26 THE AZULENE,s-TRINITROBENZENE COMPLEX

the disorder would presumably have been concealed

0,0 ~0
by errors in the molecular geometry and thermal
motion.

The specimen material was prepared by Mr Y. Lu-

pien of the Division of Pure Chemistry. Computations
were carried out on the IBM 1620 computer under the
supervision of Dr F. R. Ahmed, using his programs
and those of Dr G. A. Mair and Mrs M. E. Pippy.
zj
Drs Brown and Wallwork were kind enough to allow
study of their paper before publication. The assistance
of those mentioned, and the continued encouragement
of Dr W. H. Barnes, are gratefully acknowledged.

References

BROWN, D. S. & WALLWORK, S. C. (1965). Acta Cryst. In

the press.
BUSrNG, W. R. & LEvY, H. A. (1964). Acta Cryst. 17, 142.
CRtaCKSnANK, D. W. J. (1961). Computing Methods and the

l/
Phase Problem in X-Ray Crystal Analysis. London: Per-
gamon Press.
F~MAN, A. J. (1959). Acta Cryst. 12, 261. s
FURNAS, T. C. (1957). Single Crystal Orienter Instruction 03
Manual. Milwaukee: General Electric Company. Q Carbon
HANSON, A. W. (1964). Acta Cryst. 17, 559.
HANSON, A. W. (1965). Acta Cryst. 18, 599.
HODGSON, L. T. & ROLLET,J. S. (1963). Acta Cryst. 16, 329. Nitrogen
MAre, G. A. (1963). Private communication.
ROBERTSON,J. M., SHEARER,H. M. M., SIM,G. A. & WAT- Oxygen
SON, D. G. (1962). Acta Cryst. 15, 1.
SUNDARALINGHAM, M. • JENSEN, L. H. (1963). Acta Cryst. Fig. 7. Projection along b of the asymmetric unit, showing
16, A61. overlap.

Acta Cryst. (1965). 19, 26

Multiple Phase Formation in the Binary System Nb2Os--WO3.
I. Preparation and Identification of Phases
BY R. S. ROTH* AND A. D. WADSLEY
Division of Mineral Chemistry, C.S.LR.O., Melbourne, Australia

(Received 26 October 1964)

An equilibrium study at 1300-1400°C of the system Nb205-lNb205:1W03 led to the identification

of five compounds in what had been previously reported as a region of solid solution. Three phases,
rWNb30078, WNb12033 and WsNb16055 are monoclinic, while the remaining two, W3Nb14044 and
WsNb18069, are tetragonal. Each phase is characterized, and it is shown that the existence of not more
than two of them could have been predicted by crystallochemical arguments.

that are formed when elements with about the same

Introduction
ionic radii, and of the same or lower valency, are sub-
We have recently discussed the crystal chemistry of stituted for niobium (Roth & Wadsley, 1965a). The
niobium pentoxide and the compounds, related to it, structures all contain metal-oxygen octahedra form-
ing sub-units or blocks of the ReO3-type, of finite sizes
in two dimensions but infinite in the third, which con-
* Permanent address: National Bureau of Standards, Wash-
ington, D.C., U.S.A. dense upon themselves by edge-sharing in such a way
 R. S. R O T H A N D A. D. W A D S L E Y 27

