Перовскіт – це мінерал переважно чорного або червонувато-коричневого кольору з оригінальною структурою кристалічної решітки.

1. Mat. Res. BuU., Vol. 17, pp. 1245-1250, 1982. Printed in the USA.
0025-5408/82/101245-06503.00/0 Copyright (c) 1982 Pergamon Press Ltd.
FABRICATION OF PEROVSKITELEADMAGNESIUMNIOBATE
S.L. Swartz and T.R. Shrout
Materials Research Laboratory
The Pennsylvania State University
University Park, PA 16802
(Received July 7, 1982; Refereed)
ABSTRACT
The perovskite relaxor ferroelectric lead magnesium niobate
(PbMgl/3Nb2/303) is an important material because of its high
dielectric constant and correspondingly large electrostrictive
strains. However, it is difficult to prepare in polycrystalline
ceramic form because of the formation of a stable pyrochlore
phase. The reaction kinetics during calcining were investigated
and an improved fabrication scheme was developed.
Introduction
It is well known that the perovskite relaxor ferroelectric lead magnesium
niobate (PbMgl/3Nb2/303, designated PMN) exhibits unusually high dielectric
properties (I)~ It is also known that the dielectric properties of PMN can be
further enhanced by additions of PbTi03, thus making the PMN-PbTi03 system
promising for various dielectric and electrostrictive applications (2-6).
Until now, however, ceramic specimens in the PMN rich portion of the PMN-
PbTi03 system have proved to be quite difficult to synthesize using conven-
tional solid state reaction methods. This is due to the inevitable appearance
of a stable pyrochlore phase. Other investigators (2,7) have reported that
PMN can be successfully fabricated using conventional ceramic techniques but
which require repeated calcinations at high temperatures (>900°C) and long time
periods (~24 hrs). This approach has been found to be time consuming with
problems in reproducibility and the control of PbO stoichiometry.
The purpose of this investigation was to develop a simplified fabrication
method in which single phase PMN can be readily produced. In order to achieve
this, a better understanding of the perovskite-pyrochlore problem was required.
Thus, the initial stage of this investigation was to study the reaction kine-
tics of the formation of perovskite PMN from mixed oxides.
1245

2. 1246 $. L. SWARTZ, et al. Vol. 17, No. 10
ReactionSequence and Mechanisms - Conventional Mixed Oxides Processing
Experimental
Throughout this investigation, the raw materials used were readily avail-
able reagent grade PbO (yellow)*, food grade MgO (light powder)*, and crystal
grade Nb205**. The oxides were weighed out according to the PMNstoichiometry
and then wet ball milled in ethanol for 12 hrs. The resultant wet slurry was
then dried.
Two types of reaction studies were carried out:
I. A calcination study was performed wherebyclosed alumina crucibles
containing unreacted powderwere placed in a furnace at various temera-
tures (700-1000°C) and removedafter various times (I/2 - 24 hrs.).
2. Pellets I/2" in diameter and approximately I/4" in thickness were
pressed using the unreacted powder. The pellets were then placed in
a furnace and heated at a rate of 400°C/hr with pellets being removed
at 50°C intervals between 500 and lO00°C.
The resultant calcines and reacted pellets were then analyzed for phase
content and relative amounts using x-ray diffraction.
Results and Discussion
The results of the calcine study can be summarizedby the following
observations:
I. Increasing temperature increases the yield of perovskites phase,
but even at high calcine temperatures (950-I000°C), a relatively
large amountof pyrochlore phasewas still present.
2. Calcining for periods longer than 4 hours does not appreciably
improve the yield for perovskite.
3. Excesslead and magnesiumoxides were always detected indicating
incomplete reaction.
As expected, the above results confirm that even for a wide range of calcina-
tion cond4tions, single phase PMNcannot be easily fabricated.
The above results are also supported by the results of the pellet study,
in which the following reaction sequencewas deduced. No reaction of the
mixed oxides was observed up to ~650°Cwith the appearanceof a pyrochlore
phase being detected at approximately 700°C. As the temperature was increased
tO 800°C, the amountof pyrochlore increased with a corresponding decrease in
niobium oxide. Associatedwith the formation of the pyrochlore phasewas the
volume expansionof the pellets. This is attributed to the pyrochlore phase
having a larger molar volumethan any of the starting oxides.
The PMNperovskite phase first appeared around 850% at which timeniobium
oxide could no longer be detected. The amountof perovskite phasewas found
to increase until 950°Cwhere it appeared that no further reaction took place.
Lead oxide volatilization may have caused the reaction to end prematurely.
Results from the calcination study indicate, however, that it is doubtful that
the reaction would have gone to completion (i.e., complete transformation of
pyrochlore to perovskite) had the lead oxide not volatilized. As observed in
*Baker Chemical Co., Philadelphia, PA.
**Herman C. Starcke, Inc., New York, NY.

