Surface Technology, 26 (1985) 93 - 10 5 93 A REVIEW ON PHOTOLUMINESCENCE AND ELECTROLUMINESCENCE OF SEMICONDUCTORS FOR PHOTOELECTROCHEMICAL CELLS L. PERALDO BICELLI Departm ent of Applied Phy sical Chemistry, Milan Polytechnic, Research Centre on Electrode Processes, Consiglio Nazionale delle Ricerche, Piazza Leonardo da Vinci 32, 20133 Milan (Italy) (Received January 22, 1985) Summary This review is concerne d with the photoluminescent and electrolumi- nescent properties of semiconductor electrodes used in photoelectroch emi- cal cells for optical-to-electrical energy conversion. Luminescence not only provides insight into the excited-state deactivation routes which are crucial to efficient energy conversion, but also allows one to obtain in a relatively simple manner valuable information about bulk and surface imperfections that negatively influence photoelectrochemical performanc e of the semiconductor. 1. Introduction Solar energy represents a major source of ene rgy for the Earth and it can be transformed into different kinds of energy that are useful to human- ity. One of the various possible transformation methods is the conversion of solar energy into electrical energy by utilizing photovoltaic and photo- electrochemical (PEC) cells. Both devices involve semiconductors, the former is based on a homo- or hetero-solid-solid junction, while the lat ter is based on a solid-liquid junction. There are numerous reviews (e.g., ref. 1 and references cited therein) that provide a general picture of the PEC solar cells that have been studied and discuss the difficulties connecte d with photostability and low efficiency of semiconduc ting electrodes. In order to obtain materials whose perfor- mance allows interesting applications, it is imperative to control all surface and bulk defects that act as recombination centres of the electron-hole pairs. One method of doing this is to characterize the electronic structure of the semiconductor and to examine the excited -state processes as well as their deactivation routes, i.e. to study the photoluminescen t and electrolumi- nescent properties of the semicond uctor itself. In this review, luminescence @ Elsevier Sequoia/Printed in The Netherlands
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
A REVIEW ON PHOTOLUMINESCENCE ANDELECTROLUMINESCENCE OF SEMICONDUCTORS FOR
PHOTOELECTROCHEMICAL CELLS
L. PE RALDO BICELLI
Departm ent of Applied Phy sical Chemist ry, Mil an Poly technic, Research Centre on
Electrode Processes, Consiglio N az ional e delle Ricerche, Piaz za Leonardo da Vinci 32,
20133 Milan (Italy)
(Received January 22, 1985)
Summary
This review is concerned with the photoluminescent and electrolumi-
nescent properties of semiconductor electrodes used in photoelectrochemi-cal cells for optical-to-electrical energy conversion. Luminescence notonly provides insight into the excited-state deactivation routes which arecrucial to efficient energy conversion, but also allows one to obtain in arelatively simple manner valuable information about bulk and surfaceimperfections that negatively influence photoelectrochemical performance
of the semiconductor.
1. Introduction
Solar energy represents a major source of energy for the Earth and it
can be transformed into different kinds of energy that are useful to human-ity. One of the various possible transformation methods is the conversionof solar energy into electrical energy by utilizing photovoltaic and photo-
electrochemical (PEC) cells. Both devices involve semiconductors, theformer is based on a homo- or hetero-solid-solid junction, while the latteris based on a solid-liquid junction.
There are numerous reviews (e.g., ref. 1 and references cited therein)that provide a general picture of the PEC solar cells that have been studiedand discuss the difficulties connected with photostability and low efficiencyof semiconducting electrodes. In order to obtain materials whose perfor-mance allows interesting applications, it is imperative to control all surfaceand bulk defects that act as recombination centres of the electron-holepairs.
One method of doing this is to characterize the electronic structureof the semiconductor and to examine the excited-state processes as well astheir deactivation routes, i.e. to study the photoluminescent and electrolumi-nescent properties of the semiconductor itself. In this review, luminescence
from semiconductors will be discussed in connection with their PEC behav-
iour, and the basic principles will be considered by which such excited-state
processes occur. We then concentrate on the actual knowledge of and the
problems relating to photoluminescence (PL) and electroluminescence (EL).
