A novel characterization of the structure of 'Cossa's salt' Kx(NH4)1-x[PtCl3NH3] · H2O by crystallographic comparison with the stoichiometric compounds K[PtCl3NH3] · H2O and NH4 [PtCl3NH3] · H2O

Article abstract of DOI:10.1016/S0020-1693(96)05532-6

The present understanding of 'Cossa's salt' with two crystalline forms in single crystals, which describes the yellow parts of the crystal as the pure compound K[PtCl₃NH3] · H₂O and the red parts as K[PtCl₃NH₃]_1-ω[PtCl₅NH ₃]_ω (0 < ω < 1) with randomly situated Pt(IV) impurities in the lattice is proved to be erroneous according to the crystallographic evidence given in this paper. A more obvious explanation is that 'Cossa's salt' forms mixed-occupancy crystalline material K_x(NH₄)_1-x[PtCl₃NH₃] · H₂O containing mixed counter ions with different composition of the K⁺ and NH₄⁺ ions. A sample of the 'red' form with x ≈ 0.25 (2) and the 'yellow' form with x ≈ 0.75 (3) are compared with the stoichiometric compounds NH₄[PtCl₃NH₃] · H₂O, (1) published earlier, and K[PtCl₃NH₃] · H₂O (4). The unit cell volumes decrease with increasing concentration of K⁺ ion, being 825.2(3) A³ for (1), 813(1) A³ for (2), 786.6(8) A³ for (3), and 774.2(8) A³ for (4).

Full text of DOI:10.1016/S0020-1693(96)05532-6

Inorganica Chimica Acta 260 (1997) 53–60
A novel characterization of the structure of ‘Cossa’s salt’
Kx(NH4)1yx[PtCl3NH3]PH2O by crystallographic comparison with
the stoichiometric compounds K[PtCl3NH3]PH2O
and NH4[PtCl3NH3]PH2O
Anneli Oksanen
Department of Chemistry, Inorganic Chemistry Laboratory, Box 55, FIN-00014 University of Helsinki, Helsinki, Finland
Received 25 June 1996; accepted 15 October 1996
Abstract
The present understanding of ‘Cossa’s salt’ with two crystalline forms in single crystals, which describes the yellow parts of the crystal as
the pure compound K[PtCl3NH3]PH2O and the red parts as K[PtCl3NH3]1yv[PtCl5NH3]v (0-v-1) with randomly situated Pt(IV)
impurities in the lattice is proved to be erroneous according to the crystallographic evidence given in this paper. A more obvious explanation
is that ‘Cossa’s salt’ forms mixed-occupancy crystalline material Kx(NH4)1yx[PtCl3NH3]PH2O containing mixed counter ions with different
composition of the Kq and NH4q ions. A sample of the ‘red’ form with xf0.25 (2) and the ‘yellow’ form with xf0.75 (3) are compared
with the stoichiometric compounds NH4[PtCl3NH3]PH2O, (1) published earlier, and K[PtCl3NH3]PH2O( 4). The unit cell volumes decrease
3
3
3
3
with increasing concentration of Kq ion, being 825.2(3) A for (1), 813(1) A for (2), 786.6(8) A for (3), and 774.2(8) A for (4).
˚
˚
˚
˚
Keywords: Platinum complexes; Mixed-occupancy crystalline material; Crystal structures
1. Introduction
Also this route often tends to produce mixed crystals [2].
Two crystal structures of the hydrated ‘Cossa’s salt’ 1 have
been published previously [5,6]. The crystal structure of
the ‘yellow’ form based on the molecular formula
K[PtCl3NH3]PH2O has been published in 1970 by Jeannin
and Russell [5]. Studies of the polarized crystal spectra pub-
lished in 1973 by Fanwick and Martin [2] provided evidence
in favour for the purity of the ‘yellow’ form while the ‘red’
form was supposed to contain randomly situated Pt(IV)
impurities. Earlier in 1965 Kukina and Bokii [6] had deter-
mined the structure of ‘Cossa’s salt’ in which one of the
crystal axes was tripled as a result of a superlattice, but the
data were reduced during the calculations to essentially sim-
ilar to those determined by Jeannin and Russell. Later it was
suspected that Kukina and Bokii had actually studiedacrystal
of the ‘red’ form [2]. This present study offers some new
aspects to reconsider the above-mentioned conclusions con-
cerning the purity of the crystals and the character of the
assumed impurities.
