2. Article Edri and Regev
aggregate; (2) a “stable” state, after sonication (and cen- dispersion mechanism of lysozyme and BSA are different: while
trifugation), where large bundles and most impurities are re- the lysozyme protein stabilization mechanism is considered to be
moved from the solution and the supernatant contains mainly mainly electrostatic,30 the BSA one involves also steric effects
individual SWNT or small bundles covered by dispersant and free resulting from its bulkier structure that can stabilize the dispersion
dispersant molecules. A stable dispersion can be further manipu- even at the IEP, when electrostatic repulsion is zeroed.24
lated to induce instability (a precipitation): for instance, salt has A kinetic study on the adsorption process of BSA on SWNT
been used to enrich a stable solution with SWNT of specific suggested that at pH values close to the IEP BSA adsorbs on the
chirality27 while pH has been employed to switch between stability SWNT wall more rapidly, and the final surface coverage is higher
and instability of pH-sensitive polymer-or proteins30-NT solu- than away from the IEP.40 It has been suggested that better
tion. However, it is not clear if these properties affect the surface coverage is an important parameter in CNT dispersion as
dispersion dynamics. For example, is pH change of an already it increases the energetic barrier to coagulation.41
dispersed solution equivalent to setting the pH of a predispersed To study the dynamics of the dispersion process, we make use
solution on the same value? of a technique reported by Grossiord et al.,25,42 who monitored
CNTs have been recently shown to perform as efficient drug SWNT dispersion process during sonication by UV-vis absor-
transporters to cell.31-34 With the intention of turning CNT bance at a specific wavelength values. Since metallic SWNTs in a
biocompatible, a dispersing agent of biological origin is usually bundle quench electron excitation in “neighboring” SWNTs in
used to eliminate rejection by the body.4 As one possible route to the bundle,43 the absorbance intensity is related to the SWNTs
achieve such prospect, proteins have recently become widely used exfoliated in solution (individual or small bundles). The same
as CNT-dispersing agents.15,35-37 At the same time, the properties concept was earlier used by Jiang and co-workers44 to monitor the
of most proteins are pH responsive, hence offering a leverage dispersion stability of a centrifuged solution over time. The
point for system manipulation. prevalence of individual over bundled SWNTs has also been
In this paper we study nanotubes dispersed by bovine serum studies by near-IR fluorescence.43,45 It was found by UV-vis
albumin (BSA), which is known to adopt pH-dependent con- measurements25,42 that the absorbance intensities at several
formations. At acidic pH values (pH < 4) the protein adopt an different wavelengths show the same trend and reach a plateau
elongated conformation (E form). At 4 < pH < 8 the protein has after rather long sonication time is applied; the plateau is
a bulky and packed conformation (“heart”-shaped, N form), and attributed to maximum exfoliation and dispersion of SWNT.
at basic pH values (pH > 8) a more “loose” yet still bulky Furthermore, a similar trend was reported for SWNT of different
conformation is adopted (B form).38,39 We recently found that sources or impurities.25,42 Nonetheless, the amount of actually
each conformation shows a different SWNT recovery.24 Thus, we stable SWNT in the solution and its relation to this “dynamic
ask whether the pH affects also the dispersion dynamics and could picture” has not been addressed yet.
