H. E. SPENCER AND 111. W. SCHMIDT
2986
of iron group elements, ions must be available in quantities of, for instance, about 0.01, 0.1, 0.35, 0.8, and 1.6 pequiv (or 0.4, 3, 10, 25, and 50 pg) for bead diameters of 0.2, 0.4, 0.6,0.8, and 1.0 mm, respectively.
(iii) However, the method is inapplicable to ion species, such as polyelectrolytes, which may not be adsorbed by the resin in sufficient quantity to saturate the exchange capacity of the resin.
Photochemical Studies of Solid Potassium Trisoxalatoferrate( 111) Trihydrate by H. E. Spencer* and M. W. Schmidt Research Laboratories, Eastman Kodak Company, Rochester, New Yorlc 14660
(Received April 19, 1971)
Publication costs aseisted by the Research Laboratories, Eastman Kodak Company
To investigate some of the principles of solid-state photochemistry, three kinds of crystals of Ks[Fe(C20d)a]. 3Hz0, large crystals, powdered large crystals, and microcrystalline coatings, have been studied. A correction for the optical filtering by solid photoproducts is applied to obtain the quantum yield for production of Fe(II), 4. The value of 4 increases appreciably upon evacuation. Addition of water vapor reverses the effect of evacuation, whereas the addition of 02, air, COz, or HZ causes only slight change. Heating immediately before exposure increases +. The value of 4 for the large crystals is lower than that for the smaller crystals. Carbon dioxide is a photoproduct for all crystals, and carbon monoxide is also for the coatings. A mechanism with electron-hole pairs and radical ions as intermediates is suggested. Comparison is made to the photochemical mechanism of acidic solutions of the compound.
There are two major reasons for studying the solidstate photochemistry of potassium trisoxalatoferrate(111)trihydrate, Ka [Fe(Cz04)3] -3Hz0. First, an acidic solution of the compound is a widely used actinometer for the blue, violet, and ultraviolet wavelength regions. Extensive studies of the solution mechanism have been made; it is worthwhile to study the mechanism in the solid state and compare it with the better understood solution mechanism. Second, the compound serves as a model for studying the principles of solid-state photochemistry. Effects of crystal size, optical filtering by accumulated solid photoproducts, and consequences of electronic band structure, for example, are unique to the solid phase. Also, the general interest in solid-state photochemistry of coordination compounds is increasing, as is shown by a number of recent papers dealing with K3[Fe(CZ04)3].3Hz0,1-4as well as with related cobalt and manganese corn pound^.^^^ Here we report studies concerned mostly with the effects of evacuation and of crystal size upon the photo.3Hz0.The composition chemistry of K3[Fe(C204)3] of both gaseous and solid photoproducts is studied, also. Experimental Section Three types of crystals were used in this study: coatings, large crystals, and powdered large crystals. The coatings, prepared by pouring a water solution of K3 [Fe(C~04)8] -3Hz0into an organic solvent, miscible The Journal of Physical Chemistry, Vol. 76, N o . 19, 1972
with water but in which the compound is insoluble (e.g., ethanol, isopropyl alcohol, or acetone), have been described previously. a Large crystals were grown from aqueous solution by imbedding a seed crystal in a piece of paraffin and floating the combination upon the surface of a saturated .3Hz0. After several aqueous solution of K3 [Fe(Cz04)8] hours or days of evaporation, large crystals with edges 1 or 2 cm long were formed.6 Seed crystals were made by evaporation of water from a beaker containing an aqueous saturated solution. To form the ground crystals, large crystals were ground first in a ball mill with (1) J. N. Pitts, Jr., J. K. S. Wan, and E. A. Schuck, J . Amer. Chem. Soc., 86, 3606 (1964).
