan, S.; Ventura, J. M.;
Ferreira, J. M. F. Ceram. Int. 2007, 33, 637. (d) Kwon, S.; Jun, Y.;
Hong, S.; Kim, H. J. Eu. Ceram. Soc. 2003, 23, 1039. (e) Bow, J.;
Liou, S.; Chen, S. Biomaterials 2004, 25, 3155.
(20) Koutsopoulos, S. J. Biomed. Mater. Res., A 2002, 62, 600.
(21) (a) Posner, A. S.; Betts, F. Acc. Chem. Res. 1975, 8, 273. (b) Betts,
F.; Blumenthal, N. C.; Posner, A. S.; Becker, G. L.; Lehninger, A. L.
Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 2008. (c) Levi-Kalisman, Y.;
Raz, S.; Weiner, S.; Addadi, L.; Sagi, I. AdV. Funct. Mater. 2002, 12,
43. (d) Levi-Kalisman, Y.; Raz, S.; Weiner, S.; Addadi, L.; Sagi, I.
J. Chem. Soc., Dalton Trans. 2000, 3977. (e) Politi, Y.; Levi-Kalisman,
Y.; Raz, S.; Wilt, F.; Addadi, L.; Weiner, S.; Sagi, I. AdV. Funct.
Mater. 2006, 16, 1289.
(22) Saida, J.; Matsushita, M.; Inoue, A. J. Appl. Phys. 2001, 90, 4717.
(23) Grassmann, O.; Neder, R. B.; Putnis, A.; Lo¨bmann, P. Am. Mineral.
2003, 88, 647.
(24) Jiang, X.; Herricks, T.; Xia, Y. AdV. Mater. 2003, 15, 1205.
(25) (a) Sugimoto, T.; Sakata, K. J. Colloid Interface Sci. 1992, 152, 587.
(b) Sugimoto, T.; Sakata, K.; Muramatsu, A. J. Colloid Interface Sci.
1993, 159, 372. (c) Sugimoto, T.; Muramatsu, A. J. Colloid Interface
Sci. 1996, 184, 626. (d) Sugimoto, T.; Wang, Y. J. Colloid Interface
Sci. 1998, 207, 137. (e) Sugimoto, T. J. Colloid Interface Sci. 2007,
309, 106.
(26) Ostwald, W. Z. Phys. Chem. 1897, 22, 289.
(27) (a) Kuznetsov, Y. G.; Malkin, A. J.; McPherson, A. J. Cryst. Growth
2001, 232, 30. (b) Vekilov, P. G. Cryst. Growth Des. 2004, 4, 671.
(c) Chen, X.; Samia, A. C. S.; Lou, Y.; Burda, C. J. Am. Chem. Soc.
2005, 127, 4372. (d) Lutsko, J. F.; Nicolis, G. Phys. ReV. Lett. 2006,
96, 046102. (e) Zhang, T. H.; Liu, X. Y. J. Am. Chem. Soc. 2007,
129, 13520.
(28) Ten Wolde, P. R.; Frenkel, D. Science 1997, 277, 1975.
(29) (a) Wu, W.; Giese, R. F., Jr.; van Oss, C. J. Langmuir 1995, 11, 379.
(b) Wu, W.; Nancollas, G. H. AdV. Colloid Interface Sci. 1999, 79,
229.
(30) Berkovitch-Yellin, Z. J. Am. Chem. Soc. 1985, 107, 8239.
CG801130W
Table 3. Lattice Planes with the Lowest Attachment Energy, Where
Eatt is the Attachment Energy; These Planes Were Used to
Construct the Surface of Crystals Together with the HRTEM
Lattice Images
crystal face index Eatt (kcal/mol)
(010), (01j0)
(11j0), (1j10)
(100), (1j00) 45.19
(110), (1j1j0)
(12j0), (1j20)
(21j0), (2j10) 57.97
(001), (001j) 60.96
(101j), (1j01) 100.27
(11j1), (1j11j) 100.29
(011), (01j1j) 100.31

Structural Components and Anisotropic Dissolution Behaviors in
One Hexagonal Single Crystal of β-Tricalcium Phosphate
Jinhui Tao, Wenge Jiang, Halei Zhai, Haihua Pan, Xurong Xu, and Ruikang Tang*
Department of Chemistry and Center for Biomaterials and Biopathways, Zhejiang UniVersity,
Hangzhou, 310027, P. R. China
ReceiVed August 27, 2007; ReVised Manuscript ReceiVed NoVember 20, 2007
ABSTRACT: Large-scale β-tricalcium phosphate (β-TCP) hexagonal single crystals were synthesized at a relatively low temperature
(150 °C) by using a solution-phase method. The solvent, ethylene glycol, played an important role during the formation of the
homogeneous submicron-sized crystals. Unlike the conventional understanding of a single crystal, the wall of the formed β-TCP
hexagonal was well crystallized, showing different physicochemical properties from the bulk part. The dissolution spots were
anisotropically distributed throughout the single crystal. The bulk part dissolved readily from the top and bottom planes in the
undersaturated solutions, but the thin hexagonal wall could be stable against any dissolution even in pure water. These differences
between the wall and the bulk part were attributed to the different crystallinities and defect densities in their structures. It was
suggested that the low defect number might stem from the solvent-interface exchange that was allowed the edge surfaces in contact
with the solution. And the rapid growth of the particles resulted in the randomly distributed defects in the bulk part, which induced
a selective dissolution along the c-axis of β-TCP. Furthermore, the stability of wall could be explained by a size effect during the
nanodemineralization. It was interesting that both the wall and the bulk part shared the exact same lattice fringes under the transmission
electron microscope. This phenomenon implied that both components were crystallographically identical so that they were constructed
into an integral single crystal of β-TCP. The distinct dissolution behaviors of these two parts in one single crystal resulted in the
formation of porous, gearlike, and ringlike single crystals at different demineralization stages, which demonstrated an easy control
of crystal morphology patterns by using the anisotropic dissolution behavior.
Introduction
Because of the dependence of physical and chemical proper-
ties on the size, morphology and microstructure of materials,1–3
controllable synthesis of nanocrystals with various shapes and
structural complexities with high precision presents a great
challenge in nanosized materials synthesis.4–7
The morphology
control of single crystals of natural minerals such as calcium
carbonates and calcium phosphates is also an essential charac-
teristic of biomineralization.8–11
The precise control of crystals
is intensively investigated in biominerals.12–16
Many organisms
shows exceptional control over the gross morphology, physical
properties, and nanoscale organization of biomaterials, creating
shapes that defy strict geometrical restrictions.8,10,13–16
A
remarkable category of biominerals is the single crystal with
complex form although they have the complicated structures.8,14,15
Inspired by biomineralization, various approaches have been
developed to the large-scale control of structures and morphol-
ogiesofnanoparticles,mainlybyalteringadditivesorsolvents,17–19
template-aided synthesis,20–24
and self-assembly.25
These meth-
ods usually include relatively complicated operations, low yields,
or poor controllability in uniformities and shapes. Besides
biomineralization, it is also noted that biodemineralization is
another useful strategy in the control of single crystals in living
systems. Here, we demonstrate that polymorph control of
β-tricalcium phosphate (β-TCP) can be conveniently achieved
by an anisotropic dissolution behavior of the hexagonal single
crystals. A series of the derivative morphologies including
porous, gearlike, and ringlike are achieved at different time
scales of demineralization.
β-TCP is an important biomineral since it has potential
applications in bone grafting, calcium phosphate cements and
surgical implants.26
In the present work, we report that the
hexagonal single crystals of β-TCP are ﬁrst synthesized by using
a solution method under a relatively low temperature. Unlike
the conventional understanding of a single crystal, the crystal-
linities of six edges and the bulk part in as-prepared β-TCP are
different although their chemical compositions, phases, and
crystallographic structures are exactly identical. The improved
crystallinity and thin thickness of the edge wall can protect this
part against dissolution reaction in water even though the bulk
part is completely etched. It shows that the anisotropic dissolu-
tion of the structural complex in one single crystal can result in
an easy but effective control of morphologies of the single
crystal.
The hexagonal β-TCP plates were synthesized by a solution-phase
method. Ethylene glycol was used as solvent and CaCl2 and Na2HPO4
were used as calcium and phosphate sources for the precipitation,
respectively. 0.10 g CaCl2 ·2H2O was mixed with 50 mL of ethylene
glycol and the slurry was heated to 150 °C under vigorous magnetic
stirring. A mixed aqueous solution of 1.36 mL of 0.3 M Na2HPO4 and
120 µL of 1.3 M NaOH was added to 20 mL of ethylene glycol at a
temperature of 95 °C. The phosphate-containing ethylene glycol solution
was added dropwise into the calcium containing ethylene glycol solution
at a rate of 20 mL/min. The mixture was bathed at 150 °C for 90
min and then was cooled in air. The solids were separated by
centrifugation at 2000g and were washed using ethanol and water
alternatively 3 times to remove the residual solvent or other impurities.
The products were dried under a vacuum condition at 30 °C. The
chemical compositions and structures of the solids were characterized
by chemical analysis (atomic adsorption for calcium and UV for
phosphate). The molar conductivities of CaCl2 and Na2HPO4 in water
and in ethylene glycol were also examined to discuss the roles of solvent
in the formation of β-TCP.
In the demineralization experiments, 1.5 mg of solids was dispersed
into 50 mL of water (pH ) 7.0) under a stirring condition. One milliliter
slurry samples were withdrawn at different experimental periods. The
* Corresponding author: Department of Chemistry, Zhejiang University,
Hangzhou, 310027, China, Tel/fax: +86-571-87953736. E-mail: rtang@
zju.edu.cn.
CRYSTAL
GROWTH
DESIGN
2008
VOL. 8, NO. 7
2227–2234
10.1021/cg700808h CCC: $40.75  2008 American Chemical Society
Published on Web 06/05/2008

