Radiolysis and Energy Transfer in the Adsorbed State - The Journal of

Radiolysis and Energy Transfer in the Adsorbed State. J. G. Rabe, Birgit Rabe, and A. O. Allen ... S. A. Ruetten and J. K. Thomas. Langmuir 2000 16 (1...
1 downloads 0 Views 1014KB Size
J. G. RABE,B. RABE,AND A. 0. ALLEN

1098

Radiolysis and Energy Transfer in the Adsorbed State

by J. G. Rabe, Birgit Rabe, and A. 0. Allen Chemistry Department, Brookhaven iyational Laboratory, Upton, New York 11973

(Received September 22, 1966)

The radiolysis of azoethane adsorbed on various solids was studied as an indicator of the energy-transfer processes occurring within the solids and a t the surface. On the insulators Sios, NgO, and various alkali haIides considerable decomposition occurred, which was assumed to result from energy transfer to the adsorbed molecules. The semiconductors ZnO, TiOsJ NiO, and graphite gave little or no sensitization, as expected, since the available quantum energy should not be much larger than the band gap. Some phenomena peculiar to the MgO system are described and are attributed to closer coupling between excited states of the solid and of the adsorbed molecules than occurs in other systems. A chemical method is described for determination of the number of negative color centers in an irradiated sample of alkali halide. The decomposition of azoethane occurring on contact with irradiated solids was studied and is discussed. I n radiolysis, a considerable decrease in yield of decomposition of adsorbed azoethane with increase in the total dose was generally found and was attributed to hindrance of energy transfer by radiation-produced defects.

The study of radiolysis of substances in a state of adsorption on chemically inert ~ u p p o r t s l - ~ has shown clearly that excitation energy, delivered to the solids from the radiation, can move to the surface and cause decomposition of the adsorbate. The yield of the decomposition depends very strongly on the properties of the solid. These radiolytic processes can serve as a tool for the study of excitation in solids and may perhaps indirectly shed useful light on the fundamental processes occurring in surface-catalyzed reactions. The organic molecule may be regarded as an indicator of the processes of energy migration which are occurring in the solid support. I n the present work, we chose as adsorbate azoethane, (C2H5)2r\'zJ since it readily decomposes to nitrogen which is easily separated and measured. Small yields of hydrogen and methane, which are measured together with the nitrogen, serve as indicators of other modes of decomposition which probably require more energy than needed for the production of nitrogen. The quantum of energy available for delivery to the adsorbed molecule is not likely to be larger than the energy gap between the ground state and conduction band in the solid. One object of the present study was to see if a correlation existed between the yield of decomposition of the organic molecules and the size of the band gap in the solid support. It was hoped that The Journal of Physical Chemistry

the yields of the processes would lead to some idea of the distance over which excitation energy can migrate without dissipation in solids. A preliminary account of this work appeared as a Communication to the Editor.5

Experimental Section Azoethane, synthesized by the method of Renaud and LeitschJ6 was purified by pouring it through a column of activated alumina, followed by gas chromatography on diethylene glycol. The purity was better than 99.8%. The solids were prepared and were outgassed by heating under vacuum, as shown in Table I. The sample preparations were carried out in a grease-free system using metal and Teflon valves. Azoethane was measured as vapor in a calibrated volume and was condensed onto the solid in a glass ampoule fitted with a break-seal and sealed off under vacuum. The irradiations were all made a t 23" with cobalt-60 (1) J. M. Caffrey, Jr., and A. 0. Allen, J . Phys. Chem., 62,33 (1958). (2) J. W.Sutherland and A. 0. Allen, J . Am. Chem. Sac., 83, 1040 (1961). (3) R. R. Hentz, J . Phys. Chem., 68, 2889 (1964). (4) Yu. A. Kolbanovskii, L. S. Polak, and E. B. Shlekhter, Dokl. Akad. Nauk SSSR, 136, 147 (1961). (5) J. G. Rabe, B. Rabe, and A. 0. Allen, J . Am. Chem. Sac., 86, 3887 (1964). (6) R. Renaud and L. C. Leitsch, Can. J . Chem., 32, 545 (1954).

RADIOLYSIS AND ENERGY TRANSFER IN THE ADSORBED STATE

~~~

1099

~

Table I: Preparation and Properties of Solids -Heated Temp, "C

SP surface, Name

m2/g

Si02 gel A Si02 gel B MgO 600 MgO 800 MgO 1140 ZnO NiO TiOz Graphite NaCl

775" 500" (-100)" 29" 29"

NaF LiF

KI

... ... ... 3.9" 4.7" 8" 12" 5.1"

Starting material

14-20 mesh gelb SiOz gel A Basic MgC036 Same Same ZnCOad Pptd NiC03 (b)

hr

...