(2) W. W.Wendlandt and E. L. Simmons, J . Inorg. Nucl. Chem., 28, 2429 (1966). (3) H. E. Spencer, J . Phys. Chem., 73, 2316 (1969). (4) R. Ballardini and M.T. Gandolfi, A n n . Chim. (Rome), 60, 272 (1970). ( 5 ) (a) D. Klein, C. W. Moeller, and R. Ward, J . Amer. Chem. SOC., 80, 265 (1958); (b) W. W. Wendlandt and J. H. Wooklock, J . Inorg. Nzicl. Chem., 27, 259 (1965); (c) W. W. Wendlandt and E. L. Simmons, ibid., 27, 2317 (1965); (d) E. L. Simmons and W. W. Wendlandt, ibid., 27, 2325 (1965); (e) V. Balzani, R. Ballardini, N. Sabbatini, and L. Moggi, Inorg. Chem., 7, 1398 (1968); (f) D. A. Johnson and J. E, Martin, ibid.,8, 2509 (1969); ( 9 ) H. E. Spencer, Photogr. Sci. Eng., 13, 147 (1969); (h) H. E. Spencer and M. W. Schmidt, J . Phys. Chem., 74,3472 (1970); (i) G. Lohmiller and W. W. Wendlandt, Anal. Chim. Acta, 51, 117 (1970); ( j ) G. D’Ascenzo and W. W. Wendlandt, J . Inorg. Nucl. Chem., 32, 3109 (1970); (k) S. T. Spees, Jr., and P. Z. Petrak, ibid., 32, 1229 (1970); (1) A. C. Sarma, A. Fenerty, and 5 . T. Spees, J . Phys. Chem., 74,4598 (1970). (6) A. J. Fishinger, J . Chem. Educ., 46, 486 (1969).
PHOTOCHEMICAL STUDIESOF SOLIDPOTASSIUM TRISOXALATOFERRATE(III) TRIHYDRATE Burumdum media for several hours; the resulting compact mass was then forced through an 88-p sieve with a pestle. The 200-W superhigh-pressure mercury source and monochromator used to carry out the irradiations were described p r e v i ~ u s l y . ~The vacuum system consisted of a forepump, a mercury diffusion pump, and an allglass line with a McLeod gauge and manometer. A liquid nitrogen trap was always used between the diffusion and forepumps. The glass vacuum cell was provided with a flat quartz window. Large crystals were mounted with paraffin with the large (110) faces perpendicular to the radiation beam. Ground crystals were evacuated and irradiated by placing the material in a 1.0 mm thick spectrophotometer cell covered by a plastic cap through which small holes were bored to allow gas to escape. Slow initial evacuation was necessary to prevent scattering of the powdered sample. After irradiation of an evacuated coating, air was admitted to the cell, the slide was removed immediately, and the coating was washed into a weighed glass beaker. Added to this were 5 ml of an acetate buffer solution (0.6 mol of sodium acetate and 360 ml of 1 N &so4 diluted to 1 1.) and 10 ml of a 0.1% 1,lO-phenanthroline solution. Enough water was added to bring the weight of the solution to a given value, usually 20 g. After the solution had stood for 30 min, its optical density at 5100 A was measured against a reference solution which was identical except that an unirradiated coating was used in its preparation. The concentration of Fe(I1) was calculated using a value of 1.11 X lo4l./mol cm as the molar extinction coefficient.s Ground crystals were analyzed in much the same manner as the coatings. For irradiation of large crystals, a fresh crystal was used for each exposure. Before exposure, the crystal was soaked for 5 min in water. After removal of the crystal, the solution was analyzed for Fe(I1) in the same way described for coatings. The crystal was rinsed in acetone, dried, mounted, evacuated, exposed, removed from the cell, soaked in water for 5 min, rinsed again in acetone, and dried. Again the aqueous solution was analyzed for Fe(I1). Finally the dried crystal was soaked a third time in water for 5 min, and the solution was analyzed for Fe(I1). This elaborate procedure was used to ensure that all the Fe(I1) produced by irradiation was removed during the second soak. The criterion used to judge this was the agreement between the amounts of Fe(I1) found in the first and third soak. The pressure of the gas produced by irradiation was measured with a McLeod gauge, often with a Dry Ice cooled trap between the irradiation cell and the gauge. After mass spectrometric determination indicated that only CO and COz were present, the composition was found by determining the pressure with and without a liquid nitrogen trap in the line. Such a trap condenses COZ but not CO.