solids were separated by centrifugation (10000g). In order to investigate
the effects of undersaturation on the dissolution of β-TCP, a parallel
experiment was performed by using a low content (0.015 mg) of seeds
to increase the final undersaturation level. Some synthesized β-TCP
crystallites were also heated to 500 °C in the presence of flowed air to
examine the influence of calcination and organic residuals on the
dissolution kinetics.
All the solids were examined by using a JEM-200CX (JEOL, Japan)
transmission electron microscope (TEM) and a JEM-2010HR (JEOL,
Japan) high resolution TEM (HRTEM). Scanning electron microscopy
(SEM) was performed using a S-4800 field-emission scanning electron
microscope (HITACHI, Japan). The samples were also measured by a
Nanoscope IVa atomic force microscope (AFM, Veeco). The phase of
the solids was examined by a D/max-2550pc XRD (Rigaku, Japan)
with monochromatized Cu KR radiation at the working voltage of 40
kV, and the scanning step was 0.02°.
Results and Discussion
The phase of the obtained solid was examined by X-ray
diffraction (XRD, Figure 1). All the peaks could be well indexed
by using the standard card of β-TCP (JCPDS: 09-0169, a ) b
) 10.42 Å, c ) 37.38 Å; R ) β ) 90°, γ ) 120°; space group
of R3jc (167), Figure S4, Supporting Information). The result
of chemical analysis showed that the atomic molar ratio of
calcium to phosphate of the solids was 1.51 ( 0.02, which was
consistent with the stoichemical value of ideal β-TCP, 1.50.
These results confirmed that we obtained β-TCP crystals by
using a feasible, large-scale, and controllable synthesis method
in the laboratory. β-TCP was widely used as the calcium
phosphate bone cement in biomedical areas. The other important
applications of this compound included fertilizers, polishing,
dental powders, porcelains, pottery, and animal food supple-
ments. In the previous literature,26
it was widely accepted that
β-TCP could only be obtained by calcination of calcium
deficient hydroxyapatite at temperature above 800 °C. The
previously synthesized β-TCP crystallites had the irregular
morphologies and nonuniform sizes.27
However, our preparation
was performed at a much lower temperature (150 °C) in ethylene
glycol. The formed β-TCP crystals were hexagonal plates, and
their sizes could be well controlled. This new method provided
a convenient but effective pathway to prepare β-TCP crystallites.
It was believed that the solvent, ethylene glycol, played a key
role in the crystallization. The molar conductivity of CaCl2 and
Na2HPO4 in water and in ethylene glycol was measured (Figure
S1, Supporting Information). The curves indicated that the
amounts of free calcium and phosphate ions in the aqueous
solution were significantly greater than those in ethylene glycol.
Besides, the influence of electrolyte concentration on its molar
conductivity in ethylene glycol was negligible since the molar
conductivities of CaCl2 and Na2HPO4 were almost unchanged
in Figure S1. This result indicated that ethylene glycol provided
a medium for the controlled release of free calcium and
phosphate ions from their electrolyte solids. Thus, a low but
stable driving force was maintained during the precipitation of
β-TCP in the ethylene glycol solvent, which promoted the
formation of the well-crystallized crystals.
The obtained β-TCP were examined by SEM, TEM, and
AFM. A typical SEM of the as-prepared samples is shown in
Figure 1a. It can be seen that the hexagonal plates had the size
distribution of 750-800 nm. The thickness of the plates,
200-250 nm, was measured by their side view (Figures 1 and
S2, Supporting Information). The result of selected area electron
Figure 1. SEM micrographs of samples extracted at different time scales. (a) SEM of the synthesized hexagonal plates of β-TCP, the side view of
plates could give thickness information (white circle and inset); the other inset is the magnified image of the plate indicated by the white arrow,
which shows the pits on the surface (arrows). (b) Samples after demineralization for 21 h. The density and size of the pits increased obviously;
some of them even passed throughout the plate to form the holes. The inset image is the magnification of the plate denoted by the white arrow. (c)
Samples after demineralization for 12 days. Only the rings survived, and they had the same dimensions as the solid plates. The magnified graph of
the single ring indicated by the white arrow is shown as the inset. (d) XRD pattern of hexagonal solids, all the peaks could be assigned to β-TCP.
The XRD pattern of hollow rings was exactly the same (Figure S4).

