1098
BARNEY L. BALESAND LARRYKEVAN
range it becomes an extrinsic semiconductor because of the doped NiO. Therefore it does not satisfy the general semiconductor equation, cr = Ae-E/RT,l 1 but shows a higher electrical conductivity. Sample 6 and 7, which show little change in conductivity over the temperature range 400-900”, show Fermi degeneracy because of a very high concentration of nickel oxide and conduction electrons in the system of NiO-a-FezOa. Above 900” the carrier concentration is constant because of the complete ionization of the donors; on the other hand the mobility of the carriers decreases with rising temperature due to the “impurity scattering”12 because of the excess nickel oxide. Therefore the electrical conductivity is reduced.
The values of the electrical conductivity measured both as the temperature was raised and lowered are similar. Therefore the sample was in a state of thermal equilibrium a t the moment of measurement. It appears that the new contact method devised in this laboratory is satisfactory.
Acknowledgment. The authors are grateful to the Graduate School of Yonsei University and Ministry of Science and Technology of Korea for support of the experimental work. (11) L.G.Uitert, J . Chem. Phys., 23,1883 (1955). (12) N. B. Hannay, “Semiconductors,” Reinhold Publishing Corp. New York, N . Y., 1959.
Electron Paramagnetic Resonance Studies of Silver Atom Formation and Enhancement by Fluoride Ions in ?-Irradiated Frozen Silver Nitrate Solutions by Barney L. Bales’ and Larry Kevan2 Department of Physics and Department of Chemistry, University of Kansas, Lawrence, Kansas (Received July 14,1909)
060.44
The y radiolysis of AgNOa ices a t 77°K produces trapped Ago, OH, and NOz. Addition of fluoride ion increases Ago, prevents NO2 formation, increases the total number of observable spins, and increases the linear range of the dose-yield curve for Ago. The initial Ago yields are G(Ago) = 1.2 in 1M AgNOs and G(Ag0) = 3.2 in 1 M AgNOa-0.5 M KF. These effects indicate that fluoride ion acts as an efficient hole trap for HzO+and prevents electron-hole recombination. Low fluoride concentrations are eff eotive; thus some HzO+ is mobile.
Introduction The radiolysis of ice has been the subject of much researcha and is understood in terms of the initial reaction scheme given in (1) and (2). The details of the
+ OH HzO+ + em-
€ 1 2 0 --+ H
H20 --+
(1.)
(2)
fate of the electron and the hoie depend on the nature of the solutes present in the ice and, to some extent, on the phase of the ice.“ Shields6 and Zhitnikov and Orbelie have shown that 7-irradiated silver salt solutions at 77°K yield silver atoms which were attributedb~~ to the electron-capture reaction em-
+ Ag+ +AgO
(3)
The epr spectra of Ago were analyzed, but no quantitaThe Journal of Physical Chemistry
tive data on yields or on the reactions occurring in the frozen system were reported. One interesting observation was that fluoride ion enhanced the yield of A@ in the irradiated silver salt ices. Shields6 inferred that fluoride ion “promotes” the reactivity of electrons with (1) Department of Physics. (2) Department of Chemistry. Address inquiries to this author at Department of Chemistry, Wayne State University, Detroit, Mich.
48202. (3) L. Kevan in “Radiation Chemistry of Aqueous Systems,” G. Stein, Ed., Interscience Division, John Wiley & Sons, Inc., New York, N. Y., 1968,pp 21-72. (4) H. Hase and L. Kevan, J.Phys. C ~ E W 73,3290 Z., (1969). (5) L.Shields, J . Chem. Phys., 44, 1685 (1966); Trans. Faraday soc., 62, 1042 (1966); I,, Shields and M. C. R. Symons, Mol. Phys., 11, 57 (1966). (6) R. A. Zhitnikov and A. L. Orbeli, Fiz. Tverd. Tela., 7,1929 (1905); Sou. Phys. Solid State, 7,1559 (1960). (7) B.L. Bales and L. Kevan, Chem. Phys. Lett., 3,484 (1969).