that tetrahedrally coordinated atoms may, or may not, a parent phase B308, with a structure of the kind re-
be present at the junctions. For each compound the cently reported for NbaO7F by Andersson (1964),
repeat distance along the axis of symmetry is approx- which is broken up into the smaller units and then
imately 3.8 A, and the blocks are centred about two rearranged in these four characteristic ways. There is
equidistant levels in the unit cell perpendicular to it. no simple formula containing only one variable that
The six compounds whose structures have been re- encompasses the six known compounds.
ported can be classified into four groups. It appeared not unlikely that more homologues could
Group A. TiNb207 contains blocks of octahedra belong to each group B, C, and D. The substitution of
3 x 3 × oo joined at both ends to similar blocks forming lower valency ions for Nb does not give any more
infinite ribbons. These appear at both levels and are phases other than those that are listed (Roth & Wads-
joined by edge-sharing (Wadsley, 1961a). Ti2Nb10029 ley, 1965a), and the alternative, therefore, was to seek
has an identical arrangement, but the blocks are larger, higher members of the series. These could possibly be
3 x 4 x oo (Wadsley, 1961b). synthesized by reacting Nb205 in the solid state with a
Group B. In Nb205 there are two sizes of block, hexavalent metal oxide, WO3 or MoO3, and some of
3 x 5 x oo joined as in group A at one level, but cem- the compounds that might be expected to form are
ented together by single 3 x 4 x oo blocks at the other given in Table 1. Their ideal structures can be derived
level, with tetrahedral metal atoms at the junctions. geometrically from those already known from within
Nb22054 is probably related to Nb2Os, but the blocks the same group. The composition of each one is de-
of which the ribbons are composed are 3 x 4 x oo, and duced from its particular formula, while the symmetry,
the single units are 3 x 3 x oo (Gatehouse & Wadsley, space group, and close approximations to the unit-cell
1964). size are readily evaluated from the ideal model.
Group C. 3 x 4 × oo blocks are joined in pairs, ap- We have therefore examined the phase equilibria in
pearing at both levels and with tetrahedral metals at a binary system where compound formation might be
the junctions in TiNb24062 (Roth & Wadsley, 1965a). predictable. Although the approximate crystallogra-
Group D. In PNb9025 there are single 3 x 3 x oo blocks phic data for any likely compound are known, the best
at two levels with tetrahedral phosphorus atoms (Roth, possible powder diffraction patterns, used as the sole
Wadsley & Andersson, 1965). means of identification, can be misinterpreted. Where
The idea of structurally related 'homologous series' there is a sub-cell, there will be an overall similarity
of reduced metal oxides (Magn61i, 1953) has been a between the patterns of the various phases, irrespective
particularly fruitful one. In binary systems BOn-BOn-x, of their composition; and where the unit cell is large
where B is one of the elements Mo, W, or Ti, a partic- and oblique, there will be difficulties in assigning the
ular series may consist of several members, and if the correct indices to the reflections. Optical methods of
crystal structure of one is known, those of the re- phase determination are of very little use in a system
mainder, in principle, can be deduced from their chem- where the indices of refraction are very high, but may
ical composition. Conversely the unit-cell size of a help in evaluating the extent of regions of homogeneity,
member defines the position it adopts in a series, and if these exist. Consequently the use of single-crystal
its structural formula and chemical composition can techniques is of the utmost importance as a phase-
thereby be established. The two compounds of group determining criterion, while crystal structure analysis
A are the members n = 3 and 4 of the series B3nO8n-3, provides the only satisfactory proof that each com-
and no other homologues have been reported. Those pound has the arrangement of atoms claimed for it.
in group B form the series Ban+lOsn-2 which is limited A preliminary appraisal of the system Nb2Os-WO3,
to members where n is odd. Group C, containing only dealt with here, forms the introduction to the three
one compound, belongs to a potential series with the papers following immediately after. Five previously
same formula as group B but where n is even. Finally unknown compounds have been identified in the region
a series B3n+lOsn+l could possibly arise from com- NbEOs-Nb205 . WO3 at 1350 °C. The determination of
pounds with structures related to PNb9025 of group the structures of four of them, two monoclinic and
D, n in this case being 3. Each series is derived from two tetragonal phases, is described in the two follow-

Table 1. Possible compoundformation in the system Nb2Os-WO3 (below dotted line)