diffraction (SAED) indicated that the top/bottom surface of the
hexagonal plate was identical to (001) facets of β-TCP. The
diffraction dots and their 6-fold symmetry showed that the whole
plate was a single crystal (Figure 2a), which was also supported
by the direct measurements of their lattice structures (Figures
2b, 2c, and 2d).
Different from the nature of the perfect single crystals, the
structure of the β-TCP hexagonal plate was not consistent, for
each plate of β-TCP, the six thin sides acted as a wall to wrap
the inside part, the bulk. This structural complex was well
displayed by a demineralization reaction of the solids (Figure
1), which showed the distinct behaviors of two components of
the crystal. The dissolution phenomena clearly implied that the
edge wall and the bulk part might have different physicochem-
ical properties although they were in one single crystal.
The differences in contrast under bright-field TEM image
(Figure 2a) indicated that the internal texture of the bulk part
was actually not uniform, which might be caused by the different
crystallinity or thickness. By using SEM, it was noted that the
surface of the bulk part was not perfect too and some pits were
present (black arrows, inset of Figure 1a). As the previous
understanding,28,29
these pits could provide the active sites to
initiate crystal dissolution. Thus, the spontaneous demineral-
ization of the plate surface occurred spontaneously when an
undersaturated medium, e.g. water, was introduced. When the
particles were immersed into water for 21 h (free drift
dissolution), the pits extended to contribute to the dissolution
reaction. However, the kinetic rates of these pit developments
were anisotropic. It seemed that their dissolution directions were
more preferred along the c-axis to penetrate the plates. As a
result, the dissolution holes were formed (Figure 1b). Actually,
a similar selective dissolution process had been reported and
explained in a demineralization model of dental enamel,28b
in
which the etched enamel surfaces only developed along the
c-axes of hydroxyapatite. Furthermore, the resulting pits and
holes on the β-TCP were almost irregular, e.g. the density,
morphology, and size of the pits and holes, resulting in various
porous structures (Figures 1b, 3a, and 3d). This phenomenon
also implied the random and heterogeneous internal texture of
the bulk part of the β-TCP hexagonal plates. During the
dissolution process, the layered structure of the bulk during the
dissolution was also revealed (Figure 3b). It could be found
that each layer had the same crystallographic lattice structure
and orientations. The layers packed along the c-axis. This
ordered texture was another proof to confirm that the single
crystal structure was formed in the bulk part. Although the
dissolution spots on the plates were random, it was interesting
to note that no dissolution occurred on the six edges. Figure 3d
clearly showed that the wall structure was maintained well in
the partially dissolved plates. In contrast, the conventional crystal
dissolution model described that the edges should be more
readily dissolved since they provided more natural dislocation
sources.
When the dissolution reaction was extended to 62 h, the
porouslike β-TCP crystallites evolved into the gearlike rings
(Figure 4a). At this stage, most of the bulk part disappeared
Figure 2. HRTEM studies of hexagonal β-TCP plates. (a) A single hexagonal plate and its corresponding SAED recorded along the [001] zone
axis. Three different sites (circles) were used for the measurement of the lattice structures. (b) Magnified TEM image of site 1. (c) Magnified TEM
image of site 2. (d) Magnified TEM image of site 3. The lattice fringes of {110} planes (d ) 0.52 nm) and {300} planes (d ) 0.30 nm) can be seen.
Anisotropic Dissolution of β-TCP Crystal Growth Design, Vol. 8, No. 7, 2008 2229

and the hexagonal crystals became hollow. Again, it was
emphasized that the six edges and the wall structure remained
without any dissolution. Another interesting phenomenon was
that the β-TCP compounds in all six concave corners of the
hexagon were also not dissolved, implying that the demineral-
ization was somehow retarded at these sites. Actually, Figure
3d showed that the six corners were also against dissolution
reaction in the intermediate state. It could be understood by
using a thermodynamical model of the growth/dissolution on
different crystal substrates. Analogous to crystallization, the
energy barrier, ∆g*, of dissolving a crystal unit could be given
by eq 1,28c
∆g/
)
16πγSL
3
3∆gv
2
F(θ) (1)
where ∆gv was the change of free energy per unit volume before
and after dissolution, γSL, the nucleus-liquid interface energy,
Figure 3. HRTEM images of the β-TCP samples with dissolution period of 21 h. (a) Morphology and SAED pattern (along [001]) of a hexagonal
plate with partial dissolution. (b and c) The enlarged TEM image of the sites denoted by 1 and 2 in (a), respectively. The layered structure of the
bulk part was shown in (b). The detailed structure with defects of the bulk was detected on a remaining thin layer. (d) Partially dissolved hexagonal
plates; the dissolution period extended to 2 days in this case.
Figure 4. SEM image of the β-TCP samples with dissolution of 62 h in water. (a) Most materials were etched but the sites at the six vertexes of
the hexagon were still present against the dissolution. The gearlike morphology of β-TCP single crystal was formed. (b) The curves of F(θ) against
θ for the concave corner (green) and the ﬂat plane (blue).

and F(θ), a function of shape of crystal face and contact angle
of the unit and substrate. For the dissolution cases, F1(θ) on
the flat crystal face could be described by eq 2.
F1(θ) ) -
1
4
(2 - 3 cos θ + cos3
θ) (2)
At the concave corners (the angle was set as 120°), F2(θ) was
much more complicated as a description by Trivedi and
Sholl,30,31
F2(θ) ) -
1
4π{2sin2
θcosθcos-1
(√3
3
cotθ)+
2√3
3
cos2
θ
√sin2
θ -
1
3
cos2
θ - 4cosθcos-1
(√3
3
cotθ)+
cos-1
( 1
2sinθ)} (3)
therefore, a difference of the energy barrier at the concave
corner, ∆g2
/
to that on the flat surface, ∆g1
/
could be represented
by
∆g2
/
- ∆g1
/
)
16πγSL
3
3∆gv
2
{F2(θ) - F1(θ)} (4)
and a curve of F(θ) vs θ was also illustrated in Figure 4b. It
was noted that F1(θ) was always less than F2(θ) within a range
of all contact angle zone. The curves implied that, under the
same experimental condition, the dissolution barrier at the
concave corner was always greater than that in the bulk or on
the edge. Besides the wall itself, the sites around the hexagonal
corners of the wall were more difficult to be dissolved. Thus,
the formation of the gearlike structure could be understood.
Unlike the wall, which was really stable against the dissolu-
tion, the remained β-TCP at the corner sites could be dissolved
eventually with the reaction time. At the end of dissolution (12
days), the hexagonal dentations almost disappeared and only
the six edges survived, forming the hexagonal ring (Figures 1c
and 5). Most of the rings could keep their hexagonal structures
without any deformation. No obvious dissolution was detected
even that the resulting rings were redispersed in pure water.
The sizes of the hollow rings were 750-800 nm, and the heights
were 200-250 nm (Figure S3, Supporting Information), which
were in good agreement with the dimensions of the original
solid hexagons of β-TCP. The chemical composition and phase
of the remaining rings were also checked by using XRD (Figure
S4) and SAED (Figure 5b). The results confirmed that the
remaining walls were still pure β-TCP and there was no
detectable phase transformation during the reaction. Thus, it was
surprising that the wall and the bulk have different dissolution
properties even though they are identical in the crystal.
By increasing the undersaturation level in the demineralization
solution, similar dissolution results could be observed (Figure
S5, Supporting Information). However, a promoted dissolution
rate of the bulk part was detected since the hollow hexagonal
rings could be obtained within only 5 days. This experiment
indicated that the anisotropic dissolution behaviors could not
be affected by the change of undersaturation.
Figure 5. HRTEM images of the hollow rings at the end of dissolution (12 days). (a) The remaining rings. (b) The SAED pattern of the rings along
the [001] zone axis, showing that the plate had a top/bottom (001) surface and outer (100) surface. (c) The detailed structure of the boundary of
wall and bulk (white circle in a). (d) The lattice fringes of the edge wall (dark circles in a).

