]DECAY
RATESOB’
3167
TRAPPED ELECTRONS IN 3-lJETHYLPENTANE
Annealing an Other Variables Which Affect Observed Decay Rates o ions in 3-Methylpentane Glass1
avid Shooter and John E. Willard* Department of Chemistry, University of Wisconsin, Madison, Wisconsin 65706 (Received M a y 2 , 1972) Puhlkm!wn costs assisted by the U.S. Atomic Energg Commission
The rate of decay at 77°K of trapped electrons produced by ?-irradiation of 3-methylpentane (3MP) glass decreases with time of annealing of the matrix at 77°K before ?-irradiation, and, for short annealing times, is dependent on the geometry of the sample. In quench cooled samples in 2-mm i.d. tubes the initial halfMe increases from 10 min without annealing to 60 min for samples annealed for 250 hr, but there is little or no effect of annealing on the yields and absorption spectra of the trapped electrons. Decay rates measured by esr and ir have been compared and sources of error in both types of measurement considered.
Introduction The yields, decay properties, and spectra of trapped electrons produced in rigid hydrocarbon glasses by yirradiation have been studied in a number of labora1,ories to obtain B better understanding of trapping phenomena.2 In the present work we have investigated the influence of two experimental variables suspected of having previously unrecognized effects on the i*esults of such sf,udies. These are preirradiation annealing of the matrix and changing microwave power saturation of the trapped electron esr signal as decay progresses. Earlier W7Ork3 suggesting an effect of cooling rake of the matrix on trapped electron decay rates has been extended, and other experimental varjables in esr and infrared measurements of trapped electron decay have been reviewed.
Phillips pure grade 3MP was prepared for use by passage through a column of freshly activated silica gel and storage on the vacuum line over a sodium mirror, from which it was distilled into the Suprasil cells used For jrradiatioxi and exr and ir spectrometry. Cylindrical 2-mm i d . and rectangular 2 mm X 3 mm i.d. cells were used for both esr and ir measurements. Cylindrical 8-mm i d , rectangular 1 mm X 6 mm i d . , and ,square 1 cm X I cm i d . cells were also used for ir measurements. -plsradiations were made with a 60Co source a t dose rates of 3 6 X 1W8eV g-I min-I on the small cells arid 2 X 10l8 eV g-’ min-l on the large cells. During irradiation the samples were immersed in liquid nitrogen maintained at ca. 71°K by bubbling He gas.
Trapped electron concentrations were monitored at 1560 or 950 nm, depending on the optical density, using a Cary 14 spectrophotometer modified to pass the infrared analyzing beam through the monochromatos before passing through the sample. Esr measure-
ments were made with a Varian 4500 eter equipped with Fieldial, using XOO-kHz modulation, a 4531 cavity, and a modulation amplitude of 4 G or less with samples under liquid nitrogen, at 77”K, or at 71°K with the aid of a stream of heiium entering the finger of the Varian dewar 3 cm from the bottom. Microwave powers of 0.5 to 600 pW incident on the cavity were obtained with the cavity attached t o the low-power arm of the microwave bridge and a resistive terminator attached to the high-power arm. The power passing down the lowpower arm to the cavity was determined with a Hewlett-Packard power meter (Model 431c) for powers down to 30 pW. Lower powers were determined from the height, of the 3MP radical signal relative to its height in the range above 30 pW, after showing that the radical signal does not saturate in the 30- to 300-pW range (?;.e., a plot of radical signa1 height vs. (power)”’ is linear). The spectrometer sensitivity was monitored using a strong pitch sample, I n this work rates of trapped electron decay are indicated by the “initial half-lives” (Le., the time for half of the electrons t o decay starting a t a specified short time after a specified short irradiation). The kinetics of the decay2&are time-dependent first order, i.e., plots of log [e-t,] vs. t decrease in slope with time but are first order in dose (the decay curve8 for different doses are superimposable after normalization for dose). This indi(1) This work has been supported in part by the G. 9. Atomic Energy Commission under Contract No. AT(11-1)-1715 and by the TV. F. Vilas Trust of the University of Wisconsin. (2) For examples and references see: (a) Y. E. Willard in “Fundamentals of Radiation Chemistry,” P. Ausloss, Ed., Wiley, New Vork, N. U.,1968, Chapter 9; (b) A. Ekstrom, Radiat. Res. Rev., 2, 381 (1971); (c) H. Hase, M. Noda, and T. Nigashimura, .I. Chem. Phys., 54, 2975 (1971); (d) D. P. Lin and L. Kevan, ibid., 5 5 , 2629 (1971) ; (e) J. Miller, ibid., in press. (3) (a) J. Lin, K. Tsuji, and 3’.Williams, J . Amer. Chem. Soc., 90, 2766 (1968); (b) K. Tsuji and F . Williams, Znt. J . Radiat. Phys. Chem., 1, 383 (1969). (4) F.W. Lytle and J. T. Stoner, Science, 148, 1721 (1965).