2987
4 = 0.65 r: K = 2 80
+_
0.02 0.02
UNEVACUATED 4 = 017 i 0 0 3 K = 281 t 0 0 3
d
I x 10'8 10 12 14 I
I,t,
photons
Figure 1. Effect of evacuation of coatings, wavelength 366 nm.
Results and Discussion I, Evacuation and Determination of 4. Whereas evacuation of the coatings after exposure but before dissolution stops the slight decay of Fe(I1) which occurs upon holding for a period of hours, evacuation before exposure greatly increases the efficiency of Fe(I1) production. I n contrast, we find that coatings of the corresponding cobalt compound, K3[Co(C204),].3H20, are unaffected by evacuation, either before or after irradiation. Figure 1 illustrates the effect for coatings of Ks[Fe(C204)3]*3H~O.Plotted is the amount of Fe(1I) produced as a function of number of incident photons, l o t , where IOis the incident intensity and t is the time of exposure. The curves drawn are those calculated by computer; they obey the two-parameter equation [Fe(ll)] =
1
-
K
In (1
+ KC#J~)
(1)
where [Fe(lI)J is the number of Fe(I1) ions, 4 is the quantum efficiency for Fe(I1) production, and K is an attenuation factor characteristic of the solid photoproducts. This equation is identical in form with one derived previously for the photolysis of KB [Co(C204),] 3Hz0.5h It differs only in that Fe(I1) replaces Co(I1) and in that [Fe(II)1, Io, and K are not normalized to unit area. Because the irradiated area varies with experimental conditions, it is convenient not to have to correct the raw data, values of [Fe(II)] and l o t , to unit area. From the raw data, values of 4 and K are calculated by computer, as discussed p r e v i ~ u s l y . For ~ ~ the definition made here, the parameter of most interest, 4, does not depend on the size of the irradiated spot, whereas the other parameter, K , is inversely proportional t o the area. The agreement of eq 1 with the data in Figure 1 is additional justification for the simple model used to derive the equation comparable to eq 1 in a previous publica(7) H. E. Spencer and J. 0. Darlak, J . Phys. Chem., 72, 2384 (1968). (8) J. G. Calvert and J. N. Pitts, Jr., "Photochemistry," Wiley, New York, N. Y., lS66, p 785. The Journal of Physical Chemistry, Vol. 76, N o . 10,1971
H. E. SPENCER AND 14. W. SCHMIDT
2988 1
I
GROUND CRYSTAL
rot, photons
Figure 2. Effect of crystal size upon Fe(I1) yield, wavelength 366 nm.
t i ~ n .Essentially, ~ ~ eq 1 provides a means of extrapolating back to obtain the initial value of 4 a t zero exposure time, or, in other words, to correct for radiation lost by photoproduct attenuation. Also, eq 1 fits data for both large and ground crystals (Figure 2) and has proved to provide a useful and convenient method for obtaining 4. For coatings of Figure 1 evacuation increases d, by almost 400% but changes the attenuation factor, K, almost not a t all. For a given area of irradiation, K depends only upon the optical properties of the solid photoproducts and thus should not be affectedby evacuation. Another result of evacuation is that the Fe(I1) yield for all three crystal types is independent of intensity, in contrast to the slight intensity dependence reported previously for unevacuated coatings. Because of this latter intensity dependence, we were surprised to find that the data for unevacuated coatings obey eq 1. However, although the lower curve in Figure 1 fits the points fairly well, these points from the unevacuated coating do not fit so well as the upper curve fits the data for evacuated coatings. The yield does depend upon pumping time and rate of pumping, however. For a given rate of pumping, there is an optimum time of pumping for each type of crystal (see Figure 3). Experiments made with a smaller roughing pump indicate that the optimum time increases