Metal ion-sulfur(IV) chemistry. 1. Structure and thermodynamics of

Aug 1, 1988 - Martha H. Conklin, Michael R. Hoffmann. Environ. Sci. Technol. ... Speciation and role of iron in cloud droplets at the puy de Dôme sta...
0 downloads 0 Views 1MB Size
Metal Ion-Sulfur( IV) Chemistry. I. Structure and Thermodynamics of Transient Copper( I1)-Sulfur( I V ) Complexes Martha H. Conkllnt and Mlchael R. Hoffmann"

Environmental Engineering Science, W. M. Keck Laboratories, California Institute of Technology, Pasadena, California 9 1 125

rn The stability constant for the formation of CuS03has been determined spectroscopically to be K = (1.8 f 0.6) X lo4 M-I at p = 0.4 M. The magnitude of this constant is large compared to that of the stability constants of simple inorganic complexes of Cu(I1) such as CuS04, CuCP, CuClO+, and CuN03+. Infrared and Raman spectroscopicmeasurements indicate that sulfite ion binds to copper through both sulfur and oxygen. Comparisons of CuS03 complexes with other first-row transition metal-sulfite complexes are made, and the implications for the atmospheric chemistry of S(1V) are discussed. Introduction Reactions of sulfur dioxide, S(IV), with transition metal ions and complexes in the aqueous phase have been of great interest to atmospheric chemists for the last 15 years. This interest has grown since the liquid-phase oxidation of SOz has been established to be a major pathway in the biogeochemical sulfur cycle and in the production of acidity in the atmosphere. The metal-catalyzed autoxidation of S(1V) has been studied for more than a century without a consensus being obtained as to reaction rates, rate laws, or mechanisms. Our principal objective in this study was to explore the fundamental reaction between Cu(1I) and S(1V) in the absence of oxygen in order to establish a firm basis for further study of the metal-catalyzed autoxidation. When blue copper [Cu(II)] solutions are mixed with colorless sodium sulfite (Na2S03)solutions, the resultant solution immediately changes color from blue to green at pH C6.0. This color change is indicative of the formation of Cu(I1)-S(1V) complexes. These complexes appear to undergo a series of redox reactions that result in an aqueous mixture of Cu(I), Cu(II), SO?-,and SO?- in equilibrium with a red mixed-valence precipitate of copper known as Chevreul's salt ( C U " S O ~ C U ' ~ S ~ ~ - ~The H~O). stoichiometry for the reduction of Cu(I1) is 2Cu(II) + S032-+ H 2 0 = 2Cu(I) + Sod2+ 2H+ (1) Previous studies of the reaction of Cu(I1) and S(1V) have focused on reaction stoichiometry and product determination (1-4). Baubigny (21,who investigated the system under anoxic conditions with [S(IV)] 1.0 M > [Cu(II)], reported the following stoichiometry: 2CuS04 + 4Na2S03* 2Na2S04+ (Cu2S03+ Na2S03)l + NazS20B(2) 'Present address: Department of Hydrology and Water Resources, University of Arizona, Tucson, A2 85721. 0013-936X/88/0922-0883$01.50/0

Table I. Stability Constants for Sulfate and Sulfide Complexes'

metal complexb ionic strength, M AgL CdLz HgLz CeL UOZL Cu(I1)L Cu(1)L

0.0 1.0 1.0 0.0 1.0 0.0 1.0

log of stability constant MS03 MSOd

5.6 4.2 24.1 8.0 5.36 4.2c 7.85

1.3 1.6 2.4 3.6 1.8 2.36

"From Smith and Martell (32) and Sillen and Martell (41, 42). b L = SO," or SO?-. CThiswork (w = 0.4 M).

According to Baubigny, the reduction of Cu(I1) by S(1V) appeared to be similar to the reduction of Ag(1) by S(1V). The unique features of this reaction were the apparent formation of a precipitate (Cu2S03)and S20$-. Albu and Graf von Schweinitz (3) repeated Baubigny's experiments and noted that dithionate was only a product under alkaline conditions (pH -10). Ramberg (I) determined that Cu(1) initially precipitated as Cu2S03.0.5H20(Etard's salt); however, it was shown to be unstable with respect to Cu2SO3.HZ0(Rojoski's salt). Dansent and Morrison (5) showed that Cu2SO3.Hz0was actually a mixture of elemental Cu and CU"SO~CU'~SO~-~H~O, while Foffani and Menegus-Scarpa ( 4 ) ascertained that the green complex was transformed to Cu(S03)$- under anoxic conditions. Equilibrium constants for some known metal-S(1V) and metal-S(V1) complexes are listed in Table I. Substantial differences in the stability constants for certain metals are noted. These variations are due to differences in bonding for the two anions. Sulfate is limited to coordination through oxygen whereas sulfite can coordinate through either sulfur or oxygen. The reported stability constants tend to be higher for transition metal complexes with metal-sulfur bonds. Sulfur is more likely to form stronger bonds with metals than oxygen because of its stronger electron-donating ability and higher polarizability. The most pronounced example of the difference between the stability of metal-oxygen and metalsulfur bonds is given by mercury, which exhibits a difference of loz2between the stability constants of 502- and those of Sod2complexes. Mercury is known to have a much greater affinity for S than for 0; therefore, the bonding is assumed to be through S in Hg(S03)22-. Cu(1) is known to form a salt with S(IV), NH4CuSO3, which contains 0 and S bonds (6),but little is known about