Furthermore, PL and EL are examined as useful in situ techniques for thestudy of semiconductor-electrolyte interfaces.
Although research on light emission from semiconductors in contact
with an electrolyte dates back to the 197Os, at present there is a renewed
interest and several research groups are active in the field.
2. Theoretical considerations
2.1. Photogenerative cells
In PEC cells capable of transforming solar energy into electrical en-ergy (so-called regenerative cells), such as for example the n-CdSI S2-, S,
OH-IPt cell, the electrolyte contains a single dissolved redox couple. On
immersion of the semiconductor in the electrolytic solution there is an
exchange of electrons between the two phases until their Fermi levels, i.e.,
their electrochemical potentials, are equal. Owing to the small charge carrier
concentration in the semiconductor, the transfer process modifies the
electron concentration with respect to the bulk value in a small region,
typically 0.1 to 1 pm wide close to the interface, the so-called space-charge
region, thus producing a parallel bending of the semiconductor valence andconduction bands. In the case when the Fermi level of the redox couple
(representing its Nernst potential) is effectively situated within the semi-
conductor forbidden band at flat-band conditions, a depletion layer of the
majority carriers forms in the proximity of the junction. This implies a
simultaneous bending-down or bending-up of the bands for n-type or p-type
semiconductors respectively, whereas the Fermi level remains flat through-
out the whole material. Absorption of a photon whose energy is equal to or
greater than the band gap energy promotes an electron from the valence
band to the conduction band, while the electric field in the depletion layerseparates the electron-hole pairs thus produced, preventing their recombina-
tion. Thus, in an n-type semiconductor that is part of a PEC cell, the minor-
ity carriers (holes) move toward the electrolyte and oxidize the reduced
component of the redox couple, whereas the majority carriers (electrons)moire toward the bulk of the semiconductor and reach the counterelectrode
to produce the reverse reduction reaction. Therefore, the electrolyte com-
position remains (in principle) the same, and solar energy is converted into
electrical energy.
Although the device is theoretically expected to operate indefinitely,
the desired PEC reaction may be accompanied by other unwanted reactions,e.g. decomposition of the material. In fact, one of the major problems in
making practical PEC cells is to avoid photocorrosion of the semiconducting
electron with the injected hole yields emission of band gap or sub-band gap
energy [7 - 91. Thus, the emitted EL spectrum is a consequence of the
charge-transfer process across the interface, and EL continues until the
oxidizing agent in the solution is completely consumed. Usually, to avoid
EL intensity decay from the initial value, the electrode is continuouslypulsed between a potential where no current is observed and a potential
sufficiently cathodic to initiate EL, while the spectrometer is scanned at
constant speed. In addition, the electrolyte is thoroughly stirred. Some evi-
dence for competitive electrode reduction has also been presented [ 9 - 111.
EL has been observed with a variety of oxidized components of redox
couples, e.g. Fe( CN)6 3- for n-Gap [7,12], and, more recently, Fe3+, Ce4+
and again Fe(CN), 3- for n-GaAs [ 13 - 151. Using one-electron redox cou-
ples, the energetic conditions for hole injection into and/or near the valence
band can be analyzed more carefully than in cases where the injecting
species is a highly active intermediate (e.g. SO,-) whose energy level distribu-
tion is poorly defined.
2.3. Comparisons
In agreement with the above, EL takes place when the semiconductor
bands are in light accumulation conditions very close to flat-band conditions
(i.e. small upward bending for n-type materials) and the oxidizing agent
Fermi level is close to the valence band edge. In contrast, to obtain a PEC
response, the semiconductor bands have to be in depletion conditions
(i.e. strong downward bending) and the redox couple level has to be insidethe band gap. This is in agreement with electric current circulating in oppo-
site directions during EL and PEC experiments, i.e., a dark cathodic current
and a photoanodic current, respectively, are produced.