1099
EPR STUDIESOF SILVERATOMFORMATION Ag+ but did not suggest a mechanism. I n this work we investigate the radiation chemistry of AgNOa frozen solutions and the mechanisms which lead to Ago and its enhancement by fluoride ion. The relative radical yields and the magnitude of the fluoride effect change drastically with dose. The fluoride effect can be explained by hole trapping by fluoride to prevent electron-hole recombination.
Experimental Section Reagent grade chemicals and triple-distilled water were used to prepare the solutions. Concentrations were determined by the solute weight. Nondegassed solutions were used since degassing of solutions had no effect on the radical yields measured in the solid state. Drops containing 5 pl of solution were dropped directly into liquid nitrogen in which they rapidly froze to form opaque, spherical samples. Because of their opacity the samples were considered to be largely polycrystalline. This type of sample preparation avoids the use of any irradiation cells which could contribute impurities and background epr signals. I n order to have maximum sensitivity, 8-12 spherical samples were irradiated and measured as a single sample. The samples were immersed in liquid nitrogen and were irradiated in the dark in a cobalt-60 y irradiator a t a dose rate of 0.4 Mrad/hr. The dose rate was measured by ferrous sulfate dosimetry using G(Fe+3) = 15.5. After irradiation, the samples were poured into a quartz epr insertion-type dewar together with liquid nitrogen thus keeping the sample temperature at 77°K. Helium gas was bubbled through the liquid nitrogen in the dewar above the microwave cavity to eliminate bubbling within the cavity which is a source of noise. A Varian X-band reflection-type epr spectrometer using 100-kHz field modulation was used to make the measurements. Field modulation and microwave power amplitudes were kept low enough to avoid modulation broadening or power saturation. Radical yields per 100 eV of energy deposited by the y irradiation (G value) were found by comparing the doubly integrated first derivative spectra in the irradiated silver nitrate samples with the uncorrected trapped-electron spectrum in irradiated 10 M NaOH. The G value for the uncorrected trapped-electron spectrum was taken to be 2.1.3 Results Figure 1 shows epr spectra of y-irradiated 1M AgNOs (Figure la) and 1 M AgNOa-0.5 M K F (Figure lb). The dose was 3.0 Mrads and the measurements were taken a few minutes after irradiation. The spectra are similar to those reported in the literature.6re Figure l a consists of features due to A@, the groups of lines split by approximately 500 G; OH, the central doublet; and NOz, the central triplet which is partially obscured by the hydroxyl radical. The stick diagrams in the
Figure 1. First-derivative epr spectra of 7-irradiated frozen solutions at 77°K: (a) 1 M AgNOa; (b) 1 M AgNOs-0.5 M KF. Note that the spectrometer gain in (a) is a factor of 8 larger than in (b). Dose is 3.0 Mrads.
0
.
0
l 0.2~
" 0.4 " " Cos'
0.6 ' ~
" 0.8'
J
I .o
Y-DOSE (Mrad)
Figure 2. Number of Ago atoms per gram as a function of -,-irradiation dose a t 77'K: 0 , 1 M AgNOs; 0, 0.1 M AgNOs.
figure show the positions of the main features. Figure l b is similar to Figure l a except that no NO2 is detected (even at high gain) and the yield of Ago is much larger. Note thalt the spectrometer gain in Figure l a is higher by a factor of 8 than in Figure l b and that the hydroxyl radical yield in the two systems is comparable. The doubling of the features due to Ago results because naturally occurring silver has two spin ' / z isotopes which are almost equally abundant. The complexity of the epr of Ago is due to g anisotropy and to the fact that the silver atoms are trapped in several sites of different symmetry, a fact which has been interpreted7 as being due to different water dipole 0rientation.s in the solvation shell of Ago. Full details on the epr parameters of Ago in various sites are reported in ref 7. The relative concentration of Ago in various sites is time dependent a t 77"R, but the total Ago concentration depends only on irradiation dose.? Measurements of Volume '74, Number 6 March 6, 19YO
1100
BARNEY L. BALESAND LARRYKEVAN
I CosO 7-DOSE ( M r a d )
Figure 3. Number of Ago atoms per gram &g a function of yirradiation dose at 77°K: 0,1 M AgNOs-l M K F ; 0, 1 M AgNOa.