0 1988 American Chemlcal Society

Environ. Sci. Technol., Vol. 22, No. 8, 1988 883

the bonding and stability of aqueous Cu(I1)-S(1V) complexes. In C U ~ S O ~ C U ' ~ S O ~in . ~which H ~ O both , Cu(1) and the Cu(1) is bonded to the Cu(I1) are bonded to S032-, Sot- through both 0 and S bonds; Cu(I1) is bonded through 0 bonds. Copper(I1) and copper(1) have different bonding characteristics. Copper(1) is a class b metal or soft base. Copper(1) follows the trend of having increased polarizability over Cu(I1) and of having a decreased number of coordination sites (four, instead of five or six). Copper(1) preferentially complexes with "soft" sulfur-containing ligands. Copper(I1) is sometimes classified as class b (7), but it can show characteristics of both classes. Before this work, an equilibrium constant for the formation of CuSOt had not been determined. Furthermore, the nature of the bonding in Cu(I1)-S(1V) complexes had not been ascertained. First-row transition metal complexes are of particular importance in aqueous atmospheric systems (e.g., cloud droplets and haze aerosols), where trace metal catalysis of the autoxidation of S(1V) has been postulated to be an important pathway for SOz conversion (8). To elucidate the relative importance of this oxidation pathway, information about the equilibrium distribution of these metal-sulfite complexes is needed. The Cu(1I)-S(1V) system can be used as a model for first-row transition metals that are active catalysts for SOz autoxidation such as Mn(II), Co(II), and Fe(II1).

I

-,01

M CUINO&, p H 4

- - . O ~ M C U ( N O ~ ) ~ ~ . ONa2S0,,pH4 IM

-.I

i

, 300

I

500

700

WAVELENGTH (nm) Figure 1. Absorbance spectra of the Cu(I1)-S(IV) complex obtained immediately after mixing as compared to the absorbance spectrum of a Cu(NO,), solution at pH 4 with p = 0.1 M.

interference filter was used to remove stray light, and the spectral resolution was 0.25 cm-l. The frequency was calibrated by use of known frequencies for carbon tetrachloride and sodium sulfate. Solid samples were in sealed glass capillaries. Aqueous samples were introduced in a flow apparatus. Separate solutions of copper and S(1V) were fed into a junction where they were mixed approximately 30 s before they entered the sampling capillary. Copper chloride was used in the Raman studies since C1does not have a Raman signal. All the spectra were repas a summation of repetitive scans. These spectra resented Experimental Techniques were smoothed by a four-point averaging technique. Reagents a n d Materials. All reagents were of anaInfrared spectra of CU~(OH)~SO HzOwere taken in KBr lytical grade. Sodium sulfite (Mallinckrodt), NaHS03 (J. pellets, while the IR spectra of Cu$SO,CU'~SO,.~H~O were T. Baker Chemical Co.), Na2S20S(CMCB Manufacturing obtained with a Nujol mull on KBr plates. No interaction Chemists, Inc.), and Na2S206(Fischer Scientific Co.) sowith KBr was observed. Spectra were recorded with a lutions were prepared gravimetrically. The concentrations Beckman IR 4240 spectrophotometer. of the NaHSO, solutions were only approximate, as comProduct Identification. Elemental analysis was permercially available NaHSO, is a mixture of NaHS0, and formed by Galbraith Laboratories, Inc. Magnetic susNa2Sz05. These solutions were made fresh for each exceptibilities were measured on a Cahn Electrobdance DTL periment with water that had been purged with N2 7500-7 with H ~ C O ( S C Nas) ~a standard. Copper solutions were made from different salts [Cu(N03)z.2.5Hz0(J.T. Baker Chemical Co.), C ~ ( C 1 0 ~ ) ~ - 6 H ~ 0 Results and Discussion (G. Frederick Smith), CuClZ*2H20(J.T. Baker Chemical Co.), and CuS04.5H20 (Mallinckrodt)]. Copper(I1) soluThe UV/visible absorbance spectrum of the intermediate species is shown in Figure 1. The absorbance of tions were standardized by iodometric titration (9). The these complexes appears as a broad shoulder between 350 concentration of Cu(I1) in experimental solutions was and 400 nm and as a broad peak at 800 nm that is chardetermined colorimetrically (IO)with bathocuproine disulfonate, 2,9-dimethyl-4,7-diphenyl-l,10- acteristic of copper(I1) ligand-field complexes (and accounts for the green color). When the initial pH of the phenanthrolinedisulfonic acid, disodium salt (Sigma mixture is greater than 6 ([Cu(II)l0> 2 mM), a green Chemical Co.). The sulfite solutions were acidified to the Cu(I1)-S(1V) solid precipitates immediately. The Cu(1desired pH with the corresponding acid of the Cu(I1) I)-S(IV) solid reacts further over a period of hours to form counterion. Ionic strength, p (M), was held constant with a yellow solid, which in turn converts into red crystals of either NaNO, (Mallinckrodt), NaZSO4 (Mallinckrodt), within Chevreul's salt ( C U ~ S O ~ C U ~ ~ Ssee O ~Figure ~ ~ H 1) ~O , NaC104 (G. Frederick Smith Co.), or NaCl (J. T. Baker 1 day. When C1- is present, a white precipitate of CuCl Co.). High-purity (18 MQ-cm) water was obtained from is obtained before C U ~ S O ~ C U ~ ~ Sprecipitates. O ~ ~ ~ H ~This O a Millipore Milli-Q system. Cu(I1)-S(IV) solid can be used to obtain information about Spectrophotometric Methods, UV/visible absorbance the bonding of transient aqueous Cu(I1)-S(1V) species. spectra were recorded with a Hewlett-Packard dual-beam Raman and Infrared Spectroscopy Studies. Sulfite diode array spectrophotometer (HP 8450A) in 1-cm quartz can bind to metals in a variety of structures. Sidgwick (11) cells and at 21.0 f 0.4 "C. A Dionex stopped-flow spechas reviewed the structures for different sulfite complexes. trophotometer (D-100)was used for the equilibrium studies The IR frequency of the S-0 stretches yields information of short-lived species. Data were collected at 350 nm at about the structural configuration of the S(1V) complex. 25 "C in a 2-cm path length cell. Solutions with variable The S-0 stretching frequencies are expected to shift deCu(II):S(IV) ratios were examined at p = 0.4 M and at pH pending on the type of coordination. Cotton and Francis 3.6 and 4.4. Each data point represented at least three (12)observed that the S-0 stretching frequencies in sulrepeatable absorbance readings. foxide complexes shifted to higher frequencies for metRaman spectra were collected at room temperature on al-sulfur bonding and to lower frequencies for metal-oxa SPEX Model 1402 double-monochromator Raman ygen bonding; this result agreed with their theoretical spectrometer. Spectra were recorded with the 488-nm line predictions. Thiosulfate complexes (13) and sulfite comof a Spectra-Physics Model 170 argon ion laser. The laser plexes ( 1 4 1 5 ) have been shown to follow the same trend. power incident upon the sample was N 400 mW, a 10-nm 884