3. Vol. 17, No. 10 LEAD MAGNESIUM NIOBATE 1247
the calcination study, there was always small amountsof magnesiumoxide
detected even whenthe reaction appeared to have been completed.
The observations presented from the calcination study and reaction sequ-
ence appear to be similar with those reported by others (2,7). However, Inada
(7) proposed a similar but more complicated reaction sequenceduring the forma-
tion of PMN. He reported that the perovskite Pb(Mgl/3Nb2/3)O3 was formed, not
by the direct reaction of oxides, but by the repetition Of the following reac-
tions:
3PbO + 2Nb205530-650°C, 3PbO.2Nb205 (l)
3PbO.2Nb205 + PbO 600-700°C~ 2(2PbO'Nb205) (2)
2PbO.Nb205 + I/3 MgO700-800°C, Pb(Mgl/3Nb2/3)O3
+ I/3 (3PbO.2Nb205) (3)
In equation l, the initial reaction that occurs is between PbO and Nb205
to form the cubic pyrochlore Pb3Nb4013,where upon further reaction with PbO
results in the formation of a rhombohedral pyrochlore Pb2Nb207 (equation 2).
The rhombohedral pyrochlore then reacts with MgOto form the perovskite
Pb(Mgl/3Nb2/3)O3. With the reappearanceof the cubic pyrochlore phase. Inada
concludes that to get pure PMN, it is necessary to prevent the evaporation of
PbO, which would slow down the above reactions, and to repeat the process of
calcining, crushing and calcining. It was also reported that the above reac-
tions could be accelerated by excess additions of PbO and MgO, but would
result in the PMNphase being impure.
The above reaction sequencediffers from that found in this work in that
only a cubic pyrochlore phasewas observed. Anotherdifference is the compo-
sition of the observed pyrochlore phase(s) in which Inada reported were simply
lead niobates. It seemsquite feasible that MgOcan be incorporated into the
lead niobate pyrochlore structure. This was suggestedby a shift in the pyro-
chlore lattice parameter as determined by x-ray diffraction at high angles.
The exact amountof MgOwhich can enter the pyrochlore structure is unknown.
A pyrochlore with nominal composition Pb2Nbl.87Mgo.3207was reported by
Andrianova et al. (8). Whetherthis composition is similar to that found dur-
ing the reaction sequenceand/or calcination study is yet to be determined.
Thus, from the results of this investigation, the following reaction
sequence was proposed:
PbO + Nb205 + MgO 25-650°C, no reaction (4)
PbO + Nb205 + (rlgO) 700"800°~ cubic pyrochlore (5)
Pyrochlore + PbO + MgO850-900°C, perovskite + pyrochlore (6)
It is important to note that any differences in the reaction sequences
reported could be due to the reaction kinetics dependencyon various experi-
mental parameters, e.g. particle size and surface area of raw materials,
heating rates, etc.
Other Fabrication Techniquesfor the Formation of PMN
Stoichiometric Variations
From the results presented in the last section, it was shownthat the
nature of the perovskite-pyrochlore problem is a kinetic one, involving the
reactivity of PbO and more importantly MgO.

4. 1248 S.L. SWARTZ, et al. Vol. 17, No. i0
To test the importance of MgO in the formation process of PMN, a calcina-
tion study was performed in which various amounts of excess of MgOwas added.
Four compositions were batched and ball milled according to the stoichiometry
3PbO.(l+x)MgO.Nb205, with x values of O, .Ol, .02, and .05. These compositions
were designated P+M+N-STD(x = O) and P+M+N-I,2 or 5% MgO. The batches were
heated in closed alumina crucibles at furance rate (~500°C/hr) to a calcine
temperature of (800, 870, or lO00°C), followed by a four hour soak.
The relative amounts of the pyrochlore and perovskite phases were deter-
mined using x-ray diffraction by measuring the major x-ray peak intensities
for the perovskite and pyrochlore phases (for (llO) and (222), respectively).
The amount of perovskite was calculated using the following equation:
%perov. = lO0 x Iperov/(Iperov + Ipyro) (7)
and presented in Table I.
Composition
P+M+N-STD
I% MgO
2% MgO
5% MgO
800°C 4 hrs
68.4
77.0
64.3
65.3
TABLE I.
% Perovskite
870°C 4 hrs
74.0
78.5
79.0
80.0
lO00°C 4 hrs
76.3
81.3
77.5
82.8
These results indicate that the amount of perovskite increases with
increasing calcination temperature, as reported in Section I. It can also be
shown that the addition of excess MgO appears to accelerate the pyrochlore to
perovskite reaction, but not substantially, as reported by Furukawa et al. (9).
Thus the need to develop yet another approach to fabricate PMNwas still
required.
"Novel" Fabrication Technique
A possible solution to the perovskite-pyrochlore reaction problem would
be to somehow bypass the formation of the intermediate pyrochlore phase
reaction. To achieve this using the conventional mixed oxide technique, one
would have to increase the reactivity of the oxides, particularly the MgO,
such that they would simultaneously react to form perovskite. Since one of
the goals of this investigation was to develop a fabrication technique using
only commercially available starting materials the following reaction sequence
was proposed,
MgO + Nb205 ÷ MgNb206 (8)
MbNb206 + 3PbO ~ 3(PMN) (9)
whereby MgO is prereacted to form MgNb206, prior to reaction with PbO.
This approach was appealing for two reasons:
I. In order for lead oxide to react with MgNb206 to form pyrochlore,
it would have to first liberate Nb205 from MgNb206. The kinetics of
this reaction might be slow enough to prevent pyrochlore formation.
2. The columbite structure of MgNb206 is an oxygen octahedra struc-
ture, similar to the perovskite structure.