with a lower rate of pumping. Our technique has been to determine the optimum time for a given rate of evacuation of each crystal type before performing subsequent experiments. Probably the initial increase in Fe(I1) yield for short evacuation times is caused by water removal from the crystal-air interfaces, because adding water vapor, after evacuation but before exposure, decreases the yield below that for the unevacuated crystals. Addition of 0 2 , air, COS, or H2 under the same conditions only slightly lowers 4. The drop in yield for longer pumping times may be caused by loss of water of hydration. Thus two kinds The Journal of Physical Chemistry, Vol. 76, No. 19,1971
of water need be considered, molecules which are an integral part of the crystal structure, water of hydration, and another kind which is adsorbed to the intesfaces. II. Crystal Size. Figure 2 clearly shows that the magnitude of 4 increases with decreasing crystal size. This indicates a dependence upon surface-to-volume ratios, commonly found for solid-state reactions. Yet there must be other important unknown factors because q5 for coatings always exceeds 4 for ground crystals, although microscopic examination indicates that the microcrystals in the coatings are larger. For instance, localized surface topography and the way the crystals are packed together may be very important. I I I . Stoichiometry. Wendlandt and Simmons originally reported the products of photolysis to be oxalate, FeCz04, and COS.^ Unfortunately, the stoichiometry is not that simple. Two unpublished these^^,^^ reported that CO as well as C02 is produced. We have confirmed that CO comprises 20-30Oj, of the gaseous photoproducts from the coatings, but little or none is found for the large and the ground crystals. These facts suggest that a t least some of the organic solvents used for precipitation of the microcrystals are retained in the coatings and that this organic material leads to CO. Although for coatings we have always found CO among the photoproducts, we have also found that the CO/CO2 ratio increases with increasing amount of photolysis. Perhaps the initially formed solid photoproduct, which absorbs some of the r a d i a t i ~ n ,reacts ~ , ~ ~ with the adsorbed organic solvents. We find equal amounts of Fe(I1) and C o t produced by photolyzed large crystals. For coatings, not as much COz as Fe(I1) is detected, presumably because some of the gas is trapped inside the microcrystals. For the coatings in which CO is a photoproduct, there must be another product; we have been unable to identify it. We have attempted unsuccessfully to identify the solid Fe(I1) photoproduct from its X-ray diffraction spectrum. An unidentified product, not FeC204, along with KzC204, was found. Presumably the unknown compound is M6Fe2(C204)5,identified by Bancroft, Dharmawardena, and Maddock" in Mossbauer studies of the thermal and 7-ray decomposition of K3[Fe(CZO&] .3H20. However, there has to be some other product; the simplest stoichiometry is the formation of 1 mol of K6Fe2(C20&,2 mol of COz, and 6 mol of HzO from 2 mol of K3[Fe(Cz04)3].3Hz0. Neither K2C204nor CO is accounted for this way. I V , Mechanism. First, we consider the essential features of the generally accepted mechanism for photol(9) J. B. Holden, Jr., Thesis, Princeton University, 1961. (10) W. M. Riggs, Thesis, University of Kansas, 1967. (11) G. M. Bancroft, K. G. Dharmawardena, and A. G. Maddock, Inorg. Chem., 9, 223 (1970); J . Chem. SOC. A , 2914 (1969).
[:,
p
PHOTOCHEMICAL STUDIESOF SOLIDPOTASSIUM TRISOXALATOFERRATE(III) TRIHYDRATE
-
: .' CRYSTAL : , F E CRYSTAL a) >
.c
-0 E
A .2
oo
80
40
0
40
COAT ING
80
Time of evacuation,
2989
0
40
80
I20
minutes
Figure 3. Variation of Fe(1I) yield with evacuation time.