The impure crystals of K[PtCl3NH3]PH2O known as
‘Cossa’s salt’ show peculiar yellow/redcolourchangeswhen
rotated under polarized light [1,2]. Attempts to prepare
K[PtCl3NH3]PH2O crystals from aqueous NH4[PtCl3NH3]
solution containing KCl by direct evaporation produces
simultaneously two crystalline forms in single crystals, often
a red hourglass figure on a yellow background. The interface
between the different forms is not always sharp and the inter-
face zone is then formed by stripes of the ‘red’ form running
through the ‘yellow’ form [2]. Historically these forms have
been known as ‘red Cossa’s salt’ and ‘yellow Cossa’s salt’
[1].
A very popular method to isolate ‘Cossa’s salt’ involves a
reagent consuming synthesis route via a sparingly soluble
‘Magnus’ type double salt [Pt(NH3)4][PtCl3NH3]2 [3,4].
2NH [PtCl3NH ]q[Pt(NH ) ]Cl
343 2 4
™[Pt(NH34) ][Pt3Cl 3N2H ] xq2NH Cl (1)
4
[Pt(NH34) ][Pt3Cl 3N2H ] xqK [PtCl ]
2
4
1 Some authors use the name ‘Cossa’s salt’ when referring to the anhy-
drous compound, others when referring to the hydrated forms.
™2K[PtCl33NH ]q[Pt(NH ) ][PtCl ]x (2)
4 3
4
0020-1693/97/$17.00 q 1997 Elsevier Science S.A. All rights reserved
PII S0020-1693(96)05532-6

54
A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
Kraker et al. [7] have introduced an improved method to
prepare analytically pure potassium compound by an anion
exchange method.
K[PtCl3NH3]PH2O( 4). Method I: A fresh solution of
compound 1 prepared from 0.500 g (1.66 mmol) of cis-
[PtCl2(NH3)2] according to Ref. [8] was treated with
0.227 g (0.83 mmol) of [Pt(NH3)4]Cl2 as described in Ref.
[5]. [Pt(NH3)4]Cl2 was purified by recrystallizing several
times from water. The resulting yellow [Pt(NH3)4]-
[PtCl3NH3]2 precipitate was washed carefully several times
first with hot water and then with ethanol. The thorough
washing was essential at this stage to obtain the pure end
product. 0.5 M solution of K2[PtCl4] was added dropwise
with careful warming to the precipitate until only a little of
the yellow precipitate remained unreacted. The product was
obtained as large yellow plates when the resulting mixture
was filtered and the filtrate evaporated slowly to dryness. The
crystals exhibited distinct extinction and blacked out sharply
in 908 intervals when rotated under polarized light with the
cross filters on, a similar behaviour to that of crystal 1.
Method II: After the removal of the unreacted cis-
[PtCl2(NH3)2], a reaction mixture of compound 1 was puri-
fied by an anion exchange method (Dowex 1-X8 resin, 4 M
HCl). The resulting H[PtCl3NH3] solution was treated with
a stoichiometric amount of KCl. The product was obtained
as very thin lime yellow plates by slow evaporation of the
solvent. All of the crystalsappearedtwinnedanddidnotblack
out sharply under polarized light. Thus the crystals were not
suitable for crystallographic studies.
anion exchange
NH [PtCl NH ]3 3
343
™
qKCl
H[PtCl NH ]
™ K[PtCl33NH ]qHCl
(3)
No crystal structure studies of this product have been so far
presented. A more detailed discussion of other synthetic
methods is given in Ref. [8].
Additional reasons for the interest in this compound have
been, for example, to study the Pt–Cl bond lengths with
reference to the trans effect [5], and to study it as a model
compound which displays ligand field excitations where
interionic effects are a minor factor [2]. The compound has
also served as a simple model in single crystal Raman studies
[9]. Spectrophotometric and potentiometric studies of the
kinetics and equilibrium behaviour of its dilute aqueous sys-
tems [3] have been carried out in connection with a study of
the chloride exchange reactions [10].
According to the synthetic work that has been carried out
in our laboratory using impure ‘Cossa’s salt’ the idea of small
amounts of Pt(IV) impurities was not consistent with the
chemical behaviour observed. Therefore the chloride excess
in the lattice cannot be the source, or at least not the only
source, of the impurities. Crystallographic evidence was
sought to reveal theimpurities. Thedifferentcrystallineforms
were studied carefully in order to find out how to produce
pure K[PtCl3NH3]PH2O crystals.