Envlron. Scl. Technol., Vol. 22, No. 8, 1988

Table 11. Vibrational Frequencies (om-') of Sulfite Ion'

Ramanb (solution) Ramane (solid) IRb(solution)

V1

V2

v3

v4

967 s 981 s 1002 m

620 w 639 w 632 w

933 m 949m 954s

469 m 497 m

3.5

3

2.5

7

6

( ' I , ,

~

7

'

l

~

'

8 ~

'

I

15

10 '

"

'

I

"

"

~

20 25304C '

~

"

"

"

'

'

" s = strong, rn = medium, w = weak. Nyberg and Larsson (15). CThiswork.

Free sulfite ion has CSvsymmetry, which gives rise to four IR and Raman active fundamental modes: v1 (symmetric stretch), v2 (symmetric bend), u3 (asymmetric stretch), and v4 (asymmetric bend) (see Table 11). The two asymmetric modes are doubly degenerate. This symmetry is essentially preserved if the sulfite bonds to a metal through the sulfur. The sulfite symmetry will be lowered to C3, C,, or C1, if it is bound through the oxygen or is bound as a bidentate ligand. For the lower symmetry groups, the degenerate modes, v3 and v4 will split, giving six vibrations. Nyberg and Larsson (15)found that the structures of metal-sulfite compounds can be divided into three groups on the basis of sulfite coordination. They classified these groups as I (compounds without sulfur coordination), I1 (compounds with sulfur and oxygen coordination), and I11 (compounds with dominant sulfur coordination). Compounds in group I have spectra that resemble the sodium sulfite spectra, with at least one lower stretching mode, and perhaps splitting of u3 and vq. Spectra of Na2S03, (NH4)2S03.H20,NiS03.6H20, and ZnS03-2.5H20fall into this group. Compounds in group I1 with known metalsulfur and metal-oxygen bonds are NH4CuS03,Ag2S03, C U ~ ~ S O ~ C U ' ~ S O and ~.~H T~,[CU(SO~)~]; ~O, these have stretching frequencies of high intensity above and below 975 cm-l. Group I11 compounds have strong stretching frequencies above 975 cm-l; known species include PdS03("3)3 and C O ( ~ ~ ) ~ S O ~ N C S - ~ H ~ O . As mentioned above, we observed an unstable green solid formed when Cu(I1) and S(1V) were mixed at pH 16. This solid continued to react to form a yellow solid. The composition of the yellow solid has been studied previously by Dasgupta et al. (16),who determined that the composition of the solid varied with the initial pH of the system. The solid composition varried according to the formula CuS03.zCu2S03,where 0 I z 5 1. Above pH 5, they verified by elemental analysis that the solid was Cu2S03. However, we found that the green precipitate could be stabilized if it was dried immediately upon formation by washing sequentially with absolute ethanol and anhydrous ether. The stoichiometry of the green solid corresponded most closely to C U ~ S O ~ ( O H ) ~ .(see H~O Table 111). Simple Cu(I1) complexes, those lacking Cu-Cu interactions, have magnetic moments in the range of 1.75-2.20 PB. The magnetic moment of CU~SO~(OH)~.H~O has been measured and found to be 2.07 f 0.04 pB; a molecular weight of 257.9 for CU~SO~(OH)~.H,O was assumed. The IR spectrum of C U ~ S O ~ ( O H ) ~ .is H ~shown O in Figure 2a. The reported frequencies are listed in Table IV together with the frequencies observed for the other

3800 3400

3000

"

1800

1600

1400

1200

IOW

800

6W

400

2

WAVENUMBER CM-' 2.5

3

6

3.5

1 ~ " " " ' " '

\I*

7

10

15

' " I " " 1 " ' " ' " ~ ~ ' ~ ' " ' " ' ~ " ~ - ' " ' ' '

,.. ,

,

u,,

20 253040

I

Table 111. Elemental Analysis of Cu(I1)-S(1V) Precipitate element

measured"

H cu

1.6 0.5 42.2f 0.4 14.2 1.0 40.9 1.7

S 0

* * *

C U ~ ( O H ) ~ S O ~ . H ~ OCU~(OH),SO~

1.56 49.04 12.37 37.94

0.84 52.70 13.29 33.17

cuso3

cu2so3

44.25 22.32 33.42

61.35 15.48 23.17

"An average of three determinations. Environ. Scl. Technol., Vol. 22, No. 8, 1988 885

"

"

~

Table V. Infrared Bands of Metal Sulfites (cm-l)pI Vll

*a

*P

V2

1015 m, 960 s, 890 s 942 m 980 sh, 942 s, 900 s 970 s, 950 s

MnS03 MgSO3 FeS03 CaSO,

645 s 625 m 620 s 665 m, 650 m MnS03-H20 1040 sh, 975 m, 880 s 635 s FeSO3-3HP0 1020 sh, 950 8,880 s 630 s MgSOa.3HzO 940 s 620m 668 m, CaS03.1/zHz0 975 s, 950 s 650m

470 w 480 w 470 m 520 w, 480 w, 458 w 470 w 480 s 480 w 520 w, 485 w, 455 w

'Harrison et al. (19). s = strong, m = medium, w = weak, and sh = shoulder.

