Influence of spectator ions on the reactivity of divalent metal salts in molten alkali metal nitrates Morphology of the resulting metal oxides

Article abstract of DOI:10.1016/j.materresbull.2004.10.001

This work focuses on the investigation of the reaction of alkali metal nitrates (LiNO₃, NaNO₃ and KNO₃) with divalent metal salts (Cu²⁺, Ni²⁺ and Zn²⁺). Thermogravimetric analysis (TGA) was employed to study the kinetics and mechanisms of the above reactions, which led to the formation of the corresponding metal oxides. The reaction temperature was found to depend not only on the alkali metal but also on the metal salt (MCl₂ or M(NO₃)₂) involved in the reaction. SEM observations show that the spectator ions present in the reacting medium have varying degrees of influence on the morphology of the powder; the growth directions, sizes and the homogeneity of their distribution, are modified. KNO₃ generates the most significant differences compared to LiNO₃ and NaNO₃.

Full text of DOI:10.1016/j.materresbull.2004.10.001

Materials Research Bulletin 40 (2005) 1–11
www.elsevier.com/locate/matresbu
Influence of spectator ions on the reactivity of divalent
metal salts in molten alkali metal nitrates
Morphology of the resulting metal oxides
a,
P. Marote , C. Matei , C. Sigala , J.P. Deloume
b
c
d
*
a
´
UMR Sciences Analytiques 5180, Universite Claude Bernard Lyon 1, CPE Bat
308, 43 Bd du 11 novembre 1918, 69122 Villeurbanne Cedex, France
bDepartment of Inorganic Chemistry, Politehnica University of Bucharest,
1 Polizu Street, 78126 Bucharest Romania, P.O. Box 12-205, France
c
´ ´
Laboratoire des Multimateriaux et Interfaces, UMR 5615, Universite Claude Bernard Lyon 1, France
d
´
LACE, UMR 5634, Universite Claude Bernard Lyon 1, France
Received 10 December 2002; received in revised form 27 September 2004; accepted 5 October 2004
Abstract
This work focuses on the investigation of the reaction of alkali metal nitrates (LiNO3, NaNO3 and KNO3) with
divalent metal salts (Cu2+, Ni2+ and Zn2+). Thermogravimetric analysis (TGA) was employed to study the kinetics
and mechanisms of the above reactions, which led to the formation of the corresponding metal oxides. The reaction
temperature was found to depend not only on the alkali metal but also on the metal salt (MCl2 or M(NO3)2) involved
in the reaction.
SEM observations show that the spectator ions present in the reacting medium have varying degrees of influence
on the morphology of the powder; the growth directions, sizes and the homogeneity of their distribution, are
modified. KNO3 generates the most significant differences compared to LiNO3 and NaNO3.
# 2004 Elsevier Ltd. All rights reserved.
Keywords: A. Oxides; B. Crystal growth; C. Thermogravimetric analysis (TGA); C. Electron microscopy
* Corresponding author. Tel.: +33 04 72 43 1001; fax: +33 04 72 44 8319.
E-mail address: marote@univ-lyon1.fr (P. Marote).
0025-5408/$ – see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2004.10.001