it be used for control or as an optimizing parameter for protein- Recently, we applied a classical least-squares method of a full
assisted nanotube dispersion. In this study we make use of UV-vis spectrum chemometric technique to analyze SWNT and
UV-vis spectroscopy complemented by cryogenic transmission BSA concentrations of centrifuged solutions (stable state). We
electron microscopy (cryo-TEM) to investigate the interaction of demonstrated that this technique is efficient for cases where the
BSA and SWNT, focusing on the dispersion dynamics. constituents’ absorbance signals overlap.24
As mentioned above, BSA has three main NT recovery regions: In the current work we study the dispersion dynamics of the
acidic, near isoelectric, and basic.24 These are related to the BSA-SWNT system before and after centrifugation, which we
protein conformation: the less globular and dense the protein is, term “shaken” and “stable”, respectively. To monitor the
the less SWNT are dispersed. Also, within each conformation dynamics of this process, we combine a few techniques: full
range, with increasing electrostatic charge on the protein less spectrum chemometric (FSC) based on full spectrum UV-vis24
SWNT are recovered. In contrast, in another protein-SWNT to evaluate the SWNT recovery, single wavelength absorbance
system, that is, lysozyme, solutions are characterized by a single (SWA) to monitor the exfoliation process, and cryo-TEM to
transition point in solution stability;the isoelectric point;in visualize the dispersion process. We explore the effect of both pH-
which the SWNTs precipitate from the solution or are not induced protein conformation and the dispersant-to-nanotube
dispersed at all.15,30 This suggests that the stabilization and ratio on the dispersion dynamics and NT recovery.
(27) Niyogi, S.; Boukhalfa, S.; Chikkannanavar, S. B.; McDonald, T. J.; Heben, Experimental Section
M. J.; Doorn, S. K. J. Am. Chem. Soc. 2007, 129, 1898.
(28) Grunlan, J. C.; Liu, L.; Regev, O. J. Colloid Interface Sci. 2008, 317, 346–
Materials. HiPco SWNTs (Carbon Nanotechnologies Inc.,
349. >90% purity) were used when specified. Otherwise, as produced
(29) Park, C.; Lee, S.; Lee, J. H.; Lim, J.; Lee, S. C.; Park, M.; Lee, S. S.; Kim, J.; (AP)-SWNT (Carbolex, diameter ∼ 1.4 nm, catalyst Ni and Y,
Park, C. R.; Kim, C. Carbon 2007, 45, 2072–2078. purity 70 vol %) prepared by arc-discharge method were used.
(30) Nepal, D.; Geckeler, K. E. Small 2006, 2, 406–412.
(31) Dhar, S.; Liu, Z.; Thomale, J.; Dai, H.; Lippard, S. J. J. Am. Chem. Soc.
Bovine serum albumin powder was purchased from Sigma-Al-
2008, 130. drich (A3803 and A7906, Cohn Fraction V, 98%, 1% (w/v) in
(32) Kam, N. W. S.; Liu, Z. A.; Dai, H. J. Angew. Chem., Int. Ed. 2006, 45, 577– H2O) and cetyltrimethylammonium bromide (CTAB) from
581.
(33) Liu, Z.; Chen, K.; Davis, C.; Sherlock, S.; Cao, Q. Z.; Chen, X. Y.; Dai, H.
J. Cancer Res. 2008, 68, 6652–6660. (40) Valenti, L. E.; Fiorito, P. A.; Garcia, C. D.; Giacomelli, C. E. J. Colloid
(34) Liu, Z.; Sun, X. M.; Nakayama-Ratchford, N.; Dai, H. J. ACS Nano 2007, Interface Sci. 2007, 307, 349–356.
1, 50–56. (41) Shvartzman-Cohen, R.; Nativ-Roth, E.; Baskaran, E.; Levi-Kalisman, Y.;
(35) Goldberg-Oppenheimer, P.; Regev, O. Small 2007, 3, 1894–1899. Szleifer, I.; Yerushalmi-Rozen, R. J. Am. Chem. Soc. 2004, 126, 14850–14857.
(36) Dutta, D.; Sundaram, S. K.; Teeguarden, J. G.; Riley, B. J.; Fifield, L. S.; (42) Grossiord, N.; Loos, J.; Regev, O.; Koning, C. E. Chem. Mater. 2006, 18,
Jacobs, J. M.; Addleman, S. R.; Kaysen, G. A.; Moudgil, B. M.; Weber, T. J. 1089–1099.
Toxicol. Sci. 2007, 100, 303–315. (43) O’Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.; Strano, M.