B3n+108n-2
^ n3n+108n+1
Group B, n odd Group C, n even Group D
,~ ~ ^ ^

n Formula Composition n Formula Composition n Formula Composition

7 B22054 Nb22054
9 B28070 Nb205 8 B25062 TiO2.12Nb205 3 B10025 PNb9025

11 B34086 WO3 . 8Nb205 10 B31078 WO3.15Nb205 4 B13033 WO3.6Nb205

13 B400102 2WO3.9Nb2Os 12 B37094 3WO3.17Nb205 5 B16041 2WO3.7Nb205
28 MULTIPLE PHASE FORMATION IN T H E B I N A R Y S Y S T E M Nb205-WO3. I.

ing papers, while the crystal chemistry of the whole Table 2 (cont.)
group is discussed in the fourth (Roth & Wadsley, Composition Heat treatment
^ ^
1965b, c, d).
(mol ratio)" "Temp. Time Cooling~ Results
Previous work in the system Nb2Os--WO3 11 1 1350 24 An 15:1+'6:1'
8 1 1350 24 An '6:1'
Two previous attempts have been made to determine 6 1 1350 24 An 6:1
the phases in the system Nb2Os-WO3. Goldschmidt 17 3 1300 24 An 6:1
(1960) surveyed a number of binary systems of Nb205 7 2 1300 24 An 6:1+7:3
1400 18 Q 6:1+7:3
with other oxides, and concluded that more than 50 8 3 1300 24 An 6:1+7:3
mol.yo WO3 could enter into solid solution with high- 7 3 1350 24 An 7:3
temperature Nb2Os. He also reported a phase at about 1425 4 Q 7:3
75 mol.Yo WO3, Nb2Os. 3WO3, and a small amount 1500 1 Q 7: 3 (melted)
8 5 1350 24 An 8:5
of solid solution of Nb205 in WO3. Kovba & Trunov 1350 18 Q 8:5
(1962) considered that this compound was tetragonal 3 2 1350 24 An 8: 5 + unidentified
with the lattice parameters a=12.19 and c=3.934 A, phase or phases
and with a structure possibly related in some way to 9 8 1350 24 An 9: 8 + unidentified
phase or phases
that of tetragonal tungsten bronze. Fiegel, Mohanty & 1350 18 Q 9:8
Healy (1964) confirmed this compound, and suggested 1 1 1100 72 Q '1:1'
that two more were present at the compositions 1350 18 Q 9:8+'13:24'
3Nb205 • 2WO3 and Nb2Os. WO3. The first was form- 10 11 1350 18 Q 9:8+'13:24'
1l 15 1350 18 Q 9:8+'13:24'
ed above 1200 °C, the second was stable only at lower 12 19 1350 18 Q '13:24'+9:8
temperatures, and neither was prepared as a single 13 24 1350 18 Q '13:24'
phase. They reported that up to 33 mol.Yo WO3 en- 14 29 1350 18 Q '13:24'+'4:9'
tered into solid solution with Nb2Os. 3 7 1350 18 Q '4:9'
16 41 1350 18 Q '4:9'+'1:3'
17 48 1350 18 Q '4:9'+'1:3'
Experimental 1 3 1350 18 Q '4:9'+'1:3'
19 63 1350 18 Q '1:3'+'4:9'
The component oxides were weighed out to the nearest t An=annealed at 1000°C for one to three days, Q=quen-
milligram in approximately one-gram batches to give ched by dropping from furnace into water.
the required composition, and then thoroughly mixed * Ratio enclosed by quotation marks when the exact com-
position of the phase has not been established.
by hand in a pestle and mortar. A portion of each was
sealed inside a small platinum capsule in order to
eliminate volatilization of WO3 upon heating, as well
Description of phases
as to promote crystal growth. The tubes were heated WNb30078 (15Nb205 • WO3)
in a platinum-wound vertical tube furnace, and either A single phase WNb30078, appearing at the predicted
quenched by dropping into water, or annealed at some composition, is separated by two-phase regions from
lower temperature. both Nb205 and WNba2033. No measurable changes
X-ray powder diffraction patterns were taken with in the positions of lines in the Guinier patterns could
copper radiation on an accurately aligned Guinier- be found, and it therefore cannot have more than a
type focusing camera, internally calibrated with pot- small region of homogeneity. It forms minute colorless
assium chloride and where appropriate, single crystals monoclinic needles with the powder pattern given in
were examined in a Weissenberg camera to establish Table 4, to which we could not give satisfactory indices.
the unit-cell size and space group. The approximate If the phase is truly the member n = 10 of the series
dimensions were then refined with the Guinier data, B3n+lOsn-2, it will have the approximate unit-cell dim-
the complementary use of the two techniques having ensions
considerable merit in equilibrium studies of the present a=39.0, b=3.82, c=21.5/~, fl=91.5
kind. Densities were measured with a torsion micro-
balance with toluene as the displaced liquid. and the space group C2, Cm or C2/m.
The results of the partial phase analysis are sum- We have shown elsewhere that TiNbz4062, n = 8 of
marized in Table 2. the same series, could be expected to have a dimorph
where the ReO3-type blocks are joined up in pairs in
a different way (Roth & Wadsley, 1965a). This method
Table 2. Experimental data for the system Nb2Os-WO3
of joining could also be adopted by the phase
Composition Heat treatment WNb30078, and the unit cell would then be
^ ^

