A. PROCK, M. DJIBELIAN, AND S. SULLIVAN
3378
Photoinjection from Rhodium into Hydrocarbon Liquids under Visible Excitation
by Alfred Prock, Mihran Djibelian, and Susan Sullivan Department of Chemistry, Boston University, Boston, hifassachueetts 02816
(Received November 11, 1966)
Photoinjection from a rhodium electrode into hydrocarbon liquids and solutions of aromatic molecules has been observed to occur under sufficiently intense visible radiation. Benzene, cyclohexane, and benzene solutions of naphthalene, phenanthrene, anthracene, and pyrene were studied. The photoinjection wavelength threshold varied from that of the ultraviolet (for cyclohexane) to the deep red region (solutions of aromatic molecules). These differences are related to the electron affinities of the dissolved species and to the solvation energies of the radical anions produced.
This paper reports observations of apparent photoinjection which occurs from a rhodium electrode into hydrocarbon media under the stimulation of visible light. The liquids studied were cyclohexane, benzene, and benzene solutions of naphtnalene, phenanthrene, anthracene, and pyrene; the light was from a short-arc xenon lamp with suitable combinations of liquid and interference filters. The effect is unusual in that light of low energy, l.SS ev, is capable of producing photocurrents in the case of benzene solutions of the fusedring aromatic compounds. Photoinjection is the name given to the photoelectric effect when liquid rather than vacuum fills the space between electrodes.' Lack of photocurrent saturation, along with some small dependence of threshold wavelength on applied voltage, and other phenomena distinguish the photoinjection case from the photoelectric effect. Photoinjection has been studied using an aluminum cathode immersed in the liquid aliphatics, n-hexane and n-decane, under ultraviolet illuminat i ~ n l -where ~ it is found that the yield of photocurrent under given conditions is 2-3 orders of magnitude smaller than with vacuum present, but it is still 3 orders of magnitude larger than the dark current in the case of n-hexane. The photothreshold of aluminum, 4.1 ev, was found to remain nearly the same when immersed in n-hexane under low applied fields. Changes in threshold wavelength have been observed, on the other hand, as reported in a recent review by Adamczetv~ki,~ where results are discussed for various elecThe Journal of .Physical Chemistry
trodes immersed in liquids and gases and some estimates of work function changes are given. More recently, response of metal surfaces in gases at high pressure has been r e p ~ r t e dand , ~ the effect of chemisorbed gases6 in changing work function appeared to depend on the electron affinity of the adsorbed molecule. In the present work the apparent lowering of the work function of rhodium is regarded to arise from the electron affinity of the dissolved species as well as from solvation energy of the ion formed. The analysis used here is similar to the first few steps in Matsen's analysis of reduction potentials for these fused-ring aromatics in a dioxane-water mixture.' Along with some more recent estimates of electron affinity, these photoinjection results provide estimates of the solvation energies of these molecules in benzene solution.
Experimental Section Apparatus. Light was supplied by an Osram 450-w xenon lamp with reflector and quartz condensers. The desired band was isolated using combinations of (1) M. J. Morant, Nature, 187, 48 (1960). W.Swan, ibid., 190, 904 (1961). (3) J. Terlecki and 0. Gzowski, Acta Phys. Austriaca, 15, 337 (1962). (4) J. Adamczewski, Brit. J . A p p l . Phys., 16, 759 (1965). (5) D. H.Howling, J. A p p l . Phys., 37, 1844 (1966). (6) R.Kh. Burshtein and N. A. Shurmovskaya, Russ. Chem. Rev., 34, 746 (1965). (7) F. A. Matsen, J . Chem. Phys., 24, 602 (1956). (2) D.
PHOTOINJECTION FROM RHODIUM INTO HYDROCARBON LIQUIDS
3379
Table I: Filter Combinations Used to Select Wavelength Regions
Blue, 440 mp
Green, 540 m p
Orange, 600 mfi
Red, 660 mp Red, 745 mp
-
Filters
Region
Half band width, mr
2 cm circulating water
+
1 . 5 cm saturated NaNOz
40
t CuC12, CaCl?"