(37) Matsuura, K.; Saito, T.; Okazaki, T.; Ohshima, S.; Yumura, M.; Iijima, S. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.; Kittrell, C.; Ma, J. P.;
Chem. Phys. Lett. 2006, 429, 497–502. Hauge, R. H.; Weisman, R. B.; Smalley, R. E. Science 2002, 297, 593–596.
(38) Carter, D. C.; Ho, J. X. Adv. Protein Chem. 1994, 45, 153-203. (44) Jiang, L. Q.; Gao, L.; Sun, J. J. Colloid Interface Sci. 2003, 260, 89–94.
(39) Peters, T., Jr. All about Albumin: Biochemistry, Genetics, and Medical (45) Tsyboulski, D. A.; Bachilo, S. M.; Weisman, R. B. Nano Lett. 2005, 5,
Applications; Academic Press: San Diego, CA, 1996. 975–979.
10460 DOI: 10.1021/la901386y Langmuir 2009, 25(18), 10459–10465
3. Edri and Regev Article
Aldrich Chemicals (Cat. 855820). All chemicals were used as were prepared as follows. In the controlled environment box of a
received without further purification. Deionized water (18.2 vitrification robot (Vitrobot), a drop of the solution was deposited
MΩ 3 cm) was used in all samples preparation. on a glow-discharged TEM grid (300 mesh Cu Lacey substrate;
Thermogravimetric Measurements. Mettler-Toledo (model Ted Pella, Ltd.). The excess liquid was automatically blotted with a
TGA/STDA851e Modul) was used for TGA measurements. All filter paper, and the specimen was rapidly plunged into liquid
TGA experiments were performed in a 100 μL aluminum crucible ethane and transferred to liquid nitrogen where it was kept until
with a 70 μL of a sample to be tested. The experiments were used. The samples were examined below -175 °C using an FEI
performed under a N2 environment. The sample was heated to Tecnai 12 G2 TWIN TEM operated at 120 kV in low-dose mode
80 °C for 1 h and then to 280 °C for 1 h while the weight loss is and with a few micrometers under focus to increase phase contrast.
being recorded. Analysis of the results was performed with stare The images were recorded with a Gatan charge-coupled device
software, supplied by the manufacturer. camera (model 794) and analyzed by Digital Micrograph software,
UV-vis. Double-beam UV-vis spectrophotometer (Jasco Version 3.1.
V-530) was used for UV-vis measurements. The chemometric- Sample Preparation. Two sets of samples were prepared. The
based, full spectrum composition evaluation were performed first set was of four different BSA concentrations, namely 0.5, 1, 2,
using a commercial software, OPUS 6.5 .46 The software possesses and 4 mg mL-1 with fixed SWNT concentration of 2 mg mL-1 at
a calibration optimization options for e.g. finding the most pH 6. The BSA solution and SWNT powder were mixed, and the
suitable preprocessing data treatment or finding the most appro- obtained solutions were sonicated for certain given times.
priate wavelength regions from the spectrum to use, mainly by The second set included four distinct BSA conformations,
minimizing the root-mean-square error of cross-validation.47 namely, pH 3 (expanded conformation), 5, 7 (neutral), and
Specifically, the Quant 2 method, general A, was used to optimize 10 (basic).38,39 The pH was adjusted with 1 M HCl and NaOH
the calibration set. When SWNT percent recoveries48 were calcu- solutions. The solution pH was stable over time with and without
lated, a calibration procedure was performed as described else- buffering. Initial BSA and SWNT concentrations for all experi-
where.24 ments in this set were 0.5 and 2 mg mL-1, respectively.