2
P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
1. Introduction
Molten alkali metal nitrates (AMN) have received much attention during the last decade owing to
their potential application as solvents and reaction media [1]. Numerous studies have been carried
out using the so-called nitrate eutectics (1NaNO3/1KNO3) [2,3]. The main advantages of these
mixtures include their low melting temperature and their practical interest for alkali nitrate reactions.
Despite these reports, the effect of the alkali metal ion requires more careful consideration. Indeed, it
is well known that the choice of the alkali metal ion can influence the melting point of the AMN.
The choice of the metal ion (Mn+) can also modify the Lux-Flood acido-basic equilibrium Eq. (1)
[4,5].
1
2ANO3 @ 2Aþ þ 2NO2 þ O2 þ O2À
(1)
2
According to Eq. (2) the oxide ion, O2À, readily leads to the formation of metal oxides and therefore
Mn+ has to be considered as a Lux-Flood acid.
n
M
nþ þ O2À ! MOn=2
(2)
2
Clearly, a better understanding of the nature of the metal ion effect on the final product is essential. The
metal ion acidity depends on the formal charge and also influences the basicity of the molten salts. Since
tetravalent metals [6–8] are considered as strong acids, the difference of basicity between the three
AMNs disappears [9]. Therefore, it would be necessary to use metal ions having an acidity of the same
order of magnitude as the basicity of AMN to reveal their differences. Ni, Cu and Zn nitrates and
chlorides were chosen because of the two plus (2+) charge of the cations, and also because they are not
able to undergo oxidation in the nitrate medium. Zinc is often compared to the alkaline-earth metals
(weak acids); therefore the Zn2+ ion will be weakly acid. Nickel and copper ions will therefore also be
weak to medium acids.
The aim of this work is to put in light both the influence of the alkali metal ion and the role of the
spectator anion of the metal precursor on the morphology of the oxides obtained by reaction in the molten
AMN.
2. Experimental
2.1. Materials
The metal precursors CuCl2ꢀ2H2O (Merck, 99%); Cu(NO3)2ꢀ3H2O (Merck, 99%); NiCl2ꢀ6H2O
(Prolabo, 98%); Ni(NO3)2ꢀ6H2O (Merck, 99%); ZnCl2 (Prolabo, 98%); Zn(NO3)2ꢀ4H2O (Merck,
98.5%) were used without further purification. The alkali metal nitrates LiNO3 (Aldrich 98%), NaNO3
(Prolabo P.A), and KNO3 (Prolabo P.A) were dried at 120 8C and stored in an oven at 100 8C before being
used.

P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
3
2.2. Procedure
The reaction for the formation of a metal oxide starting from an anhydrous metal salt is as follows [1,4]:
n
excess ANO3
MXn þ nANO3 À! MOn=2 þ nAX þ nNO2 þ O2
(3)
2
Neither A+ nor XÀ take a part in the reaction and so they could be omitted (from aÀstrictly chemical
point of view). The gaseous compounds evolve from the medium while AX (X = NO3 , ClÀ) as well as
the excess of ANO3, are soluble in water. The metal oxide is isolated by filtration. The AMNs have been
used in excess to provide the liquid phase throughout the experiment. The ratio AMN/metal precursor
was 8/1 [10,11]. The progress of the reaction was monitored by following the evolution of mass loss: the
gases generated by the reaction. This was performed with a modified Setaram B70 thermobalance with a
heating rate of 150 8C/h from room temperature up to 500 8C. The mass of precursor involved
corresponded to about 1.5 ꢁ 10À4 mol.
The metal oxide used for the characterizations was obtained by reaction in test tubes. About two grams
of metal oxide were prepared with the following thermal schedule: ramp from room temperature to
100 8C: 1 h; level out at 100 8C: 1 h; ramp to 450 8C: 3 h and level at 450 8C: 2 h; cooling at the cooling
rate of the furnace.
The characterization was performed by X-ray diffraction (XRD) using a D500 Siemens diffractometer
´
located at the Center of Diffraction Henri Lonchambon, Universite Lyon1. Microscopy was carried out
using a Hitachi S800 Scanning Electronic Microscope at the Centre des Microstrctures, Universite
´
Lyon1. The chemical analyses were performed at the Central Service of Analysis, C.N.R.S, in Vernaison,
France, using standardized techniques.
3. Results
3.1. Characterizations
The oxides obtained have been characterized by X-ray-diffraction and chemical analysis. For reasons
of clarity we will write the origin of the oxide in the form of MO[X–A] where X is the anion of the metal
precursor and A is the alkali metal of the nitrate used in the reaction. Therefore, the notation NiO [Cl–K]
means that the NiO is obtained by the reaction of NiCl2 with KNO3.
3.2. X-ray diffraction
For each metal oxide obtained, XRD shows one single phase in agreement with ICDD files 44-1159,
45-0937 and 326-1451 for NiO, CuO and ZnO, respectively. The systematic addition of silicon as the
angular position standard allows us to conclude that there is no shift in the d spacings and consequently
that no solid solution is formed with the alkali metal ion.
The relative intensities fit the ICDD files for NiO and CuO but not for ZnO. The differences and
changes are resumed in Table 1.
The figures in Table 1 show that the alkali metal and the anion play a role in the crystal growth. In
addition, the narrow peak widths prove the high degree of crystallisation and the large size of the crystals.