(mol ratio)" 'Temp. Time Cooling]" Results a=22.4, b=3.82, c=38.0 A, fl= 106 °,
Nb205 WO3 (°C) (hr)
with the same space group alternatives.
30 1 1350 24 An Nb205+ 15:1
15 1 1350 24 An 15:1 Several crystals were mounted, but in all cases the
13 1 1350 24 An 15:1+'6:1'* films could be interpreted only by assuming they were
 R. S. R O T H A N D A. D. W A D S L E Y 29

Table 3. C r y s t a l l o g r a p h i c data f o r c o m p o u n d s in the region Nb2Os-lNb2Os: 1W O 3

Mol. ratio Symmetry Possible Unit-cell dimensions in (A)
^ ^

"Nb205 WO3' space groups a b c fl

oo - Monoclinic P2 21"16 3.822 19"35 119-8°
15 1 Monoclinic . . . .
6 1 Monoclinic C2/m, C2, Cm 22"37 3"825 17"87 123"6°
7 3 Tetragonal I4/m, 14, I~ 21.02 3.824
8 5 Monoclinic C2/m, C2, Cm 29.79 3.820 23"08 126.5°
9 8 Tetragonal I4/m, 14, 12[ 26"25 3.813

derived from two crystals, either a twin of the one It forms minute pale blue tetragonal needles averaging
structure, or an intergrowth of the dimorphs within 20 microns long. The unit cell dimensions and powder
the one crystal. This latter phenomenon was noted by pattern are given in Tables 3 and 6.
Evans & Mrose (1960) for the case of structurally This compound is not a member of any of the
related vanadium minerals doloresite and montroseite, homologous series in Table 1, and its structure is re-
and is commonly found in sulphosalt minerals. In the ported in the third paper of this group (Roth & Wads-
present case an unambiguous decision between the two ley, 1965c). It should be noted that its composition
alternatives cannot be given from the material avail- corresponds to the end of the solid solution of WO3 in
able, as the unit cells of both expected dimorphs are NbzO5 reported by Fiegel, Mohanty & Healy (1964).
very large and not dissimilar, while the diffraction data WsNb16055 (8Nb205 • 5WO3)
provided by the crystal are scanty and dominated by The composition 3Nb2Os:2WO3 was found in the
the sub-cell reflexions. present study to consist of two phases (Table 2), and
We should point out that the zero level Weissenberg 8Nb2Os:5WO3, separated from this by less than 2
film, as well as the powder pattern, could be indexed mol.yo, was essentially a single phase, consisting of
with the unit cell well-formed blue needles up to 0.1 mm long.
a=21.62, b=3.824, c = 18.77 A, fl= 125.8 °
Table 4. U n i n d e x e d p o w d e r p a t t e r n o f 15Nb205. WO3