In all experiments UV-vis absorbance was measured from a
20 μL samples diluted Â75 in 2 mg mL-1 CTAB solution, unless Results and Discussion
specified otherwise. Since we are monitoring a rapidly changing
dynamic process, which is exposed to precipitation, we are taking In this work the effect of two parameters on the exfoliation
a “snapshot” of the aggregation state during the dispersion dynamics is explored: (1) BSA-to-SWNT ratio and (2) pH. In
process by diluting in a CTAB solution. CTAB is used only after both cases the SWNT initial concentration is 2 mg mL-1.
sonication and does not take part in the dynamic process of However, in the first case the initial BSA concentration is
exfoliation. In control experiments we have not observed change increased at a constant pH, while in the second case, the pH is
in absorbance between BSA solution and BSA-CTAB due to changed while keeping both the SWNT and BSA initial concen-
BSA-CTAB interaction. The dilution to below 0.1 mg mL-1 of trations constant.
SWNT is mandatory to avoid scattering effects from inter- Two states are defined (vide supra): the unstable state, during
vening.49 The UV-vis absorbance spectrum of a sample was sonication and before centrifugation, is termed “shaken”. Here,
measured either at 500 nm (for a single wavelength measure-
the dispersion could be a result of the sonication energy and
ment-dynamic measurements) or in the 250-600 nm range
(recovery evaluation) with 1 nm intervals (for the full spectrum phase-separate when it is turned off. The stable state is defined for
analysis) using quarts cuvette and the appropriate blank the supernatant, after centrifugation, and is termed “stable”.
(2 mg mL-1 CTAB solution unless specified differently). An The tools we used are time-dependent single wavelength
example of a full UV-vis spectrum is given in the Supporting absorbance measurement of “shaken” samples and full spectrum
Information. chemometric analysis of “stable” samples. We complement these
Sonication. Glass vials were placed in a glass frame (not a measurements by cryo-TEM imaging of samples at different pH
plastic one) to arrest their movement during sonication, and the values and sonication times. In cryo-TEM the imaging takes place
location of the frame within the bath was limited. Bath-sonicator in a vitrified state, thus avoiding the need for drying, as the drying
(Elma sonic model S10; 30 W, 37 kHz, Sonics & Materials Inc.) changes the concentration of components and may result in a
was used in all experiments. The level of water in the bath (and distorted picture of aggregation state.
frame) was adjusted to be the same as the vials’ level.
The effect of BSA concentration (BSA-to-SWNT ratio) on the
Since the energy is not applied directly to the vials and part of it
is lost to the surroundings, the energy input is only estimated. We dispersion process of SWNT, as depicted by the absorbance
preferred using bath over probe sonication since the probe itself intensity, is presented in Figure 1A.
was found to degrade during sonication, and the degraded We find the “shaken” state dynamics of the BSA-dispersed
materials (from the probe) affect the absorbance spectrum and SWNT, i.e., before centrifugation (Figure 1A), to resemble the
hamper recovery calculations. “shaken” state dynamics of an ionic surfactant-dispersed SWNT
The solutions are sonicated under the specifications described reported previously.25,50 The dynamics were characterized by
above while along the sonication, samples are extracted. The level initial gradual increase in absorbance values and proceeded by
from which a sample is extracted was kept constant in all leveling off (a “plateau”). Such trend is observed for all cases
experiments. Eventually, the sample is centrifuged, and a sample (at different energy values) except for the 4 mg mL-1 BSA case.
from the supernatant is extracted.
The gradual increase was attributed to exfoliation and the plateau
Centrifugation. Megafuge 1.0 (Heraues) was used to spin
to maximal exfoliation degree.25 We note that for ionic surfactant
down nondispersed materials and was operated at 6240g for 30
min whenever centrifugation is applied. dispersion system, the plateau is obtained after ca. 5 kJ/g,25 while
Cryo-TEM. Samples for direct imaging of the aqueous disper- here the plateau is obtained only after ca. 70 kJ/g for the fastest
sions using cryo-transmission electron microscopy (cryo-TEM) case (0.5 mg mL-1 BSA). However, the experimental setup used
in their study is different, since they used a probe sonicator that
induces the pressure waves directly to the solution, with much
(46) http://www.brukeroptics.com/opus.html.