2.2. Chemicals
K2PtCl4 99.99%, Johnson-Matthey
25% NH3 p.a., Merck AG
KCl p.a., Merck AG
KI p.a., Merck AG
2. Experimental
Dowex 1-X8 anion exchange resin, 20-50 Mesh, J. T. Baker
Chemicals B.V., Deventer, Holland.
2.1. Synthesis
NH4[PtCl3NH3]PH2O( 1) was prepared as described in
Ref. [8]. Kx(NH4)1yx[PtCl3NH3]PH2O( 2) and (3) were
prepared from the aqueous solution of 1 by slow evaporation
with a stoichiometric amount of KCl. The crude product was
recrystallized from water. The prismatic crystals were pleo-
chroic with red and yellow colouring and often with a regular
pattern of a red hourglass figure on a yellow background.
When the crystals were rotated under plane polarized light
the red regions appeared and faded away in turns. The other
parts of the crystals remained constantly yellow during the
rotation. When observed under polarized light with the cross
filters on, the different sections retained their colour and
blacked out simultaneously. A set of colour photographs
exhibiting the behaviour of the crystal under polarized light
is presented in Fig. 1. The observation with the polarization
microscope revealed that the refractive index of the red sec-
tions was higher than that of the yellow sections. Samples of
the red section 2 and of the yellow section 3 were cut from a
crystal.
2.3. Structure determinations
The structures were determined by single crystal X-ray
diffraction methods. The data for crystals 2, 3 and 4 were
measured at 193 K using a Rigaku AFC7S diffractometer.
Table 1 contains the crystal data and details of the data col-
lection including the information from the already published
reference crystal 1. The crystals were stable and the standard
reflections showed no significant variation in the intensities
during the data collections. The intensities were correctedfor
absorption through the c-scan technique. The data were cor-
rected also for Lorentz and polarization effects, and disper-
sion. The atomic coordinates of crystal 1 from Ref. [8] were
used as starting parameters in the refinements. To achieve
reliable comparison, identical refinements were carried out
using anisotropic thermal parameters for Pt, Cl, K and O
atoms and isotropic parameters for N atoms. The occupation

A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
55
Fig. 1. A prismatic crystal of Kx(NH4)1yx[PtCl3NH3]PH2O observed under polarized light. (a) and (d) two orientations in plane polarized light. (c) and (b)
two orientations in polarized light with the cross filters on.
factors for K(1) and N(2) atoms were refined for crystals 2,
3 and 4 and the results compared with the results obtained
from the calculations with the occupation factors fixed to
1.00. The data for crystal (2) was refined with the counter
ion assigned both as K(1) and N(2). The influence of the
temperature on the cell dimensions was checked with crystal
(3), and since no significant change was observed, crystal
(1) was not remeasured at reduced temperature. The refine-
ments were carried out with the XTAL3.0 program [11]. The
neutral-atom scattering factors were those included in the
program and the corrections for anomalous dispersion for all
the non-H atoms were taken from the International Tables
for X-ray Crystallography [12]. The positional and thermal
parameters for the crystals (2), (3), and (4) together with
the previously determined crystal (1) are given in Table 2.
