Inorg. Chem.

1983, 22, 3323-3326 3323

oxygen atom transfer may then occur via either a nitrito scrambling of the l80and l 6 0 isotopes in C 0 2 was observed.
(Scheme 11) or a nitro isomer (Scheme I). The oxygen atom Acknowledgment. The authors gratefully acknowledge the
transfer step would appear to be facilitated by the larger ring financial support of the National Science Foundation, the
formed by a nitrito ligand in the transition state in Scheme University of Arizona, and NATO for this research. We also
11, while thermodynamics would favor the direct formation thank Drs. M. Dartiguenave and G. Elbaze for many helpful
of the N-bonded isomer of the nitrosyl ligand in Scheme I. discussions and the Colorado State Regional NMR Center
At the present time, there is insufficient experimental evidence funded by National Science Foundation Grant No. CHE
available to distinguish between these two possible reaction 78-18581 for furnishing some of the 31PNMR spectra reported
routes. In spite of the uncertainty regarding which of the two herein.
NO2 linkage isomers is responsible for the oxygen atom
transfer to CO, it is clear that the reaction proceeds via the Registry No. Ni(N02),(PMe3),, 20663-81-4; Ni(N02)2(PEt3)2,
formation of a five-coordinate carbonyl intermediate. 86941-01-7; Ni(NOz)2(PMe2Ph)2,25456-46-6; Ni(N02),(PMePh2),,
86886-05-7; Ni(N02)2(PCy3)2,86886-06-8; Ni(N02)z(P-n-Bu3)2,
Moreover, once the oxygen atom is transferred from the NO2 24510-66-5; Ni(N02)2dppe, 71604-08-5; Ni(NO&(cis-vpp),
ligand to the CO ligand, COz is lost before it can exchange 86886-07-9; Ni(N02),dppp, 86886-08-0; Ni(NO,)(NO)(PMe,),,
with the resulting N O and NOz ligands. C 0 2 and Ni(N- 79499-32-4; Ni(NO,)(NO)(PEt,),, 86886-09-1; Ni(N02)(NO)-
02)(NO)L2are produced quantitatively, but at rates that are (PMe2Ph),, 86886-10-4; Ni(NO,)(NO)dppe, 86886-1 1-5; Ni-
dependent upon L. Our studies also show that l80is not (N02)(NO)(cis-vpp), 86886-12-6;Ni(NO,)(NO)dppp, 86886-13-7;
incorporated into the nickel products, even though some Ni(N02)(NO)(PCy3)2,86886-14-8; CO, 630-08-0.

Contribution from the Department of Chemistry,

Carleton College, Northfield, Minnesota 55057

Copper(I1) Chloride Complex Equilibrium Constants

RICHARD W. RAMETTE* and GRACE FAN
Received March 17, 1983
The solubility of recrystallized copper iodate hydrate was determined, by using high-accuracy controlled-potential coulometry,
in solutions of copper perchlorate and in solutions of sodium chloride, all at ionic strength 5 with sodium perchlorate as
the inert electrolyte. The association constant for the CuI03+species and the solubility product for copper iodate were
determined at 25 and 35 OC. From the effects of chloride ion on the solubility at 25 OC, the four successive j3 values for
copper chloride complexes were determined and compared with results of spectrophotometric measurements. Revised enthalpy
changes for the complexation steps were calculated.