The .Journal of Physical Chemistry, Vol. 76, N o . 22, 1972
D A V ~SHOOTER D AND JOHN E. WILLARD
3168 W,EACWED --D
SUPRASjL
i A
SUPRAXIL
WwICM.S
A
00 u w
Figure 1. Em spectra at 71°K of 7-irradiated 3MP a t powers of 0.6 and 400 r W before and after bleaching: dose, 1.7 X 10'0 eV g-1~
L__L--__I_I-.-
0
5
I
IO
I
15
-i-"/ pw
I
20
I
25
when the signal was exhaustively photobleached, under conditions where the background was small (Figure 1B). Curves A, C, and E of Figure 2 show normalized signal heights of the trapped electron esr signal as a, function of the square root of the microwave power for three samples of 3MP glass in 2-mm i d . tubes which had received different y doses. The power range over which the signal is unsaturated and the power at; which the maximum of the saturation curve is reached are both greater the higher the dose. These effects are attributable to a decrease in the spin-spin relaxation time with increasing spin concentration. Reduction in the modulation amplitude from 2 to 0.4 G increases the power a t which the saturation curve reaches its maximum (curves C and D). Curve I3 for a modulation amplitude of 1.7 G at 400 Hz is from the work of Williams, et ~ 1 . ~ ~ Curves A and B of Figure 3 indicate a shift toward saturation at lower powers as trapped electron decay
BEFORE DECAY
Figure 2. Dose and modulation amplitude dependence of esr power saturation curve of trapped electrons in 7-irradiated 3MP a t 71°K. All data are normalized to the low power, nonsaturated data of curve C. Curve B shows results of Williams, et aZ.,a obtained at a modulation frequency of 400 Hz and a modulation amplitude OS 1.7 G or less after a y dose of 1.2 X lo**eY g-l.
Figure 3. Esr power saturation curves of trapped electrons in 3MP at 71°K: A, 7.2 X 1018 eV g-l, before decay; B, same as A but after 88% decay; C, 9 X 1Ol8 eV g-l, before decay, the trapped electron concentration in C was the same as in B.
cates that essentially all e-tr combine with a positive ion or radical within the parent spur, and possibly with the geminate positive ion. Effects of Microwave Power, Dose, and Modulation Amplitude on Esr Signal of Trapped Electrons in 3iUP Glass. Figure 1 shows typical (normalized) esr spectra from 3MP glass observed at 71°K following a 1.7 X l O I 9 eV g-l -y dose at 71°K using microwave powers of 0.6 and 400 pW. Of the three signals contributing to the spectra, (i.e~,the radical signal, the Suprasil signal, and the trapped electron signal), only the trapped electron signal was observably affected by the output of a 25-W tungsten lamp. The line width and magnetic field position of the electron signal did not vary in these investigations. The electron concentration was taken a8 proportional to the change of the vertical distance between the points of zero slope of the electron signal
progresses. This is attributable to a decrease in spinspin interaction with decreasing concentration. Curve C of Figure 3 is for a sample which received a dose l/* that of the curve A to produce an electron concentration the same as that in the sample of curve B. The higher powers required for saturation of the partially decayed sample (B) which had received the higher dose may be attributed to spin-spin interaction of the trapped electrons with the trapped radicals. The trapped electron decay plots of Figure 4, obtained from identical 2-mm i.d. samples monitored a t 0.7 and 10 pW, show the type of error which results from measurements at a microwave power at which the degree of saturation changes with decay (Figure 3). It is not sufficient to use a power where the signal is initially unsaturated. A power sufficiently low so that the normalized decay plots are independent of power
The Journal of Phgsical Chemistry, VoE. 76, No. $9, 1978
DECAY RATES OF
T~APPED ELECTRONS I N 3-METHYLPENTANE
3169
P 4
3 .P
2
T
I
-0 -L TIME, MIN
Figwe 4. Dependence of decay of trapped electron esr signal from yirra,diated 3MP at 77°K on microwave power: solid points show linear plot; open points reciprocal plot; open circles 10 fiW; open triangles 0.7 pW.