The refinementsleave severalrelativelylargeresidualelec-
tron density maxima for every crystal, which is rather normal
for platinum compounds. The maxima are situated mostly

56
A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
Table 1
Summary of crystallographic data for Kx(NH4)1yx[PtCl3NH3]PH2O, 0FxF1
Compound
(1) xs0.00
Ref. [8]
354.61
(2) xf0.25
this work
(3) xf0.75
this work
(4) xs1.00
this work
375.67
Formula weight (g moly1
)
Space group
orth. Cmc21
4444
orth. Cmc21
orth. Cmc21
orth. Cmc21
Z
˚
˚
˚
˚
a (A)
b (A)
c (A)
4.6925(9)
21.583(5)
8.148(1)
825.2(3)
2.86
4.650(5)
21.517(5)
8.125(6)
813(1)
4.585(4)
21.142(3)
8.115(4)
786.6(8)
4.546(3)
20.99(1)
8.114(5)
774.2(8)
3.22
3
V (A )
Density (calc.) (g cmy3
)
m (Mo Ka) (mmy1
)
18.8
640
19.8
672
y3
˚
F(000) (e A
)
Crystal size (mm)
Temperature (K)
l (Mo Ka)(A )
Diffractometer
Monochromator
0.075=0.075=0.05
298
0.71073
Syntex P21
0.15=0.1=0.1
193
0.71073
0.15=0.1=0.1
193
0.71073
0.1=0.1=0.1
193
0.71073
Rigaku AFC7S
˚
Rigaku AFC7S
Rigaku AFC7S
highly oriented graphite crystal
Data collection mode
v
v–2u
v–2u
v–2u
2u limits (8)
3.0–60.0
1.0–10.0
0.7033
63011
669
5.0–60.0
3.0
0.7033
63011
721
5.0–60.0
3.0
0.7033
5.0–60.0
3.0
0.7033
Scan speed (8 miny1
)
y1
˚
(sin u/l)max (A
(h k l)max
Total no. of reflections
)
62911
62610
697
650
No. of reflections used in refinement, I)3s(I)
641
703
680
607
min./max. trans. factors
No. of parameters
0.41/1.00
36
0.80/1.00
37
0.84/1.00
40
0.47/1.00
39
R
Rw
S
0.035
0.036
2.527
0.107
1.1
0.073
0.077
1.681
0.025
4.8
0.054
0.041
1.657
1.608
2.0
0.041
0.049
3.259
0.041
2.8
a
(D/ s)max
Drmax (e A
Drmin (e A
y3
˚
˚
)
)
y3
y1.2
y2.8
y2.3
y2.0
a R s8w(NF NyNF N)2, ws1/s2(F ).
c wo
o
˚
close to the Pt(1) atom (within 0.5–1.3 A distance). A few
maxima lie on the coordination plane close to the Cl atoms
and N(1) atom. The high residual electron density maxima
in crystals 2 and 3 may well be due to the inhomogeneity of
the crystals, bearing in mind how the samples were cut off
from a larger crystal containing both forms. A more accurate
information about the maxima are availableassupplementary
material (S1).
3. Results and discussion
3.1. Crystallographic solutions
All the compounds crystallize in the orthorhombic space
group Cmc21. The coordination around platinum is square-
planar and all of the non-H atoms are located in the xs0 or
xs1/2 mirror planes. The partially occupied axial chloro
ligand of the Pt(IV) impurities, if present, shouldrevealitself
among the residual electron maxima. The residual electron
density maxima are randomly situated in crystals 1, 3 and 4
on such locations where the axial chloride cannot appear. The
only maximum that may deserve some further attention is in
In the data of crystal 2 with the counter ion assigned as Kq
the refinement with fixed occupation factoryieldveryunsym-
metrical thermal parameters for O(1) and K(1) atoms. The
very high values of U11[O(1)], U22[K(1)], and U33[K(1)]
decrease when the occupation factorforK(1) atom isrefined.
When the counter ion is assigned as NH4q the thermal para-
meters are acceptable both with fixed and refined occupation
factors. The refinements yield the occupation factors0.52(5)
for K(1) and 1.45(12) for N(2). The data from crystal 3 can
be refined only when the counter ion is assigned as Kq.
Refining the occupation factor reduces the value of the max-
imum electron density residue and the reasonably high value
of U11[O(1)]. The occupation factor for K(1) is 0.77(2).
In the data of crystal 4 refining the occupation factor of K(1)
gives the value 1.06(2) and the refinements do not affect the
values of the parameters or the temperature factors.