Ago concentration require double integrations of rather complex spectra rather than peak-to-peak line intensity measurements and this is the main source of error in determining silver atom yields. It is estimated that relative silver yields are accurate to f15% and absolute silver yields (as determined by comparison with the trapped-electron yield in irradiated 10 M NaOH) are accurate to f30%. The Ago yield as a function of dose is given in Figure 2 for 1 M AgN03 and 0.1 M AgNOa. These results are not completely reproducible when solutions of AgNOa
I .o
0.8
'D '0
;0.6 OI
\
m -1 a
0
2 0.4 K
0.2
0.0
Figure 4. Number of radicals per gram in 1 M AgNOa aa a function of ?-irradiation dose a t 77'K: 0,OH; 0 , NOz. The Journal of Physical Chemistry
which have been allowed to "age" are used. The reproducibility of several samples from the same solution is *lo% while samples prepared from freshly made solutions give less intense signals. The reported results are all for freshly made solutions. The yielddose curve in Figure 2 is linear only to doses less than 0.05 Mrad. Figure 3 shows the yield-dose curve over a wider dose range in 1 M AgNOa and in 1 M AgN03-1 M KF. I n the presence of fluoride, the linearity of the yielddose curve for Ago is increased to about 1.0 Mrad. Also in the fluoride-containing system the silver atom concentration reaches a plateau at high dose, whereas in the AgN03 system without fluoride, the Ago concentration goes through a maximum and decreases a t very high doses. At these high doses the nonfluoride-containing samples change from bright yellow to brown.
Table I: Initial G Values of Paramagnetic Species" at 77°K Ago
Ice 1 M KNO; 1 M KNOa0.5 M KFb 1 M AgNOs0.5 M K F 1 M AgNOa a
... ...
N OB
OH
Referenoe
... 3 0.36 I 0.10 This work ... 0 . 0 9 I 0 . 0 3 This work 3.2 i 0.5 1 . 5 I 0.4 0.0 This work 1.2 f 0.2 1.5 f 0.4 0.20 I 0 . 0 6 This work
.
0.8
. ... ,,
The errors quoted are relative to G(et-)
* Irradiated and measured in the dark.
=
2.1 in 10 M NaOH.
Figure 4 gives the yield-dose curve of NO2 and OH in 1 M AgN03. As is illustrated in Figure 1, the OH yield is not a function of fluoride concentration while NO2 is not produced in fluoride-containing silver nitrate ice. The yield of OH at low dose was estimated by comparing the peak-to-peak height of the low-field line of the apparent doublet in ice and silver nitrate ice. The NOz yield was determined by measuring the peakto-peak height of the high-field line. A factor was determined relating the peak-to-peak height to the integrated area by doubly integrating the NO2 spectrum in 7-irradiated HNOa ice and correcting for a small linewidth difference. The results were compared with the trapped electron in irradiated 10 A4 NaOH to calculate the yield. The yield-dose curves of OH in 1M AgNOa and ice are similar except that the absolute yield is larger in the former. The yield-dose curves of NO2 in 1 M AgN03 and 1 M KN03 are somewhat different: the initial yield is less in the former (Table I), but the linearity extends over a longer dose range. The curve is linear in potassium nitrate to -0.05 Mrad while in silver nitrate the curve is linear to -0.25 Mrad.