+
infrared filter CS169 Corning
+
60
90-5-600 45-1-520 90-5-660 45-1-520 Schott Type AL
60 65 23
J. G. Calvert and J. N. Pitts, "Photochemistry," John Wiley and Sons, Inc., New York, N. Y., 1966, p 739.
liquid and interference filters so as to obtain sufficiently intense light. Table I lists the combinations for each band; the order of filters is the same as that in which they appear in the beam. The Bausch and Lomb (B and L) filters are normal incidence and dichroic interference filters. and the Schott filter is a normal incidence interference filter. These combinations were found to be very efficient in removing ultraviolet and undesired visible stray radiation and in reducing the infrared content to the point where heat effects were negligible. Adequate reduction of heat is seen experimentally in the lack of drift of the photocurrent; removal of ultraviolet is seen to be complete in that insertion of an extra saturated Xai\02 filter in the beam causes only the small decrease in effect consistent with the visible transmission of the filter. For intensity variation experiments, neutral density filters of transmission 6.2, 24.4, and 51.7% (Industrial Optics, Bloomfield, X. J.) supplied with calibration curves mere used. The light was focused on the cell and intensity was measured by substituting a Cd8 photoresistor for the cell; resistance was measured to within 1.5% error. Dark currents and photocurrents mere measured using a Cary Model 31 vibrating-reed electrometer with a shunt; the signal was recorded by a Leeds and Xorthrup Azar recorder. Soise level was 6 X 10-15 amp in favorable cases. Voltage was supplied by a Zener diode regulated supply ; the usual applied potential mas 112 v. The cell is shown in Figure 1. The quartz stannous oxide coated electrode at the bottom presses against the ground and polished Pyrex base to make the seal. The surfaces f i t well enough so that liquid leakage is no
problem. The rhodium-plated electrode passes through precision-bore tubing and is adjusted by the top micrometer arrangement. Stopcocks are glass "KOLUB" type, and glass ball joints are used as connectors. A small pressure of oxygen-free nitrogen throughout the system prevents the entrance of significant amounts of oxygen or moisture. The cell can be heated prior to distillation up to 110" using a heat gun while nitrogen passes through it. Use of this kind of cell where liquid touches no gasket material was found necessary to eliminate the very large scatter of response noted when Viton and even Teflon gaskets came into contact with the solutions. The cell was supported on a spring-mounted plate to reduce niicrophonics and was enclosed by an aluminum shield, except for a hole a t the base where shield and cell met for the light to enter. n'itrogen was allowed to flow into the shield throughout the duration of the experiment with the given solution. Chemicals. Benzene and cyclohexane starting materials were fluorometry grade, Mat,heson Coleman and Bell. These were percolated through silica and alumina gel of chromatography grade, refluxed, and distilled over lithium aluminum hydride in a stream of nitrogen which was first passed over hot copper. As measured in the photocell a t 23" (spacing 0.036 cm, constant 9.2 cm, applied voltage 112 v), apparent dark ohm-' cm-' specific conductance was 1.5-2.5 X for benzene* and less than 2 x lo-'' for cyclohexane. Naphthalene, phenanthrene, and anthrecene were obtained as zone refined, 99.99% purity. from James Hinton Co., Valparaiso, Fla. These were further chro~
~~
~~
(8) E. 0. Forster, J. Chem. Phys., 37, 1021 (1962).
Volume 71, Sumber 1 I
October 1967
A. PROCK, M. DJIBELIAN, AND S. SULLIVAN
3380
matographed on a silica gel column using spectroquality pentane as the eluent, by the method of Sangster and Er~ine.~ This was a nontrivial step since it produced a lowering of dark conductivity by as much as a factor of 5 in the case of anthracene, and it removed a rectification effect of a factor of about 3 in the case of anthracene. This latter effect means larger conductivity for rhodium as the negative electrode. It is an effect which we had noticed before in connection with less extensively purified materials. Pyrene was prepared by the chromatography method mentioned before, zone refined, and rechromatographed to give a totally white product. This is the minimum extent of treatment which was found to remove rectification effects and to produce low dark currents in solution which were comparable with those of the first three aromatics. For all solutions studied the apparent dark conductivities were bel’ow 2 X 10-l5 ohm-’ cm-l a t 23”.