800 600

16 16 16 16 16 4

O C

...

...

...

...

450 450 450 450 450 400 525 430 800 130

... ... ...

... ... ...

130 130 130

...

Ground in mortar 45 min," sieved to -300 mesh As receivedd As receivedb Same treatment as N a C I

-OutgassedTemp,

Time,

... 800 140 600 380

(f)

in air-

...

Time, hr

65 65 16 16 16 16 16 16 16 16 16 16 16

'

a Determined by BET method, using nitrogen. Fisher Scientific Co. Estimated from the reports of R. M. Dell and S. W. Weller, Trans. Faraday Soc., 55, 2203 (1959); 59, 470 (1963); and R. I. Razouk and R. Sh. Mikhail, Actes Congr. Intern. Catalyse, $e, Paris, 1960,2,2023 (1961). Baker and Adamson Chemical Gorp. e Determined by the supplier. National Carbon Co. Baker Analyzed.

'

y rays a t a dose rate of about 0.25 Mrad/hr. Total doses ranged from 0.25 to 40 Mrads. The fraction of azoethane decomposed was less than 1% in the great majority of the runs. After irradiation, the ampoule was sealed to the vacuum line, the break-seal was opened, and the permanent gas fraction was pumped (by a Toepler pump) through a liquid nitrogen trap into a McLeod gauge; the gas was then transferred to the gas chromatograph for analysis. Separation was accomplished over a 4-m column of Linde Molecular Sieve 5A a t room temperature with argon as carrier gas. The sample was passed successively through a thermal conductivity detector (for measurement of hydrogen and nitrogen) and a flame ionization detector (for measurement of methane). Gas volumes were determined from peak heights as calibrated with known samples; the total amount of gas was always checked against the amount determined in the NcLeod gauge, and the results to be acceptable had to agree within 2%. No attempt was made to determine decomposition products other than the permanent gases. To determinc: the amount of electrons trapped as color centers in irradiated salt samples, the ampoule containing the salt was opened to the vacuum system by the break-seal, and previously degassed water or aqueous solution was poured into the salt. The apparatus was agitated by hand until all of the salt was dissolved. The hydrogen gas produced was then pumped out and measured in the McLeod gauge and again in the gas chromatograph.

Results Radiolysis of pure liquid azoethane gave G(N2) -3.69, G(H2) = 0.46, G(CH.J = 0.076. Most of the present results are represented by graphs showing the amount of product formed per unit amount of solid, by a constant dose of radiation, as a function of the amount of azoethane present per unit amount of solid (here called the "coverage"). It is assumed that the azoethane covers the available surface with reasonable uniformity and that to form a monolayer would require about 4 pmoles of azoethane/m2. All of the present data were obtained a t coverages of less than one monolayer. On many graphs we have plotted the amount of product that would be expected to result from energy absorbed directly by the organic material from the radiation, assuming that the energy is taken up in the two phases in proportion to their electron fractions and that the specific energy yield G is the same as for the pure liquid organic compound. This quantity is indicated on the graphs as the "liquid line." Of course, the assumptions of energy partition between phases, and equal efficiency as between the liquid state and the adsorbed state, should not be expected to hold exactly, but an amount of product more than twice that given by the liquid line presumably provides an indication that energy taken up by the solid is being transferred to the adsorbate and is utilized for its decomposition. Figure 1 shows the nitrogen obtained a t a dose of 2 Mrads on various alkali halides. Some yields obtained on silica gel a t 1.5 and 5.9 Mrads are shown for comVolume YO, Number 4

April 1966

1100

J. G. RABE,B. RABE,AND A. 0. ALLEN

I

I

I

I

I

14

5

I

5.9

M

I

,

Y SILICA

,

/

GEL

12

O.I0

SI IO

I

1

t

I

1

3

4

N

E

$

8

$

4

.08 .O9I

$ 2 ‘0

0.2 0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

w

a

pmole AZOETHANE PER m e SOLID

Figure 1. Nitrogen production from azoethane on alkali halides at 2 Mrads and on silica gel A at 1.5 and 5.9 Mrads.