4
P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
Table 1
Comparison of the relative intensities of the three most intense peaks of the six zinc oxides prepared
Relative intensity of hkl
1 0 0
0 0 2
1 0 1
ICDD ZnO
57
100
100
100
9
44
8
100
29
38
12
27
67
53
ZnO [Cl–Li]
ZnO [Cl–Na]
ZnO [Cl–K]
ZnO [NO3–Li]
ZnO [NO3–Na
ZnO [NO3–K]
15
34
100
100
100
78
95
3.3. Chemical analyses
The results of the chemical analyses of the transition metal oxide products obtained for impurity ions
are shown in Table 2. We could observe that the amounts of impurities in all the oxides (Table 2) are
small. The atomic alkali metal content varies from a maximum of 4% for Ni(NO3)2–LiNO3 and a
minimum of 0.4% for Ni(NO3)2–KNO3. However, the zinc oxide obtained from the ZnCl2 salt, shows a
greater impurity content. This difference is evident in the high value of the chlorine fraction present in
sodium nitrate. The potassium and chloride content is greater in ZnO than in CuO and NiO, for the oxides
obtained in KNO3. The same observation was made in the two other media, however the differences
between the impurity contents in NiO and in CuO are smaller.
3.4. Thermal stability of the precursors
We systematically draw the loss of mass in g molÀ1 of transition metal ion. This allows direct
comparisons on the reactivity of the different salts whatever their molar weight. The thermogravimetric
(TG) curves of the thermal decomposition of the six precursors are similar to those described in the
literature [11,19]. That is, among the metal nitrates, only Zn(NO3)2ꢀ4H2O dehydrates immediately at
about 270 8C [11–13]. The two others, Cu and Ni, form basic nitrates [11–15], weakly stable at 200 and
240 8C for copper (Fig. 1) and nickel, respectively. The next mass loss leads to the metal oxide, at about
Table 2
Chemical analyses of the impurities in the oxides of Ni, Cu and Zn prepared from different precursors in the different molten
media (at%)
Precursors
Product
Medium
LiNO3
Li
NaNO3
Na
KNO3
K
N
Cl
N
Cl
N
Cl
NiCl2ꢀ6 H2O
Ni(NO3)2ꢀ6 H2O
CuCl2ꢀ2 H2O
Cu(NO3)2ꢀ3 H2O
ZnCl2
NiO
NiO
CuO
CuO
ZnO
ZnO
1.18
3.98
0.21
0.28
1.18
1.18
0.53
0.69
0.56
0.56
0.58
0.58
0.85
0.52
0.55
0.55
0.2
0.53
1.23
0.56
0.56
0.9
1.23
0.011
0.003
0.016
0.17
0.53
0.53
0.56
0.56
0.58
0.58
0.61
0.45
0.92
0.63
8.01
0.47
0.92
0.73
0.53
0.52
0.42
Zn(NO3)2ꢀ4 H2O
0.58

P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
5
Fig. 1. TGA of Cu(NO3)2ꢀ3H2O,
in LiNO3,
in NaNO3,
in KNO3,
precursor alone.
300 8C, with a yield of 100%. Zinc chloride is anhydrous and stable up to 450 8C. CuCl2ꢀ2H2O undergoes
a one step dehydration (Fig. 2) ending at 120 8C [16,17]. NiCl2ꢀ6H2O is dehydrated in two steps; the first
one (110 8C) leads to NiCl2ꢀ2H2O and the second produces NiO (250 8C) in agreement with the literature
[18,19].
Fig. 2. TGA of CuCl2ꢀ2H2O,
in LiNO3,
in NaNO3,
in KNO3,
precursor alone