(i.e., superimposable over the whole decay range of interest) must be used. Bfects of Xample Geometry and Preirradiation Annealing of SMP Glass at 77°K on Decay Rate of Trapped Electrons. Early in the study of the decay of trapped electrons in 3hIQ glass it was noted that rates observed in 1-cm2cells by is transmission were slower (half-lives of 35 to 45 m n ) than those observed in 3-mm i.d. tubes by esr (5 to 10 minf. Using an esr cavity which would accommodate a t-cmz cell, Tsuji and Williams3 showed that the rates observed by ir and esr are similar when similar sized samples are used, and suggested that the faster decay with small. samples is related to a faster rate of cooling of the 3 MP. We have confirmed that the rate of cooling of the 3MP t o the glassy state affects the decay rates of trapped electrons produced by subsequent 7-irradiation. Two samples cooled in 2-mm i.d. tubes to 102°K in 45 min and thcn to 77°K at ca. 1" min-I gave initial half-lives of 28 and 27 min, while four identical samples prepared hy fast quenching in liquid nitrogen to 77 or 74°K gave half-lives of 10 i 1.0 min. In each case the half -life was independent of whether measurement was made by ir or esr. Fnrther experiments have shown that annealing of quench-cooled 3MP at 77"K, like slow cooling, decreases the decay rate of trapped electrons. The initial half-lives in a 2-mm i.d. tube increased from 10 min t o over 60 min as the preirradiation annealing time at 77°K mas increased from 5 min to 250 hr (Figure 5). For half-lime >30 nnin, the half-lives given in the last paragraphs 1% ere measured starting 2 min after the end of a 5-min Irradiation at 77°K. For shorter halflives the sample was transferred to an esr dewar at
71°K after irradiation at 71°K. Zero time for the decay was taken as the time a t which the back-extrapolated decay curve reached the electron concentration observed a t 71°K. The half-lives from ir measurements in 1-cm2 cells irradiated a t 77"K, given below, were all >30 min and were measured starting 2 min after a 1.5-min irradiation. The half-life (as defined above) of the trapped electrons in quench-cooled (i.e., less t>han5 min from immersion to beginning or irradiation) 1-cm2 samples as measured by ir under our most reproducible conditions was 33 min. This value rises with increasing time of preirradiation annealing to an apparently constant value of slightly over 60 min after approximately 50 hr of annealing (Figure 5B). Consistent with the trend of decay times described, the half-life for a quench-cooled sample in an 8-mm i.d. cylindrical tube is ea. 20 min. Surprisingly, however, the decay time in a quench-cooled sample in a rectangular 2 mm X 3 mm i.d. cell is 35 min, as measured by either esr or ir. Likewise surprising is the 49-min half-life in samples in flat "paddle'p cells with internal dimensions of 6 mm X 1 mm and several centimeters high. These results (Figure 6) indicate that trapped electron decay rates in incompletely annealed samples are dependent not only on the thickness (and hence rapidity of cooling) of the sample when it is quenched, but also on its geometry. However, it appears (Figures 5 and 6) that in completely annealed samples the decay rates are independent of the geometry. The data of Figure 6 indicate, somewhat unexpectedly, a tendency for the rate of approach to the completely annealed state to be faster in the samples formed closer to equilibrium than in those farther removed, a feature which may, again, be geometry dependent. Some additional observations have been made on the decay kinetics of the trapped electrons in unannealed and annealed samples. (1) Plots of log [e-tr] us. time The Journal of Physical Chemistry, Vol. 76,N O .$3, 19Yd
DAVID SHOOTER AND JOHNE. WILLARD
3170
DECAY TIME,
MIN.