1101
EPRSTUDIESOF SILVER ATOMFORMATION Table I gives the initial G values of the paramagnetic species in ice, and in ice containing AgNOa, AgNO3KF, KNO3, and KNOa-KF. At low doses, Ag+ increases the production of OH and decreases the yield of NOz. I n 1 M AgNOa, fluoride does not affect the OH yield, eliminates NOS, and enhances Ago. The epr spectrum of irradiated 1 M AgN03 a t low dose is composed of approximately 41% AgO, 52% OH, and 7% NOz. I n the presence of 0.5 M fluoride a corresponding spectrum consists of 68% Ago and 32% OH. At higher doses, due to the differences in the dose-yield curves of the various paramagnetic species, the difference in total spin concentration and per cent Ago concentration can vary vastly as is evident in Figure 1 where the total spin concentration is larger in Figure l b by 200% while the silver yield ratio in Figure l b is 20 times that in Figure la. The effect of fluoride ion concentration on the Ago yield in irradiation of 1 M AgNOs ice is shown in Figure 5 for a low dose (0.03 Mrad) and a high dose (4.5 Mrads). The fluoride ion is quite effective in enhancing Ago even at low concentrations. A yield of NO2is observed a t low dose up to a concentration of W ~ Q - M~ KF. Above loWaM K F only OH and no NO2 is observed. However, a t concentrations above about
25 20
15
IO
‘0 m
10-4
16‘ IO-[ I m 0 les/ Ii t er
IO
I
I
I
I
I
I
IO4 5xIOm3 5 ~ 1 65xIO-l ~
[Ag+]
moles/iiter
Figure 6. Initial G(Ago) as a function of silver nitrate concentration: 0 contains 1 M KF; 0 contains no KF. Dose is 0.04 Mrad.
0.5 M KF, a new three-line spectrum whose perpendicular features are 91 =2.005 and A1 = 63 f 2 G appears. The parallel features are obscured somewhat, by the overlapping lines of the various radicals around g = 2. This spectrum is similar to those attributed in the l i t e r a t ~ r eto * ~N~ O P and is identified here as such. Further confirmation is provided by the fact that irradiation by ultraviolet light converts our spectrum assigned to NOa2- into NOz. This conversion by ultraviolet is characteristic of N032-. Note that the Ago concentration reaches a maximum at about 0.5 M K F and decreases at higher fluoride concentration. A search was made for the spectrum of F2- in 1 M AgN03-3 M KF by looking for perpendicular epr features at gL = 2.023 and A L = 296 G,l0but no evidence for this species was found. The yield of silver atoms as a function of silver nitrate concentration is given in Figure 6, in the presence and in the absence of fluoride ion. The silver atom yield has a very nonlinear dependence on the silver nitrate concentration and also shows a very strong enhancement due to fluoride ion even at very low concentrations of Ag+. The yields of OH and NO2 show similar nonlinear behavior with silver nitrate concentrations. The effect of added fluoride ion on the radicals produced in the radiolysis of 0.5 M KNOa was also briefly investigated (Table I). The addition of 0.5 M K F to 0.5 M KNOs causes the NO2 radical to be reduced by a factor of about 4 and causes the NOa2- radical to be increased by a factor of 2-3.
[F-]
Figure 5 . Number of Ago atoms per gram in y-irradiated 1 M AgNOs a t 77°K as a function of fluoride concentration: (a) y dose 0.03 Mrad; (b) y dose 4.5 Mrads. The solid line in (a) is a smooth curve through the experimental points. The solid line in (b) is a plot of [F-]1/2.