ysis of Fe(Cz04)3in acid ~ o l u t i o n . For ~ ~ ~each ~ ~ photon absorbed, two Fe(I1) ions and two COZ molecules
+ Fe(Cz04)33.C204-+ Ice(C204)33hv
--f
+
(2)
+ C2042-
(3)
Fe(C~04)2~- vCzO4-
----f
+
Fe(Cp04)22- 2C02
are produced. Over a considerable range of wavelengths, the experimental quantum yields are not 2, however, but are about 1.2, indicating some deactivation not included in the above reaction scheme. The important point is that quantum yields greater than unity dictate a reaction sequence such as given by eq 2 and 3, in which radicals formed in the first step produce additional products in the second. For the solids studied here, 4 never exceeds unity; the highest values, for evacuated coatings a t 313, 335, 366, and 405 nm, are 0.56 0.01, 0.67 0.02, 0.68 0.03, and 0.52 =t 0.02, respectively. For solid K3[Fe(Cz04)3]. 3H20 irradiated in KBr matrices, Ballardini and Gandolfi also found 4 to Ke less than but almost unity,4 although for the same type of matrix system a preliminary value of 1.3 was reported by Pitts, Wan, and Schuck.' If we presume that the more recent and extensive work of Ballardini and Gandolfi is correct, there is no need for the reaction sequence represented by eq 2 and 3 in the solid-state mechanism. It is appropriate to digress a t this point and discuss the almost unity quantum yield values for KBr maThe technique of forming the KBr disks requires the use of relatively high pressure. Ballardini and Gandolfi state that this high pressure leads to a high enough temperature to cause an 8% thermal decomposition of the iron compound. Presumably, this heat must also dry the crystals and increase the value of 4. We have found that heating the coatings immediately before exposure appreciably increases 4. If the heated coatings remain in room air at room temperature for more than 15-20 min the value of 4 begins to decrease and eventually approaches that found with no heating. Ballardini and Gandolfi find that higher pressures, presumably with concurrent higher temperatures, decrease the value of 4. Although this decrease could be a direct result of the pressure, we wonder if the
*
*
greater heat is removing some of the water of hydration in the same way that long times of evacuation (Figure 3) remove water of hydration and produce lower values of 4. Returning now to the mechanistic discussion, we must explain additional experimental facts. These are photoconductivity,3 the large effect of evacuation, and the intensity dependence of r#~ for unevacuated coating^.^ The photoconductivity indicates hole-electron pair production in the electronic bands of the solid hv+h+e
(4)
where h and e represent holes and electrons, respectively. The fates of these electrons and holes occupy much of the subsequent mechanistic considerations. We write Cz042-,Fe(III), and Fe(I1) to represent species which are oxidized or reduced, although, in general, these species are coordinated to other species. Our designation is merely a shorthand notation. Also, the radicals written as C z 0 4 - -and COz.- possibly are coordinated to Fe(lI1). The holes cause oxidation of oxalate
+ cz04'- + .czo4-
h
(5)
This oxalate radical may react by two alternative paths h
+
czo4- +2C02
(6)
or *c204-
h
'coz- + coz
+ *C02- +COz
(7) (8)
each producing two COz molecules. There is esr evidence that free radicals such as .CzO4and .COz- are produced at low t e m p e r a t u r e ~ ~and ~,'~ we assume here that radicals are important as intermediates at rocm temperature (reactions 5-8). I n (12) C. A. Parker, Trans. Faraday Soc.. 50, 1213 (1954). (13) C. A. Parker and C. G. Hatchard, J. Phys. Chem., 6 3 , 22 (1959). This paper contains much more detail about t h e mechanism than given by eq 1 and 2 which give only the features essential for our argument. (14) D. J. E. Ingram, W. G. Hodgson, C. A. Parker, and W. T. Rem, Nature (London),176, 1227 (1955).