El

m m m m m

former case, the strong peaks for vl and v3 occurred below 975 cm-l; this result was consistent with metal-0 bonding. Lutz et al. (20) determined crystal structures for NaM20H(S03)2.H20(M = Mg, Fe, Co, Mn, Ni, Zn). Metal-0 bonding was proposed for all of these solids (and the IR sretching frequencies for these solids show the major stretching frequencies to be below 975 cm-'1. Raman spectroscopy was used to determine the mode of sulfite bonding to Cu(I1). However, Raman spectroscopy is not sufficiently sensitive to probe the Cu(I1)-S(1V) system at low concentrations; concentrations of the reactants that were used to obtain well-resolved spectra were [S(IV)]T= 0.1 M and [cu(II)]T = 0.25 M. Copper(I1) was used in excess of S(IV)to promote the formation of dimeric Cu(I1)-S(1V) complexes. At these concentrations, both Cu(I1) and S(1V) are expected to dimerize. Therefore, the speciation of the system at pH 2.5 may have had an added degree of complexity. To clarify which S(1V) species were present before Cu(I1) was added to the system, Raman spectra of S02.H20,HS03-, and S032-were taken (see Figure 3). In these spectra, the peaks are very broad; this is probably due to solvent effects, such as hydrogen bonding with water. Davis and Chatterjee (21) recorded spectra of the three S(1V) species in D20 and H20, and they found that the peak half-width decreased in D20. This was consistent with hydrogen bonding as the cause of peak broadening. The peak assignment for HS03-is not straightforward. Bisulfite can exist, theoretically, in the form of two tautomers, HO-S02- or H-S03-. Golding (22) argued that both species coexist. He suggested that at higher concentrations ([S(IV)] > M), the two bisulfite tautomers form a dimer that eventually dehydrates to form pyrosulfite, S2052-: HO-S02HO-SO1

+

K

H-SOC

&

ffr &SO3-

-0, ,S,o

(3)

O / -H,

K

H-0-S

U

n

/o '0-

U

On the basis of Raman and IR measurements, Simon and Waldman (23) and Simon et al. (24) analyzed the structure of aqueous HS03- and S2052-. They suggested a CZvsymmetry group and S-0-S bonding for S2052-. Lindqvist and Mortsell (25) determined the crystal structure of K2S205 and found S-S bonding, agreeing with an earlier study of Zachariasen (26). The spectrum of crystalline NaHS03 (or Na2S205,as commercially prepared NaHS03 is Na2S205) is displayed in Figure 4. Using IR and Raman data, Herlinger and Long (27) agree with Golding as to the 888

Envlron. Scl. Technol., Vol. 22,

No. 8,

1988

3800,

1'

In

SO,

a

400

H20

P

I

3700 -

NO HSO,

3600 -

350 m

53

3500-

0

0

3400 -

300

3 300 m

3200

, IO00

I 950

I

I

1050

I150

I I00

250

WAVENUMBER (crn-l) I

I

1

I

f

2700

I

o_

IO

30

WAVENUMBER (cm'll 2600

+

Figure 4. Raman spectrum of crystalline Na,S,O,. 6900v

2500

!

,

I

I

I

25M C u C I 2 + I M NaHSO,

2

0

2400

6800 w

2300

3 2

2 200

< 1

400

6700

600

!

1

800 IO00 WAVENUMBER ( c m - ' )

I

1

1200

6600

6500 900

950

1000

1050

1100

1150

WAVENUMBER (crn-l)

Figure 5. Raman spectrum of the aqueous Cu(I1)-S(1V) system at

pH 2.54.

WAVENUMBER

(cm-0

Flgure 3. Raman spectra of 0.2 M (a) H,0.S02, (b) NaHS03, and (c) Na,S03.

probable existence of S-S bonds or hydrogen bonds for the S2052-species, on the basis of the 15 IR and Raman peaks they observed. This was the expected number of peaks if the symmetry group for S2052-were C,. Using polarization data, Herlinger and Long have assigned 1028 cm-' to HSO, and 1055 cm-l to Sz0?- (the two peaks displayed in Figure 3b). Recent Raman and UV measurements by Connick and co-workers (28) substantiate Golding's arguments for the coexistence of both tautomers in aqueous solution (Le., [HO-S02-]/[H-S03-] 4.9 at 25 "C). In addition, they determined that the overall formation constant, Kp' (Kp' = KTK&p) is 0.088 M-' compared to Golding's value of 0.07 M-l, and Bourne's et al. (29) value of 0.115 at p = 0.1 M. Rhee and Dasgupta (30) have determined an overall dimerization constant, Kp' (Kp' = KTKD) of 1.75 X M-l at 25 "C. Assuming Kp' is correct, the dimeric form would

be the major S(1V) species for the experiments in this study. The Raman spectra of HSOc obtained in this study and displayed in Figure 5b can be compared to the spectra of solid Na2S205(as shown in Figure 4). The aqueous and solid species show substantial differences in the frequency of the S-0 stretching peaks, which indicates that the SzO$ undergpes a structural change, perhaps partially dissociating in aqueous solution. The Raman spectra shown in Figures 3b and 4 compared well with the spectra of Herlinger and Long (see Table VI). The lack of detail in the weaker peaks for the aqueous species obtained in this study was probably due to the lower concentrations that were employed. The Raman spectrum for the Cu(I1)-S(1V) aqueous system is shown in Figure 5. The S-0 stretching frequencies in this case are similar to the HS03- spectrum except for the appearance of a shoulder at 1033 cm-l on the 1024 cm-' peak. The pH of the latter solution was lower than that of the NaHS03 solution spectrum presented in Figure 3b. The presence of a 1152-cm-l peak indicated that an equilibrium with SOZ.H20had been established. Low-pH conditions were necessary to prevent precipitation. The existence of a peak at 983 cm-l indicated the presence of Sod2as a result of the oxidation of S(1V) by Cu(I1). No evidence for S202-,which has a peak at 1090 cm-l, was obtained. The lack of change of the S-0 stretching frequencies would indicate that the Cu(I1)-S(1V) Environ. Sci. Technol., Vol. 22, No. 8, 1988 887

Table VI. Raman Vibrational Spectra of Metal Bisulfite (om-')

NazS205* aqueous so1ution solid (0.1 M)

K!2S206"

aqueous solution (1 M)

solid 147 s 218 s, 195 w 245 vs 317 s 433 8 517 w, 507vw 558 m 569 w, 564 w 645 w 653 s

