NOTES
section a t an ion energy of 1.0 ke.v. for F- is 2.5 X lo-’’ cm.2 while that for C1- is 5.5 X lo-’’ cm.2. These are rough values since the pressure of CC&F in the analyzcr tube was known only approximately. The ratio of the two cross sections, 2.2, has greater validity and it can be compared to the ratio of 2.0 for C1- and F- in k r y p t ~ n . ~ The cross section for electron detachment for 0- in KzO is found to be about 2.2 X 1O-I’ cm.2. Of the three ions at 1.0 kev., 0- was most effective in producing secondary electrons, possibly because of its higher velocity. One 0- ion produced 0.55 electron, on the average; one F- produced 0.46, and one C1produced 0.22. Kerwin and hIcGowan5 have pointed out how a mass spectrometer can be used to study charge exchange of positive ions a t high pressure, but they did not use the secondary electron emission as a tool. With equipment designed specifically for this purpose, it should be possible to use this method to obtain accura,te values for cross sections for charge exchange. Electron loss from negative ions produces a beam of neutral particles which might also be useful to studies of the chemistry of high energy molecular beams using the proper negative ion as a precursor. Equipment could be patterned after some of that used for the charge-changing cross section measurements described by A l l i ~ o n but , ~ ions could be fed into it from a masa spectrometer. In this way, the proper ion can he selected and a gas may be used that is easy to handle. For example, the CC13F is easier to use as a source of C1- or F-- than would be Clz or F2 or the corresponding hydrogen halides. Kegative ions would seem preferable to positive ions for the production of inolecular beanis for two reasons. First, the production of a neutral atom from a positive ion is a resonant process, and the electron must be captured into a particular state of the atom. The stripping of an electron from a negative ion should be more probable. Secondly, many negative ions can be produced in large amounts by very low energy electrons. Only one or perhaps two types of ions will be produced, and these usually are in their ground state. To produce positive ions, greater ionizing energies are required, and a multitude of positive fragments is produced. The latter may have to be separated out, and the excited states of some ions may introduce complications. I n CC13F, for example, C1- and F- are produced by a resonant process a t 2.5 and 5.0 e.v., respectively.6 0- is produced efficiently from K 2 0a t 2.2 e.v.’
3907
Acknowledgment. The writers are indebted to Professor Herbert Berry of the physics department of Syracuse University for his helpful comments and suggestions.
-
(4) 3. B. Hasted, Proc. Rou. SOC.(London), A212, 235 (1952). (5) L. Kerwin and W. McGowan, Can. J . Phgs., 41, 316 (1963). (6) Unpublished work from this laboratory; manuscript in prepa~a-
tion. (7) R. K . Curran and IR. E. Fox, J . Chem. Phys., 34, 1590 (1961).
Paramagnetic Resonance Study of Fermi Level Motion and Defect Formation in HighResistivity Cadmium Sulfide Crystals
by G. A. Somorjai and R. S. Title ZBiM Watson Research Center, Yorktown Heights, New York (Received August 97,1964)
In this Note we wish to report how paramagnetic resonance measurements may be used to monitor Fermi level motion and the kinetics of defect formation in high-resistivity crystals and illustrate the technique with measurements on high-resistivity, sulfur-dopcd CdS crystals. Electrical measurements in crystals with resistivities greater than about los ohm-em. are difficult to make because of the low current levels involved. Changes in the resistivity, however, imply a change in the position of the Fermi level that can drastically affect other physical and chemical properties such as photoconductivity’ or evaporation. A high resistivity implies that the Fermi level IS close to the center of the band gap. A defect whose energy is near the center of the band gap will gain or lose electrons as the Fermi level passes through the level. If any of the charged states of the defect are paramagnetic, e.p.r. techniques can be used to monitor the motion of the Fermi level in the vicinity of the defect. We have found iron to be a suitable defect to monitor the motion of the Fermi level in sulfur-doped, high-resistivity CdS. Iron is present in the crystals (supplied by Eagle-Picher Co. , Miami, Okla.) wle studied in concentrations of rv5 X 1016/cm.3. In untreated crystals no paramagnetic resonaiice spectrum was observed a t 77°K. In crystals fired in a sulfur ~~~
(1) K. H. Bube, “Photoconductivity of Solids,” John Wiley and Sons, Inc., New York, N. Y., 1960. (2) G. A. Soniorjai, Proceedings of the International Conference or1 the Physics and Chemistry of Solid Surfaces, Providence, R. I., 1964.