parison. Figure 2 shows nitrogen yields obtained a t 5.9 Mrads in the two preparations of silica gel, on three preparations of MgO, and on graphite. It is seen that the efficiency of energy transfer is similar a t very low coverages for the salt, silica, and magnesia; as the coverage is increased, the yield of product per molecule of azeoethane present rapidly decreases with the first two solids but remains constant up to much higher coverage for the magnesia, so that much higher total yields are obtained with the latter substance. The curves for the magnesia remain linear up to a sudden maximum, beyond which the yield suddenly declines, showing that above a certain coverage additional amounts of adsorbate protect the material already there against radiolytic decomposition-a phenomenon not found with the other solids. Figure 3 shows some results obtained with nickel oxide, zinc oxide, and titanium dioxide, expressed here as amount of product obtained per unit weight of solid, rather than unit surface as in the preceding graphs. As may be seen by comparing the results with the liquid lines shown, the yields in these semiconducting oxides are much smaller than in the insulating substances. Indeed the zinc oxide is seen to exert a definit,e protective effect for the azoethane with respect, to decomposition to nitrogen although the yield of hydrogen appears to be somewhat increased. (This hydrogen might however arise from adsorbed water or surface hydroxide groups, rather than from the azoethane.) The band gap in MgO is 8.7 ev,’ in the alkali halides it is thought to be around 10 ev,* and it is probably equally as high in silica, while the corresponding quantity in ZnO is only 3.2 e ~ in, TiOz ~ 3 e ~ in, NiO ~ probably about the same, and in graphite close to zero.9 Thus, as expected, the insulators show a large transfer of energy to the The Journal of Physical Chemistry

p

.03

.o I 0

0

2

I

p

~

o

~

~

AZOETHANE PER m p SOLID

Figure 2. Nitrogen produced from azoethane on various solids a t 5.9 Mrads. The points for graphite (labeled C) coincide with the liquid line.

adsorbate, while the semiconductors give a relatively small effect. The magnesium oxide data at 5.9 Mrads are shown in more detail in Figure 4. The same data were given in our preliminary p~blication.~At low coverages the yields of methane and nitrogen are practically equal although on the other solids the methane yield was much smaller than the nitrogen, usually by a factor of 10 or more. However, on magnesia the methane reaches its maximum at a lower coverage than does nitrogen. There is some indication for the 800’ preparation that a maximum may also be reached at extremely low coverages for the hydrogen yield. The energy transfer is seen to attain higher values, the higher the calcination temperature of the hlgO. Figure 5 shows the radiolysis of a saturated and an unsaturated hydrocarbon on MgO. In both cases the evolution of hydrogen is suppressed by the presence of MgO although the formation of methane seems to be somewhat sensitized. Figure 6 shows the results of irradiating mixtures of azoethane and 2-methylpentane. (7) G. H. Reiling and E. B. Hensley, Phys. Rev., 112, 1106 (1958). (8) J. H. Schulman and W. D. Compton, “Color Centers in Solids,” The Macmillan Co., New York, N. Y., 1962, p 6. (9) W. C. Dunlap, Jr., “An Introduction t o Semiconductors,” John Wiley and Sons, Inc., New York, N. Y., 1957.

RADIOLYSIS AND ENERGY TRANSFER IN THE ADSORBED STATE

1101

1

I

I

I

/(2

I

-ME

- PENTANE)

/

/

0.3 LL

0

I a a 0

a W

p

0.2

I-

o 3

D 0 [r

n

W -1

g

0.1

3.

20

0

p

40 60 80 100 M O L E AZOETHANE PER GRAM OF S O L I D

40

100

80

60

pmole HYDROCARBON / g SOLID

20

Figure 3. Products from aeoethane on semiconducting oxides: open symbols, ZnO ; half-filled symbols, TiOz; solid symbols, NiO; circles, Nz; triangles, Hz; squares, CHa.

Figure 5. Hydrocarbon radiolysis on MgO 800. Hexene-1 a t 4.4 Mrads; 2-methylpentane a t 5.9 Mrads: squares, CHd; triangles, Hz; open symbols, 2-methylpentane; solid symbols, hexene-l.

8 2.6-

3 2.4 -

g 2.2 -

600’

0

a

2

i $ &

fi

v

1.61.41.21.0-

-

18

IO

-pMOLE 2.0 -EtzN,PERq )

28

35

44

I,

-

1.6

%2c2

n

g 0.8 a2

-0 ’ 40 ’

do ’ I20 0 40 80 120 0 40 80 MICROMOLES OF AZOETHANE/GRAM OF SOLID

120

Figure 4. Products from azoethane radiolysis at 5.9 Mrads on MgO heated to different temperatures.