In order to reveal the structural difference of the different
parts in the β-TCP single crystal, the solid hexagonal plates,
hollow rings, and their intermediate states were studied by
HRTEM. The lattice parameters at the different sites on the
top/bottom surfaces were examined. However, they had the same
crystallographic structure and orientation as shown in Figure
2b. The interplanar distance (d-spacing), 0.52 nm, was attributed
to the (110) face of β-TCP. Together with the SAED pattern,
the orientation of the single crystal could be confirmed. The
d-spacings of two different edges of the hexagon (sites 2 and 3
in Figure 2a) clearly showed that all six side faces of the walls
were assigned to the {100} crystal face group. Since β-TCP
has the space group of R3jc, the marked faces, (100) and (11j0),
were actually equivalent. Besides, the planes of (21j0) and (1j20)
belonged to the {110} group too, and (300) was identical to
(33j0). The typical included angles of the hexagonal structure,
120°, could be obtained by using these lattice directions (Figures
2c and 2d). It could be found that the two neighboring edges
shared an integral and continuous lattice structure as their lattice
fringes could match with each other well. The study of the other
sides reached the same conclusion. Thus, the whole hexagonal
wall was constructed by six equivalent {100} thin crystal planes
of β-TCP, and it could be treated as a complete hollowed
hexagonal single crystal. This suggested model was also
confirmed by the SAED result of the rings (Figure 5b). The
6-fold-symmetry of the diffraction patterns of the wall showed
a typical pattern of the hexagonal single crystal of β-TCP.
The lattice structure of the bulk part (Figure 2b) coincided
in that of the edges (Figures 2c and 2d) too. Both the wall and
the bulk part shared the identical crystallographic structure and
orientation in a hexagonal plate, e.g. the in situ measured (110)
faces in the bulk part (Figures 2b) and that in the wall (Figure
2d) were exactly the same, which agreed with the features for
a single crystal of β-TCP. This conclusion was also confirmed
by the HRTEM image recorded from the inner edge (Figure
5c). The coexistence of wall (dark area) and the remaining part
(light area) provided an opportunity to study their interface in
detail. Although the boundary of the wall and the bulk was
obvious, their lattice structure (d-spacing) could be attributed
to (110) and (12j0) in one single crystal, respectively. The lattice
structures of the wall and bulk under HRTEM clearly showed
that the complex of them was an integral single crystal. In some
cases, the distinct dissolution behaviors were due to the different
crystallographic orientation of the crystals. However, this
explanation could not be applied in the present case of β-TCP
dissolution as the wall and bulk part had the same crystal-
lographic structure.
It had been mentioned that the internal texture of the bulk
part was not uniform, which implied that the bulk part was not
perfect. During the dissolution, the detailed structure of the
center part could be studied by their remaining thin layer. A
high density of defects of the bulk was demonstrated in the
lattice fringe image of these thin layers (Figure 3c). The
dislocation lines and the lattice-disordered regions were marked
by the lines and the arrows. In some domains, there was no
lattice fringe and it was an indication of the uncontinuous crystal
structures. However, such defects were rarely detected in the
wall structure. Figure 5c showed the consistence of the wall
structure and some remaining bulk fragments. The domains with
the discontinued lattice structure were separated by the dotted
lines. All the marked lines were in the bulk part (light region).
In contrast, the lattice structure of the wall (dark region) was
almost perfect. Moreover, the continuous and complete lattice
fringes at the other sites of the wall were demonstrated clearly
in Figure 5d, which confirmed the perfection of the wall
structure.
In order to observe the overall dislocation distributions in
the whole hexagonal plates, the dark field TEM images along
the [001] and the [100] zone axes were recorded (Figure 6),
and the diffracted beam of (110) indicated by the white arrows
in SAED patterns was used for the imaging. A perfect single
crystal should be shown by a uniformly bright image due to its
consistent lattice structure. However, dark lines or dark regions
appeared if the crystal contained dislocations for the bending
of lattice planes in the strain field, which caused the local
changes in the Bragg conditions. It was noted that such
dislocations were frequently observed in the bulk part and on
the border between the bulk part and wall (indicated by the
arrows). The distribution of these dislocations was also random
in the bulk part. This feature could explain why the dissolution
process initiated randomly on the face of bulk (Figure 1b). The
relative uniformity in brightness in the wall structure suggested
the low density of the dislocations. It was also interesting to
find that the width of this bright region, 30-40 nm, was similar
to the thickness of the resulting rings after the demineralization.
The difference in the crystallinities of the wall and the bulk
part might be caused by the fast formation of hexagonal β-TCP
during the preparation. The nuclei of the hexagonal plates were
formed within only two minutes (Figure S6, Supporting
Information). During such a rapid process, the internal structure
Figure 6. Dark-field TEM images of the β-TCP plate along the different zone axes: (a) side view, (b) top view. The insets show their corresponding
diffraction pattern, “o” indicates the transmitted beam, and the white arrows indicate the diffracted beam of the (110) face, which was used for the
dark-field imaging.

of the plate could not be well organized and the defects resulted.
However, as the outer surfaces contacted with the reaction
medium, the precipitated ions on the surfaces had the op-
portunity to exchange with the reaction solution at the solid-liquid
interfaces. The lattice structure could be reorganized during an
aging period so that the crystallinity of the wall could be
improved. Figure S6 shows that the smooth edges of the plates
evolved within five minutes. However, this reorganization effect
only occurred at the interface and it could not penetrate into
the bulk. Thus, the formed defects were proposed to be
“kinetically trapped” within the bulk part. This rapid growth
induced defect formation phenomenon had been previously
observed in other crystal system such as KDP.32,33
The dark
field TEM image recorded by the diffraction of (110) faces in
Figure 6b indicated that the six side surfaces were different from
the central part. The brightness of the side surfaces was much
stronger and more uniform than the top/bottom surfaces,
indicating the well-crystallized structure of the edge wall. The
curves and holelike lines in the bulk part demonstrated the
distortions of crystal faces, which were caused by the existed
dislocations and defects. The difference in face flatness between
side faces and top/bottom faces was also confirmed in the bright
field TEM image of side view of the hexagonal plates (Figures
S2 and S3). That the side faces had different crystallinities from
the top/bottom faces could be understood by the intrinsic
structural features of the β-TCP.34
Only three calcium ions were
distributed in the different ways over the six sites lining from
bottom to top along the [001] direction. The incomplete
distributions of calcium ions over these sites could inevitably
generate calcium vacancies, which led to the local residual
charge or the dangling bond along the [001] direction. The top/
bottom (001) facets were the polar ones of β-TCP. The surface
energies calculation also confirmed that the surface stability of
the {100} side faces was greater than the {001}.35
The polar
surface (001) was usually considered as an energetically
unfavorable one in the solution where the dislocations were more
readily generated on it than on the six equivalent nonpolar
surfaces {100}. A similar effect was also observed in the case
of ZnO dissolution.6
Furthermore, the strain field of these
dislocations in the bulk could induce the formation of etch pits
much more readily than the defect-free wall.29
These differences
of dislocation distribution between the bulk part and the wall,
the side faces and the top/bottom faces, might result in the
anisotropic dissolutions in one single crystal.
Besides, the size effect was the most important factor for the
abnormal stability of the wall. It had been suggested, and
confirmed by experiment, that demineralization of sparingly
soluble salts such as calcium phosphate was generally initiated
and accompanied by the formation and development of pits on
the crystal surfaces and that the dissolution rates were also
determined by the pit densities and spreading velocities.28
However, only the large pits (greater than a critical size) could
provide the active dissolution sites, contributing to the reaction.
The anisotropic behavior of the hexagonal β-TCP dissolution
had already been described. It implied that the dissolution along
[001] was initiated by the large pits on the top/bottom surface
of the plate, or the (001) crystal facet as shown by SEM (Figure
1), TEM (Figures 2 and S2) and AFM surface height profiles
(Figure 7). The wall had a relatively defect-free structure, and
the initiation of dissolution was more difficult than that of the
bulk. In order to dissolve the wall, the active pits on the (001)
narrow surface of the wall were required. The dimension (width)
of this facet was less than 40 nm. However, the critical size for
the active pit for β-TCP dissolution was of tens of nanometers.27a,28c
Thus, the active pit was extremely difficult to be produced on the
limited dimensions. As the nanodissolution model proposed,28a
the
thin edge wall could be dynamically self-presevered by the size
effect.
A similar size effect was also found in biodemineralizaiton
of tooth enamel.28a,b
However, they were not single crystals
but polycrystallites. The identical chemical and crystal properties
of apatite in cores and on walls were observed in the rods.
Analogous to the present work, the demineralization of the
enamel cores around the rod c-axis was privileged as the core
was always emptied while the wall remained. However,
the dissolution inhibition of the wall of the enamel rod may be
explained by the presence of some organic residuals in the
frame. In order to examine the possible effect of the remaining
organic solvent on the abnormal dissolution, the hexagonal plates
were calcinated at 500 °C for 2 h to remove the organic
compounds. TEM characterizations showed that the size,
morphology, and the structure of the plates were almost
unaffected after the calcination. Furthermore, they underwent
the same demineralization to form the hexagonal rings (Figure
S7, Supporting Information) eventually. Therefore, the interest-
Figure 7. AFM height image of hexagonal plates. (a) AFM image shows the smooth edge (wall surface) of a single hexagonal plate. (b) The rough
top (001) surface of the bulk contains many domains in the size of 20-60 nm.