The Zournal o j Physical Chemistry, Vol. 76, No. 19, 1971
H. E. SPENCER AND M. W. SCHMIDT
2990 fact, Gross reported flash photolysis studies of K3[Co(CzO4) ] solutions in which he detected GO4- a t room temperature. The primary effect of the other half of the electronhole pair, the electron, is to induce reduction of Fe(II1) to Fe(I1)
-
e
+ Fe(II1) +Fe(I1)
(9)
The increase in t$ caused by evacuation and the subsequent decrease caused by admitting water vapor t o the system indicates that adsorbed water somehow induces a recombination reaction.I6 We list two possibilities. The first is that adsorbed water molecules create electron-trapping centers which we designate as X H ~ OThe . X indicates that we do not know if the center is water itself or is some hydrated ion a t the interface. The reaction which competes with reaction 9 is e
+
(10)
X H a O +X H a O -
Upon evacuation, this adsorbed water is removed and reaction 10 becomes unimportant. We emphasize that we are discussing adsorbed water and not water which is an integral component of the crystal. Recombination occurs when X H reacts ~ ~ with a hole h
+ LO-X H ~ O --t
(11)
The other possibility is that the adsorbed water promotes the surface mobility of the oxalato radicals such that the radicals have increased probability of oxidizing Fe(I1) .GO4-
+ Fe(1l)
+ Fe(II1)
-+ Cz042-
(12)
This explanation has the attractive capability of explaining the lack of effect when Ka[Co(Cz04)3].3Hz0is evacuated. The photoproduct Co(I1) is more stable than Fe(II), and we presume that oxalato radicals would react little with Co(I1). However, in dilute solid solution, Sarma, Fenerty, and Spees61 gave conclusive evidence for back-reaction of photoproducts of K3[Co(CZO4),] -3HzO. Of course, our conditions are different, and it is not surprising to find different results. At present there is no contradiction between this second explanation and the experimental findings. Other recombination steps undoubtedly are irnportant; several may be written, but two will suffice for this discussion
+ .CzO4- +Cz0d2Fe(I1I) h + Fe(I1)
e
-+
Conclusions Some comment is warranted about the several techniques for measuring t$ with seemingly discordant results. The interesting situation of greater-than-unity values for solution experiments vs. less-than-unity values for solids has already been discussed. Additional attempts should be made with solids to find values of t$ greater than unity. It is clear that experimental conditions such as crystal size, solvent history, pressure, and ambient conditions are important. The dilute-solid-solution technique has not been reported for K3 [Fe(Cz04)3].3Hz0 yet, but low values of t$ presumably would be found, analogous to the low values of quantum yields for production of Co(I1) found for dilute solid solutions of Ka [Co(C~04)3]3H20.61 Apparently, back-reaction between photoproducts, which was actually demonstrated,61is of overriding importance for this type of materials. The complexes are molecularly dispersed; the cage of the surrounding solid greatly retards separation of photoproducts and greatly enhances back-reaction. I n solution, on the other hand, the cage is far less rigid, photoproducts can separate, and high values of t$ result. I n the solids reported here, separation of photoproducts is achieved by the electronic carriers. If this were not so, photoconductivity would not be detected. Thus high values of d, are found. The mechanism proposed here is greatly expanded over the brief discussion given earlier.3 It is stated mostly in chemical terms. Physical factors such as sites of reaction, atomic arrangements, trapping probabilities, lifetimes of various entities, and mobilities through the bulk and along interfaces are highly significant. At the present time, practically nothing is known about these factors; future research will deal with them, undoubtedly. +
Acknowledgments. We appreciate the X-ray analyais made by Mr. S. Donley and the mass spectrometric analysis by Mr. G. P. Happ.
(13)
(14)
High values of t$ for evacuated crystals and for KBr matrices4 indicate that these recombination steps can be minimized, especially for small crystals.
The Journal of Physical Chemistry, Vol. 76,N o . 19, 1971
For the unevacuated coatings, the intensity dependence of t$ reported earliera could arise from recombination reactions such as (11) or (12) competing withreaction 6. Evacuation removes adsorbed water, eliminates either reaction 11 or 12, and eliminates the intensity dependence.
(15) R . C.Gross, Abstracts, 156th National Meeting of the American Chemical Society, Atlantic City, N. J., Sept 1968, No. PHYS 50. (16) Recombination induced by water adsorbed to semiconductors is discussed in A. Many, Y. Goldstein, and N. B. Grover, “Semiconductor Surfaces,” North-Holland Publishing Co., Amsterdam, 1965, Chapter 9.