Figure 7 . Dependence of second-order plots of trapped electron decay in 3MP on time of preirradiation annealing of l-cm2 samples a t 77°K.
Q
I I I !&-A 10 20 50 TIME A T 77OK BEFORE IRRADIATION, HR ,
Figure 6 . Dependence of initial half-lives of trapped electrons in 3MP glass a t 77°K on time of preirradiation annealing a t 77°K and on sample geometry. The cross sectional shape of each of the five cells used is depicted with the dimensions adjacent to each curve. The horizontal scale is contracted between 25 and 50 hr.
for decay a t 77°K are always curved but, consistent with observations in several laboratories, are first order in dose (i.e., the decay curves are superimposable after normalization for dose) over the range of 3 X 10l8to 1 X l0l9 eV 8-l and 80% decay for which we have followed them. This is true for both annealed and unannealed samples. ( 2 ) Plots of l/[e-tr] vs. time for samples with short trapped electron half-lives (e.g. , in quench-cooled, 2-mm i d . tubes) are always linear, with slopes which vary with the initial concentration ( i e . , with the dose). ( 3 ) Plots of l/[e-t,] for unannealed samples with half-lives >30 min (e.g., quench cooled in l-cm2 cells) are curved (Figure 7 ) . However, the curvature decreases as the preirradiation annealing time is increased. For samples in which annealing is nearly complete, the l/[e-tr] US. time plots are linear from the start. Curvature seems to be observed only when appreciable decay and appreciable annealing (Le., change in. matrix characteristics) occur in the same time period. Thus it is only when the sample is completely annealed QP anneals slowly relative to the decay rate that decay i s not influenced by simultaneous changes in the matrix. Within experi"a1 error, the microwave power saturation properties of thc trapped electron are the same In a sample in a 2-mm i.d. tube irradiated immediately after quench cooling to 77"K, as in a similar sample irradiated after 200-hr annealing at 77°K. Likewise the trapped electron yield was the same for annealed and unannealed samples, and the trapped electron spectra from 600 to 2160 nm in 1-cm2 cells were unaffected by preirradiation annealing. The Journal of Physical Chemistry, Vol. 76,N o . E?,1979
Measurement of Trapped Electron Concentration by I r Absorption. Significant differences in apparent decay rates of trapped electrons observed in identical samples by infrared absorption can result from such factors as (1) changes in optical density resulting from smoothly changing concentrations of ice crystals as more are produced or as the37 settle out during an experiment and (2) systematic changes in the rate of liquid nitrogen bubbling with resultant changes in "noise" which affects the apparent baseline. Surrounding the top of the dewar with a box of Aowing nitrogen gas minimizes ice formation. Bubbling can sometimes be reduced by rinsing the dewar with liquid nitrogen before the final filling. In work at lower temperatures (produced by bubbling helium gas through the liquid nitrogen) nitrogen bubbling is reduced. In a previous study in our laboratory5 the effects of both ice and bubbling on the measurement of light transmission by samples at 77°K have been avoided by using Suprasil rods as light pipes to conduct the light through the liquid nitrogen to and from the cell. I n some cases the rods were sealed to the optical faces of the cell and in ot)hexs the cell fit ciosslp between the inner ends of the rods. The outer ends of the rods were sealed through the sides of a styrofoam-lined metal box of liquid nitrogen which could be positioned in the sample compartment of the Cary spectrophotorrreter. The light pipe technique was completely successful in eliminating ice and bubbling effects and should be generally useful for spectrophotomeric studies under liquid nitrogen.
Discussion This work has shown that the yields of trapped electrons produced by y-irradiation of 3nIP glass are independent of the time of annealing of the glass at 77°K prior to irradiation and of the sample geometry, but that the decay rate of the electrons i s dependent on both of these €actors. This dependence is lost when samples have been completely annealed. The absorp( 5 ) R. Fass and J. E. Wdlard, unpublished results.