ing dissolution behavior of these hexagonal β-TCP had no direct
relationship with the involvement of organic additives, which
should be eliminated by calcination. However, it could be
contributed to the unique structural complex of the single crystal
as indicated by HRTEM and dark-field TEM images. Actually,
the size effect of bulk β-TCP particles had already been revealed
in our previous constant composition dissolution study.27a
Based on the collected structural information, a scheme of
β-TCP nanoplate was suggested as Scheme 1: the two parts,
the wall and the bulk part (displayed by blue and green,
respectively), had different dissolution features despite their
being integrated in one single crystal. The dark circles repre-
sented the defects in the bulk. The schematic structure was also
supported by the surface morphology information, obtained by
AFM (Figure 7). The thin wall had a relatively smooth facet;
on the surfaces of the bulk, many tiny domains in the size of
20-60 nm were separated by the block boundaries, irregularly
shaped holes, which represented a higher density of the defects.
Conclusion
By using ethylene glycol as the solvent, we have succeeded
in the synthesis of a uniform hexagonal submicron single crystal
of β-TCP phase at relatively low temperature. However, this
single crystal has a complex structure, a well-crystallized wall
and a poorly crystallized bulk part. These two components have
different physicochemical properties, resulting in anisotropic
dissolution behaviors. This abnormal but interesting feature can
be used to produce various structures, porous, gearlike, and
hexagonal rings of β-TCP single crystals by controlled dem-
ineralization reaction. The technique presented here might be
regarded as an effective and feasible approach to synthesize
complicated structures of functional materials without the
involvement of template and complicated operations.
Acknowledgment. We thank Profs. Jianguo Hu, Ying Chen
(Fudan University) and Dr. Yaowu Zeng for their help in
HRTEM and Drs. Youwen Wang and Jieru Wang for their help
in TEM and SEM. This work is supported by National Natural
Science Foundation of China (20571064 and 20601023) and
Changjiang Scholar Program (RT).
Supporting Information Available: Supporting figures: conductiv-
ity measurement of CaCl2 and Na2HPO4 in water and in ethylene glycol
(Figure S1), the side views of the solid (Figure S2) and hollow (Figure
S3) β-TCP single crystals, XRD of the hollow hexagonal crystals
(Figure S4), dissolution of β-TCP at a higher undersaturation level
(Figure S5), fast formation of hexagonal single crystals (Figure S6),
and hexagonal plates calcinated at 500 °C and their dissolution results
(Figure S7). This material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Aizpurua, J.; Hanarp, P.; Sutherland, D. S.; Ka¨ll, M.; Bryant, G. W.;
Garcıá de Abajo, F. J. Phys. ReV. Lett. 2003, 90, 57401.
(2) Li, F.; Xu, L.; Zhou, W. L.; He, J.; Baughman, R. H.; Zakhidov, A. A.;
Wiley, J. B. AdV. Mater. 2002, 14, 1528.
(3) Li, F.; He, J.; Zhou, W. L.; Wiley, J. B. J. Am. Chem. Soc. 2003, 125,
16166.
(4) Sano, M.; Kamino, A.; Okamura, J.; Shinkai, S. Science 2001, 293,
1299.
(5) Kong, X. Y.; Ding, Y.; Yang, R.; Wang, Z. L. Science 2004, 303,
1348.
(6) Li, F.; Ding, Y.; Gao, P.; Xin, X.; Wang, Z. L. Angew. Chem., Int.
Ed. 2004, 43, 5238.
(7) Caruso, F.; Caruso, R. A.; Mo¨hwald, H. Science 1998, 282, 1111.
(8) So¨llner, C.; Burghammer, M.; Busch-Nentwich, E.; Berger, J.;
Schwarz, H.; Riekel, C.; Nicolson, T. Science 2003, 302, 282.
(9) Sańchez-Romań, M.; Rivadeneyra, M. A.; Vasconcelos, C.; McKenzie,
J. A. FEMS Microbiol. Ecol. 2007, 61, 273.
(10) Bazylinski, D. A.; Frankel, R. B. ReV. Mineral. Geochem. 2003, 54,
217.
(11) Frankel, R. B.; Bazylinski, D. A. ReV. Mineral. Geochem. 2003, 54,
95.
(12) Estroff, L. A.; Hamilton, A. D. Chem. Mater. 2001, 13, 3227.
(13) Dujardin, E.; Mann, S. AdV. Mater. 2002, 14, 775.
(14) Wucher, B.; Yue, W.; Kulak, A. N.; Meldrum, F. C. Chem. Mater.
2007, 19, 1111.
(15) Meldrum, F. C.; Ludwigs, S. Macromol. Biosci. 2007, 7, 152.
(16) Mann, S. Angew. Chem., Int. Ed. 2000, 39, 3392.
(17) Yin, Y.; Alivisatos, A. P. Nature 2005, 437, 664.
(18) Wang, X.; Zhuang, J.; Peng, Q.; Li, Y. Nature 2005, 437, 121.
(19) Wiley, B.; Herricks, T.; Sun, Y.; Xia, Y. Nano Lett. 2004, 4, 1733.
(20) Hobbs, K. L.; Larson, P. R.; Lian, G. D.; Keay, J. C.; Johnson, M. B.
Nano Lett. 2004, 4, 167.
(21) Zhu, F. Q.; Fan, D. L.; Zhu, X. C.; Zhu, J. G.; Cammarata, R. C.;
Chien, C. L. AdV. Mater. 2004, 16, 2155.
(22) Yan, F.; Goedel, W. A. Nano Lett. 2004, 4, 1193.
(23) Xu, H.; Goedel, W. A. Angew. Chem., Int. Ed. 2003, 42, 4696.
(24) Zhao, S.; Roberge, H.; Yelon, A.; Veres, T. J. Am. Chem. Soc. 2006,
128, 12352.
(25) (a) Liu, B.; Zeng, H. C. J. Am. Chem. Soc. 2005, 127, 18262. (b)
Zhou, W. L.; He, J.; Fang, J.; Huynh, T. A.; Kennedy, T. J.; Stokes,
K. L.; O’Connor, C. J. J. Appl. Phys. 2003, 93, 7340.
(26) Dorozhkin, S. V.; Epple, M. Angew. Chem., Int. Ed. 2002, 41, 3130.
(27) (a) Tang, R.; Wu, W.; Hass, M.; Nancollas, G. H. Langmuir 2001,
17, 3480. (b) Pan, Y.; Huang, J.; Shao, C. Y. J. Mater. Sci. 2003, 38,
1049. (c) Engin, N. O¨ .; Tas, A. C. J. Am. Ceram. Soc. 2000, 83, 1581.
(28) (a) Tang, R.; Wang, L.; Orme, C. A.; Bonstein, T.; Bush, P. J.;
Nancollas, G. H. Angew. Chem., Int. Ed. 2004, 43, 2697. (b) Wang,
L.; Tang, R.; Bonstein, T.; Orme, C. A.; Bush, P. J.; Nancollas, G. H.
J. Phys. Chem. B 2005, 109, 999. (c) Tang, R.; Nancollas, G. H.;
Orme, C. A. J. Am. Chem. Soc. 2001, 123, 5437. (d) Dove, P. M.;
Han, N.; De Yoreo, J. J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102,
15357. (e) Wang, L. J.; Tang, R.; Bonstein, T.; Bush, P.; Nancollas,
G. H. J. Dent. Res. 2006, 85, 359. (f) Wang, L. J.; Nancollas, G. H.;
Henneman, Z. J.; Klein, E.; Weiner, S. Biointerphases 2006, 1, 106.
(29) Lasaga, A. C.; Luttge, A. Science 2001, 291, 2400.
(30) Trivedi, R. Scr. Mater. 2005, 53, 47.
(31) Sholl, C. A.; Fletcher, H. Acta Metall. 1970, 18, 1083.
(32) Zaitseva, N.; Carman, L.; Smolsky, I. J. Cryst. Growth 2002, 241,
363.
(33) Demos, S. G.; Staggs, M.; Radousky, H. B. Phys. ReV. B 2003, 67,
224102.
(34) Yin, X.; Stott, M. J.; Rubio, A. Phys. ReV. B 2003, 68, 205205.
(35) Yin, X.; Stott, M. J. J. Chem. Phys. 2006, 124, 124701.
CG700808H
Scheme 1. Schematic Representation of a Single Hexagonal
Plate of β-TCPa
a
The thin edge wall (blue) had well-crystallized structure; but the bulk
part (green) contained lots of defects. The blue part could be stabilized by
size effect against dissolution, and the green part could be dissolved readily
in water.