The methylpentane caused a reduction of the nitrogen yield, compared to that obtained with the azoethane present alone, when the azoethane coverage was low; but at high coverages, beyond the maximum in the nitrogen yield curve, adding methylpentane tended to cause an increase in the nitrogen yield. Thus, the effect of adding methylpentane was similar to the effectof decreasing the amount of azoethane present. Figure 7 shows the effect of total dose on the specific energy yield of nitrogen, G(Nz), for different coverages on silica gel. There is a marked decrease of the yield

IO 30

IO

p MOLE

30

IO

30

IO

30

IO

3(

2-METHYLPENTANE PER g

Figure 6. Product yields from mixtures of aaoethane and 2-methylpentane on MgO 800 at 5.9 Mrads.

with increasing dose. Figure 8 shows the effect of total dose for MgO, which is seen to behave in this respect quite similarly to the silica. Figure 9 shows that a similar effect occurs for the yield of total gas on sodium chloride. Irradiated salt must contain an equal number of negative and positive centers, usually referred to as “trapped electrons” and “trapped holes,” respectively. When irradiated salt is dissolved in water, one might expect the trapped electrons to react to yield hydrogen Volume 70,Number 4

April 1966

1102

J. G. RABE,B. RABE,AND A. 0. ALLEN

3.5 32.5 .

a 32

3.2

28

2.8

24

2.4

2! 0

k

\

c

8 20 a 12

\

0

2.0

16

b

00

- 1

.a .4

W

A ; 1;;;;

b

DOSE, Mrod

12

Iblll

Figure 7. Effect of total dose on the yield of nitrogen G(N2) (in molecules per 100 ev delivered to the total system of solid plus adsorbate) from azoethane on silica gel A. pmoles of azoethane/g of solid: 0, 105; X , 284; A, 598; 0 , 1100; 0 , 1810.

x=.

4

sw 0.5

2

1.2

- r 8

W

o

1 / , , , 1 , , , , I , , , , ( , / , , I , , , , 1 , , , , I , , , , , , , , ,I , , , j

0

5

IO

15

20

25

30

35

40

;

1.6

z

a

0

DOSE, Mrod

Figure 9. Effect of total dose on the excess yield of permanent gas (right-hand scale) from azoethane adsorbed on NaCl (10 pmoles/g) (yield corresponding to “liquid line” has been subtracted from total yield) and dose effect on the number of negative centers in the irradiated salt as determined by the hydrogen formed on dissolution in an acid aqueous solution containing 3 M ethanol: solid circles, salt irradiated under vacuum without adsorbate; open circles, irradiated in presence of azoethane.

MgO 800

0.5

PRE-IRRADIAT

0

1

2

3

4

5

DOSE, M r o d

0

1

2 3 4 DOSE, M r o d

5

6

Figure 8. Effect of total dose on product yields from aeoethane on MgO 800.

atoms or solvated electrons, while the holes should yield OH, C1, or other oxidizing radicals. The oxidizing and reducing radicals would then be expected to combine to a great extent with one another. Some hydrogen gas should form by combination of reducing radicals with one another, but its amount should be less than half the amount of trapped electrons present. If some acid and a sufficient concentration of an organic substance such as ethanol were added to the water, however, we would expect that all the negative centers would be rapidly converted to hydrogen atoms, which would then react with the alcohol to form a molecule of Hz and an organic radical, while all the positive centers would eventually react with the alcohol to abstract hydrogen, forming HzO or HC1 and another alcohol radical. Thus, when sufficient ethanol is present, a yield of hydrogen gas should be obtained The Journal of Phyaical Chemistry

that would serve as a measure of the number of electrons trapped in the salt. Figure 10 shows the hydrogen yields obtained when 1 g of irradiated sodium chloride was dissolved in acid solutions containing different amounts of ethanol. The yield is seen to reach a constant value at ethanol concentrations above 2 M . These determinations were made routinely using 0.1 IM H2S04and 3 M ethanol. Different samples of ground Baker Analyzed SaC1, irradiated separately to the same dose, gave hydrogen yields reproducible to within 5%, provided the bottles in which the salt was obtained bore the same lot number. Samples from different lots might differ as much as 20%. A saturated solution of hydrogen sulfide in water was found to give the same yield of hydrogen as the acidic alcohol solution, as was expected, since each negative center would form an H atom which would then abstract H from an €12S molecule to give a molecule of hydrogen. An acidic alcohol solution saturated with oxygen was tried and was expected to give a yield of hydrogen peroxide equal to that of the hydrogen formed in the absence of oxygen, provided that reactive negative and positive centers were present in equal numbers. It was thought that each center whether negative or positive should either react directly with oxygen to give a radical HOz or with the alcohol to give an alcohol radical which in its turn would react with oxygen to give HOz. Two HOz radicals would then react together to give a molecule of peroxide. I n fact, the yield of peroxide found was only 70y0of the yield of hydrogen produced in the absence of oxygen. The difference may be due in part to the low

RADIOLYSIS AND ENERGY TRANSFER IN THE ADSORBED STATE

i

;0 . 0 8 9

1

0.04

"

0

I

2

3

4

1103

m c

5

mole ETHANOL PER LITER

I

DOSE, Mrad

Figure 10. Effect of ethanol concentration on hydrogen formed when KaC1, irradiated to 2 Mrads, was dissolved in 0.1 iV HzSOa.