the ‘red’ form, crystal 2, the maximum with 2.6 e Ay3 which
is about one-half of the intensity of the largest maximum near
the platinum atom. It lies quite far (2.72 A) from the over-
lapping platinum atoms on the xs0 and xs1 planes, and it
is located near the Cl(1) atom (distance, 2.49 A) with about
˚
˚
˚
378 deviation from the axial position. It is surrounded by the
K(1) [or N(2)] and O(1) atoms with the closest distances
˚
˚
of 2.64 A and 2.99 A respectively. If there were any chlorine
atoms occupying this position the short distancesfromCl(1),
K(1) [or N(2)], and O(1) atoms would cause severe dis-
tortion of the Cl(1) and N(1) atoms from the coordination
plane and disorder among the O(1) and K(1) [or N(2)]

A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
57
Table 2
2
2
˚
The positional and thermal parameters (=10 A ) for Kx(NH4)1yx[PtCl3NH3]PH2O, (0FxF1). The z-coordinate for the position of the Pt atom has not
been refined. Q denotes the counter ion Kq or NH4q
x
y
z
U11 or Uiso
U22
U33
U23
1, xs0.00
Pt(1)
Cl(1)
Cl(2)
Cl(3)
O(1)
N(1)
QsN(2)
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.33972(3)
0.4476(2)
0.3358(6)
0.3374(7)
0.0740(8)
0.2446(7)
0.046(1)
0.2500
0.244(3)
0.530(2)
y0.031(2)
0.118(2)
0.269(5)
0.457(3)
3.39(3)
6.1(3)
10(1)
8(1)
7(1)
4.4(4)
4.7(6)
3.09(3)
3.6(2)
3.6(7)
7(1)
3.17(3)
3.8(5)
4(1)
2.6(7)
5(1)
0.1(2)
y1.5(6)
2.3(6)
1.5(5)
0.8(9)
4(1)
2, xf0.25
Pt(1)
Cl(1)
Cl(2)
Cl(3)
O(1)
N(1)
QsN(2)
Pt(1)
Cl(1)
Cl(2)
Cl(3)
O(1)
QsK(1)
N(1)
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.33989(5)
0.4473(3)
0.3367(9)
0.3380(7)
0.076(1)
0.246(1)
0.044(1)
0.33989(5)
0.4474(3)
0.337(1)
0.3380(7)
0.076(2)
0.044(1)
0.246(1)
0.2500
0.255(5)
0.535(2)
y0.028(4)
0.116(4)
0.26(2)
2.24(5)
4.2(4)
6(2)
8(2)
3(2)
4.3(6)
4.5(9)
2.24(5)
4.2(4)
8(3)
2.28(5)
2.4(3)
5(1)
1.3(7)
4(1)
2.83(6)
3.1(4)
0.5(5)
7(1)
y0.1(2)
0(1)
y0.5(6)
y0.3(8)
y1(1)
4(2)
0.455(4)
0.2500
2.27(5)
2.4(3)
2(1)
3(1)
3(1)
2.83(6)
3.2(4)
7(2)
0.6(5)
6(2)
0.0(2)
0(1)
y0.2(9)
0.1(6)
y1(2)
0(2)
0.257(5)
0.535(4)
y0.028(2)
0.117(4)
0.454(4)
0.26(2)
6(2)
4(2)
6(2)
4.2(7)
6(2)
9(2)
3, xf0.75
Pt(1)
Cl(1)
Cl(2)
Cl(3)
O(1)
QsK(1)
N(1)
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.34200(4)
0.4513(2)
0.3367(8)
0.3408(6)
0.0755(9)
0.0411(4)
0.2440(8)
0.2500
0.260(2)
0.545(2)
y0.019(2)
0.130(2)
0.462(1)
0.282(3)
2.32(4)
4.1(3)
7(2)
12(2)
4(1)
2.19(4)
2.4(2)
7(1)
2.3(5)
4(1)
3.45(5)
3.1(3)
4.1(8)
2.6(6)
3(1)
0.2(1)
y2.2(5)
4.3(7)
y0.5(4)
3(1)
3.5(4)
2.3(4)
3.5(4)
3.2(4)
0.4(4)
4, xs1.00
Pt(1)
Cl(1)
Cl(2)
Cl(3)
O(1)
QsK(1)
N(1)
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.34333(4)
0.4537(2)
0.3357(7)
0.3443(6)
0.0729(9)
0.0391(3)
0.2475(8)
0.2500
0.260(2)
0.523(3)
y0.036(3)
0.132(2)
0.4645(9)
0.290(3)
1.60(3)
3.0(2)
11(1)
10(1)
5(1)
1.23(3)
1.5(2)
4.4(9)
1.3(4)
3(1)
3.97(6)
2.7(3)
3.2(8)
3.9(8)
3(1)
0.0(2)
0.3(6)
1.1(6)
y0.6(4)
1.5(9)
0.3(3)
3.0(3)
2.2(5)
2.5(3)
2.9(3)
atoms. Thus no evidence of the assumed partial axialchloride
was found to support the formula K[PtCl3NH3]1yv
[PtCl5NH3]v that has been suggested for the ‘red’ form in
Ref. [3].
Some interesting results were obtained when the data of crys-
tals 2 and 3 were refined with the counter ion assigned both
as Kq and as NH4q with and without the refinements of the
occupation factors. The parameters selected from the refine-
ments are collected in Table 3.