(8) P. B. Ayscough and R. G. Collins, J. Phys. Chem., 70, 3128 (1966). (9) B. G. Ershov, A. K. Pikaev, B. Ya. Glaaunov, and V. I. Spitsyn, Dokl. Akad. Nauk SSSR, 149,363 (1963). (10) P. W.Atkins and M. C. R. Symons, “The Structure of Inorganic Radicals,” Elsevier Publishing Go., Amsterdam, 1967, p 115. Volume 74, Number 6 March 6, 1970
BARNEY L. BALESAND LARRYKEVAN
1102
Discussion A . Primary Reactions in AgN03 Ices. The primary species produced in the radiolysis of ice are given in reactions 1 and 2 . At 77°K the directly produced OH radicals are trapped, but the H atoms are mobile. The H atoms back-react to some extent with OH radicals and also react with each other to form Hz. In this work, we are primarily concerned with the reactions of the electrons and holes produced in reaction 2, both of which are considered to be mobile. I n NaN03 ice, the mobile electrons react with the anion as shown in reaction 4.89g From the suggested correlation of mobile electron reaction rates with solutes in ice,” electrons are expected to react several times faster with Ag+ than with NO-3 in AgNOa ices. The reduced NOz yield and the absence of N03’- in AgN03 ice compared to NaN03ice bear this out. Some electrons must recombine with holes as portrayed in reaction 5. I n addition to reaction 5, holes react by reactions 6 and 7 in NaNOa ice.12 e,-
+ NOa- +NO?-
-+
+ HzO+-+
e,-
NOZ
+ 20H-
HzO
+ H20 +H30+ + OH HzO+ + NosNO2 + 0
H2O+
-+
(4) (5) (6)
(7)
As shownin Figure 2 the yield-dose curve is linear to less than 0.05 Mrad. This suggests that there is another reaction which destroys Ago a t relatively low concentration. We suggest that the reaction is (8). The postulate that mobile holes reduce the Ago yield at low concentrations is supported by the fluoride effects discussed below. If reaction 8 is primarily H2O+
+ Ago--+ Ag+ + HzO
(8)
responsible for the shape of the yield-dose curve, then it implies that some HzO+can travel considerable distances in ice without participating in reaction 6. B. Hole Trapping by Fluoride Ion. The addition of fluoride ion to silver nitrate ices has several dramatic effects. It (a) increases the Ago yield, (b) prevents the formation of NOz at low y doses, (c) increases hhe total number of observable trapped paramagnetic species, and (d) increases the linear range of the yield-dose curve for Ago. All of these effects can be explained by the assumption that fluoride ion acts as an efficient hole trap for H 2 0 + as indicated by reaction 9. The comHzO+
+ F- +HzO + F
(9)
petition of (9) and (8) increases the Ago yield and increases the linear range of the yield-dose curve of Ago. The competition of (9) with ( 7 ) prevents the formation of NOz. This, in fact, supports the mechanism of NO, formation by hole reactions.lZ Finally, competition of (9) with (5) increases the total number of observable trapped paramagnetic species by allowing more electrons to react with Ag+. The Journal of Physical Chemistry
Reaction 9 is postulated to form trapped fluorine atoms. If the fluorine atoms are trapped as such, they would not be expected to show an epr spectrum in the solid phase because of a strong spin-orbit interaction which probably broadens the lines beyond detection. A search was made for the known epr spectrum10 of Fz-, but no evidence for it was found. The fact that fluoride ion exhibits its effects at very low concentrations also argues against the involvement of FZ-. I n work on frozen chloride solutions, GIz- is formed only a t concentrations above 0.5 M.13 The possibility that a silver fluoride complex is responsible for the effects observed can also be ruled out because the fluoride effects occur at such low fluoride concentrations. If fluoride ion indeed acts as a hole trap, it should also do so in other systems. To test this, the effect of fluoride on the radical yields in irradiated KNOl ice was examined (Table I). It was predicted that the NOz yield, which is mainly formed by reaction 7 , would be decreased while the N032- yield would be increased. This is precisely what was observed as shown in Table I. The decrease in the NO2 yield was expected because of competition of reaction 9 with 7. Also, competition of (9) with ( 5 ) allows more electrons to participate in reaction 4 in the absence of silver ions. The NOZyield is slightly smaller in AgN03 ice compared with KNO, ice. This is expected because in AgN03ice most of the electrons react with Ag+, whereas in the KNO, ice most, of the electrons react with NOa- and form a small amount of NOz. C. Radical Yields. The radiolysis of pure ice a t 77°K produces G(0H) = O.tJe3 Addition of causes reaction 3 to compete with ( 5 ) . This causes the OH yield to increase by reaction 6 relative to pure ice because fewer holes undergo reaction 5. The NO3 arises largely from (7). In 1 M AgT\T03,Table I shows that the total yield of trapped-electron species (Ago)is a little smaller than the total yield of trapped-hole species (OH NO2). Some Hzmay be formed as is postulated in pure ice.3 However, within the experimental error the yields of electron and hole species are approximately equivalent. Figure 6 shows that the increase in Ago with Agn’O3 concentration is quite sublinear and that the Ago yield plateaus at 0.5 M AgN03. The Ago yield is approximately linear with either the square root or cube root of to 5 X the silver ion concentration from 5 X M Ag+ but deviates considerably at higher concentrations. This concentration dependence is consistent with scavenging electrons by Ag+ from recombination reaction 5 if H20+ and e,- are considered to be diffusfrom an initial inhomogeneous distribution in
+
L. Kevan, J . Amer. Chem. Boc., 89,4238 (1967). L.Kevan, J . Phys. Chem.. 68,2590 (1964). D.M.Brown and F. S. Dainton, Nature, 209, 195 (1966).