EL is a typical surface phenomenon being initiated at the semicon-
ductor-electrolyte interface, the hole diffusion length L, (we are still
considering an n-type semiconductor) in practice giving the thickness of the
surface layer involved. However, excitation wavelengths used to induce PL
will typically have penetration depths a--’ (where cx is the incident photon
absorption coefficient) exceeding lo-’ cm. Therefore, when CW-’ s muchgreater than L,, PL originates from the bulk of the material, whereas EL
gives a typical picture of the semiconductor surface, i.e., PL and EL have a
different spatial origin. Since hole diffusion lengths generally decrease with
increasing charge carrier density, the above conditions are met better the
higher the semiconductor is doped. Conversely, very pure and lightly doped
materials show similar PL and EL spectra, provided the same emissive
excited state is populated.
3. Experimental results and discussion
Significant experimental PL and EL results of various authors will be
discussed, whereas those regarding photoelectrochemistry (examined in a
previous review [ 11) will be considered in connection with luminescence
phenomena only.
3.1. Photoluminescence spectral distribution
Figure l(a), taken from ref. 5, is an example of PL spectra due ton-type CdS,Se, _ X single crystals in 5 M OH- electrolyte. After illumination
with ultra-band gap (457.9 nm) light, green, yellow-green, orange, red and
red luminescences are obtained corresponding to the x-values of 1.00, 0.74,
0.49, 0.11 and 0.00 respectively, i.e. the emitted light is red shifted with
increased selenium content.1 I I I
i
(b) 0I I 1 1 J
500 600 700 800
WAVtLENtTH, nm
Fig. 1. (a) Un corr ected PL spectra of CdS,Se_, sam ples irr adia ted in 5 M OH elec-
tr olyte while th ey were held at -1.50 V (SCE). Th e ar ea of th e sam ples was about 0.25
cm2 a nd t hey were excited with 457.9 nm light of about 1.0 mW power (excita tion spike
sh own at l/100 th e scale of th e PL spectru m). Em ission inten sities are not directly com-
par able because of differen ces in geomet ry. (b) Uncorrected EL spectra of th e sam plesin (a) obtained without cha nging their geometr y. The electr olyte in th e EL experiments
was 5 M OH--O.1 M SzOs2-. Electr odes wer e cont inu ously pulsed between 0.00 V (11 s)
an d -1.50 V (SCE) (1 s) while the emission spectr ometer was scann ed at 12 nm min-‘.
For both PL an d EL, a spectra l resolut ion of 2.0 nm was employed. (From ref. 5.)
High resolution PL spectra occasionally show variation in breadths
depending on excitation wavelength owing to self-absorption effects, the
spectral mismatch occurring almost exclusively in the high energy tail of
the emission band where the probability for emitted light absorption is
greatest.As expected, PL spectra often contain more than one peak. An example
is shown in Fig. 2, taken from ref. 16. The PL spectrum of aluminium-
doped n-type ZnSe is dominated by a broad peak, but a weaker and sharper
one is also present. The energetic proximity of the latter to the band gap,
together with its temperature dependence (a marked short-wavelength shift
on cooling at 77 K is observed) allowed the designation as edge emission.
The broader sub-band gap emission was, however, assigned to a transition
involving a deep trapping level arising from a complex (VznAlzn) based on
a zinc vacancy and on a next-neighbour donor impurity, i.e. aluminium
on a zinc site [ 17 - 201.
I L 1 I
500 600 700
WAVELENGTH. nm
Fig. 2. Corrected, front&-face PL spectra of n-ZnSe:Al at 295 K (broken line) and 77 K(full line). Both spectra were taken at the same sensitivity (a loo-fold increase in sensi-
tivity was used in scanning the 410 - 480 nm region). The sample was excited with iden-
tical intensities of 405 nm light without disturbing the experimental geometry. (Fromref. 16.)