I

I

I

I

I

assignment

S2062-

168 s 200 s 235 vs 309 s 395 w 424 s 467 m 510m 558 w

s2052-

s202S206" HS03S2062S032-, HSO;

971 w

587 w 637 w 655 s 685 vw 709 w 740 w 933 vw 966 w

1059 s 1088 m 1178 m

1021 s 1052 vs 1085 s 1170w

1202 w

1196 w

515 w 533 m, 555 m 569 w

s2052-

S2062S2OSz, HSOc S20:-, HSO;

659 s

SIE) 0.25

b

'

I

I

X ~ O ~ I M )

I

I

I

I

I

s2052-

HSOaHSOf HS03-

soa2-

S2052-, SOB2-

979 w, 1005 vw 1065 vs 1087 m 1171 w, 1179 m 1203 m

1022 s 1054 vs

HSO; S20E2-

s202- (SzOsZ-) 532052-

s2052-

"Herlinger and Long (27). vs = very strong, s = strong, m = medium, w = weak, and vw = very weak. *This work. Copper Sulfite (X.596 nm)

S (Ip1 x to3 (MI Flgure 7. Absorbance at X = 350 nm of the Cu(I1)-S(IV) complex as a function of [S(IV)] at (a) pH 3.6 and (b) pH 4.4.

(see Figure 1)Cu(I1)-S(1V) solutions. The S(1V):Cu ratio was increased by varying [S(IV)]; [Cu(II)lT,pH, and ionic strength were held constant (31). In this case we have assumed that the reaction can be written as Cu2++ nS032-+ Cu(S03):-2n

(5)

where n = 1,2,3, .... If the total copper concentration is held constant and Beer's law holds for the system, then A = -K-l(A - A,)/[S(IV)]" 2400

600

800

1000

1200

1400

WAVE NUMBERS (crn-l)

Flgure 6. Raman spectrum of Cu,SO,(OH),~H,O in the laser beam.

888

Environ. Sci. Technol., Vol. 22, No. 8 , 1988

(6)

a

where A = absorbance; K-l = [CU~+][S(IV)]~/[CU(SO~)~-~]; and A, = absorbance of complex, if all the copper was complexed. The SO:- concentration is related to [S(IV)] by the following relationship:

after decomposltlon

complex, at the concentration the spectra were taken, has a symmetry very similar to that of aqueous HS03-, with either Cu-S bonding or a dimeric form with Cu-0 and Cu-S bonding through a sulfite bridge. The other possibility is that the Cu(I1)-S(1V) complexes are simple ion pair complexes, but this is unlikely since the experimental stability constant for CuS03 was too high for an ion pair complex (vide infra). Attempts to take the Raman spectra of Cu2S03(OH),-H20 were unsuccessful; it decomposed upon irradiation. The decomposed solid had a very strong peak at 729 cm-l (Figure 6) that did not correspond to the S-0 stretching frequencies of any known metal-sulfite complex. Determination of Equilibrium Constant. Equilibrium studies on the Cu(I1)-S(1V) system were performed with a Dionex stopped-flow spectrophotometer. The concentrations used in these studies were much lower than those used for Raman spectroscopy. An equilibrium constant was calculated on the basis of observed changes in the absorbance at 350 nm measured 20 ms after mixing

+ A,

[S032-]

= a2[S(IV)]

(7)

where a2 =

1

[H+12/Ka1Ka2+ [H+I/Ka2 + 1 where Kal and Ke2are the first and second acid dissociation constants for S02.H20. Since pH was constant in this experiment, a2was a constant. A plot of A vs ( A - AJ/ [S(IV)]" should give a straight line (if n = 1)with a slope of -K-l and an intercept of A,. Figure 7 shows absorbance as a function of [S(IV)] for two values of pH (3.6 and 4.4). Data were plotted according to eq 6 (see Figure €9, which was found to be valid at intermediate ratios of S(1V):Cu(11). As shown in Figure 8, deviations from eq 6 were apparent at low S(IV):Cu(II) ratios. This departure from eq 6 a t low S(IV):Cu(II) ratios was expected. Uncertainty about the free [S(IV)] as well as competition from dimeric Cu(I1)-S(1V) species contributed to this deviation. The Cu(I1)-S(1V) system consisted of several Cu(I1)-S(1V) species: CuS03, C U ~ S O ~and ~ +Cu20HS03+. , The latter

Table VII. Calculated Equilibrium Concentrations for the Cu(I1)-S(1V) System Using the Determined Equilibrium Constant for CuSOa4 PH predicted 6.3 6.2 6.0 5.8 4.7 4.2 4.0 3.5

measured 6.1 6.0 5.9 5.8 5.2 4.9 4.7 3.8

cu2+

98.5 98.3 97.7 96.3 72.7 45.2 35.6 15.0

1.3 1.5 2.1

3.4 24.8 49.8 58.5 77.2

S(1V) distribution, %

Cu(I1) distribution, % CuN03+ cuso3

2.5 5.0 5.8 7.7

HSOc

SO?-

cuso3

66.7 69.7 75.2 81.3 91.6 94.9 95.8 97.2

23.3 20.3 14.8 8.9

10.1 10.0 10.0 9.8 6.8 4.6 3.6 1.5

‘Concentrations and stability contants (for ~1 = 0.0 M) used are as follows: ST = 9.9 mM,CUT= 1.01 mM,p = 0.1 M, log K C ~ O H = +6.03 (47), log KcuNos+= -0.4 (32),log Kcuso3= 5.5 (this work),log KHso3- 7.18 (32).