(47) For further elaboration on the software properties and chemometrics, we
refer the reader to refs 20 and 46 and references therein.
(48) SWNT percent recoveries are defined as [SWNT]final/[SWNT]initial  100. (50) Grossiord, N.; Loos, J.; Regev, O.; Koning, C. E. Chem. Mater. 2006, 18,
(49) Priya, B. R.; Byrne, H. J. J. Phys. Chem. C 2008, 112, 332–337. 1089–1099.
Langmuir 2009, 25(18), 10459–10465 DOI: 10.1021/la901386y 10461
4. Article Edri and Regev
Figure 1. Effect of BSA concentration on the dispersion dynamics of SWNT. (A) The absorbance progress (“shaken”) along the dispersion
process of SWNT ([SWNT] = 2 mg mL-1) at four different BSA concentrations, namely 0.5, 1, 2, and 4 mg mL-1 (circles, triangles, squares,
and diamonds, respectively). (B) SWNT recovery at the end of the process, after centrifugation (“stable”) calculated by full spectrum
chemometrics. The error is evaluated as ∼15% for all samples. (C) Comparison between the absorbance progress of shaken (full symbols) and
stable (empty symbols) solutions at [BSA] = 4 mg mL-1. The nanotube source in panels A-C is Carbolex. (D) Absorbance progress along the
dispersion process at [BSA] = 4 mg mL-1 for SWNT from a higher purity source (HiPCO).
higher efficiency than our bath sonicator (for details see the nanotubes are arranged in a weblike structure, which supports
Experimental Section). faster dispersion in comparison to the more bundled initial
In Figure 1A two effects are pronounced with decreased BSA structure of HiPCO nanotubes.25
concentration: (1) faster increase rate of absorbance intensity with The origins of the above-mentioned depletion force is a balance
energy and (2) higher plateau level (when obtained). This suggests between the translational and orientational entropy of SWNT to
that for lower BSA concentrations (1) the exfoliation process is the translational entropy of the protein molecules.51 Such phe-
faster and (2) eventually more SWNT are exfoliated. These results nomena have been described for the surfactant-SWNT system
are complemented by the chemometrics results in Figure 1B, (surfactant: sodium dodecyl sulfate).44,52,53 Jiang et al.44 reported
indicating that with lower BSA concentration the SWNT recov- a maximum in SWNT recovery near the critical micellization con-
ery after centrifugation is increased. We find that higher plateau centration (cmc) of the surfactant. It was suggested that above the
levels indicate higher exfoliation degree, which ultimately results cmc additional micelles induce intertubes attractive forces, hence
in higher SWNT recovery, in line with previous studies.25 The reducing the dispersed-SWNT concentration. On the other hand,
increase in SWNT recovery and exfoliation rate with the reduc- below certain dispersant concentration all the SWNT precipitate.
tion in BSA concentration are both explained by depletion Similar findings and considerations were computer simulated51
interactions (vide infra). and employed52 for lowering the SWNT percolation threshold in a
Before further discussing the above results, we ask whether the polymer matrix. Despite the above, the effect of dispersant
dynamics of the absorbance of the “shaken” state (Figure 1A) concentration on the dispersion dynamics has not been addressed
correlates with those of the “stable” state. In Figure 1C the yet, neither has such an effect been considered for proteins.
progress of SWNT dispersion (as indicated by absorbance in- Since the diameter of a BSA molecule and a micelle in solution
tensities) is plotted before (“shaken”, full symbols) and after is comparable, it is argued that the reported micellar depletion
(“stable”, empty symbols) centrifugation for the 4 mg mL-1 BSA phenomenon occurs also here. It is suggested that depletion
case. Indeed, in both cases a gradual increase in the absorbance induces an opposing “force” to the exfoliation-stabilization
intensity is detected. process, which results in both slowing down the process and
An additional issue of importance, as stated earlier,25 is the reducing the maximum exfoliation degree.