if certain weak reflexions in both sets of data were
ignored. For some time we believed this to be the true l(obs) sinZ0 (obs)
unit cell with a structure related to that of WNb12033, w 0.00509
vw 0"01500
but with excess metal atoms in the tetrahedral positions. ms 0.02288
This model did not refine, and the study was aban- m 0"02747
doned. vs 0"04242
vw 0-04300
s 0-04492
WNbx2033 (6Nb205 • W O 3 ) vw 0.04531
Although the compositions ranging from 8: 1 to 17: 3 w 0"04699
Nb2Os:WO3 all showed a single phase with no detect- w 0"04814
able differences in the unit-cell dimensions, it is debat- s 0"04862
m 0.05274
able whether the experimental techniques are good w 0"05877
enough to classify this region as a solid solution.
This compound forms colorless monoclinic needles Table 5. P o w d e r p a t t e r n of WNb12033 (6Nb205 . WO3),
up to 0.04 mm long, providing single-crystal data of Cu Kua
high quality. Unit-cell dimensions and indexed powder
patterns are given in Tables 3 and 5. All of the evidence l(obs) sinZ0 (obs) sin20 (calc) hkl
suggested that the compound was WNb12033, the pre- m 0.00268 0"00268 001
m 0.00478 0"00478 20i
dicted homologue n = 4 of the series B3n+108n+1 , group vw 0.01073 0.01076 002
D, which contains PNb9025 as the member n = 3 (Table w 0.01922 0.01916 402
1). This was subsequently confirmed by the structure mw 0.02326 0.02313 40~
analysis reported in the following paper (Roth & Wads- vw 0"02423 0"02421 003
mw 0-02706 0"02708 202
ley, 1965b). The Guinier pattern was very much like ms 0"04219 0"04225 110
that of WNb30078, and if the films of preparations at ms 0"04254 0"04256 llI
the compositions 13:1 and 11 : 1 NbEO5:WO3 had not W
0"04318 [ 0"04304 004
shown two sets of reflexions side by side, it would have L
0.04311 60]
vw 0"04397 0"04388 602
been logical to assert that the two compounds were s 0"04526 0"04527 203
part of the same solid solution. mw 0"04726 0"04730 111
s 0.04802 0"04772 602[
W3NbI4044 (7Nb205.3WO3) 0.05002 60I
vw 0.05011 0.05014 203
A single phase W3Nb14044 with no range of homo- m 0.05249 0.05247 312
geneity was found at the composition 7Nb205.3WO3. mw 0.05772 0.05775 112
30 MULTIPLE PHASE FORMATION IN T H E B I N A R Y S Y S T E M Nb2Os-WO3. I.

Table 6. P o w d e r pattern ofWaNb14044(7Nb205.3WO3), Table 8. P o w d e r pattern of W8NblsO69(9Nb205.8WO3),

Cu K~I Cu K~I
I(obs) sin20 (obs) sin20 (calc) hkl I(obs) sin20 (obs) sin20 (calc) hkl