Volume 68, Number 1.2 December, 1964.
NOTES
3908
atmosphere a resonance characteristic of Fe3+ was r n e a ~ u r e d . ~Samples fired in Cd vapor showed no Fe3+ spectrum. The heat treatment in sulfur vapor causes iron, which is present in CdS as Fez+, to convert to Fe3+. The intensity of the Fe3+paramagnetic resonance spectrum is found to be dependent on the sulfur pressure and on the temperature at which the sample is doped. The heat treatment in sulfur vapor results in a motion of the Fermi level through the Fez+ level converting it to Fe3+. The resonance detection of Fe3+ is therefore a convenient method of monitoring the Fermi level motion in high-resistivity, sulfur-doped CdS crystals. Besides the merit of being able to monitor Fermi level motion when other techniques are difficult to employ, this method is nondestructive and requires no electrodes on the sample. The technique, by careful choice of paramagnetic impurity, should be of general use in studying the kinetics of defect formation in high-resistivity solids.
Experimental The ultra-high-purity grade crystals were cut (3 X 3 X 6 mm.) so that the c-face (0001) was perpendicular to the direction of the long axis. These samples were fired a t 700-1100° in 5-30 atm. of sulfur for 48-72 hr.4 to assure uniform dopant distribution throughout the crystal and were quenched to room temperature in order to freeze in the defect concentrations characteristic of the firing temperature. The doped crystals were orange in color5 in contrast to the yellow color of the untreated specimens. The band gap at room temperature is 2.42 e.v. The resistivities of the lreated samples were, as mentioned previously, high, ranging to beyond 1O1O ohm-cm. The paramagnetic resonance measurements were carried out at 77 and 300'K. The only spectrum observed in any of the crystals was due to Fe3+,and this appeared only in crystals heat-treated in sulfur. The resonance spectrum of Fe3+in CdS has been observed by Lambe, et aL3 They found that the resonance was observable at 4.2'K. and only with light of wave length 550 nip irradiating the crystal. They could observe no resonance a t 77 or 300'K. We find similar results in untreated crystals. However, in crystals treated in a sulfur atmosphere the resonance due to Fe3+is observable a t 77 and 300'K. The resonance is not photosensitive a t these temperatures. The parameters characterizing the resonance are as reported by Lambe, et aL3 The intensity of the resonance absorption due to Fe3+was found to depend on both the temperature and the pressure of the sulfur atmosphere in which the crystals were treated. The Journal of Physical Chemistry
16' -E
\
1
17 \
[
,A50.7
SULPHUR PRESSURE 20 ATMOSPHERES
I
0.8
I
I
0.9
I.o
lO3/T"K
Figure 1. FeS+ resonance intensity of sulfur-doped CdS single crystals as a function of the temperature of heat treatment a t a constant sulfur pressure (20 atm.).
Results and Discussion The dependence of the intensity of the Fe3+ resonance absorption on the temperature of the heat treatment for a constant sulfur pressure of 20 atm. is shown in Fig. 1. The intensity of the Fe3+ resonance signal is expressed in t e r m of the number of Fea+ centers/ ~ 1 1 1 .that ~ give rise to the signal. The logarithm of the intensity is plotted against the reciprocal of the temperature. At the highest temperatures there is an indication that the intensity of the Fe3+absorption approaches a constant value. The other points fall more or less on a straight line. The slope of the straight-line portion may be used to determine an activation energy, AE = 1.2 f 0.1 e.v. This activation energy is the net energy required for the over-all reaction that occurs during firing. It involves the sums of many energies such as the energy to remove an electron from Fez+ in the CdS lattice, the energy involved in incorporating sulfur atoiiis into the lattice (energy gain), the energy required to create a Cd vacancy, as well as other energies. None of these (3) J. Lambe, J. Baker, and C. Kikuchi, Phys. Rev. Letters, 3 , 270 (1959). (4) G. A. Somorjai and D. W. Jepsen, J . Chem. Phys., in press. ( 5 ) F. A. Kroger, H. J. Vink, and J. vonden Boomgaard, 2. physik. Chem., 203, l(1954).