source and purity of SWNT used. We find the same trend for To examine the depletion hypothesis BSA molecules were
SWNT from high-purity (HiPCO) and low-purity sources added (final [BSA] = 10 mg mL-1) to BSA-dispersed SWNT
(Carbolex) (Figures 1D and 1C, respectively, and Experimental solution (4 and 2 mg mL-1, respectively) after sonication and
Section). We note, however, an order of magnitude difference in
absorbance values for the same energy inputs between the two (51) Schilling, T.; Jungblut, S.; Miller, M. A. Phys. Rev. Lett. 2007, 98.
sources. For Carbolex the absorbance intensity is higher, indicat- (52) Vigolo, B.; Coulon, C.; Maugey, M.; Zakri, C.; Poulin, P. Science 2005, 309,
920–923.
ing enhanced exfoliation.25 One possible explanation to this (53) Vigolo, B.; Penicaud, A.; Coulon, C.; Sauder, C.; Pailler, R.; Journet, C.;
difference is the initial structure of a given SWNT: Carbolex Bernier, P.; Poulin, P. Science 2000, 290, 1331–1334.
10462 DOI: 10.1021/la901386y Langmuir 2009, 25(18), 10459–10465
5. Edri and Regev Article
Figure 2. Effect of pH on the dispersion dynamics of BSA-dispersed SWNT. (A) Absorbance progress along the dispersion process of SWNT
at four different pH values, namely, 3, 5, 7, and 10 (circles, triangle, squares, and diamonds, respectively); [SWNT] = 2 mg mL-1, [BSA] =
0.5 mg mL-1. (B) SWNT recovery, calculated by chemometrics, after centrifugation (“stable”).
decantation (“stable” state). It was found that simply adding BSA Table 1. Dynamics of SWNT Dispersion by BSA
molecules to a stable SWNT solution (postdispersion) does not
BSA/NT increase ratea SWNT recovery
result in SWNT coagulation whereas mixing the same total pH [w/w] ratio [g/kJ] (Â10-3) [%, (15]b
amount of BSA molecules with SWNT prior to sonication
(termed predispersion) does result in precipitation, indicating a ∼6 0.25 9(1 110
path-dependent process. This path dependency is reasoned by the ∼6 0.5 2.9 ( 0.2 90
∼6 1 3.6 ( 0.3 79
strong BSA-SWNT binding energy52,54-56 that turns ∼6 2 1.06 ( 0.06 37
BSA-SWNT complex to a single entity. For the BSA-dispersed 3 0.25 N/A 10
SWNT to bundle (in the postdispersed solution), one has to first 5 0.25 2.1 ( 0.1 90
remove the adsorbed BSA molecules from the tubes to enable 7 0.25 3.6 ( 0.1 110
10 0.25 1.1 ( 0.1 74
their bundling. An alternative explanation could relate to the a
Calculated from a linear fit of absorbance to energy applied at initial
thick diameter of the bundles in the predispersed state compared phase of the dispersion process, e.g., for BSA/NT ratio = 0.5 ([BSA]=
to the thin diameter of the individual SWNT in the postdispersed 1 mg mL-1, Figure 1A) the absorbance values up to 150 kJ/g were taken;
state. Since the depletion force is proportional to the size ratio R2 > 0.95, for all samples, except pH 3. b Calculated using chemometrics
between BSA and SWNT, the processes in these solutions are from samples after centrifugation.
different.