-
The alternative solution is based on the similarities of the
K
q and NH4q ions, which are comparable in size. The ionic
In the data of crystal 2 with the counter ion assigned as Kq
the refinement of the occupation factor improves the very
unsymmetrical thermal parameters for the O(1) and K(1)
atoms. When the counter ion is assigned as NH4q refining the
occupation factor for N(2) does not affect the solution sig-
nificantly. The data from crystal 3 can be refined only when
the counter ion is assigned as Kq. Refining the occupation
factor improves the precision of the solution. In the data of
crystal 4 refining the occupation factor of K(1) does not
affect the solution significantly.
q
q
˚
˚
radii are 1.33 A for K and 1.43 A for NH4 [13]. The
synthesis route also supports the formation of mixed-occu-
pancy material. Related replacements are known in platinum
chemistry. For instance chloride impurities can replace the
bridging bromides in [Pt(en)2][Pt(en)2Br2](ClO4)4 [14].
In fact several samples with varying constitution of bromide
and chloride bridges in the chain structure [Pt(en)2]-
[Pt(en)2Cl2yfBr2f](ClO4)4 (0FfF1) have been pre-
pared and studied by solid-state15NNMRspectroscopy[15].

58
A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
Table 3
Some relevant parameters from the refinements of Kx(NH3)1yx[PtCl3NH3]PH2O(0 FxF1) with fixed or refined occupation factors for the counter ions
S
(D/ s)max
Drmax
˚
Drmin
˚
R
Rw
Ueq
y3
y3
(e A
)
(e A
)
[N(2),K(1)]
(=10y3
)
y2
˚
A
(1)
(2)
pop[N(2)]s1.00
2.527
0.1073
1.1
y1.2
0.035
0.036
4.7(6)
pop[N(2)]s1.00
pop[N(2)]s1.45(12)
1.701
1.681
0.02503
0.02456
4.7
4.8
y2.9
y2.8
0.074
0.073
0.078
0.077
2.2(6)
4.5(9)
(2)
(3)
(4)
pop[K(1)]s1.00
1.703
1.688
1.554
0.1508
5.1
4.9
y2.8
y2.8
0.076
0.073
0.078
0.077
17(2)
7(1)
pop[K(1)]s0.52(5)
pop[K(1)]s1.00
pop[K(1)]s0.77(2)
1.800
1.657
1.561
1.608
3.3
2.0
y2.4
y2.3
0.058
0.054
0.045
0.041
5.6(3)
3.4(3)
pop[K(1)]s1.00
pop[K(1)]s1.06(2)
3.259
4.162
0.04059
0.03638
2.8
2.8
y2.0
y2.0
0.041
0.041
0.049
0.049
2.8(2)
3.2(2)
Ueqs(1/3)[U22q1/sin2b(U11qU33q2U13 cos b)]s(1/3)(U22qU11qU33) for K(1) UeqsUiso for N(2).
No disorder between the position of the K(1) atom and
the crystal water was found and any other types of disorder
are ruled out on the basis that the Pt(IV) compounds are
known to have invariably octahedral six-coordination [16].
The relative shares of the counter ions calculated from the
changes in unit cell volumes in Eqs. (7) and (8) share the
same trend with those estimated from the occupation factors
according to Eqs. (4)–(6). The counter ion composition is
about 1/4 Kq and 3/4 NH4q in the ‘red’ form 2 and 3/4 Kq
3.2. Determination of the Kq/NH4q ratio
and 1/4 NH4q in the ‘yellow’ form 3. The unit cell volume
3
˚
determined by Kukina and Bokii is 772 A [6] and it is closer
Those cases in which refinements of the occupation factors
significantly improved some ofthe parameterswithoutaffect-
ing the others and improved the precision of the solution are
regarded here as evidence of partial occupation among the
counter ions. The observation of crystals 1 and 4 with a
polarization microscope revealed a distinct extinction behav-
iour which is typical of pure compounds. These results with
the decrease in the unit cell volume from crystal 1 to crystal
4 indicate that the crystals 1 and 4 are pure compounds while
crystals 2 and 3 are mixed-occupancy material which contain
both NH4q and Kq ions. The formerion isdominantincrystal
2 and the latter in crystal 3. The correlations between the
counter ion (Q) composition and the occupation factors
(Eqs. (4)–(6))) or the growth in unit cell volume (Eqs. (7)
and (8)) have been studied in order to estimate the counter
ion composition of crystals 2 and 3. Both correlations has
been assumed to be linear.
to the volume of the pure compound 4 than the unit cell
volume of 781 A determined by Jeannin and Russell [5].