200
a I
Q. ?!
52
100
x H
Results and Discussion The photoelectric threshold for the rhodium electrode was determined in a nitrogen atmosphere using a short-arc mercury lamp, Osram HBO 200 W/2, and liquid filters with known cutoff points. The quartz condenser, a front-surface aluminum mirror, and a quartz container for liquids were the only components in the beam. A voltage of 112 v was applied between electrodes with rhodium negative; electrode spacing was 0.036 CM. The leakage current remained constant to within 6 X amp down to the cutoff point of 1 cm of CC1, (Baker Spectrophotometric Quality, 265 mp) but rose “instantly” with a photocurrent of 5.3 X 10-l2 amp when the filter liquid was CHCl, (Baker Spectrophotometric Quality, 245-mp cutoff). The effects were reproducible and are the result of several trials. A tenfold reduction in applied voltage produced
A’
Figure 1. Photoconductivity cell: A, micrometer; B, Teflon seal; C, rhodium-plated rod; D, aluminum ring; E, quartz flat wit,h stannous oxide coating.
The Journal of Physical Chemistry
cwhth. 0
4
AFigure 2. Photoresponse of pyrene (0.285 M ) and of naphthalene (0.36 M ) in benzene. Polarity of rhodium electrode is indicated a t left. Vertical bars a t lower left show dark currents of pyrene and naphthalene, respectively.
no detectable change in threshold wavelength and reduced photocurrent by a factor of 2.5. Reversal of polarity reduced the photocurrent by a factor of lo2. The apparent work function was thereby determined to lie between 4.7 and 5.1 ev, in reasonable agreement with the literature values’O for rhodium, between 4.6 and 4.9 ev for thermionic and photoelectric work functi ons. With cyclohexane as the medium and the rhodium electrode negative no photocurrent was detected down to an incident wavelength of 315 mp, the lower limit for the optical system used. However, with benzene as the medium a photocurrent was observed a t 440 mp which was about as large again as the dark current and easily measurable. That heating was not responsible for the effect was seen by the virtual absence of response when the rhodium was made positive; stray (9) R. C. Sangster and J. W. Ervine, Jr., J. Chem. Phys. 24, 670 (1956). (10) “Handbook of Chemistry and Physics,” 45th ed, Chemical Rubber Publishing Co., Cleveland, Ohio, 1964-1965.
PHOTOINJECTION FROM RHODIUM INTO HYDROCARBON LIQUIDS
3381
0
A-
AFigure 3. Photoresponse of phenanthrene (0.053 M ) in benzene.
80
z"4 e
40
52 X
H
0
0 L
1
+.
4
I
te
R 700 A-
F
Figure 4. Photoresponse of anthracene (0.050 M ) in benzene.
80
a
5
40 X
H
0
0 4
e
I
G
-
R 700
0 A-
Figure 5. Photoresponse of pyrene (0.089 M ) in benzene.
R'
Figure 6. Photoresponse of pyrene (0.074 M) and of naplithalene (0.071 M ) in cyclohexane.