However, unlike before (Figure 1A), the amount of energy
We now turn to present the results of the second tested
applied in this experiment was insufficient to reach a plateau.
parameter, that is, the dynamic effect of pH on dispersion of
From linear fitting to the plots, we find the following trend: at pH
SWNT by BSA. Previously,24 we have tested the pH effect on
7 the highest rate of dispersion is observed, the slope is found to be
SWNT dispersion by BSA. In short, BSA conformations were
(3.6 ( 0.1) Â 10-3 g/kJ (R2 =0.9898), and the maximal SWNT
found to disperse SWNT to different extents: bulkier BSA
recovery is obtained, in line with previous reports.24 At pH 5
conformation induces higher SWNT recovery. Additionally, at
and 10 the SWNT recovery is reduced, and so is the rate (slope =
each conformation, as the electrostatic charge the protein mole-
(2.1 ( 0.1) Â 10-3 and (1.1 ( 0.1) Â 10-3 g/kJ, respectively).
cule carries is higher, less SWNT are recovered. Here, we ask
These are summarized in Table 1. However, at pH 3 the
whether this collective pH effect applies also to the dynamics of
absorbance plot shows no gradual increase along the process,
the exfoliation process.
possibly indicating that SWNT are not exfoliated in this case. In
In this Experimental Section, we keep the [BSA] constant (0.5
accord, at pH 3 nearly no SWNT are dispersed.24
mg mL-1) and the pH is changed. Representative pH values are
Cryo-TEM micrographs of samples at pH 7 in the beginning,
selected from each region of BSA conformation, namely, pH 3 for
the end of the exfoliation process, and after centrifugation are
the expanded form, pH 5 and 7 for the normal form, and pH 10
presented in Figure 3 (see caption for details). These are compared
for the basic form.38 In addition to the previous experimental
to a sample at pH 3 at the end of the dispersion process (before
approaches, cryo-TEM was used to monitor the exfoliation
centrifugation).
process. Figure 2A shows the progress of absorption as more
The cryo-TEM images in Figure 3 support the spectroscopy
sonication energy is applied. After the sample is centrifuged, and
findings from a supramolecular point of view. At pH 7, with
the large bundles and unstable material are spanned down, the
increasing sonication energy the SWNT are exfoliated: from
SWNT recoveries at each pH value are evaluated (Figure 2B).
bundles (Figure 3A, ∼20 nm, double white arrows) to individuals
A general trend of a gradual increase in absorbance with
(Figure 3B, ∼2 nm, single white arrows), as is indicated in the
sonication energy is observed (Figure 2A), except for pH 3
UV-vis absorbance plot (Figure 2A). After centrifugation im-
(Figure 2A, circles), indicating an exfoliation process. We find
purities and large bundles are spanned down leaving mostly
that, indeed, at pH values where the exfoliation rate is faster (see
individual BSA-dispersed SWNTs (Figure 3C). At the same time
Table 1), the final SWNT recovery (i.e., after centrifugation) is
at pH 3, for the same sonication energy history (as in Figure 3B),
higher, similar to our previous findings for BSA-to-SWNT ratio.
the SWNT remain mostly in bundle form (Figure 3D), and when
centrifugation is applied, these bundles are spanned down to-
(54) Nakanishi, K.; Sakiyama, T.; Imamura, K. J. Biosci. Bioeng. 2001, 91, 233– gether with the impurities leaving a clear, SWNT-free supernatant
244. (not shown).
(55) Raffaini, G.; Ganazzoli, F. J. Biomed. Mater. Res., Part A 2006, 76A, 638–
645. The SWNT recovery can also be manipulated postdis-
(56) Raffaini, G.; Ganazzoli, F. Macromol. Biosci. 2007, 7, 552–566. persion. A similar approach was adopted in the section regarding
Langmuir 2009, 25(18), 10459–10465 DOI: 10.1021/la901386y 10463
6. Article Edri and Regev
Figure 3. Cryo-TEM micrographs of the solutions at several points along the dispersion process. At pH 7, in the beginning of the process (A),
the end of the process (B) (∼60 and ∼230 kJ/g, respectively), and after centrifugation (C), and at pH 3, at the end of the dispersion process
before centrifugation (D). For all samples [BSA] = 4 mg mL-1 and [SWNT] = 2 mg mL-1. Single white arrows indicate exfoliated SWNT.