The thermodynamic characterization of the reactions that for the equilibrium constants and for the molar absorptivities.
occur when chloride ion is added to a copper(I1) perchlorate Bjerrum and Skibsted4 suggested reinterpretation of
solution is an inorganic chemistry problem of long standing. Schwing-Weill's data on the basis of data in very concentrated
The color changes gradually from blue to green with increased chloride media. Ashurst and Hancock5 used their own data
absorption both in the ultraviolet (250-275 nm) and at the under selected conditions to obtain another set of equilibrium
high end of the visible spectrum (700-800 nm). Absorption constants. The lack of agreement in these studies gives support
spectra of the solutions have been repeatedly examined with to the closing paragraph of Bjerrum and Skibsted:
the goal of determining the equilibrium constants for the "The conclusion of this paper is that attempts to obtain
formation of the successive complexes quantitative results in studies of consecutive weak complex
+
M j L = MLj formation in solution necessitate the use of many approxi-
mations, and studies of such systems can therefore be expected
where j = 1-4 and M and L represent the tetraaquocopper(I1) to give only semiquantitative results."
ion and the chloride ion, respectively. The system may be Given the inherent problem of the spectrophotometric ap-
described in terms of /3 values, which are overall equilibrium proach to this system, namely that the data must be treated
constants Pj = [MLj]/[M] [L]', or in terms of the individual with both the equilibrium constants and the molar absorp-
equilibrium constants for each step Qj = [MLj]/[MLr,] [L]. tivities as unknown parameters, it seemed very desirable to
A recent paper by Arnek et al.' presents a calorimetric study study the copper-chloride system by using a method com-
of the system and includes a useful summary of previous pletely independent of spectrophotometry, i.e. by measuring
equilibrium research under a variety of conditions of ionic the effect of varying chloride concentration on the solubility
strength. of copper(I1) iodate. Accurate equilibrium constants are also
Three studies of this system have been reported for an ionic required for the interpretation of the calorimetric study,' which
strength of 5 , permitting the high chloride concentrations also was performed at ionic strength 5.
needed for appreciable formation of the higher complexes. The theoretical and computational basis for the use of
Schwing-WeilP3 obtained a comprehensive set of ultravio- solubility of a metal iodate as a chemical probe for finding
let-visible spectra for a large number of copper-chloride metal-ligand complex formation constants has been described
mixtures and applied least-squares analysis to estimate values for an analogous study of cadmium bromide complexes.6 The

(1) Arnek, R.; F'uigdomenech, I.; Valiente, M. Acta Chem. Scand., Ser. A (4) Bjerrum, J.; Skibsted,L. H.Actu Chem. Scand., Ser. A 1977, A31,
1982, A36, 15-9. 673-7.
(2) Schwing-Weill, M. J. Bull. SOC.Chim. Fr. 1973, 823-30. (5) Ashurst, K. G.; Hancock, R. D.J . Chem. Soc., Dalton Trans. 1981,
(3) Khan, M. A,; Schwing-Weill, M. J. Inorg. Chem. 1976, 15, 2202-5. 245-50.

0020-1669/83/1322-3323$01.50/0 0 1983 American Chemical Society

3324 Inorganic Chemistry, Vol, 22, No. 22, 1983 Ramette and Fan

present work is concerned with the effect of varying sodium Table I. Solubility and Thermodynamic Quantities for Copper
chloride concentration on the solubility, S , of copper iodate. Iodate in Copper Perchlorate
T h e key relationships a r e 25 "C 35 "C