m 0.00267 0.00269 110 vw 0"00688 0"00689 220

m 0"01074 0"01075 220 vw 0"01548 0"01550 33O
w 0.01339 0.01343 310 w 0.01715 0.01722 420
mw 0.02417 0.02418 330 m 0"02927 0.02927 530
s 0.02686 0"02687 420 s 0-04169 0.04166 101
vs 0.04189 0-04192 101 vw 0"04305 0"04305 710
m 0.04302 0"04299 440 vs 0"04477 0.04472 640
vvs 0.04576 0.04568 530 w 0.04510 0.04511 211
w 0.04827 0"04836 600 w 0"05199 0"05199 321
m 0"05801 0"05804 321 vw 0-05859 0"05855 820
vw 0"06716 0"06717 710 ~ 0-06199 660
w 0.06997 0.06986 640 w 0"06220 t 0.06233 501
s 0.07422 0"07416 501 m 0"07603 0.07610 541
w 0.07803 0.07792 730 vw 0"07743 0.07749 930
s 0.07948 0"07954 521 ms 0"07945 0"07955 631
w 0"09029 0.09028 611 vw 0.08642 0.08643 721
ms 0-09145 0.09135 820 mw 0"08947 0"08954 10t2t0
w 0"09571 0"09566 541
w 0-10104 0"10103 631
m 0"11189 0"11178 721 nealed for three days at 1000 °C. Its unit-cell dimen-
ms 0.14263 0.14240 950 sions and powder pattern are given in Tables 3 and 8.
ms 0.16238 0.16231 020 This high-temperature compound crystallizes as
small dark green tetragonal needles, turning yellow
Part of its Guinier powder pattern is given in Table after several months, and like the two preceding phases
7. The unit cell is large and oblique (Table 3), and it its structure was not predictable with the original as-
would have been well nigh impossible to index the sumptions prompting this study. It has now been de-
pattern unambiguously if good single-crystal films had termined, and is reported together with that of
not been available for cross-reference. The powder pat- W3Nb14044 (Roth & Wadsley, 1965c).
tern, furthermore, could have been dismissed as a mix-
ture of the two adjacent phases if no attempt had been
made to examine single crystals. The compound does Additional phases in the system l~b205-WO 3
not belong to any of the predicted series of compounds,
and its structure is reported in the following paper All of the previous workers have agreed upon the com-
(Roth & Wadsley, 1965b). position of tetragonal Nb205.3WO3, although Mo-
hanty & Fiegel (1964) reported that single-crystal pat-
Table 7. P o w d e r pattern WsNb16055 (8Nb205 • 5WO3),
of
terns contained superstructure reflexions differing from
Cu Kcq one crystal to the next. We found three and possibly
I(obs) sin20 (obs) sin20 (calc) hkl
four phases in the composition region 65-80 mol.yo
mw 0.00266 0.00268 20I WO3. The Guinier patterns were remarkably similar to
vvw 0"00687 0-00689 002 each other as well as to that reported for Nb205: 3WO3,
m 0"01073 0-01079 40~ but none of them indexed satisfactorily with the re-
vw 0.01305 0.01303 403 ported lattice parameters. Single crystals have not yet
vvw 0"01540 0.01551 003
vvw 0"01702 0.01739 202 been examined, and therefore no assignments of unit
mw 0.02427 0.02417 60] cell size and symmetry are included in this paper.
ms 0"02688 0.02670 60~ The compositions at which the phases are found are
mw 0.02911 0.02918 203 approximately 13Nb2Os: 24WO3, 4Nb205: 9WO3, while
0.04169 110
vvs 0.04184 0.04182 llI the third is richer in tungsten than NbzOs:3WO3. The
0.04296 80~ three preparations are characterized simply enough by
w 0.04307 0.04308 005 striking colour differences between them; the first is
s 0.04441 0.04442 204 cream, the second which is dark grey contains two
mw 0.04505 0.04500 111
vs 0.04576 0.04576 803 phases, colourless and black, while the third is bright
m 0.05006 0.04997 310 yellow. The crystal structure determination of these
m 0.05824 0.05820 51~ compounds may prove to be the only satisfactory way
of deciding their chemical formulae, and this is now
WsNblsO69 (9Nb205 • 8WO3) being examined by one of us (A.D.W.).
The composition 9Nb205.8WO3 was found to be Most of the preparations have been confined to the
single phase when quenched from 1350 °C, but under- interval 1300-1400 °C. The formation at 1100 °C of
went a polymorphic change, or decomposed to some the compound at or near NbzO5. WO3 was confirmed,
other phases that are as yet unidentified, when an- but it did not have the B308 structure that is the pro-
 R. S. R O T H AND A. D. W A D S L E Y 31