~ O D ~ ~ I C A T ITO O ~ASPRECISION ADSORPTION APPARATUS
tion spectrum of the trapped electrons which decay with a half-life of 30 min is indistinguishable from that of those which decay with a half-life of 60 min. These facts suggest that annealing does not change the numlber or depth of thc dectron traps but rather involves (a change in matrix rigidity on the molecular scale which decreases the rate a t which the trapped electrons can (diffuse. This decrease is presumed to be due to a (decrease in free volume in the glass as it anneals at a temperature ab or behw the glass transition. Current work of 8. L.I-Iager in our laboratory indicates that ~ n ~ ~ r a 3MP ~ i a glass t e ~shows an endothermic differential thermal analysis peak at 82°K when warmed from 77°K. ‘The m ~ ~ ~of ~thist peak ~ dincreases e with time of annealing of the sample a t 77°K for time intervals similar to tho%! required to produce maximum trapped electron decay times, By analogy wit)h DTA studies6 on polymeric systems the peak is presumed to
3171
reflect the energy necessary for regaining the free volume which is lost while annealing the sample a t 77°K. Preirradiation annealing has been found to change postirradiation phenomena other than electron decay in 3MP glass. It causes a decrease in the luminescence intensity,’ a decrease in the 3MP radical decay rate,8 and a lengthening of the observed phosphorescence lifetime of deuteriobenzene solute.9 ~ ~ a n t i tcorrela~ t ~ ~ e tion of the effects relative to annealing times is not possible because of differences in sample size, sample geometry, and rate of cooling. (6) (a) a. E. McLoughlin and A. V. Tobolsky, J . Polymer Sci., ’7, 658 (1951); (b) Yu. A. Sharonov and M~V. Vol’kenshtein, Sov, Phye. Solid State, 5, 429 (1963). (7) K. Funabashi, P. IF’. Herley, and M. Burton, J. Chem. Phys., 43, 3939 (1965). (8) W. K.Kam and J. E. Willard, unpublished results. (9) T. E, Martin and A. H. Kalantar, 1.Phys. Chem., ?2, 2265 (1968).
Modifications to a Precision Adsorption Apparatus. Interaction of the Inert
8
with Boron Nitride
.N. Ramsey, H. E. Thomas, and R. A. Pierotti* School of chemi$try, Georgia Institute of Technology, Atlanta, Georgia 90852 (Received M a y 8, 19r8) Fub2iea.tion costs assisted by the U.S. A r m y Research Ofice-Durham
A. modified high-precision volumetric adsorption apparatus is described. The apparatus incorporates a precision. cryostat, the precise pressure-measuring techniques of gas thermometry, and a null capacitance manom-
eter. The use of the capacitance manometer to separate the mercury manometer from the adsorption system is shown to eliminate a number of apparatus corrections and to simplify the acquisition of data. The apparatus is suitable for the investigation of the low-coverage, high-temperature physical adsorption region in which Henry’s law behavior and deviations therefrom occur. Measurements are reported for the adsorption of neon, argon, krypton, and xenon on the hexagonal modification of boron nitride a t 273°K. The results are analyzed using the gas-solid virial coefficient treatment to give energies of interaction of these gases with boron nitride.
~ ~ t ~ ~ a u ~ t i ~ ~ Studies of the physical adsorption of the inert gases on graphitized carbon blacks at low surface coverage have yielded important information on the nature of gas-solid interactions.’ I n the analysis the adsorption isotherm data are fitted to a virial equation in which the number of moles of gas adsorbed per gram of adsorbent, naris given by1
where f is the fugacity, R is the gas constant, T is the isotherm temperature, and B2sl Bas,.~. are the gassolid virial coefficients. The present analysis is concerned only with Bzs, the second gas-solid virial coefficient which is related to the Henry’s law constant, and which is characteristic of the interaction between an isolated adsorbate molecule and the adsorbent. The determination of values of Bzs requires accurate adsorp(1) R. A. Pierotti and H. E. Thomas, “Surface and Colloid Science,” Vol. IV, E. MatijeviE, Ed., Wiley-Interscience, New York, N. Y,, 1971, pp 93-259.
The Journal of Physical Chemistry, Vol. 76,N o . 22, 1978