MA2(s) = M + 2A Ksp = [M][AI2 (1) CL mol/L

Sobsd3
mmol/L
Scaled,
mmol/L
Sobsd3
mmol/L
Scalcd9
mmol/L
0.010392 2.011 2.016 2.085 2.090
0.021 091 1.530 1.524 1.590 1.584
0.031631 1.291 1.290 1.344 1.343
0.042115 1.146 1.149 1.195 1.198
0.052801 1.053 1.052 1.098 1.098
where A represents the iodate ion and ionic charges are omitted KSP 1.893 (0.009) X 2.034 (0.009) X
for simplicity. T h e solubility was determined by controlled- 10-7 10-7
potential coulometry, using reduction of dissolved iodate a t K, 2.23 (0.07) 2.35 (0.07)
a platinum cathode. The data were corrected for the slight AG, Jimol AH,Jim01 AS,J/(mol K)
formation of the MA ion pair to give values for CM. T h e
reaction 1 +38370 +5484 -110
concentration of t h e uncomplexed copper(I1) ion was calcu- reaction 2 -1992 +3906 +19.8
lated from the iodate ion concentration by using the solubility
product for copper iodate. T h e formation function, Fo = Table 11. Solubility of Copper Iodate and Overall Equilibrium
CM/[M], follows. The Fo values are related to the equilibrium Constants for Copper Chloride Complexes in Sodium Chloride
constants and to the equilibrium concentration of chloride: Solutions at 25 'C
[cl-l equib Sobsd, Scaled,
CL,mol/L mmol/L mmol/L mmol/L Fo
An iterative least-squares procedure that includes correction 0.099 00 0.098 16 3.944 3.939 1.261
of the total sodium chloride concentration, C
,, for the chloride 0.19960 0.19797 4.219 4.216 1.549
used in forming the complexes yields t h e set of /3 values that 0.298 07 0.295 67 4.477 4.480 1.856
best describe the data. 0.401 41 0.398 19 4.739 4.756 2.206
0.502 33 0.498 27 5.003 5.026 2.602
Experimental Section 0.698 7 5 0.692 98 5.555 5.559 3.573
0.798 63 0.791 94 5.864 5.835 4.209
Copper iodate was prepared by slow addition of sodium iodate and 1.296 4 1.284 8 7.302 7.270 8.163
copper sulfate solutions simultaneously to rapidly stirred hot 0.1 M 1.696 7 1.680 8 8.513 8.492 12.96
nitric acid. The thoroughly washed precipitate was placed in a Soxhlet 2.4967 2.471 8 11.06 11.07 28.46
extractor, using a fritted-glass crucible in place of the usual thimble, 3.5032 3.4666 14.40 14.49 62.94
and with 0.02 M nitric acid in the boiling flask, a recrystallized product 4.9904 4.9360 19.86 19.78 165.2
was obtained over a period of 1 week. The copper iodate thus obtained
p values: 2.49 (0.09), 1.14 (0.21), 0.90 (0.13), 0.025 (0.020)
consisted of clear, greenish blue crystals in contrast to the opaque
clumps of microcrystals formed by direct precipitation. When samples
of this product were heated to constant weight at 300 OC, the mass Table 111. Solubility of Copper Iodate and Overall Equilibrium
loss was precisely in accord with the formula ~ C U ( I O ~ ) ~ . ~inH ~ O ,Constants for Copper Chloride Complexes in Sodium Chloride
agreement with the findings of Nassau et a1.J who examined a variety Solutions at 35 "C
of preparations and concluded that copper iodate formed from nitric
acid solutions was identical in composition with the mineral bellingerite.
The preparation and sampling of saturated solutionsof copper iodate
0.098 70 0.097 76 4.060 4.062 1.278
were accomplished by the techniques described*for a study of cadmium 0.19899 0.197 1 1 4.384 4.377 1.616
iodate solubility. However, the coulometry cell solution also contained 0.297 17 0.294 37 4.694 4.685 1.990
0.02 M EDTA to mask the copper(I1) against reduction either by 0.400 20 0.396 41 5.001 5.008 2.412
the iodide ion or by the cathode. This was important to eliminate 0.500 80 0.496 03 5.305 5.325 2.886
any interference by dissolved oxygen, which would reoxidize Cu(I), 0.696 62 0.689 87 5.936 5.951 4.057
and made it unnecessary to deaerate the solutions. 0.796 19 0.78840 6.290 6.273 4.836
All solutions were prepared and analyzed on a mass basis to 1.2924 1.2791 7.952 7.929 9.814
eliminate volumetric errors. Later conversion of data to a molarity 1.691 5 1.673 5 9.337 9.320 15.92
basis required determination of solution densities by weighing precise 2.4889 2.460 7 12.21 12.24 35.71
volumes delivered by a standardized and siliconized pipet. Density 3.4921 3.4500 16.13 16.15 82.37
data were used in smoothed form after a least-squares fit. All solutions 4.9742 4.9097 22.35 22.32 219.3
had an ionic strength of 5.0 and were prepared with water redistilled p values: 2.66 (0.091, 1.99 (0.21), 0.88 (0.13), 0.091 (0.020)