3
˚
3.3. The environments of the counter ions
The changes on the unit cell volumes are reflected on the
relatively large changes on the lengths of the a- and b-axes
while the c-axis is less affected. In general the packing and
the coordination around platinum are essentially similar. In
the packingofthemoleculespresentedinFig. 2onecaneasily
see how shrinking of the cell along the directions of the a-
and b-axes is possible but repulsion between the Cl(2) and
Cl(3) atoms hinders the trend to tighten the packing along
the direction of the c-axis. The changes are difficult toexplain
on the level of short term interactions between the atoms
because of the large standard deviations. Some selectedinter-
atomic distances and bond angles are given in Table 4.
The most distinct differences are found as expected along
the x- and y-directions at the environments of the counter ions
and around the water molecules. There are significant differ-
ences between the Q∆O(1)i distances, varying from
(2) QsNH4q x=18q(1yx)=7
s1.45=7´xs0.29´Kqs29% (4)
(2) QsKq x=18q(1yx)=7s0.52
=18´xs0.21´Kqs21% (5)
(3) QsKq x=18q(1yx)=7s0.77
=18´xs0.62´Kqs62% (6)
825.2y813
˚
˚
2.91(3) A for 1 to 2.72(2) A for 4 while the Q∆O(1) sep-
arations, which run mostly along the z-direction, are constant
within the standard deviations. The same trend is seen on the
separations between the Cl(1) and Q atoms of which the
Cl(1)∆Qii distances have constant values but the change in
iii
˚
˚
the Cl(1)∆Q distances [3.61(2) A for (1) and 3.31(1) A
for (4)] is highly significant. The decrease in the separation
between the atoms is noticeable also on some of the longer
(2)
s0.239´Kqs24%
s0.757´Kqs76%
(7)
825.2y774.2
distances, e.g. the Cl(1)∆Oiii distances are 3.74(2) A for
825.2y786.6
825.2y774.2
˚
(3)
(8)
ii
˚
(1) and 3.54(2) A for (4) and the Cl(3)∆O distances are

A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
59
Table 4
Selected interatomic distances and bond angles of Kx(NH4)1yx
-
[PtCl3NH3]PH2O
˚
Distances (A)(
1)(
2)(
3)(
4)
xs0.00
xf0.25
xf0.75
xs1.00
Pt(1)–Cl(1)
Pt(1)–Cl(2)
Pt(1)–Cl(3)
Pt(1)–N(1)
Q∆O(1)
2.329(5)
2.28(2)
2.29(2)
2.06(1)
2.82(3)
2.91(3)
3.32(2)
3.61(2)
3.74(2)
3.13(2)
3.44(2)
4.16(2)
2.311(7)
2.32(2)
2.26(3)
2.02(3)
2.84(4)
2.89(4)
3.37(4)
3.53(3)
3.78(3)
3.07(2)
3.44(2)
4.15(2)
2.312(5)
2.39(2)
2.19(1)
2.09(2)
2.79(2)
2.82(2)
3.34(1)
3.40(1)
3.64(2)
3.03(2)
3.39(1)
4.06(2)
2.318(5)
2.22(2)
2.32(2)
2.04(2)
2.79(2)
2.72(2)
3.33(1)
3.31(1)
3.54(2)
3.10(2)
3.34(1)
3.93(2)
Q∆O(1)i
Cl(1)∆Qii
Cl(1)∆Qiii
Cl(1)∆O(1)iii
Cl(2)∆O(1)iv
Cl(3)∆Qii
Cl(3)∆Oii
Bond angles (8)(
1)(
2)(
3)(
4)
Cl(1)–Pt(1)–Cl(2)
Cl(1)–Pt(1)–Cl(3)
Cl(2)–Pt(1)–N(1)
Cl(3)–Pt(1)–N(1)
93.4(7)
90.0(7)
84(1)
91(1)
92(1)
87(5)
90(5)
90.7(6)
92.7(5)
80.1(7)
96.6(7)
92.0(6)
91.6(5)
76.7(7)
99.6(7)
93(1)
i x, yy, zq1/2; ii xq1/2, yyq1/2, zy1/2; iii xq1/2, yq1/2, z; iv xq
1/2, yyq1/2, zq1/2.