two species were expected to exist as their stoichiometry was similar to the green precipitate C U ~ S O ~ ( O H ) ~ . H ~ O . These species were expected to be of greater importance at lower S(1V) concentrations for which the 2:l Cu:S stoichiometry was satisfied. When these species were present, eq 5 was an invalid representation of the system. At pH 3.6, the absorbance was very low for the complex a t low S(IV):Cu(II) ratios. The error bars on Figure 8 represent instrument noise. The signal-to-noise ratio decreased at high S(IV):Cu(II) ratios. Data points for which the noise was greater than 10% of the signal were not used in the analysis. At both pH values, the solutions were unbuffered in an attempt to avoid the introduction of competitive complexes into the solution. For the experiment at pH 4.4,the formation of the complex at S(1V): Cu(I1) ratios above 4:l caused the pH of the solution to decrease significantly (from pH 4.4 to pH 4.0);data points for S(IV):Cu(II) ratios below this value were used in the analysis. This decrease in pH would account for the deviation of the data in Figure 8b from eq 6 at high S(1V):Cu(II) ratios. The conditional stability constant as defined for formation of CuS03 is

For the pH range of this study, a2 = K,/[H+]; therefore, we can write

\

[a

0’05 0.04

I

I

t%

I

I

tu,. 2.6mM pH.3.6 p.0.4M Abro=0.131 t 0 . 0 2 3

I

0.00 O.O1 I.5

2.0

2.5

3.0

3.5

PI-pO [SIIEI] IM-”

0.25

b



I

I

I

7,

0

9

I

I

1

0.20

p 51 m

-

0.10

1

O

Il5.9l”

0 0.005

6

A-A, [Sllzl]

10

0

1

1 122.01?

I 25

IM-’I

Flgure 8. Application of eq 6 to the data of Figure 7 for (a) pH 3.6 and (b) pH 4.4. Error bars are shown only when they exceed the symbol size.

log K& [pK, = 7.18 (32)] was adjusted by use of the Davies equation (33): log

( fi

= - 0 . 5 ~ ~-- 0.21)

1+2/7

(10)

where z = charge of the ion; I = ionic strength (M); and y = activity coefficient. This yielded log Ka2= -6.6 ( p = 0.4 M). The stability constant was obtained by application of eq 6 to the data to obtain initial estimates of K and A,. These estimates were then substituted back into eq 6 until the result converged on a value of K. Once K was obtained, K1was calculated. The constants determined by this method were K1 = (2.1f 0.4)X lo4 M-l at pH 3.6 and Kl = (1.3f 0.1)X lo4 M-‘ at pH 4.4. The best estimate that can be given here is K1 = (1.7 f 0.5)X lo4 M-l, p = 0.4 M. The above model is valid only if eq 5 is valid, or in other words, there are no other Cu(II)S(IV)species contributing significantly to the absorbance measurements that were

made. An estimation of the relative importance of the other postulated Cu(I1)-S(1V) species, C U ~ S O and ~~+ Cu2S030H+,equilibrium computations were made with MINEQL, a computer program designed to solve chemical equilibrium problems (34). The computations were done on the Cu(I1)-S(1V) system with the equilibrium constant determined in this work and adjusted to p = 0.0 M (i.e., log K = 5.5). The computational cases were simulations of actual laboratory experiments when the composition of the Cu(I1) and S(1V) solutions were known before mixing, and the pH was measured immediately after mixing. As was noted previously, the pH was lowered upon mixing. This was attributed to a shift in the S0;-/HSO3- equilibrium, which was perturbed by the formation of the complex. Thus, the pH drop was a measure of the degree of complexation. Formation of CuS03 adequately accounted for the observed drop in pH upon mixing Cu(1I)-S(IV) for the conditions of the calculations [10:1,S(IV):Cu(II)]. The predicted and calculated pH values are presented in Table VI1 along with speciation. Envlron. Sci. Technol., Vol. 22, No. 8 , 1988 889

~

Table VIII. Calculated Stability Constants for MSOS Complexesa metal cu2+ Ni2+ Zn2+ co2+

Fe2+ Mn2+ "For SO?-, EA Marsicano (38).

EA 1.27 1.20 1.43 1.33 1.59 1.64

CA

0.466 0.300 0.312 0.276 0.256 0.223

DA 6.0 4.5 4.0 3.0 2.0 1.0

Table IX. Stability Constants for Cu(I1) Complexes (25 OC)"

log K 3.62 1.33 1.30 1.28 0.77 0.48

-1.94; C B = 18.2; DB = 0.4. Hancock and

+ CI'Cgig - DADB

(11)

where E and C represent the tendency to form ionic and covalent bonds of Lewis acid A and base B, and D parameterizes desolvation and steric hindrance effects. This equation is applicable to monodentate ligands. Values used in the calculation with the predicted stability constants are presented in Table VIII. The predicted stability constants are small when compared to the measured stability constant of CuS03. The predicted stability constant for Cu(I1) is significantly larger than those that are predicted for the other divalent metals (0.48 < log K < 1.33)) although the predicted value (log K = 3.62) is smaller than the experimental value (log K = 5.5). The relative magnitude of these stability constants follows the order predicted by the Irving-Williamsseries, which is the order that the metals would follow for an oxygen-bonded ligand. Measured first-row transition metal-sulfate complexes have stability constants that are larger than the predicted constants for metal-sulfite complexes with the exception of Cu(I1). The occurrence of metal-sulfur bonds in the CuS03may account for its enhanced stability compared to CuS04. Sulfur bonding is likely to occur in the metal-sulfite complexes for class b metals. When compared to the class b metals listed in Table I, the difference in stability between CuS03and CuS04is of roughly the same order of magnitude (excluding Hg). Even though the stability constant of CuS03 is relatively large compared to most other Cu(I1) inorganic complexes (see Table IX), CuS03is not expected to be a major S(IV) species in aqueous-phase atmospheric systems unless formaldehyde is not present. Bisulfite adducts, such as the a-hydroxyalkylsulfonates,which are formed with aldehydes, have larger stability constants ( K = [HMSA]/ ([HSO3-][HCHOlT) = lo7 (39). To form equivalent amounts of CuSOs and a-hydroxyalkylsulfonates (HMSA) in the presence of 1 mM S(IV), the ratio of total Cu(I1) to total formaldehyde must be lo7 (i.e., [cu'+],/[HcHo]~, e lo7) at pH 2 or [CU"]T/[HCHO]T ei 6.5 at pH 5. Aldehydes are usually found in much higher concentrations than Cu(I1) in cloudwater, fogwater, and rainwater (39,40); 890

log K

(2032HPOS2SO?NSSO4'NCS-

6.75 4.57 4.2b 2.86' 2.36 2.33

0

3.5 0.4 0 0 0

ionic strength, ligand

log K

M

NO< FNO, C1-

2.02 1.2

0 0

0.5

0 0 0

Br-

0.4 0.3

"From Smith and Martell (32). This work. '20 O C .