6
P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
3.5. Reactivity in the molten salts
All the metal precursors except ZnCl2 react in the media with a 100% yield according to the overall,
mass loss determined from the general equation:
1
excess ANO3
MX2 ꢀ xH2O þ 2ANO3 À! MO þ 2AX þ 2NO2 þ O2 þ xH2O
(4)
2
The mass of gas phase produced in the reaction (4) is Dm = 18 x + 108 g molÀ1M2+.
We have observed three kinds of TGA curves: the first concerns the metal nitrates; the second, the
copper and nickel chlorides; and the third, zinc chloride.
3.5.1. Reactivity of the metal nitrates
The reaction of the metal nitrates generally follows the curves shown in the Fig. 1 which represent the
reaction of Cu(NO3)2ꢀ3H2O. We observe a modification of the dehydration step, mainly in KNO3. This
salt (the most ionic and the least basic) prevents the formation of the basic nitrate and instead results in the
anhydrous Cu(NO3)2 which reacts at about 300 8C to give CuO. In the other AMN, the TG curve
corresponds, for the temperatures below 250 8C, to the decomposition curve of the precursor. The
differences observed between 200 and 250 8C could correspond to the stabilization of the intermediate
species [3Cu(NO3)2ꢀCu(OH)2ꢀH2O] by the nitrates. This has been already described in the literature [11–
15]. The formation of this intermediate requires the release of ten H2O and two HNO3 which corresponds
to Dm = 306/4 = 76.5 g molÀ1 in agreement with the observed mass loss (Fig. 1, NaNO3). The
stabilization of this species causes an increase in the temperature of formation of the metal oxide
(Table 3).
Dehydration of nickel nitrate and zinc nitrate both lead to similar mass loss, whatever the AMN (ca.
75 g molÀ1 for nickel and 108 g molÀ1 for the zinc salt). Therefore, the oxide formation curves look very
similar too. The temperature of the maximum rate of formation of the metal oxide increases according to
the AMN in the order Li, Na and K (Table 3).
Copper nitrate appears to be the least stable of the three metal nitrates whereas zinc nitrate reacts at the
highest temperature, which confirms its trend to behave as an alkaline earth metal. It is worth noting that
the reaction goes to completion with this zinc salt (vide infra zinc chloride).
3.5.2. Reaction of copper and nickel chlorides
The copper and nickel chlorides (di- and hexahydrate, respectively) undergo dehydration to
the anhydrous chlorides, which are stable until relatively high temperatures. Fig. 2 shows that CuCl2
begins to be oxidized by the ambient oxygen at about 400 8C. For this reason the oxide formation
Table 3
Temperature of the maximum rate of reaction of metal oxide formation starting from a metal nitrate in the alkali metal nitrates
Maximum rate temperature (8C)
Cu(NO3)2ꢀ3H2O
Ni(NO3)2ꢀ6H2O
Zn(NO3)2ꢀ4H2O
Precursor only
269
350
376
392
304
354
388
423
323
366
418
441
Precursor in LiNO3
Precursor in NaNO3
Precursor in KNO3

P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
7
Table 4
Temperature of the maximum rate of the reaction of formation of a metal oxide starting from a metal chloride in the alkali metal
nitrates
Maximum rate temperature (8C)
CuCl2
NiCl2
In LiNO3
In NaNO3
In KNO3
390
368
380
418
418
439
TG curves start from a well-defined plateau (from 120 8C to about 250 8C) and over a narrow
temperature range. The maximum rate temperatures remain similar as shown in Table 4. Moreover,
the temperature domain is comparable to the reaction temperature of the respective transition metal
nitrate in KNO3.
3.5.3. Reaction of zinc chloride
Zinc chloride is particularly sensitive to the reaction medium. In every case, the TG curve reaches a
plateau; the oxide is not totally formed as the thermal decomposition of the AMN begins to affect the
reaction. At 500 8C the mass losses are 138, 109 and 51 g molÀ1 in Li, Na and KNO3, respectively.
Consequently, the reaction yields (450 8C for 2 h) decrease in the same order: 98, 85 and 70%. These
results show the differences in reactivity due to the medium and the metal salt.
In the case of the metal nitrates, the reactive species are present from the beginning. When
the precursor is heated alone, it cannot transfer the thermal energy to other entities, therefore
the cation and the nitrate ions are directly involved, and so the reaction temperature is low.
The presence of the AMN could account for an energy radiator, however the thermal conduc-
tivities of the AMN decrease from Li to KNO3 [20] and the order of the reaction temperatures
should be the inverse of the observed one. Complexes can be formed with the nitrate ions of the AMN
as shown by Liu et al. [21] and Eweka et al. [2]. These complexes are all the more stable since the
mediumislessbasicandthenitrateionmoresymmetric;thetemperatureofreactionincreasesfromLito
KNO3.
When the metal chlorides are involved, the nitrate ions have to enter the coordination sphere before
reacting. As the stability of the chloro species depends only on the metal and chloride ions the energy
required is approximately the same whatever the alkali metal. The reaction temperatures systematically
increase, eventually they are close to that for reaction of the metal nitrate in KNO3. The uncompleted
reaction of ZnCl2 related to the total one of Zn(NO3)2 helps this assumption.
3.6. Scanning electron microscopy
This technique was used to demonstrate relationships occurring between the reactivity and the
crystallite morphologies. The space groups of the oxides are different: rhombohedral with a rock salt
structure for NiO, monoclinic with a distorted rock salt structure for CuO and hexagonal with a wu¨rtzite
structure for ZnO. The comparisons are only valid for a single metal with respect to one precursor and one
medium; we choose to begin with the lowest symmetry oxide.
Fig. 3 presents the micrographs obtained with CuO. CuO[Cl] forms systematically agglomerates of
about 20 mm. The crystallites show different shapes according to the medium, and an irregular growth