RESEARCH ARTICLE
The luminescent enhancement of LaPO4:Ce3+
,Tb3+
nano
phosphors by radial aggregation
Xin JI1
, Fei-Jian ZHU2
, Ha-Lei ZHAI1
, Rui-Kang TANG (✉)1
1 Department of Chemistry, Zhejiang University, Hangzhou 310027, China
2 Research and Development Department, Hangzhou Daming Fluorescent Materials Co. Ltd., Hangzhou 311200, China
© Higher Education Press and Springer-Verlag Berlin Heidelberg 2010
Abstract The rare earth nano phosphors can meet the
challenging demand for new functional devices but their
luminescence is always poor. Here we report on a simple
method to prepare uniform LaPO4:Ce3+
,Tb3+
sphere-like
nano aggregates from the precipitated nano phosphor
crystallites without using any additive. The spontaneous
aggregation is induced and controlled only by the
suspension pH conditions. It is found that the 100 nm
spherical aggregates can significantly improve the green
emissions of the LaPO4:Ce3+
,Tb3+
nano particles. The
intensity of the aggregates can be about 10 times as that of
the 80 nm-sized individual ones. This study may provide a
useful yet convenient strategy in the improvement and
application of nano phosphors.
Keywords nano phosphor, lanthanide phosphate, aggre-
gation, luminescence, enhancement
1 Introduction
Nowadays the development and performance of the new
generation of energy saving lighting, flat displays with
liquid crystal display, and biologic marker and signal have
been credited a lot to luminescent properties of rare earth
materials [1–4]. Nanometric materials have attracted great
interests because they may serve as active components in
new functional devices. However, synthesized nano
phosphors always have extremely lower luminescent
efficiency than the corresponding bulk materials [5].
Many approaches have been tried to improve emission of
nano phosphors. Among these studies, a significant
research interest is toward the control of nano particle
size, morphology and aggregate by using various organic
templates [6,7]. It is noted that the template assembly of
nano particles into spatially well-defined architectures can
offer new properties to the functional materials, which
seem distinctly different from the isolated ones [8,9].
Structural characteristics of these assembled nano particles
like lanthanide(Ln)-doped materials endow them with a
wide range of potential applications, such as for phosphors,
optical amplifiers, biochemical probes, and medical
diagnostics [10,11]. Unfortunately, the template-directed
method needs some special instruments and harsh condi-
tions, and usually leads to impurities due to the incomplete
removal of the templates [12]. Here we describe a facile
precipitation approach for the preparation of LnPO4
sphere-like nanostructures by isolated nano-particle aggre-
gation in the absence of any template agent. And the
aggregation can greatly enhance the luminescence of the
nano phosphors.
LnPO4 (Ln = La, Ce, and Tb) is an excellent light-
emitting phosphor, which has been extensively used in
luminescent lighting industry. The green luminescence of
the terbium ions is observed after a UV excitation of the
cerium ions at the optimum wavelength of 272 nm. The
excitation can further migrate from cerium to cerium until
it reaches a terbium luminescent center. Although the
quantum yields of nano-particles are always lower than the
corresponding bulk materials, nano materials may increase
the luminescence of the 5
D4 – 7
F5 of Tb3+
via energy
transfer of Ce3+
! Tb3+
due to hindrance of boundary and
size [5]. So the preparation and characteristics of
LaPO4:Ce3+
,Tb3+
nano-particles are of great importance,
and one of the most effective strategies to improve
emission is to construct architectures. However, the
presence of organic additives such as surfactants in the
templated nano-assembly often results in an increased
luminescence quenching [12,13], which leads to a negative
effect on theluminescence. Therefore, it is a challenge to
obtain the nano-architecture without any additive.
Received October 21, 2010; accepted November 2, 2010
E-mail: rtang@zju.edu.cn
Front. Mater. Sci. China 2010, 4(4): 382–386
DOI 10.1007/s11706-010-0115-z