from alkaline permanganate. Copper perchlorate stock solutions were
standardized by visual EDTA titration using murexide, by poten- to an uncertainty of less than 1 ppt. After results were obtained at
tiometric EDTA titration using a mercury electrode, and by con- 25 OC, the solubilities were determined at 35 " C .
trolled-potential coulometric deposition and stripping using both
mercury-pool and platinum-gauze cathodes. The accuracy and Results
precision of the coulometric determination of iodate were checked T h e solubility of copper iodate was determined in five so-
with solutions of recrystallized potassium iodate, with standard de- lutions of copper perchlorate, and the results are shown in
viations of about 0.03%. All solutions were prepared with reagent
grade sodium chloride, with use of sodium perchlorate to adjust the Table I. T h e d a t a were interpreted with the program
ionic strength to 5.00. KSPKiFIT. T h e uncertainties were obtained by simulating the
Experience showed that the solid phases of copper iodate were slow experiment 10 times with the program GENKSPK1, which
in reaching their final values of solubility. Fresh portions of solutions generated d a t a with t h e imposition of a specified standard
were added repeatedly until the observed solubilities were reproducible deviation of 2 ppt in the solubility measurements.
T h e solubility of copper iodate was determined in 12 solu-
tions of sodium chloride ranging from 0.1 to 5 mol/L. T h e
(6) Ramette, R. W. Anal. Chem. 1983, 55, 1232-6. slight increase in solution volume due to the dissolving of the
(7) Nassau, K.; Shiever, J. W.; Prescott, B. E. J . Solid State Chem. 1973, solid was calculated by assuming that the molar volume of
7, 186-204.
(8) Ramette, R. W. Anal. Chem. 1981, 53, 2244-6. aqueous copper iodate is t h e same as that of the pure solid.
Copper(I1) Chloride Complex Equilibrium Constants Inorganic Chemistry, Vol. 22, No. 22, 1983 3325
Table IV. Equilibrium Constants and Thermodynamic Quantities of the copper is converted to the ML4 complex. This makes
complex the value of o4 much less reliable than the other values.
However, when the data are treated on the assumption that
ML ML, ML, ML, only three complexes are formed, the least-squares fit is less
u 4.0 (0.7) 4.7 (0.6) 2.0 (0.3) 0.23 (0.05) satisfactory.
b 2.49 (0.09) 1.14 (0.21) 0.90 (0.13) 0.025 (0.020) For reasons presented below, we consider the detailed work
c -2.26 (0.09) -0.32 (0.42) 0.26 (0.33) 9.1 (1.6) by Schwing-Weill to be the only other experimental study
d 12.3 (0.1) 23.0 (1.5) 11 (4)
e 9.0 (0.1) 15.5 (1.7) 19.8 (0.9) capable of determining the set of four equilibrium constants.
f 37.7 (0.5) 53 (5) 66 i3) She used 37 copper solutions ranging from 0.005 to 5 mol/L
in chloride concentration and determined absorbances at 12
a p values from ref 3. p values from present work. Gibbs
free energy change (kJ/mol) from p values of present work. different wavelengths. An iterative least-squares algorithm
Enthalpy change (kJ/mol) from ref 1 with p values from ref 4. treated the 324 data points in terms of 52 adjustable param-
e Revised enthalpy change (kJ/mol) from values of present eters (48 molar absorptivities and 4 equilibrium constants) to
work with calorimetric data from ref 1. !Entropy changes obtain the best fit to the data. The success of this approach
(J/(mol K)) corresponding to lines c and e. is strongly dependent upon the accuracy of the raw data, as
we have found by computer simulations (program GENSPEDAT)
that included a specified standard deviation in absorbance
measurements. In the presence of normal spectrophotometric
I\ L
error, imposed randomly on a number of different simulations,
the data yield a wide range of equilibrium constants, with sets
of diverse values serving equally well to fit the data. This is
because the programs must simultaneously optimize both the
equilibrium constants and the absorptivities, and errors in the
constants may be essentially offset by errors in the absorp-
tivities. This problem is especially severe for the determination
of P 4 .
The contribution by Bjerrum and Skibsted presents a very
different set of j3 values based largely on speculation rather
Figure 1. Distribution (fraction) of CuClt-’ as a function of chloride than experiment and cannot be considered a valid revision of
+