The Cu(I1)-S(IV) complexes appear to be different from the other metal-sulfite complexes in the Irving-Williams series. Stable metal-sulfite crystals have been isolated for all of these metals except Cu(I1) (19,35,36). In all of the first-row transition metal-sulfite salts, which have been isolated, crystal structure, IR, and Raman studies have shown metal-oxygen bonds. The Cu(I1)-S(1V) complexes are different in that Raman spectra of the transient Cu(11)-S(1V) complexes suggest the presence of metal-sulfur bonds. Stability constants for the divalent first-row transition metal-sulfite complexes have not been determined. However, they can be calculated with the modified equation of Drago et al. (37) that can be applied to aqueous systems (38): log K1 = EIqEbq

ligand

ionic strength, M

Environ. Scl. Technoi., Vol. 22, No. 8 , 1988

therefore, S(1V) speciation is likely to be dominated by RC(OH)S03- rather than CuS03. Acknowledgments

We are indebted to Eric C. Betterton (Caltech) for his expert assistance throughout the latter stages of this work. We also acknowledge the personal support of our program officers, D. Alan Hanson (EPRI) and L. Swaby (EPA). Registry No. CuUS03CumS03.2H20,13814-81-8;Cu2S03(OH)2*H20,114927-79-6; SOZ, 7446-09-5. Literature Cited (1) Ramberg, L. 2.Phys. Chem., Stoechiom. Verwandtschaftsl. 1910,69, 512-522. ( 2 ) Baubigny, M. H. C. R. Hebd. Seances Acad. Sci. 1912,154, 701-703. (3) Albu, H. W.; Graf von Schweinitz, H. D. Ber. Dtsch. Chem. Ges. B 1932, B65, 729-737. (4) Foffani, A,; Menegus-Scarpa, M. Gazz. Chim. Ital. 1953, 83, 1068-1081. (5) Dansent, W. E.; Morrison, D. J.Inorg. Nucl. Chem. 1964, 26,1122-1125. (6) Nyberg, B.; Kierkegaard, P. Acta Chem. Scand. 1968,22, 581-589. (7) Gray, H. B. Chemical Bonds: An Introduction to Atomic and Molecular Structure; Benjamin/Cummins: Menlo Park, CA, 1973. (8) Hoffmann, M. R.; Jacob, D. J. In SO2,NO and NO2 Oxidation Mechanisms: Atmospheric Considerations; Acid Rain Precipitation Series; Calvert, J. G., Ed.; Butterworth Boston, 1984; Vol. 3, pp 101-172. (9) Hammock, E. W.; Swift, E. H. Anal. Chem. 1949, 21, 975-979. (10) Blair, D.; Diehl, H. Talanta 1961, 7, 163-174. (11) Sidgwick, N. V. The Chemical Elements and Their Compounds; University Press: Oxford, 1951;Vol. 2, pp 909-910. (12) Cotton, F. A.; Francis, R. J. Am. Chem. SOC.1960, 82, 2986-2991. (13) Freedman, A. N.; Straughan,B. P. Spectrochim. Acta, Part A 1971,27A, 1455-1465. (14) Newman, G.; Powell, D. B. Spectrochim. Acta 1963, 19, 213-224. (15) Nyberg, B.; Larsson, R. Acta. Chem. Scand. 1973,27,63-70. (16) Dasgupta, P. K.; Mitchell, P. A.; West, P. W. Atmos. Enuiron. 1979, 13, 775-782. (17) Earwicker, G. A. J . Chem. SOC.1960, 2620-2626. (18) Lebedinskii, V. V.; Shenderetskaya, E. V. J. Znorg. Chem. (USSR) 1957,2, 1768-1774. (19) Harrison, W. D.; Gill, J. B.; Goodall, D. C. Polyhedron 1983, 2, 153-156. (20) Lutz, H. D.; Eckers, W.; Buchmeier, W.; Engelen, B. Z. Anorg. Allg. Chem. 1983, 499, 99-108. (21) Davis, A. R.; Chatterjee, R. M. J. Solution Chem. 1975,4, 399-412. (22) Golding, R. M. J. Chem. SOC.1960, 3711-3716. (23) (a) Simon, A.; Waldman, K. 2.Anorg. Allg. Chem. 1955, 281, 113-134. (b) Simon, A.; Waldman, K. 2.Anorg. Allg. Chem. 1955,281,135-150. (c) Simon, A.; Waldman, K. 2. Anorg. Allg. Chem. 1956, 284, 36-46. (d) Simon, A.;

Envlron. Sci. Techno/. 1988. 22, 891-898

Lutz, H. D.; El-Suradi,S. M.; Mertins, C.; Engelen, B. 2. Naturforsch. B: Anorg. Chem., Org. Chem. 1980, 35B, 808-816. Drago, R. S.;Vogel, G. C.; Needham, T. E. J . Am. Chem. SOC.1971,93,6014-6026. Hancock, R. D.; Marsicano, F. Inorg. Chem. 1980, 19, 2709-2714. Munger, J. W.; Tiller, C.; Hoffmann, M. R. Science (Washington, D.C.) 1986,231,247-249. Munger, J. W.; Jacob, D. J.; Waldman, J. M.; Hoffmann, M. R. J . Atmos. Chem. 19848,1,335-350. Sillgn,L. G.;Martell, A. E. Stability Constants of Metal-Ion Complexes; The Chemical Society: London, 1964. Sillh, L. G.;Martell,A. E. Stability Constants of Metal-Ion Complexes, Supplement No. 1; The Chemical Society: London, 1971. Evans, J. C.; Bernstein, H. J. Can. J. Chem. 1955, 33, 1270-1272. Pannitier,G.;Djega-Mariaclasson,G.; Bregeault, J. M. Bull. SOC.Chim. Fr. 1964,1749-1756. Baldwin, M. E. J . Chem. SOC.1961,3123-3128. Baggio, S.;Becka, L. N. Acta. Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1969,B25, 946-954.