3.2. PL quenching
The effect on PL properties of the passage of a photocurrent was
examined with different n-type semiconductors in aqueous solutions con-
taining a redox couple, that is in the same conditions of a PEC experiment
[ 5, 21 - 231. In passing from open-circuit to in-circuit conditions the inten-
sities of the PL peaks are quenched by the applied electric field. This is
shown in Fig. 3(a), taken from ref. 2, for an n-CdSe single-crystal electrode
in an alkaline sulphide-polysulphide solution. Since the spectral distribution
is not appreciably altered, changes in PL are monitored at a single wave-
Fig. 3. (a) Uncorrected PL spectrum of an n-CdSe single-crystal electrode excited with514.5 nm light in a 1 M OH--l M S2--0.5 M S electrolyte. Curves A, B and C were taken
at -1.50 V (open circuit), -1.45 V and -1.30 V us. SCE, respectively; all spectra wereobtained with an identical sample geometry. Photocurrent (b) and emission intensity (c)monitored at &,a, _ 720 nm us. potential for an n-CdSe single-crystal electrode excitedat 514.5 nm in a 5 M OH--O.11 M Se2--0.004 M Se 22- electrolyte. The photocurrent
quantum efficiency, f#~,, at -0.3 V (SCE) was measured and found to be about 0.9.
Related experiments with n-CdSe electrodes are given in ref. 6. (From ref. 2.)
Simultaneous measurements of photocurrent and PL intensity as a
function of electrode potential allow one to obtain the so-called ILV curves
reported in Figs. 3(b) and 3(c). There is a typical inverse correlation betweenphotocurrent and PL intensity, as the former increases the latter decreaseswhen increasingly anodic potentials are scanned. In fact, as more anodicpotentials are applied to the semiconductor, downward band bending inthe space-charge region is increased. Therefore, electron-hole separation ispromoted and radiative electron-hole recombination inhibited. Conse-quently, the photocurrent is enhanced and photoluminescent intensitydiminis)led.
Quenching of semiconductor PL by applied electric fields has been
quantified by a dead-layer model [ 25 - 271 which has recently been foundto be applicable also to PEC-based systems [ 2, 5, 16, 28 - 301. In its simplestform, the model states that electron-hole pairs formed within some fractionof the depletion width are swept apart by the field and do not contributeto PL.
In Fig. l(b), taken from ref. 5, EL spectra of n-type CdS,Sei_, single
crystals in a 5 M OH--O.1 M S,Os 2- electrolyte are reported. The electrodes
were continuously pulsed between 0 and -1.5 V (SCE), thus avoiding EL
intensity decay.Another example is shown in Fig. 4, taken from ref. 15, where the
relationship between the EL spectrum and the oxidized component of the
redox couple in the electrolyte is considered. For n-GaAs, both edge (at
0.87 ,um) and sub-band gap (at 1.1 pm) emissions are observed. The latter
is related to a deep acceptor level due to gallium vacancies. The remarkable
difference between EL intensity of Ce4+- or Fe3+-containing solutions has
been explained on the basis of the higher redox potential of the Ce3+/Ce4+
redox couple assuring a better overlapping of the electrolyte empty energy
levels and both the valence band and the deep acceptor level. Therefore,
Ce4+ ions easily inject holes into the semiconductor, whereas Fe3+ ions
do not [ 151.
EL spectra are typically recorded under conditions which facilitate
comparisons with PL spectra. As Figs. l(a) and l(b) clearly show, there is
a striking similarity between the PL and the EL spectra, thus indicating
that an identical emissive excited state is involved in both experiments.
In other cases, however, particularly when a high resolution is em-
ployed, spectral differences in the high energy tail of EL and PL spectra
are observed [ 5, 6, 151. This spectral mismatch has been attributed to
greater self-absorption of the emitted light in the PL experiment [ 5, 61.
WAVELENGTH (pm)
Fig. 4. Uncorrected EL spectra of n-GaAs in 0.01 M Fe,(SO& + 1.0 M HzS04 and in0.03 M Ce(SO& + 1.0 M HzSO 4. The electrode potential was continuously pulsed be-
tween 0 (20 s) and -1.75 V (2 s), while the emission spectrum was scanned at 5 w 6-l.
Note the tenfold increase in sensitivity in order to record the spectrum in the Fe,(SO&
As on average EL is significantly more surface sensitive than PL,
several semiconductors show some EL emission peaks below the band-gapenergy which do not appear in the PL spectrum (compare curves a and c ofFig. 5, taken from ref. 31). It has been suggested that these additional peaks
originate from transitions involving hole-accepting, i.e. filled, surface states[4, 5, 7, 9, 31, 321.