concentration. The marks are the experimental values of A, the Schwing-Weill’s results. The list of assumptions deserves
average ligand number. The smooth curve through the points was comment. They assumed that the copper was nearly com-
calculated by using the values determined in the present work. pletely converted to the ML4 complex in 6.2 M calcium
Curves were plotted by the program BETAPLOT. chloride solution. But, leaving aside the possibility of a severe
The data and calculated values are given in Tables I1 and medium effect in this solution as compared to 5 M sodium
111. The calculated solubilities were found by using the chloride, Bjerrum and Skibsted’s own proposed value for Q4
previously determined Kspand Kl values, together with the j3 would lead to the prediction that about 50% of the copper
values, in the program GENSOLDAT. The standard deviation would still be in the form of the ML3 complex. They assumed
in the solubility values is about 4 ppt, which corresponds to that only the ML4 complex will absorb light at 436 nm, but
8 ppt in the Fo values. This uncertainty is due to slight dif- there is no direct experimental evidence for this. They also
ferences in behavior of the 12 solid phases in the equilibrium assumed that the molar absorptivity of this complex would be
system, not to errors of equilibration, sampling, or coulometric the same in Schwing-Weill‘s solutions as it appeared to be in
analysis. For comparison, these uncertainties are equivalent the 6.2 M calcium chloride, an unlikely event. They assumed
to a hypothetical potentiometric study, using a copper ion that the equilibrium constants would be interrelated in accord
electrode, with a standard deviation of only 0.1 mV in the with a statistically calculated ligand effect, which is interesting
measurement of cell potential. To obtain an estimate of the but not a valid substitute for experiment. They then picked
uncertainties in the derived values, the experiment was a set of equilibrium constants consistent with the foregoing
simulated 10 times with the program GENSOLDAT, with the assumptions and with the data and assumptions for the calcium
specification of 4 ppt standard deviation in the solubility data. chloride solutions.
Random errors were imposed on the data by a subroutine that Finally, the experimental study by Ashurst and Hancock
provides a Gaussian distribution. The simulated data were must be discussed. First, these authors attempted to find a
treated by the same calculation procedures, using the programs value for Qlby measuring the absorbance of six solutions
FIXSOLDAT (to correct for the MA ion pair) and FORMSOLVE,
containing 0.03 mol/L chloride and varying concentrations
and 10 sets of j3 values resulted. The standard deviations are of copper(I1) up to 0.1 mol/L. A computer program, SPEFO,
shown in parentheses in the tables. was used to calculate a value of Q1= 1.43 on the valid as-
Table IV shows the P values from this and the earlier works. sumption that negligible amounts of the ML2 complex would
Figure 1 is the conventional type of distribution diagram and be present. However, an examination of the data shows that
includes the values of the average ligand number ri calculated this attempt to determine Q1is not successful.
from the data, ri = (CL- [L])/CM, along with the smooth The data, based on the assumption that only one complex
curve for ri calculated from the j3 values. is present, must be in accord with the equation9
Discussion QI= E/[LI(E1 - E) (5)
The values of Kspand Kl were accurately determined at two where E is the apparent absorptivity of the copper-chloride
temperatures. Although the temperature range was only 10 solution, El is the unknown molar absorptivity of ML, and [L]
OC, it seems valid to calculate the thermodynamic quantities is the equilibrium molarity of chloride. When the data (from
shown in Table I. There are no other literature data on the Figure 1 of the Ashurst and Hancock paper5) are examined
values of Kspand K1 at ionic strength 5 and no previous de- in light of this equation, it is clear that they are ill-conditioned.
terminations of K, under any conditions.
From the distribution diagrams it is clear that, even when
the chloride concentration is 5 mol/L, only a small fraction (9) Ramette, R. W. J . Chem. Educ. 1967, 44, 647-54.
3326 Inorganic Chemistry, Vol. 22, No. 22, 1983 Ramette and Fan