~i~~~~11. Total negative centers of irradiated salt, as determined by hydrogen evolution, and F centers determined optically, by assuming 0.9 oscillator strength.

solubility of oxygen in concentrated salt solution and also perhaps to the fact that some of the negative centers determined in this way are really hydride ions which are not accompanied by an equal number of reactive positive centers. It might be expected that the number of trapped electrons indicat,ed by this method should be nearly equal to the number of F centers produced by radiation in the solid. To test this point, we irradiated single-crystal specimens of NaCl, cleaved from large synthetic crystals obtained from the Harshaw Co., and determined the number of F centers present from the area under the F band in the absorption spectrum, assuming an oscillator strength of 0.9. These crystals, carefully protected from light and mechanical strain, were then dissolved in the acidic alcohol solution, and the resulting hydrogen yields were determined. Figure 11 shows that the total hydrogen yield was much greater than the number of F centers present. A similar large discrepancy was shown to exist between the F center concentration and the hydrogen yield from an additively colored crystal prepared in sodium metal vapor by the method of van Doorn.lC The spectra of these crystals all showed a peak a t 1900 A, the position reported in the literature for the U band. At low doses in our irradiated crystals, the U band actually appeared larger than the F band though it became relatively smaller a t higher doses. The U band is supposed to indicate the presence of hydride ion H- in the crystal, probably derived from radiation decomposition of impurity OH- ions or perhaps from traces of occluded water or hydrogen. Continued irradiation is known to cause the decomposition of the hydride ion into a trapped electron (F center) plus an interstitial hydrogen atom." One would expect, on dissolution in the

alcohol solution, to obtain one molecule of hydrogen by reaction of this hydrogen atom with the alcohol, while the original hydride ion would react with water to give only one molecule of hydrogen, Thus, the difference between the total hydrogen yield and the number of F centers should, at high doses, remain constant and equal to the number of hydride ions present at low dose. Figure 9 shows that radiolysis of adsorbed azoethane decreases with dose in approximately the same way that the number of electron centers increases with dose, in the case of sodium chloride. A plausible assumption is that in all cases the decrease in decomposition yield with dose results from a decreasing efficiency of energy transfer within the solid, owing to the buildup of radiation-produced defects. When the solids were first irradiated under vacuum and then azoethane was added without further irradiation, small yields of gas were noted, amounting to a few per cent of the quantity that mould have been produced had the azoethane been present throughout the irradiation, Silica gel is colored blue under irradiation. It was noted by KohnI2 that this blue color could be bleached simply by the addition of various gases to the system. We observed that irradiated silica gel would lose its color when azoethane vapor was added to it, while small quantities of nitrogen were formed at the same time. When a limited quantity of azoethane vapor was let into a tube containing irradiated silica gel at room temperature, only the gel at the top of the column was bleached; apparently no (10) C. Z. van Doorn, Rev. Sci. Znstr., 32, 755 (1961). (11) C. J. Delbecq, B. Smaller, and P. H. Yuster, Phys. Rev.,104, 599 (1956). (12) H. W. Kohn, Nature, 184, 630 (1959).

Volume 70, .\'umber

4 April

1966

1104

J. G. RABE,B. RABE,AND A. 0. ALLEN

Table 11: Preirradiation Effects in Sodium Chloride Preirradn dose, hlrads

Irradn dose, Mrads

2.0 17.25

...

...

2.00 17,25

...