ultraviolet radiation was ruled out also by showing that insertion of an extra ultraviolet filter which transmits less than 1% below 440 mp reduced the effect by only about 20%. There was no photocurrent evident for green light illumination. The photoresponse of the benzene solutions of fusedring aromatics is shown in Figures 2-5. (Figure 6 shows results for a cyclohexane solution.) The polarity shown on each graph is for the rhodium electrode. The vertical height of a mark indicates the scatter of data taken over a few days for each solution to observe whether there were any steady drifts in dark current or photocurrent; no such drifts were noticed. The vertical bar, or bars, a t the lower left of each graph shows the average dark current under 112 v applied with a cell constant of 9.2 cm. As mentioned earlier, no significant dependence of dark current on electrode polarity was observed. The points have been normalized for an which is the incident flux of 10'' photons sec-' actual measured flux for the 4 4 0 - m ~region, where it has been assumed that photoresponse varies linearly with light, intensity for all regions. This linearity is borne out for pyrene in the blue and green regions as shown in Figure 7 . The photocurrents for all bands are seen to be larger with the rhodium as the negative electrode; for anthracene and naphthalene, the asymmetry of response is large in all regions, for pyrene it is largest in the long-wavelength region. Because the photocurrrent shows this kind of polarity dependence but the dark current does not, the photocurrent is attributed to photoinjection, at least in the long-mavelength region near the threshold. In the case of pj.r.ene the relatively large photocurrent seen in the blue and green regions with rhodium positive cannot be overlooked. A step in the mechanism which produces it may be excitation to the triplet state which lies 48.7 Volume 71.Number 11
October 1967
A. PROCK, M. DJIBELIAN, AND S. SULLIVAN
3382
3.0
0.47
2.0
E
+
0.58
H 2.0 c3
2
s
t .H
0 0
J
1.0
0
/+
/ 6.2
24.4 5i.7 io0
LOG % TRANSM I SS\ON Figure 7 . Plot of log of photocurrent us. log of incident light intensity for pyrene (0.285 M ) in benzene, for the blue- and green-wavelength regions, and for rhodium positive and negative.
kcal/mole (384 mp) above the ground state.” The question is under study and will be reported on in the future. The role played by the solvent is a major one. Figure 6 shows the results for naphthalene and pyrene dissolved in cyclohexane. Photocurrents are 2 orders of magnitude smaller than those for benzene solutions and no effects are produced for light of longer wavelength than that of the blue region. The large photocurrents produced with pyrene solutions under blue and green light excitation with rhodium positive appear to be intrinsic to pyrene and not due to impurities. Evidence for this comes from studies of light intensity dependence of photocurrent for this system and others. Results for pyrene are shown in Figure 7 . The abscissa refers to the transmission of neutral density filters placed in the path of the beam. The intensity dependence is seen to be not far removed from a linear one for the conditions of the experiment. In contrast, Figure 8 shows the response under the same experimental conditions of a benzene solution of a pyrene sample kindly supplied to us by the National Bureau of Standards. This was chemically purified to a purity between 99 and 99.9% and then chromatographed on receipt by the method mentioned previously. Photocurrent, is an order of magnitude larger for this material and depends more nearly on the square root of The Journal of Physical Chemistry
1.0
0
2.0
LOG % TRANS. Figure 8. Plot of log of photocurrent us. log of incident light intensity for a sample of chemically purified pyrene (0.142 M ) in benzene. Rhodium polarity is positive: A, blue light; B, green light.
2.0
I
0.48
-+
N
H
* 1.0
s
0
2.0
1.0
LOG %TRANS Figure 9. Plot of log of photocurrent us. log of incident light intensity for pyranthrene (2.5 X M ) . Rhodium polarity is positive; the incident light is blue.
light’ intensity. The square root dependence is also seen, Figure 9, in the response of a benzene solution of pyranthrene, a fused-ring aromatic containing eight rings, in a strongly absorbing region of its spectrum.12 ~~
~~~~~
(11) W. G. Herkstroeter, A. A. Lamola, and G. S. Hammond, J . Am. Chem. Soc., 86, 4537 (1964).
(12) E.J. Clar, “Aromatische Kohlenwasserstoff Polycyclische Systerne,” 2nd ed, bpringer-Verlag, Berlin, 1952,p 380.