Fig. 2. The packing of atoms in Kx(NH4)1yx[PtCl3NH3]PH2Ovieweddown
the x-axis. The positional parameters have been taken from the data of crystal
(4).
contamination, and enables the stoichiometric potassium
compound to be produced in an essentially pure state.
ii
˚
˚
4.16(2) A for (1) and 3.93(2) A for (4). The Cl(2)∆O
distances appear short in all crystals without any significant
variation.
5. Supplementary material
More accurate information including a table of the five
most prominent maxima with the closest distances to the
neighbouring atoms for crystals (1), (2), (3), and (4) (S1),
a more complete table of the selected interatomic distances
(Table S2), and the lists of the structure factors (Table S3)
for crystals (2), (3), and (4) are available from the author
on request.
4. Conclusions
This study has achieved important results in several
aspects. The theoretical studies that have been carried out
with ‘yellow Cossa’s salt’and‘redCossa’ssalt’deservesome
critical reobservation. A new explanation for the peculiar
colour change is needed. A semiquantitative correlation
between the size of the unit cell and the counter ion compo-
sition hasbeen established. TheNH4q counterionisdominant
in the ‘red’ form while the Kq counter ion is dominant in the
‘yellow’ form. The crystals may contain a variable compo-
sition of the counter ions in different parts of the crystal, and
therefore the correlation gives at the best only an averaged
estimation of the composition in the piece of crystal under
observation. The crystals studied here contained around 3/4
NH4q4 and 1/4 Kq in the ‘red’ form and 1/4 NH q and 3/4
Kq in the ‘yellow’ form. Small amounts of the narrow stripes
of the other form among the main product in a crystal may
cause inaccuracy in the structure determination but may not
be detectable microscopically or by crystallographic meth-
ods. The knowledge of the character and source of the impu-
rities involved in ‘Cossa’s salt’ helps to avoid the
Acknowledgements
Doc. PhD Martti Lehtinen is acknowledged for taking the
colour photographs of the microscopy imagines.
References
[1] M.S. Jo¨rgensen, Z. Anorg. Chem., 24 (1900) 153.
[2] P.E. Fanwick and D.S. Martin, Jr., Inorg. Chem., 12 (1973) 24.
[3] T.S. Elleman, J.W. Reishus and D.S. Martin, Jr., J. Am. Chem. Soc.,
80 (1958) 536.
[4] A.A. Grinberg and Yu.N. Kukushkin, J. Russ. Inorg. Chem., (1958)
59.
[5] Y.P. Jeannin and D.R. Russell, Inorg. Chem., 9 (1970) 778.

60
[6]
A. Oksanen / Inorganica Chimica Acta 260 (1997) 53–60
[12] International Tables for X-ray Crystallography, Vol. IV, Kynoch,
G.A. Kukina and G.B. Bokii, Zh. Strukt. Khim., 6 (1965) 233.
A.J. Kraker, J.D. Hoeschele, W.L. Elliot, H.D.H. Showalter, A.D.
Sercel and N.P. Farrell, J. Med. Chem., 35 (1992) 4526.
A. Oksanen and M. Leskela¨, Acta Chem. Scand., 48 (1994) 485.
D.M. Adams and R.E. Christopher, J. Chem. Soc. Dalton Trans.,
(1973) 2298.
Birmingham, UK (present distributor: Kluwer, Dordrecht,
Netherlands).
[13] J.G. Stark and H.G. Wallace (eds.), Chemistry Data Book, 2nd edition
in SI, John Murray, London, UK, 1982, p. 28.
[7]
[8]
[9]
S.C. Huckett, R.J. Donohoe, L.A. Worl, A.D.F. Bulou, C.J. Burns, J.R.
Laia, D. Carroll and B.I. Swanson, Chem. Mater., 3 (1991) 123.
E.J.W. Austin, P.J. Barrie and R.J.H. Clark, Inorg. Chem., 34 (1995)
3859.
[14]
[15]
[16]
[10]
[11]
T.S. Elleman, J.W. Reishuis and D.S. Martin, Jr., J. Am. Chem. Soc.,
81 (1959) 10.
S.R. Hall and J.M. Stewart (eds.), XTAL3.0, Reference Manual,
Universities of Western Australia and Maryland, 1990.
F.A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, Wiley,
New York, 5th edn., 1988, p. 927.