genitor of the phases in Table 1, and the examination fore, that systems such as this are called 'solid sol-
was taken no further. A more comprehensive phase utions' in a normal phase-equilibrium diagram.
equilibrium study will be reported in due course by Solid solutions of WO3 in Nb205 would give a defect
R.S.R. structure, either (Nb2-xWx)O5+~ requiring excess oxy-
Discussion gen, or (Nb2_6/szWz vac 1/5z)O5 with vacated metal pos-
itions. Needless to say this is not confirmed by the
The results of the present partial phase equilibrium present study, where each new phase at the most can
diagram together with the relevant parts of the two have only a very small range of homogeneity. Where
previous studies by Goldschmidt (1960) and by Fiegel equilibrium conditions are established, the widely held
et al. (1964) are represented in Fig. 1. The differences concepts of extensive regions of homogeneity in phase
between them may be due to different methods of pre- systems, whether they are called solid solutions or non-
paration. It is important to emphasize that reactions in stoichiometric compounds, may in general be incorrect
the solid state at high temperatures, where one or more (Wadsley, 1964). The attempted prediction of phases
of the components is volatile, must be conducted in a on the other hand was only partly successful, and mod-
sealed system in order to maintain the composition. ifications to the ideas leading up to the present study
The inconsistencies may also be due to different inter- will conclude this present group of papers (Roth &
pretations of the same diffraction data, where we have Wadsley, 1965d).
emphasized the need for the use of single-crystal tech-
niques. The failure on our part to solve the problems We should like to thank J. W. Asker and W. F.
posed by the compound WNb30078 shows that the Donovan for their help with the experimental work,
value of even these techniques is uncertain when the and L. J. Fiegel of the A. O. Smith Corporation, Wis-
crystals themselves are poor. Although likely guesses consin, U.S.A., for a preprint of his phase equilibria
of structure are made readily enough, the initial pro- paper.
blems of assigning the correct unit cell prove embar-
rassingly difficult in cases where twins or intergrown References
crystals of two different structures are formed under ANDERSSON,S. (1964). Acta Chem. Scand. 18, 2339.
the experimental conditions. It is little wonder, there- EVANS,H. T. & MRos~, M. E. (1960). Amer. Min. 45, 1144.

t Nb20s S.S. +'1:3" WO~

/
Goldschmidt (1960)
/ / / wo, \
,'oo°c,/ 1/ i \
Nb20ss.s. +3:2

tN,, tl
1300 ° C
I 3:2+11:3'
Nb20s s.s.
- - LN-b-~-O~.~'~~" - '1:3' '1 : 3' + W03 Fiegel et aL (1964)
/
b2 05s.s. I 1
1000 ° C +1:1 I :1 + ' 1 : 3 '
,,
3:2 1:1

135°°c / J I 1 8:5 1
+9:8 9:8 +'13:24' I
,'', I
,'
I
6: ' 7:3 I ;~
"-- " ?- - ~ -- --? '13:24' ' 4 2 ' Present study
6:1 +7:3 +8:5 I . I, + I

1000,c
I
~j
i
~ I i
IIj
i I I II
15:1 6:1 7:3 8:5 9:8

I I I [ I I I I I I I
0 10 20 30 40 50 60 70 80 90 1 O0
Nb20s Mol o o W03 W03

Fig. 1. Phases reported for the equilibrium system Nb2Os-WO3.

The present study was confined to the temperature interval 1300-1400°C.
32 MULTIPLE PHASE FORMATION IN T H E B I N A R Y S Y S T E M NbO25-WO3. I.

FIEGEL, L. J., MOHANTY, G. P. & HEALY, J. H. (1964). J. ROTH, R. S. & WADSLEY,A. D. (1965b). Acta Cryst. 19, 32.
Chem. Engng Data 9, 365. ROTH, R. S. • WADSLEY,A. D. (1965c). Acta Cryst. 19, 38.
GATEHOUSE, B. M. & WADSLEY, A. D. (1964). Acta Cryst. ROTH, R. S. t~ WADSLEY,A. D. (1965d). Acta Cryst. 19, 42.
17, 1545. ROTS, R. S., WADSLEY,A. D. & ANDERSSON,S. (1965). A c t a
GOLDSCHMIDT, H. J. (1960). Metallurgia, 62, 373. Cryst. 18, 643.
KOVBA, L. M. & TRUNOV, V. K. (1962). Dokl. Akad. Nauk WADSLEY, A. D. (1961a). Acta Cryst. 14, 660.
SSSR, 147, 622. WADSLEY, A. D. (1961b). Acta Cryst. 14, 664.
MAGN~LI, A. (1953). Acta Cryst. 6, 495. WADSLEY, A. D. (1964). In Non-stoichiometric compounds.
MOHANTY, G. P. & FIEGEL, L. J. (1964). Acta Cryst. 17, 454. Ed. L. MANDELCORN.pp. 98-209. New York: Academic
ROTH, R. S. & WADSLEY,A. D. (1965a). Acta Cryst. 18, 724. Press.