2 Materials and methods
2.1 Materials
La2O3 (99.99%) and Tb4O7 (99.99%) were supplied by
Hangzhou Daming Fluorescent Materials Co. Ltd (China).
(NH4)2HPO4, NaOH, HNO3 (65%) and Ce(NO3)3$6H2O
were of analytical grade. All the chemicals were used
without any further purification. Double-distilled water
was used in the experiments.
2.2 Methods
LaPO4:Ce3+
,Tb3+
nano-crystals were prepared by a
solution precipitation method. Briefly, 4.07 g La2O3 and
4.68 g Tb4O7 were dissolved in 50.0 mL 2.5 mol/L HNO3,
respectively. 10.86 g Ce(NO3)3$6H2O was dissolved with
50.0 mL water. Certain amounts of these three solutions
were mixed together till the finial ratio of Ln∶Ce∶Tb =
0.59∶0.22∶0.19, which was defined as the Ln solution.
(Total concentration of Ln was 0.5 mol/L). 20.0 mL of the
Ln solution was slowly dropped into 0.15 mol/L
(NH4)2HPO4 aqueous solution. The pH of (NH4)2HPO4
solution was adjusted to 9.0 and 7.0 respectively prior to
the use. The ratio of Ln:PO4 in the reaction solution was
around 1∶1.1 and the solution pH was adjusted to a
required value by using 4.0 mol/L HNO3 or 4.0 mol/L
NaOH. The resulting lanthanum phosphate colloidal
suspension was aged for 2 d and was separated with
centrifugation. The obtained solids were washed with
water at least three times and then were dried under
vacuum condition at 35°C. To test the effect of crystallinity
of solid phase on luminescence, different experimental
temperature (5°C–45°C) was applied in the synthesis to
obtain the LaPO4:Ce3+
,Tb3+
particle/aggregates with
different crystallinity.
2.3 Characterization
The solid structure, morphology and size were examined
with X-ray diffraction (XRD) using Rigaku D/max-rA
(Japan) diffractometer with mono-chromatized Cu KR
radiation, transmission electron micrograph (TEM,
JEM200CX, JEOL, Japan) and scanning electron micro-
graph (SEM, SIRION, FEI, Holland). The fluorescent
emission spectra were recorded with RF-5301pc spectro-
fluorometer (Shimadzu, Japan). Luminescence intensities
were measured and compared at room temperature using
two parallel windows with a solid luminescence spectrum
analysis (SPM-3, Sanming, China) in which the commer-
cial LnPO4 was used as the standard so that the relative
brightness values of samples were measured directly. Zeta
potentiometer characterization was performed by
ZEN3600 (Malvern, UK).
3.1 Nano-particles and nano-aggregates
Figure 1(a) shows the isolated nano-particles LaPO4:Ce3+
,
Tb3+
from the suspension at pH = 2, which were needle-
like with an average length of about 80 nm. It could be
noted that there was no aggregation structure under such a
condition. However, the well-controlled LaPO4:Ce3+
,Tb3+
sphere-like aggregates of the nano-needles could be
Fig. 1 TEM of the resulted LaPO4:Ce3+
,Tb3+
products in the
suspension of (a) pH = 2 and (b) pH = 6; SEM of the sphere-like
aggregates synthesized at (c) pH = 6
Xin JI et al. The luminescent enhancement of LaPO4:Ce3+
,Tb3+
nano phosphors by radial aggregation 383

formed spontaneously in the suspension at pH = 6
(Figs. 1(b) and 1(c)). The nano-spheres had the uniform
morphology and size distribution; their diameters were
about 100 nm. The basic building units, the needle-like
crystallites, could be identified clearly under TEM and
SEM (Figs. 1(b) and 1(c)). The similar sphere-aggregates
could be obtained in the suspensions within the pH range
of 5.5–6.5 and no significant difference of these aggregates
was detected. These phenomena implied that solution pH
might play a key role in the spontaneous aggregations.
The XRD patterns (Fig. 2) of the isolated solids (nano-
aggregates and nano-particles) could be indexed to the
rhabdophane-type structure of lanthanum phosphate and
all the peaks were assigned by using Joint Committee on
Powder Diffraction Standards (JCPDS, No.04-0635). The
results implied that the aggregation did not alter the lattice
structure of LaPO4:Ce3+
,Tb3+
. To our expectation, the
product crystallinity was sensitive to the reaction tempera-
ture. The increase of reaction temperature resulted in the
improvement of product crystallinities. For example, the
diffraction peaks of the products synthesized at tempera-
ture of 35°C were significantly sharper than those of the
samples prepared at 5°C, indicating a higher crystallinity.
However, at the same reaction temperature, the individual
nano-particles synthesized under lower pH conditions
always had a greater crystallinity than the nano-aggregates
prepared under higher pH conditions, implying the
influence of solution pH on the crystallinity.
3.2 Zeta-potential
Zeta-potential is a key factor in the studies of particle
aggregation in solutions, which is sensitive to the solution
conditions such as pH and ionic strength. In our study, the
ionic strength in the reaction solution was relatively
constant though the conditions might differ. And we noted
that the spontaneous aggregation only occurred within the
pH range of 5.5–6.5. Thus, we examined the zeta-potential
of the isolated LaPO4:Ce3+
,Tb3+
nano needles in the water
under different pH conditions (Fig. 3). The measured
potential of the particle was about 41 mVat pH of 2.0. The
particle surface charge value decreased slightly with the
increase of pH within the pH of 2.0–5.0. At the point of
pH = 5, zeta-potential of the nano-particles was about
28 mV, but then the value suddenly dropped, which was
only – 3 mV at pH = 6. However, around pH = 7 the zeta-
potential of LaPO4:Ce3+
,Tb3+
reached the lowest value,
about – 38 mV and the value began to increase slightly
with the further increase of solution pH. The repulsive
force of particles in solution is proportional to zeta-
potential due to the electrostatic interaction. It is widely
accepted that the aggregation of particles can be effectively
dispersed when the absolute value of their zeta-potential
was greater than 30 mV. The strong electrostatic repulsive
forces between particles can prevent them from
aggregating [14]. We noted that under our aggregation
experiment conditions with the pH of 5.5–6.5, the zeta-
potential values located within the range from + 15 to
– 20 mV, providing a preferred experimental condition
for the particle aggregation.
3.3 Luminescence
The nano-crystal aggregations of LaPO4:Ce3+
,Tb3+
led to
the remarkable luminescence enhancement of the nano
phosphor. As the common limitation of nano phosphor, the
needle-shaped nano LaPO4:Ce3+
,Tb3+
exhibited a little
visible luminescence under UV excitation (l = 254 nm,
Fig. 4, left). However, under the same UV excitation, the
nano-aggregates (prepared at pH = 6) emitted much more
green lights (Fig. 4, right). A quantitative measurement by
using the solid luminescence spectrum analysis showed
that the lighting intensity of LaPO4:Ce3+
,Tb3+
aggregates
was almost 10 times greater than the corresponding
isolated nano-particles. Figure 5 shows the emission
Fig. 2 XRD patterns of LaPO4:Ce3+
,Tb3+
aggregates prepared
for 2 d at (a) 5°C, (b) 15°C, (c) 25°C, (d) 35°C, (e) 45°C, and of
(f) individual nano needles prepared at 35°C
Fig. 3 Zeta-potentials of LaPO4:Ce3+
,Tb3+
particles at different
solution pH values
384 Front. Mater. Sci. China 2010, 4(4): 382–386