5. Vol. 17, No. I0 LEAD MAGNESIUM NIOBATE 1249
To test this 'novel' approach, magnesium niobate (MgNb206) was formed by
calcination of MgO and Nb205at IO00°C and batched with PbO to form the stoi-
chiometric PMNcomposition. A reaction pellet study, similar to the one
described earlier was then performed. The results were very encouraging.
Perovskite PMNwas the first phase to form at approximately 700°C followed by
a small amount of pyrochlore phase at 800°C. The reaction appeared to be
complete by 900°C, with the resultant phase content being primarily perovskite
PNNwith less than 5% pyrochlore. Thus the proposed reaction sequence, equa-
tions 8 and 9, appeared to be correct.
To determine the effect of excess NgO on the MgNb206 (designated HN)
approach, another calcination study was performed. ExcessMgOwas added
prior to the MgNb206 calcination step. The compositions investigated were
similar to those described earlier (P+MN-STD, P+MN-I,2 and 5% MgO).
The results of the calcine study are given in Table 2. For comparison,
results from the conventional mixed oxide approach were also included.
TABLE 2.
Composition %Perovskite
800°C, 4 hrs 870°C, 4 hrs lO00°C,4 hrs
P+M+N-STD 68.4 74.0 76.3
P+MN-STD 98.9 98.5 98.2
P+MN-I%MgO 98.9 99.4 98.2
P+MN-2%MgO lO0.O* lO0.O --
P+MN-5%MgO lO0.O lO0.O --
*No pyrochlore phase was detected within the limits of x-ray
diffraction.
As presented in Table 2, the magnesium niobate approach represents a substan-
tial improvement over the conventional mixed oxides method. It was also
found that excess MgO can further decrease and even eliminate pyrochlore
formation.
Summary
The formation processes of the perovskite relaxor ferroelectric lead
magnesium niobate have been investigated using conventional mixed oxide ceramic
processing. It was found that on reaction of the oxides, a pyrochlore phase
is the first to form. On further reaction, this pyrochlore partially trans-
forms to perovskite. It appears that the reactivity of MgO plays an important
role in determining the completeness of this reaction. In this study, single
phase PMN could not be fabricated from mixed oxides, even with excess MgO and
varying calcining conditions.
A "novel" fabrication process was developed in which the formation of the
pyrochlore phase was bypassed. Pre-reacting magnesium and niobium oxides to
form MgNb206 prior to reaction with PbO led to the direction formation of the
perovskite phase. This technique resulted in perovskite phase purity substan-
tially improved over conventional mixed oxides processing. Further additions
of excess MgO reduced and eliminated the pyrochlore phase. The effect of
excess ~IgOon the dielectric properties is under investigation.

6. 1250 S.L. SWARTZ, et al. Vol. 17, No. i0
This new approach should result in improved reliability in the fabrica-
tion of single phase PMNbased ceramics and might also be applied to the
fabrication of other complex perovskite relaxors, e.g. PbNil/3Nb2/303 and
PbCol/3Nb2/303.
Acknowledgements
The authors would like to acknowledge Dr. L.E. Cross for suggesting the
research and for stimulating discussions on the topic, Dr. W.A. Schulze for
his interest and support, and especially Colleen Ottoson for her help in much
of the experimental work.
References
I. Landolt Bornstein, Ferroelectrics and Related Substances, New Series,
Vol. 16 (1981).
2. S.J. Jang, Ph.D. Thesis, The Pennsylvania State University, 1979; S.J.
Jang, et al., Ferroelectrics 27, 31 (1980).
3. W.A. Schulze, et al., J. of the Amer. Ceram. Soc. 63(9), 956 (1980).
4. S. Nomura and K. Uchino, Ferroelectrics 41, If7 (1982).
5. L.E. Cross, et al., Ferroelectrics 23, 187 (1980).
6. H. Ouchi, et al., J. Amer. Ceram. Soc. 48(12), 630 (1965).
7. M. Inada, Japanese National Technical Report, 23(I), 95 (1977).
8. Adrianova, et al., Optical Spectroscopy (English Translation), 36,
547 (1974).
9. K. Furukawa, Proceedings of the Japan-U.S. Study Seminar on Dielectric
and Piezoelectric Ceramics, pg. T4 (1982).