Waldman, K. 2.Anorg. Allg. Chem. 1956,284,47-59. Simon, A.; Waldman, K.; Steger, E. 2.Anorg. Allg. Chem. 1956,288,131-147.

Lindqvist, I.; Miirtsell, M. Acta Crystallogr. 1957, 10, 406-409.

Zachariasen, W. Phys. Rev. 1932,40,923-935. Herlinger, A. W.;Long, T. V. Inorg. Chem. 1969, 8, 2661-2665.

(a) Connick, R. E.; Tam, M.; von Deuster, Inorg. Chem. 1982,21,103-107. (b) Horner, D. A.; Connick, R. E. Inorg. Chem. 1986,25,2414-2417.

Bourne, D. W.;Higuchi, T.; Pitman, I. H. J . Pharm. Sci. 1974,63,865-868.

Rhee, T.S.; Dasgupta, P. K. J . Phys. Chem. 1985,89, 1799-1804.

Newton, T. W.; Arcand, G. M. J. Am. Chem. SOC.1953,75, 2449-2453.

Smith, R. M.; Martell, A. E. Critical Stability Constants; Plenum: New York, 1976;Vol. 4. Stumm, W.; Morgan, J. J. Aquatic Chemistry; Wiley: New York, 1981;pp 134-137. Westall, J. C.; Zachary, J. L.; Morel, F. M. MINEQL, A Computer Program for the Calculation of Chemical Equilibrium Composition of Aqueous Solutions; Department of Civil Engineering, Massachusetts Institute of Technology: Cambridge, MA, 1976;Technical Note 18. Klasens, H. A.; Perdok, W. G.;Terstra, P. 2.Kristallogr., Kristallgeochem., Kristallphys. 1936,94, 1-6.

Received for review May 1,1987.Accepted January 6,1988. This work was cooperatively supported by the U.S. Environmental Protection Agency (EPA R811496) and the Electric Power Research Institute (RP 1630-47).

Metal Ion-Sulfur( IV) Chemistry. 2. Kinetic Studies of the Redox Chemistry of Copper( 11)-Sulfur( I V ) Complexes Martha H. Conkiln+ and Michael R. Hoffmann*

Envlronmental Engineerlng Science, W. M. Keck Laboratories, Callfornia Institute of Technology, Pasadena, California 91 125

rn The redox chemistry of Cu(I1)-S(1V) complexes is examined, and the rate of reduction of Cu(I1) is determined. The reduction of Cu(I1) is shown to proceed via Cun2SOt+ and Cun2S030H+intermediates. Cop er(I), SO4", and a mixed-valence compound, Cu"S03Cu{S03.2H20, are determined to be the principal products. The rate law is consistent with consecutive first-order reactions. Results are interpreted in terms of the initial formation of an inner-sphere complex, which is followed by a rate-limiting electron-transfer step. These results are discussed in light of previously accepted mechanisms for the trace metal catalysis of the autoxidation of SO-.: Introduction Copper(I1) forms transient complexes with S(IV) in solution as discussed in our previous paper in this series (I). These complexes react to form a mixture of Cu(I), Cu(II), Sod2-,and S032-in apparent equilibrium with Chevreul's salt (CU"SO~CU'~SO~-~H~O), a mixed-valence precipitate of copper under anoxic conditions. However, the reduction of Cu(I1) by S(IV), in terms of the predominant Cu(I1) and S(IV) species, is thermodynamically unfavorable [i.e., e- + Cu2++ Cu+, 0.153V (2) and HS03-* H++ e- + SO3-- 0.84V (3)]. The unfavorable AGO for this reaction indicates that other processes must be taken into account. These processes include the stabilization of the products and the energy released by precipitation. 'Present address: Department of Hydrology and Water Resources, University of Arizona, Tucson, AZ 85721. 0013-936X/88/0922-0891$01.50/0

The kinetics of reduction of Cu(I1) by S(IV are relevant to the chemistry of SOz in clouds and haze aerosol (4). Copper, which is present in most aqueous atmospheric systems, is known to be an effective catalyst for the autoxidation of S(1V) at pH >6. Reinders and Vles (5) have investigated the catalytic autoxidation of S(IV) in the presence of Cu(II), Fe(III), Ni(II), and Co(I1). They found that Cu(I1) and Fe(II1) were active catalysts in the pH range 4-12. Fuller and Crist (6)obtained the following rate law for the Cu(I1)-catalyzed oxidation at pH 8.7: -d[SOS"]/dt

= (k + kz[C~~+])[S0:-]

(1) where K1 = 0.013 s-l and k2 = 2.5 X 10+ M-' s-l. The observed rate was found to decrease with a decrease in pH. The first term of the two-term rate law was proposed to account for noncatalytic autoxidation of S(1V). Backstrom (7)has argued that the reduction of Cu(I1) by S(IV) is the initial step in the Cu(I1)-catalyzed freeradical oxidation of S(1V): Cu(II)

-

+ S032-

Cu(1) + SOB-

(2)

Even though Backstrom did not postulate the preequilibrium formation of a Cu(I1)-S(1V) complex before the electron transfer, formation of a complex appears to be a necessary step in the mechanism for thermodynamic reasons (1). Veprek-Siska and Lunak (8)reported that cupric ion wa8 "immediately and quantitatively reduced by sulfite to cuprous ion" and they proposed that the metal-catalyzed autoxidation of S(1V) proceeded via the formation of sulfide-cuprous intermediates [such as 02Cu(S03),,-2"+1]

0 1988 American Chemlcal Society

Environ. Sci. Technoi., Voi. 22, No. 8, 1988 891