2.01 2.00 2.00

Nz

Hz

CHI

ZNz, Hz, CH4

6.7 13.2 52.0 49.0 47.2

0.06 0.21 5.8 4.7 4.1

0.24 1.01 2.8 2.2 2.0

7.0 14.4 60.6 55.9 53.3

a “Liquid line” yield is subtracted from the preceding column. Based on the numbers in the preceding column. ceding column.

azoethane penetrated to the lower parts of the column. Figure 12 shows the approximate fraction of silica gel decolorized by different amounts of added azoethane, along with the quantities of nitrogen simultaneously produced. The amount of nitrogen is, within the experimental uncertainty, proportional to the amount of decolorization. The nitrogen produced here is about 9% of what would be formed, had the azoethane been present during the irradiation. When similar experiments were tried with MgO 800, the amount of gas formed was much smaller and appeared to be approximately constant at about 0.015 pmole/g, independent of the amount of azoethane added although with the silica gel the gas formed was almost proportional to the amount of azoethane added. Table I1 shows the results of similar experiments on sodium chloride, with further experiments in which additional irradiation was given after the azoethane was added to the preirradiated salt. It is seen that the gas formed on adding the azoethane to the preirradiated salt has a much lower ratio of hydrogen to nitrogen than that formed when the azoethane is present during irradiation. The table shows that the amount of gas formed during the irradiation is considerably less when the solid has been preirradiated than when it is fresh. This effect was also demonstrated for hIg0, as shown in Figure 8. Table I1 also shows that the amount of decomposition on samples of salt preirradiated with different doses is proportional to the number of negative centers present (as shown by the amount of hydrogen generated when the salt is dissolved in the acid solution of alcohol). It seems reasonable that the decomposition of the azoethane should be :tccompanied by bleaching of color centers located near the surface although the surface is so small that the fractional decrease in total color centers present in the salt is too small to be noticed. The quantity of nitrogen shown in the upper two rows of Table I1 corresponds to a yield of one molecule for The Journal of Physical Chemistry

CHdC

Electron center/g of solid, nmoles

... ...

... ...

230 45 1

51.5 40.1 30.1

0.0247 0,0193 0.0145

...

-

Products, nmoles/g of solid--

7

7

Zoor for liq linea

Zoor for preirradnb

... ... 51.5 47.1 44.5

G(Nz

+ Hz +

... ...

Amount produced during preirradiation is subtracted from pre-

E 0.4-I

0

- 60 8 - 40 w 0 !-

w

0

20

40 60 80 100 120 140 p MOLE A Z O E T H A N E PER GRAM SOLID

Figure 12. Per cent decolorization and amount of when azoethatie is put on silica gel A, irradiated to 6 PI‘frads under vacuum.

160

N 2

180

formed

each negative center lying within 30 A of the surface (if the centers are uniformly distributed). Since essentially the entire volume of silica gel lies within this distance of the surface, it is consistent to find that the silica gel is bleached quickly and completely.

Discussion The interpretation of data on such complicated systems, involving processes, the basic nature of which is unclear, must be tentative a t best. Much more experimentation will be required to confirm any proposed models. For instance, we determine, among the decomposition products of azoethane, only the permanent gases, as we believe it more valuable to obtain yields of a few products under a wide range of environmental conditions than to obtain a more nearly complete product spectrum with less variation in conditions. Yet we cannot be sure that study of the yields of other products might not have led to a different picture. The data all seem consistent with a picture in which energetic entities, which might be electrons, holes, or excitons, are produced in the solid by radiation, migrate to the surface with a probability which is higher the more perfect the solid lattice, and transfer their energy to the adsorbed molecules which then

RADIOLYSIS AND ENERGY TRANSFER IN THE ADSORBED STATE