8
P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
Fig. 3. (a) CuO obtain using CuCl2 in LiNO3, (b) CuO obtain using CuCl2 in NaNO3, (c) CuO obtain using Cu(NO3)2 in LiNO3
and (d) CuO obtain using Cu(NO3)2 in KNO3.
characterizes the reaction in LiNO3 (Fig. 3a). In Na and KNO3 the crystallites are elongated
(1 mm2 ꢁ 5 mm, Fig. 3b), ore over, in relation to this kind of growth, large X cross-shaped crystals
(30 mm) can be observed. CuO[Cl, K] differs from CuO[Cl, Na] by the presence of very small crystallites
(20 nm) on the surface of the larger ones. In comparison to this, the aggregates formed by CuO[NO3]
look different, probably because the directions of growth are not similar to the previous ones. In LiNO3
(Fig. 3c), the size distribution is quite monomodal, whereas, in both other AMNs, the powders are
mixtures of polycrystalline (20 mm) aggregates and small (50nm) crystallites (Fig. 3d). To summarise
therefore, the crystal shape of CuO depends on the precursor, and the size and aggregation depend on the
medium.

P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
9
Fig. 4. (a) NiO obtain using NiCl2 in LiNO3 and (b) NiO obtain using Ni(NO3)2 in KNO3.
The micrographs of NiO seem on the whole rather similar. NiO[Cl,Li] and NiO[Cl,Na] are
comparable. The sizes of the agglomerates are different, but they are constituted from spheroid
crystallites (Fig. 4a) whose varies from 500 nm to 2.5 mm. In NiO[Cl,K] the shapes become more
angular, cubic shaped aggregates appear. The examination of NiO[NO3, Li] or NiO[NO3,Na] shows that
the upper limit of the crystal sizes is closer to 1 mm than to 2.5 mm as in the previous oxides. The shapes
are less spheroid and more octahedral. NiO[NO3,K] is different as can be observed in Fig. 4b: large and
compact agglomerates of cubes of 1 to 2 mm as well as the same kind of agglomerates of cubes 0.5 mm in
size. For this oxide the most significant feature is the difference, which appears when the medium is
KNO3 whatever the precursor.
Most of the micrographs of ZnO show particles of large size, isolated from each other. The
solids observed correspond to different crystal growth. According to the structure, the unit part
of these building is a thin hexagonal crystallite as shown on Fig. 5a. The nature of the
precursor influences growth and agglomeration of the crystallites. With ZnCl2 the trend is a
development along the c axis, as shown partially on Fig. 5a. Several kinds of solids are observed,
most of them with a central hole. The most complete ones obtained in NaNO3 (Fig. 5b), look like a
smooth hexagonal tube. Once more, in KNO3 the trend vanishes and a randomisation of the shapes
appears. The ZnO[NO3] powders have different crystal shapes in LiNO3 and KNO3. In the first one, the
growth and agglomeration lead to large hexagonal plates (>20 mm) like ornaments as shown on Fig. 5c.
However, in NaNO3, the plates are hexagonally drilled and needles appear. Finally in KNO3, only
needles are observed, and they are moreover covered with very small crystallites (Fig. 5d). These
differences of growth explain the differences observed in the relative intensities of the X-ray
diffractograms.
Among the three metal oxides, ZnO[NO3] shows the most significant changes according to the
precursor and the liquid medium, and in general in the case of zinc, we characterize one powder for one
medium and one precursor.