Single black arrows indicate catalyst particles. Double black arrows in image (B) indicate frost on the vitrified grid, which precipitates during
its transfer to the electron microscope. Double white arrows indicate SWNT bundles.
BSA-to-SWNT ratio (vide supra). Here, the effect of pH on a recovery (N more than B). This is indicated in the UV-vis
stable BSA-SWNT solution (dispersed at pH 6) was explored results and complemented by our cryo-TEM images. We find
(postdispersion), unlike the above where the pH was changed that the N conformation disperses SWNT faster than the B
prior to the dispersion, i.e., predispersion. conformation and recovers more SWNT.
The pH of BSA-SWNT solutions was changed from pH 6 to These finding were recently supported by Valenti and co-
pH 3 or pH 10. Hence, changing BSA conformation from normal workers,40 who found through reflectometry measurements
(pH 6) to expanded (pH 3) or basic (pH 10). We found that both that at acidic pH values BSA adsorbs much slower on SWNT
solutions remained stable after 5 days unlike the predispersion and poorly covers the surface, in comparison to neutral pH
case, where changing the pH to pH 3 resulted in poor exfoliation values. We also find that changing the pH pre- or postdisper-
and dispersion of SWNT (Figure 2). This answers the question on sion results in different SWNT recoveries, suggesting that pH
pH change effect of pre- and postdispersed solution raised in the affects the stability differently than it affects the dynamics.
Introduction. The pH effect is different in pre- and postdispersion Once SWNT are dispersed by BSA, the conformation of
since different parameters control the exfoliation and the stabi- the protein can be changed without disturbing the affecting
lization of the SWNTs. stability.
In summary, BSA conformation (dictated by pH) has a
considerable effect on the dispersion dynamics of SWNT and Conclusions
not only on the final SWNT recovery. However, we find that The dispersion mechanism of BSA-dispersed SWNT in aqu-
the two are related: faster exfoliation dynamics results in higher eous media is studied via chemometrics-assisted UV-vis spec-
SWNT recovery (Figure 2 and Table 1). At pH 3 (expanded troscopy complimented by cryo-TEM. In both time-dependent
conformation), BSA does not facilitate SWNT exfoliation as experiments performed a direct correlation between the exfolia-
indicated by spectroscopic measurements and complemented tion dynamics and SWNT recovery was found.
by our cryo-TEM micrographs, which indicate bundled In the first set of experiments, different BSA:SWNT ratios were
SWNT. This contributes to our previous assumption on the studied, and the exfoliation process was monitored by the
low dispersion efficiency of the expanded conformation.24 A absorbance values of the “shaken” solutions (before cen-
reduced rate of BSA adsorption and surface coverage, trifugation). An entropic effect that counteracts SWNT exfolia-
presumably due to specific orientation required for BSA tions was suggested to take effect, as it was found that lowering
adsorption and high electric charge, ultimately results in BSA concentration improves the exfoliation dynamics and yields
poor exfoliation and dispersion of SWNT. In contrast, we higher SWNT recovery.
find that bulky BSA conformations (N at pH values In the second set, the solution pH was changed, and as a result,
5 and 7 and B at pH ∼10) facilitate SWNT exfoliation and the BSA conformation and electrical charge were altered. It was
10464 DOI: 10.1021/la901386y Langmuir 2009, 25(18), 10459–10465
7. Edri and Regev Article
found that the bulkier BSA conformations (N and B con- Supporting Information Available: Figure showing evolu-
formations) disperse SWNT rather effectively (N conformation tion of UV-vis spectra of aqueous BSA-dispersed SWNT
more then B) while the expanded conformation does not disperse solutions. This material is available free of charge via the
SWNT at all. Internet at http://pubs.acs.org.
Langmuir 2009, 25(18), 10459–10465 DOI: 10.1021/la901386y 10465