have a certain probability to undergo various modes of decomposition. This picture is consistent with the finding that the yields on MgO increased with increasing calcination temperature of the oxide, which is known13 to be associated with increasing perfection of the crystal lattice. Such a result emphasizes the basic difference between radiolysis and heterogeneous catalysis. A catalyst is generally spoiled by treatments that tend to eliminate lattice imperfections. I n radiolysis the activation energy is supplied from outside, and the problem is to get it to the point of reaction; in catalysis the reaction arises spontaneously from the field of force resulting from a peculiar specific arrangement of atoms in the system. As remarked above, we incline to the opinion that the decrease in yield with increasing dose results from hindrance to energy transfer by radiation-produced imperfections in the solids. The most effective imperfections are no doubt the new vacancies and interstitials which are known to be produced by X-rays, probably as a result of multiple ionizations arising from inner-shell excitation of atoms in the solid. We cannot tell whether the presence of color centers (trapped electrons and holes) may also hinder energy transfer. A possible alternative explanation for the dose effect is that radiolysis products on the surface could act as inhibitors to further reaction. Another possibility is that energy transfer to adsorbate molecules is favored a t certain active sites; molecules on these sites will tend to decompose first, and if the decomposition products remain on these sites, preventing other molecules from migrating to them, the decomposition yield should decrease with time. However, the data of Table I1 and Figure 9 show that in the case of NaCl a given dose of radiation is about as effective in lowering the energy transfer when applied to the solid before the adsorbate is introduced as when the adsorbate is present throughout the irradiation. The effect must be on the solid itself. It could be thought that radiation actually destroys active sites, or potential active sites, by annealing out or breaking up peculiar surface structures. This idea is plausible in the case of the oxides, where OH groups on the surface may play a role in the adsorbate decomposition, but does not seem likely in the case of alkali halides. It may seem paradoxical that build-up of radiationproduced defect3 should decrease the decomposition yield, while on the other hand the presence of color centers causes decomposition of added materials with no direct exposure a t all of these materials to radiation. When a solid is irradiated, most of the

1105

electrons and holes recombine, but a few are trapped a t lattice vacancies, at other defects, and a t points on the surface. Though these traps appear optically to be several volts in depth, it is well known that slight straining of the lattice, such as caused by relatively gentle jarring of the crystals, will distort some of the traps enough to free some of the electrons and holes, thereby decreasing the concentration of color centers. A similar strain is produced near the surface of a crystal by the mere physical adsorption of vapor molecules on the surface14 because the change in surface energy is sufficient to change appreciably the equilibrium separation of ions in the surface layer. Hence, adsorption frees some of the electrons and holes trapped near the surface of an irradiated solid, and some of them will interact with the adsorbing molecules and may cause decomposition if sufficient energy is made available. This differs from the interaction at the surface of excited entities generated directly by radiation since the latter should on the average carry more energy than those released from traps. An indication of the difference is the relatively smaller amount of the higher energy product hydrogen, relative to nitrogen, generated from azoethane on contact with irradiated salt, as compared to the corresponding yields formed during irradiation (Table 11). The radiolysis of materials adsorbed on JIgO follows a different pattern from that found for any of the other mineral supports tried, which include’s various oxides, silicates, and alkali halides. Except for MgO, the curve showing the yield of decomposition products as a function of coverage ordinarily rises at a continually decreasing rate, bends over gradually, and becomes parallel to the “liquid line” after the first monolayer is complete. On MgO, the curve, instead of gradually bending over, remains straight, up to a certain coverage, after which it abruptly falls and reaches a relatively very low value by the time the monolayer is complete. The maximum in these yieldcoverage curves is reached a t very different coverages for different individual products of the radiolysis. A further peculiarity of MgO is shown in the profound effects on the yields of particular products at low coverages. Thus, with hydrocarbons on JIgO the yield of hydrogen was almost completely suppressed while that of methane was increased. For azoethane, the percentage of methane in the permanent gas fraction was an order of magnitude higher on JIgO than on any other solid, or in the liquid state. (13) R . Fricke and J. Lueke, 2. Elektrochem., 41, 174 (1935); R. M. Dell and S. W. Weller, Trans. Faraday Soc., 5 5 , 2203 (1959). (14) D. J. C. Yates, Advan. Catalysis, 12, 265 (1960).