spectrum of the nano-aggregates under excitation of l =
272 nm, which was exactly the same as that of the standard
LaPO4:Ce3+
,Tb3+
nano phosphors. This result demon-
strated that the aggregation actually did not alter the
luminescent properties of the nano materials. The typical
emission peaks of terbium were observed around 486, 547,
587 and 619 nm assigned to the transitions of 5
D4 – 7
FJ (J =
6, 5, 4 and 3) respectively [15]. Ce3+
ions had a relatively
broad absorption band from 200 to 300 nm with an allowed
4f–5d transition, and transfered their energy to the doped
Tb3+
ions, emitting the green light [16–18].
The previous study of bulk phosphors suggested that the
luminescence is highly dependent upon the crystallinity of
materials [19,20], which is another important pathway to
improve the luminescence. In the current study, the
crystallinity of the nano LnPO4 was increased by using
high reaction temperature (Fig. 2). However, the nano-
aggregates with improved crystallinity could not enhance
the emission intensity signiﬁcantly (Fig. 6). Although the
nano needles were even featured by the highest
crystallinity among the samples, their luminescence was
still weak. It should be noted that the crystallinity of these
nano needles were even better than most aggregates.
Therefore, it could be concluded that the aggregation
played the most important role in the luminescent
enhancement rather than the crystallinity in our case. We
supposed that the effect of crystallinity on the lumine-
scence improvement could be ignored in the LaPO4:Ce3+
,
Tb3+
nano phosphors. Actually, the nano-sized materials
limit the number of primitive cells per particle and
therefore, there are only a few traps in the nano-particles.
The energy of a luminescence center can only be
transferred resonantly within one particle since the energy
transfer is hindered by the particle boundary [4]. So
quenching occurs at high concentration in the isolated
Fig. 4 LaPO4:Ce3+
,Tb3+
powders under UV excitation (l = 254 nm): isolated needle-liked particles (left) and sphere-like aggregates
(right)
Fig. 5 Emission spectrum of LaPO4:Ce3+
,Tb3+
under UV
excitation of l = 272 nm Fig. 6 Relative luminescent intensities of different aggregates of
LaPO4:Ce3+
,Tb3+
nano-phases prepared at 5°C, 15°C, 25°C,
35°C, 45°C (top), and of individual nano phosphors prepared at
35°C (below). The luminescence intensity of a commercial bulk
material (provided by Hangzhou Daming Fluorescent Materials
Co. Ltd.) was used as the standard sample and its relative
luminescent intensity was deﬁned as 100.
Xin JI et al. The luminescent enhancement of LaPO4:Ce3+
,Tb3+
nano phosphors by radial aggregation 385

nano-sized particles, which is the main reason for the poor
luminescent characteristics for the nano phosphors.
Although the crystallinity is a key factor in the improve-
ment of bulk phosphor materials, its influences on the
nano-phase is very weak. However, such a negative effect
on luminescence by quenching may be effectively reduced
by the nano-particle aggregation even the aggregation is
very simple [6]. Thus, the new luminescence property is
conferred on the nano-aggregates. However, if an organic
template is used additionally to assist in such an
aggregation, the strong adsorption of the cross-linkers or
surfactants may also be assistant in the unexpected
quenching process. However, this negative influence can
be avoided by using a strategy of additive-free aggregation,
which is demonstrated by our current study.
4 Conclusions
We suggest a simple approach for the preparation of
uniform LnPO4 nanostructures without any assistance of
organic additive in this article. The luminescent intensity
of the spontaneously formed spherical aggregates can be
almost 10 times greater than the corresponding individual
nano-particles. And we also reveal that in the case of nano
system of LaPO4:Ce3+
,Tb3+
, the effect of aggregation may
play much more important role in the luminescent
enhancement rather than the particle crystallinity. These
findings may provide a useful strategy to improve the
synthesis and application of various nano phosphors.
Acknowledgements This work was supported by Daming Biomineraliza-
tion Foundation and the Fundamental Research Funds for the Central
Universities.
References
1. Giaume D, Buissette V, Lahlil K, et al. Emission properties and
applications of nanostructured luminescent oxide nanoparticles.
Progress in Solid State Chemistry, 2005, 33(2–4): 99–106
2. Bunzli J C G, Comby S, Chauvin A S, et al. New opportunities for
lanthanide luminescence. Journal of Rare Earths, 2007, 25(3): 257–
274
3. Jüstel T, Nikol H, Ronda C. New developments in the field of
luminescent materials for lighting and displays. Angewandte
Chemie International Edition, 1998, 37(22): 3085–3103
4. Hu H, Zhang W. Synthesis and properties of transition metals and
rare-earth metals doped ZnS nanoparticles. Optical Materials, 2006,
28(5): 536–550
5. Yu L X, Song H W, Liu Z X, et al. Remarkable improvement of
brightness for the green emissions in Ce3+
and Tb3+
co-activated
LaPO4 nanowires. Solid State Communications, 2005, 134(11):
753–757
6. Yang M, You H, Song Y, et al. Synthesis and luminescence
properties of sheaflike TbPO4 hierarchical architectures with
different phase structures. The Journal of Physical Chemistry C,
2009, 113(47): 20173–20177
7. Li L, Jiang W G, Pan H H, et al. Improved luminescence of
lanthanide(III)-doped nanophosphors by linear aggregation. Journal
of Physical Chemistry C, 2007, 111(11): 4111–4115
8. Horiuchi S, Nakao Y. Polymer/Metal Nanocomposites: Assembly
of metal nanoparticles in polymer films and their applications.
Current Nanoscience, 2007, 3(3): 206–214
9. Sun Y J, Lu Y, Yu Y, et al. Template-assemble synthesis of ZnO:Er
nanostructure and their upconversion luminescence properties.
Journal of Nanoscience and Nanotechnology, 2009, 9(2): 1316–
1320
10. Tissue B M. Synthesis and luminescence of lanthanide ions in
nanoscale insulating hosts. Chemistry of Materials, 1998, 10(10):
2837–2845
11. Arellano I, Nazarov M, Byeon C C, et al. Luminescence and
structural properties of Y(Ta,Nb)O4:Eu3+
,Tb3+
phosphors. Materi-
als Chemistry and Physics, 2010, 119(1–2): 48–51
12. Fang J, Saunders M, Guo Y, et al. Green light-emitting
LaPO4:Ce3+
:Tb3+
koosh nanoballs assembled by p-sulfonato-calix
[6]arene coated superparamagnetic Fe3O4. Chemical Communica-
tions, 2010, 46(18): 3074–3076
13. Wang L Y, Li P, Li Y D. Down- and up-conversion luminescent
nanorods. Advanced Materials, 2007, 19(20): 3304–3307
14. Rabinovich-Guilatt L, Couvreur P, Lambert G, et al. Extensive
surface studies help to analyse zeta potential data: the case of
cationic emulsions. Chemistry and Physics of Lipids, 2004, 131(1):
1–13
15. Wang L Y, Li Y D. Na(Y1.5 Na0.5)F6 single-crystal nanorods as
multicolor luminescent materials. Nano Letters, 2006, 6(8): 1645–
1649
16. Kojima Y, Doi S, Yasue T. Synthesis of cerium (III) and terbium
(III) codoped vaterite phosphor emitting by black light irradiation
and its fluorescence property. Journal of the Ceramic Society of
Japan, 2002, 110: 755–760
17. Sohn K S, Park D H, Cho S H, et al. Genetic algorithm-assisted
combinatorial search for a new green phosphor for use in tricolor
white LEDs. Journal of Combinatorial Chemistry, 2006, 8(1): 44–49
18. Ding S J, Zhang W, Xu B Q, et al. [Spectra of Ce3+
, Tb3+
and Gd3+
ions in Ln(BO3,PO4) [Ln = La, Y]]. Spectroscopy and Spectral
Analysis, 2001, 21(3): 275–278 (Ln = La, Y)
19. Mari B, Singh K C, Sahal M, et al. Preparation and luminescence
properties of Tb3+
doped ZrO2 and BaZrO3 phosphors. Journal of
Luminescence, 2010, 130(11): 2128–2132
20. Kim S W, Masui T, Matsushita H, et al. Enhancement in
photoluminescence of Gd2O2CO3:Tb3+
submicron particles by
introducing yttrium into the oxycarbonate lattice. Journal of the
Electrochemical Society, 2010, 157(5): 181–185
386 Front. Mater. Sci. China 2010, 4(4): 382–386

5. all published paper

More Related Content

What's hot

Viewers also liked

Similar to 5. all published paper

5. all published paper