10
P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
Fig. 5. (a) ZnO obtain using ZnCl2 in LiNO3, (b) ZnO obtain using ZnCl2 in NaNO3, (c) ZnO obtain using Zn(NO3)2 in LiNO3
and (d) ZnO obtain using Zn(NO3)2 in KNO3.
4. Conclusion
TG analyses have shown the differences in the reactivity of the metal precursors in the AMN, mainly in
the temperature of formation of the oxide. The SEM studies reveal that the crystallites and the
agglomerates are different when the precursor is the metal nitrate or the metal chloride. The reacting
media also influence the morphology of the oxide powders. On the whole, molten KNO3 leads to the most
significant modifications in crystal shape, size and their distribution. This feature cannot be attributed to
the temperature of reaction, which is higher in KNO3 than in the other AMNs. This high temperature,
observed with the metal nitrates, is typical of the reaction of metal chlorides whatever the alkali metal
nitrate.

P. Marote et al. / Materials Research Bulletin 40 (2005) 1–11
11
Overall, these results show that when using molten salts synthesis for metal oxides for catalytic or
sintering purposes, it is worth taking into account the initial experimental conditions. Some ions not
specifically involved in the chemical process can have a direct influence on the morphology of the
powder.
References
[1] D.H. Kerridge, in: J.J. Lagowski (Ed.), The Chemistry of Non-Aqueous Solvents, vol. VB, Academic Press, New York,
1978(Chapter 5).
[2] E.I. Eweka, D.H. Kerridge, Thermochim. Acta 290 (1996) 133–138.
[3] N.B. Singh, S.P. Pandey, Thermochem. Acta 67 (1983) 147–156.
[4] D.H. Kerridge, J. Cancela, Rey, J. Inorg. Nucl. Chem. 37 (1975) 975–978.
[5] D.H. Kerridge, S.A. Tariq, Inorg. Chem. Acta 4 (4) (1970).
[6] J.P. Deloume, J.P. Scharff, P. Marote, B. Durand, A. Abou-Jalil, J. Mater. Chem. 9 (1999) 107–110.
[7] M. Jebrouni, B. Durand, M. Roubin, Ann. Chim. Fr. 16 (1991) 569–579.
[8] A. Benamira, J.P. Deloume, B. Durand, J. Mater. Chem. 9 (1999) 2656–2662.
[9] K.H. Stern, J. Phys. Ref. Data 1 (1972) 747–772.
[10] P. Marote, B. Durand, J.P. Deloume, J. Solid Stat. Chem. 163 (2002).
´
[11] P. Marote, Thesis in Chemistry, Universite Claude Bernard Lyon1, no. 161/2001 (2001).
[12] S. Gordon, C. Campbell, Anal. Chem. 27 (1955) 1102.
[13] W.W. Wendlandt, Tex. J. Sci. 10 (1958) 392.
[14] M. Le Van, Bull. Soc. Chim. Fr. 3 (1964) 545.
[15] J. Labat, Ann. Chim. 9 (7/8) (1964) 399.
[16] D. Weigel, B. Imelik, P. Lafitte, Bull. Soc. Chim. Fr. 544 (1962).
[17] P. Lumme, J. Peltonen, Suomen Kemistilehti 37B (9) (1964) 162.
[18] G. Le Van My, P. Perinet, Bianco, J. Chim. Phys. 63 (5) (1966) 719.
[19] R. Perrot, Bull. Soc. Chim. Fr. 2 (1966) 755.
[20] J. McDonald, H.T. Davis, J. Phys. Chem. 74 (4) (1970) 725–730.
[21] C.H. Liu, J. Hasson, G.P. Smith, Inorg. Chem. 7 (11) (1968) 2244–2247.