1103
EPRSTUDIESOF SILVER ATOMFORMATION spurs.'29l4 The cube root dependence has usually been applied to scavenging in irradiated aqueous solutions16 while the square root dependence has more recently been applied to ion scavenging in irradiated organic 1 i q ~ i d s . l ~Both correlations fail a t high solute concentrations as is also found here. The effect of added K F as a hole trap has been discussed. It is striking that the optimum concentration of K F (0.5 M ) increases the maximum, initial G(Ago) from 1.2 to 3.2. This high value is reasonable for the maximum ionization yield of electrons in the system. This means that reaction 5 is nearly eliminated in 1 M AgNOr0.5 M KF. Note that both electron and hole traps are necessary to achieve this condition. The concentration dependence of the enhancement of Ago by F- is quite sublinear (Figure 5) and approximately fits square root or cube root dependencies on the F- concentration. The fit is better a t high doses than a t low doses. As discussed above for the Ag+ concentration dependence, this is expected for ion scavenging in irradiated systems. D. Mobile and Nonrnobile Holes. I n the presence of F-, NO2 is eliminated. Thus F- is a better hole trap than NOa- in ice a t 77°K; Le., F- effectively competes with NOa- in reaction 7. Fluoride ion increases the initial Ago yield by scavenging holes from reaction 5. On the other hand, the OH yield is unaffected by F- so F- does not scavenge the holes in reaction 6. w e suggest that there may be two populations of holes which are differentiated by their reactivity. One H2O+ population readily undergoes reaction 6 with an H2O
molecule in its own solvation shell, is not appreciably mobile, and is not scavengeable. The yield of these nonmobile holes is the maximum OH yield in AgN03 ice (G = 1.5). The other H 2 0 +population reacts only slowly, if a t all, by reaction 6, is mobile, and can be scavenged. This mobile hole yield is given by the increase in Agodue to F- (G = 2). If we postulate the existence of two types of holes in ice, how can they be described physically? We suggest that the mobile hole is unrelaxed; that is, it has the same geometry as a neutral water molecule so charge conduction is facile. Reaction 6 is inefficient because charge conduction is rapid. The nonmobile hole is relaxed and has attained a new equilibrium geometry consistent with the loss of one electron from the neutral molecule. The relaxed hole readily undergoes protontransfer reaction 6 but charge conduction is slow because of the change in geometry required. This physical description of two types of holes in ice is rather analogous to a recent proposal by Hamill for two types of holes (dry and hydrated) in liquid water radiolysis.ls
Acknowledgment. This research was supported by NASA, the AEC, and the Air Force Rocket Propulsion Laboratory. This is AEC, Document No. COO1528-35. We wish to thank Dr. P. Hamlet for helpful discussions. (14) J. M. Warman, K. D. Asmus, and R. H. Schuler, Advances in Chemistry Series, No. 82, American Chemical Society, Washington,
D, c., 1968, 25. (15) H. A. Schwartz, J.Amer, Chem. Soc., 77,4960 (1955), (16) w. M. Hamill, J. Chem. Phys., 49,2446 (1968).
Volume 74, Number 6 March 6, 1970