The problem is that in the six solutions the equilibrium chloride the values for Kspand K , obtained in the perchlorate medium,
concentration does not vary significantly. In effect, the six and we recognize this fundamental limitation in the solubility
data values are virtually equivalent to a single data point and, approach.
even with very slight errors in E , it is impossible to find rea- The special problem in the spectrophotometric method is
sonable values for both unknowns, Q1and E , . We have made the need to assume that the four molar absorptivities of the
an estimate of the value for E l by using the equilibrium copper chloride complexes are not affected by the change in
constants from the present work with Schwing-Weill’s ab- medium. This is also improbable, as we have shown by pre-
sorbance data at 250 nm (the wavelength used by Ashurst and paring solutions of the copper ethylenediamine complex in 5
Hancock5). We find using program PHOSOLVE that the molar M sodium perchlorate and in 5 M sodium chloride. The
absorptivity of the ML complex is 1770. If this value of E l absorption maxima are at 538 and at 551 nm, respectively,
is assumed to hold for the Ashurst-Hancock experiment, we and the molar absorptivity is about 15% higher in the chloride
calculate a value of 2.3 for Q, from their data. This is close medium. If the molar absorptivities of the copper chloride
to our value of 2.5 from the solubility data. The other complexes are also subject to such variation, then there is little
equilibrium constants deduced by Ashurst and Hancock are hope of determining the ‘‘true- p values by spectrophotometric
dependent upon the value for Ql and are subject to additional studies.
assumptions. Similar questions arise when we consider the classical po-
We turn now to the calorimetric study,’ which used a flow tentiometric method of studying metal complexes and also in
microcalorimeter and chloride concentrations ranging up to calorimetric measurements when the ionic medium must vary
about 3 mol/L. Given the small value of Q4, these data will widely. How nearly constant are the standard potential and
not yield information about the ML4 step in the complex the response slope of the metal electrode? How do the molar
system but are adequate for finding enthalpy changes for the enthalpy changes for the complexation steps vary with drastic
first three steps. Interpretation of the data requires a set of changes in medium?
equilibrium constants from other experiments, and these au- The conclusion is that our careful equilibrium and calori-
thors adopted the values proposed by Bjerrum and Skibsted. metric studies may be less a determination of fundamental
We have reinterpreted the calorimetric data using the equi- properties than a description of the combined behavior of the
librium constants from the present work to obtain values for complex system and the particular probe that is being used
the enthalpy and entropy changes, using program CALSOLVE. in the study.
The results are given in Table IV. It is encouraging that the Acknowledgment. This research was supported by a
enthalpy changes calculated with the p values from the present Northwest Area Foundation grant from the Research Corp.
work show a trend, rather than the sharp change of direction We are grateful to the New Brunswick Laboratory for the loan
(Table IV, line d) found for the ML3 complex when the /3 of the coulometry equipment.
values of Bjerrum and Skibsted are used with Arnek‘s data.
Further, the present /3 values give a somewhat better fit to the Appendix
calorimetric data, as has been verified by Professor Arnek.Io The calculations used in this research involve several com-
In conclusion, the solubility approach has yielded equilib- puter programs, written in DEC VAX-11 Basic, and further
rium constants that are independent of spectrophotometric information may be obtained from the authors. The program
measurements. Future work will be aimed at improving the are as follows:
consistency of the solubility behavior of metal iodates because, GENKSPKI generates solubilities of divalent metal iodate in
given the present techniques of sampling and coulometric specified solutions of the metal perchlorate, given values of
determination, it should be possible to reduce the standard Kspand K,. The resulting data (or actual experimental data)
deviation of the solubility determinations by a factor of 10. are interpreted by KSPKlFIT to find the values for Kspand K,.
Such an improvement would greatly decrease the uncertainty GENSOLDAT generates solubilities of metal iodate in specified
in the derived equilibrium constants. solutions of complexing ligand, given Ksp,K,, and the /3 values.
It is possible that the spectrophotometric study and the The resulting data are first treated by FIXSOLDAT to correct
solubility study will necessarily yield different sets of p values for the metal-iodate ion pair and are then interpreted by
and that neither set is “correct”. The true equilibrium con- FORMSOLVE to find the best set of /3 values.
stants for the system are those defined in terms of activities GENSPEDAT generates absorbances for solutions of metal ion
of the species, even if we adopt the 5 M sodium perchlorate in specified solutions of complexing ligand L, given the p values
solution as the reference state for ideal behavior. In both and the molar absorptivities for the complexes. The data are
methods the aqueous medium has been varied from 5 M interpreted by SPECSOLVE to find the best set of 0 values and
perchlorate to 5 M chloride, and it is highly unlikely that our absorptivities. If the values are known from other experi-
assumption of constant activity coefficients is valid. However, ments, the program PHOSOLVE will use them to find the molar
that doubt applies to both methods. The special problem in absorptivities from the data.
the solubility method is our need to assume that the activity GENCALDAT generates heat effects for specified mixtures of
coefficients of iodate ion and the copper-iodate ion pair remain metal ion and complexing ligand, given the p values and the
constant. We make this assumption implicitly when we use enthalpy changes. The data are interpreted by CALSOLVE,
given a set of p values, to find the best set of enthalpy changes.
(10) Amek, R., private communication. Registry No. C U ( I O ~ )13454-89-2.
~,