NOTES
energies has been measured in CdS. AIandeP has estimated the energy required for vacancy formation in CdS to be 4.0 e.v. However, the accuracy is probably no better than k0.5 e.v. Even if all the energies were known, a knowledge of the relative concentrations of the various defects involved would be required in order to separate the net activation energy into its various components. One can, however, state that, since the valence state of iron does change, the energy level of Fez+must be near the Fermi level which, in these high-resistivity samples, is close to the center of the band gap. The intensity of the Fe3+ absorption a t the highest temperatures, -5 >< 1016/cm.3, corresponds to all iron in the labtice being present as Fe3+. Spectroscopic analyses of the samples, both in our laboratory and by the Eagle-Picher Co., indicate an iron concentration of about 1 p.p.m. which corresponds to 5 X 1016/~n1.3. The dependence of the Fe3+ intensity on the sulfur pressure a t a firing temperature of 900’ is given in Fig. 2 . The number of Fe3+centers shows a maximum a t 10 atm. and drops off to almost zero a t 30 atm. The maximum value of 3 X 101e/cm.3is less than the total iron concentration in the lattice. This is known both from the results given in Fig. 1 and from the result given in Fig. 2 on a sample heat-treated in 10 atm. of sulfur, but ;at a higher temperature, 1100’. The maximum in the number of Fe3+ centers in the 900’ curve of Fig. 2 cannot, therefore, be correlated with the exhaustion of the number of Fez+ centers available for ionization. It is indicative of the formation of a new defect. The qualitative model we use to understand the results of Fig. 1 and Fig. 2 is given in Fig. 3. I n Fig. 3 the band gap of CdS is drawn, and the approximate positions of the various defects are indicated. I n untreated crystals the concentration of sulfur vacancies V S is assumed to be in excess of the concentration of cadniiuni vacancies V C ~ In . CdS a sulfur vacancy acts as a double donor and a cadmium vacancy as a double acceptor. At room temperature, the energy levels of V S and V Care ~ such that almost all of these centers are ionized. The excess of Vs centers over V Ccenters ~ will thereflore give rise to high conductivity n-type samples as observed.5 The Fermi level in this material is near or in the conduction band. Heat treatment of the sample in cadniium vapor can only increase the ratio of 17s to V C and ~ make the material more n-type.’ On the other hand, heat treatment of the saniples in sulfur vapor will decrease the ratio of lis to Vcd and make the material less n-type, as is observed. I n the high-resistivity samples, the Fermi
3909
*
t
16. iij 3 x I O
z W
I-
z 16 1.5xlO -
Figure 2. Fe3+ resonance intensity in sulfur-doped CdS single crystals as a function of sulfur pressure a t a constant temperature (900”) of heat treatment.
-- Fe VCd
Figure 3. A model for the qualitative interpretation of the paramagnetic resonance experiments in sulfur-doped CdS single crystals.
level approaches the middle of the gap. Iron is assumed to be present, as Fez+ in the untreated crystals with an energy level close to the center of the band gap. As the Fermi level approaches the middle of the gap in the sulfur-doped samples, Fez+ is converted to Fe3+which accounts for our resonance observations. The decrease in the intensity of the Fe3+ resonance absorption a t sulfur pressures above 10 atm. (Fig. 2), even before all Fez+-has been converted to Fe3+, indicates that a new defect is formed. The decrease in the Fe3+ intensity in this case is not accompanied by a decrease in the resistivity, indicating that the defect is not electrically active. It is likely the product of a ( 6 ) G. Rlandel, Phys. Rei)., 134, A1073 (1964). (7) F. A. Kroger and H. J . Vink, “Solid State Physics,” Vol. 3, F. Seits and D. Turnbull, Ed., Academic Press, New York. N. Y., 1056, p. 310.
Volume 68, Number 1s December, 1964;
3910
chemical reaction between Fe3+ and sulfur in the CdS crystal lattice at high sulfur pressures. Further work is required to clarify the detailed mechanism of this interaction. The use of paramagnetic resonance of a deep-lying impurity to study Fermi level motion in high-resistivity materials should have general applicability and can take over when other electrooptical measurements
The J O U Tof ~Physical Chembtru
NOTES
of transport properties become difficult. It can be used to monitor the kinetics of defect formation in these materials in a nondestructive relatively simple manner.
Acknowledgments. We wish to gratefully acknowledge the help of J. A. Kucza for heat treatments of the crystals and E. E. Tynan for aid in taking the data.