PHOTOINJECTION FROM RHODIUM INTO HYDROCARBON LIQUIDS
Finally, the square root dependence is observed in the intensity dependence of the pyrene prepared in this laboratory when it absorbs light in the near-ultraviolet region (345-400 mp), which overlaps its region of strong singlet-singlet absorption. A sufficiently serviceable filter was macle for this region using a series of cobalt blue glasses with ultraviolet filter, Corning CS 052, in addition to the circulating water and infrared filters mentioned before. It is clear that what is observed in the visible region arises from a different mechanism than that active in the ultraviolet region and that there is evidence that it is not due to a strongly absorbing impurity . The large photocurrents which are measured when the rhodium electrode is negative appear to be explainable on the basis of a lowering of the work function of rhodium by the following scheme (compare Matsen') Rh(s) -e-
Rh(s)+
+ e(g)
e(g) $- l\l(g) -3 ;LI(g)M(g)
+ M(SO1)
----f
Rl(g)
AE =
AE
=
(b
-EA
(1) (2)
+ M(so1)AE = AEsoi (3)
Rh(s)
+ lI(so1) -+Rh(s)+ + M(s01)AE = (b - E A + AEsoi = hv
where (b, EA, AEsol, and 11 are work function in vacuum, molecule electron affinity, molecular solvation energy, and the molecule being considered, respectively. hv represents the energy requirement of the reaction; it is the photoinjection threshold observed in the present experiments. AEsol is the unknown in this equation, the work function is taken as 4.7 ev, and recent values for electron affinity are assumed. Table I1 presents the results. In the first column, the last four electron affinities are experimentally determined by the electron-capture method; the benzene value is a theoretical SCF value. The second column brackets the threshold photoemission wavelengths. The third column brackets the values determined for the solvation energies of the ions in benzene, and the last column lists solvation energies determined from polarographic experiments with 73% dioxane-water solutions along with electron affinities from the first column. The solvation energies obtained for benzene solution (D = 2.28) are seen to be comparable with those for dioxane-water solutions (D = 25); that is, they do not differ by the ratio of 1 - (1/D) correction factors,l3*l4 which is 1.7 for this
3383
Table I1
EA, ev
Benzene Naphthalene Phenanthrene Anthracene Pyrene
- 1 .40" 0 . ISb 0. 31a 0 ,M b 0 ,5gb
A,
--Em1 (benzene),
-A E d ,
ml.r
ev
ev
440-540 660-745 660-745 660-745 660-745
3.3-3.8 2.6-2.8 2.5-2.7 2,2-2.4 2.2-2.4
3 . 5Savd 2.44"" 2 . 3ZbSc 2 , 56b*c 2 . 37b3c
a D. R. Scott and R. S. Becker, J . Phys. Chem., 66, 2713 R. S. Becker and E. Chen, J . Chem. Phys., 45, 2403 (1962). R. M. Hedges (1966). F. A. Mat,sen, ibid., 24, 602 (1956). and F. A. Matsen, ibid., 28,950 (1958).
case, that would be expected to apply to the model of a charged structure in a homogenous dielectric fluid. Since the present experiment deals directly with benzene solutions, the authors regard thiznear agreement as just more evidence that the 1 - (1/D) factor, with D the macroscopic dielectric constant, is not strictly applicable to solvation of ions; a considerable solvent structure may exist around the ion even in benzene solution. This is supported by the evidence of behavior of cyclohexane solutions where for naphthalene, for example, there is no clear indication of photoinjection even a t 440 mp. The solvation energy of the negative naphthalene ion in cyclohexane would then be at least 0.9 ev lower than in benzene; this is also unaccountable merely on the basis of the ratio of 1 (1/D) terms between these solvents, namely, 1.1. In conclusion, ionic solvation energy is mainly responsible for work function lowering of rhodium in benzene solutions of the fused-ring aromatic compounds reported on here. Using the known electron affinity for a given species, one should be able to determine ionic solvation energy by this simple method.
Acknowledgment. The authors express their appreciation to the National Science Foundation for support of this research under Grant G P 3526. The authors also wish to express their thanks to Dr. David H. Freeman, Chief of Separations and Purifications Section, Kational Bureau of Standards, for kindly supplying the sample of chemically purified pyrene. (13) N. S. Hush and J. Blackledge, J. Chem. Phys., 23, 514 (1955). (14) L. E. Lyons, Nature, 166, 193 (1950).
Volume 71, Number 11 October 1967