Volume 70,Number 4

April 1966

J. G. RABE,B. RABE,AND A. 0. ALLEN

1106

One may ask first why with most materials these curves bend over gradually. One possibility is that the range of action of an excitation produced in the solid is quite large; that is, a molecule adsorbed on the solid has a certain probability of receiving energy from an event occurring anywhere within a considerable distance although this probability may not be very large. Then as additional molecules are added to the surface, even though they may be at considerable distances from the first molecule, they will compete for the energy coming from a given event in the solid, and the probability of a given molecule becoming decomposed will continually decrease with increasing coverage. A model of this type would predict a hyperbolic dependence of yield on coverage, typical of competition kinetics. A second possibility is that, even if the adsorbed molecules only communicate with a rather small volume in the solid, additional molecules added to the surface go on a t random or show some tendency to cluster together. Then competition for the energy will occur only between molecules lying close together, but the proportion of molecules occurring in such close pairs or groups will increase as the coverage increases, approaching unity exponentially. Such a model mould lead to an exponential dependence of yield (corrected for liquid line) on coverage. Actually, a fair fit to the data can be made with either a hyperbolic or exponential type of expression. Probably both effects are present in most cases. Since MgO shows no gradual change in the slope, we must conclude that neither condition holds in this case. First, the molecules must be drawing energy from a rather restricted volume of solid; and, second, as they go on the solid, they must be spaced so that no two molecules are located very close together. Although we have no evidence that chemisorption is occurring in any of our experiments, we must assume that in the neighborhood of a molecule adsorbed on MgO the force of attraction for a second molecule is somewhat reduced, so that molecules effectively tend to repel one another on the surface to the extent of keeping a respectful distance away from each other. The marked enhancement of yields of certain decomposition products and suppression of others shows that electronic excitations within the lattice must activate surface states of the solid which interact with excited states of the adsorbed molecules in very specific ways. Thus, in the case of azoethane, electronic excitation within the magnesia activates a surface state which communicates its energy to a specific excited state of the molecule that results in splitting of the relatively strong C-C bond. Most modes of excitation of this molecule split the weak C-X bond, resulting T h e Journal of Physical Chemistry

in the formation of nitrogen gas and not of methane. I n hydrocarbons, excitation by direct absorption of energy usually gives rise to states which lead to the breaking of a C-H bond with ultimate formation of hydrogen gas. On MgO these states, which have rather high energy, are in communication with surface states of somewhat lower energy so that this mode of decomposition is suppressed. Instead, other modes of decomposition giving rise to breakage of C-C bonds, which have somewhat lower energy, are favored. Now we have to suppose that when a certain degree of coverage is attained, the distance between the molecules becomes comparable to the geometrical extent of the surface excitation which is responsible for the formation of a decomposition product and wh;ch presumably includes orbitals belonging both to the adsorbed molecule and to the magnesium and oxygen atoms nearby. When the second molecule comes close enough to distort the orbital of that excited state of the first molecule which leads to formation of a particular decomposition product, then this excited state will be quenched, its energy presumably divided between the two molecules, and decomposition in this particular mode will no longer be possible. I n general, we might expect that those states of higher energy which are required to break the strongest bonds in the molecule would tend to have the most extended orbitals. I n fact, it is found that hydrogen, which requires the most energy for its formation from azoethane, reaches a maximum in its yield a t a very low coverage; methane reaches its maximum at a higher coverage; and nitrogen, which requires the least energy, reaches its peak at the highest coverage. When hydrocarbon is added to the surface which already contains adsorbed azoethane at a fairly high coverage, the decomposition yield of the azoet,hane products, nitrogen and methane, is increased, as though the amount of azoethane had been decreased. A probable explanation is that the hydrocarbon molecule, which has a tendency to deliver energy to the surface states, intervenes between two azoethane molecules which would otherwise share the energy. The energy cannot flow so readily into the hydrocarbon molecule and hence remains in the azoethane molecule, which decomposes. The yield of nitrogen per molecule of azoethane on MgO at 5.9 Mrads is 20 times that from liquid azoethane, while the total yield of permanent gas ( i Y 2 Ha CH,) is about 50 times that of the liquid. The region in the crystal from which energy communicates to the adsorbed molecule may then be assumed to have a mass at least 20-50 times that of the azoethane molecule or (density of MgO = 3.65) a volume

+

+

RADIOLYSIS AND ENERGY TRANSFER IN THE ADSORBED STATE

of 800-2000 A3. This would correspond to a cubic volume 9.3-12.6 A on a side-not much greater than the length of the azoethane molecule, which would be about 8.3 A if it lay in a stable configuration but stretched out to get maximum contact with the surface. Some features of the data remain unsatisfactorily explained. Thus, it is not clear why, in Figure 7, the effect of dose on G should be small a t low coverage. Again, on comparing MgO 800 with MgO 1140, the yields at low coverage are nearly the same, showing equally good energy transfer. Why should the maximum yields then be so different? Though the present model seems to fit the major qualitative features of the data, a great deal obviously remains to be learned about the fundamental processes occurring in the radiolysis of heterogeneous systems.

1107

Acknowledgments. The surface-area determinations by the BET method were performed on his apparatus with the permission of R. N. Dietz, who also supplied valuable advice and assistance in carrying out the measurements and calculations. Preparation of the large single crystals and measurement of their absorption spectra were carried out for us by R. L. Warasila. Throughout the work we have profited by frequent discussions with Dr. Paul Levy of the physics department. The chemical method for determination of negative centers in salt originated from discussions we held with Drs. A. Appleby and W. A. Seddon. J. G. R. was supported by a NATO research fellowship during most of the time in which this work was carried out. This research was performed under the auspices of the U. S. Atomic Energy Commission.

Volume '70,Number 4

April 1966