Acta Cryst. (1965). 19, 32

Multiple Phase Formation in the Binary System Nb2Os--WO3
II. The Structure of the Monoclinic Phases WNb12033 and WsNb16055
BY R. S. ROTH* AND A. D. WADSLEY
Division o f Mineral Chemistry, C.S.LR.O., Melbourne, Australia
(Received 26 October 1964)
The unit-cell dimensions of WNb12033 are a=22.37, b=3.825, c= 17.87/~, fl= 123"6°, space group
C2. The structure consists of ReO3-type blocks of Nb-O octahedra three wide, four long, and infinite
in the third direction, which are joined by sharing edges, and with tetrahedrally coordinated W atoms
ordered at the junctions of every four blocks. WsNb16055 has the dimensions a=29.79, b=3.820,
c= 23.08/~, fl= 126.5° and the space group C2. It also contains octahedral blocks, but of a different
size, four wide, five long and infinite along the third direction. These blocks are grouped as in WNb12033,
with W occupying similar tetrahedral positions. The structures of both were deduced by trial-and-error,
and refined by two-dimensional Fourier methods.

All four compounds contain an axis of symmetry

Introduction
3.82 A long corresponding to an octahedral body
The preparation and identification of five phases in diagonal, and overlap is therefore limited to the oxygen
part of the equilibrium system Nb2Os---WO3 was re- atoms overlying the metal atoms with this particular
ported in the preceding paper (Roth & Wadsley, coordination. The answers to these questions can be
1965b). Although the ideal structures of four of them readily provided by two-dimensional methods of X-ray
in principle can be solved by geometrical reasoning, it analysis, where the objectives are to establish these
is particularly important to examine each one in some points beyond all reasonable doubt, rather than to
detail, since there are potentially a very large number attempt ultra-refinement which, for compounds of the
of related structures in the same composition range present kind, poses many problems and may have little
which might instead be adopted (see part IV of the or no immediate significance.
present series, Roth & Wadsley, 1965d). The crystal structures of the two monoclinic com-
The ideal structure of a compound of this type con- pounds WNb12033 and WsNb16055 are described in
tains perfectly regular metal-oxygen octahedra joined this paper, and the two tetragonal compounds in part
up by edge- or corner-sharing without irregularities or III (Roth & Wadsley, 1965c).
distortions of any kind. Refinement by crystallogra-
phic techniques must provide the proof, and requires Experimental
at least partial solutions to three problems posed by
this simplification: Both compounds formed very small but well-shaped
(a) The real positions of the atoms. needles, with the crystallographic constants summar-
(b) The way oxygen atoms are coordinated to the ized in Table 1. The hOl and h l l intensity data for both
metals. were recorded with filtered copper radiation on mul-
(c) The distribution of Nb and W over the metal pos- tiple films by the integrating Weissenberg method, and
itions, and in particular whether tetrahedrally co- measured with a standard scale. The crystals were
ordinated atoms are ordered. about 0.1 mm long, but only 5 microns in the average
cross-section, so that absorption Corrections could be
* Permanent address: National Bureau of Standards, Wash- neglected. Scattering curves for Nb 5+ were taken from
ington D.C., U.S.A. the data of Thomas & Umeda (1957), W 6+ from Inter-