The rather exceptional case of a PL peak having no EL counterpart(compare curve c with curve b of Fig. 5) indicates that the empty en-ergy levels of the redox couple in the electrolyte (i.e. the energy levels of
Fe(CN),3-) and the valence band do not overlap. However, such emptylevels overlap the hole-accepting surface states and, therefore, can directlyinject holes into them, thus producing the EL peak around 900 nm shownin curve b of Fig. 5 [31].
1 0 0 0 800 600 (nm)
Fig. 5. CdS PL and EL spectra in the following electrolytes: curve a, EL in i M KOH-0.1
M KzSzOs; curve b, EL in 1 M KOH-0.1 M K3Fe(CN)s; curve c, PL in 0.2 M NazS03-0.2
M NazS04. (From ref. 31.)
3.4. Efficiencies
3.4.1. PL efficiencyThe quantum efficiency of PL q& (defined as photons emitted per
photon absorbed (or per populated excited state)) is very difficult toestimate with accuracy. Therefore, an upper limit can be obtained by findingother experimental conditions (e.g. cooling the material) which markedlyenhance the emission [ 61. Another technique developed by Wrighton et al.
[33] compares light reflected from a non-absorbing material with lightreflected and emitted from the sample.
Sometimes, 4, was observed to decrease with the excitation wavelengthand such behaviour has been attributed to the presence of a near-surfacenon-emissive region which absorbs larger fractions of incident light, thusreducing 9, [ 161.
By considering the different excited-state decay routes, one mayobtain a very simple relationship that connects 4, with the photocurrent
was well correlated with the formation of an insulating CdS,Se,_, film atthe surface. By penetrating deeper into the electrode, the mixed compoundseparated into CdS and CdS,Se, _ Y which was considerably poorer in sul-phur [38].
Surface modifications due to treatments such as etching, PEC etchingas well as adsorption of metal ions to which semiconductors are submittedin order to increase their PEC behaviour have also been examined through
their effect on luminescence properties [ 391.Although surface treatments obviously affect EL spectra more than PL
spectra, evidence was found that PL and its spatial distribution may beinfluenced, too.
Both etching and photoetching of CdSe electrodes resulted in anincrease in the emitted light (up to 50 times after photoetching) and in abroadening of the high energy portion of the emission band [6]. Such an
enhancement of PL intensity suggested a substantial increase in PL efficiencyand the concomitant spectral mismatch indicated a greater contribution toPL from regions on average nearer the semiconductor-electrolyte interface.At present, there are uncertainties as to the origin of these effects and
research is still in progress [ 61.It may thus be concluded that PL and EL spectroscopies are very
important tools for surface properties investigation, as even these few
examples have shown. Therefore, it is very easy to predict an increasingdevelopment of these techniques in the near future.
References
1 L. Peraldo Bicell i, Surf. Technol., 20 (1983) 357.
2 A. B. Ellis, J. Chem. Ed., 60 (1983) 332.
3 R. E. Halsted, in M. Aven and J. S. Prener (eds.), Physics and Chemistry of II-VI
4 H. H. Streckert, B. R. Karas, D. J. Moran0 and A. B. Ellis, J. Phys. Chem., 84 (1980)3232.
5 H. H. Streckert, J. Tong, M. K. Carpenter and A. B. Ellis, J. Electrochem. Sot., 129
(1982) 772.6 H. H. Streckert, J. Tong and A. B. Ellis, J. Am. Chem. Sot., 104 (1982) 581.7 K. H. Beckmann and R. Memming, J. Electrochem. Sot., 116 (1969) 368.
R. Memming, J. Electrochem. Sot., 116 (1969) 785.
8 F.-R. F. Fan, P. Leempoel and A. J. Bard, J. Electrochem. Sot., 130 (1983) 1866.9 B. Pettinger, H. R. S chijppel and H. Gerischer, Ber. Bunsenges. Phys. Chem., 80
(1976) 849.10 D. M. Kolb and H. Gerischer, Electrochim. Acta, 18 (1973) 987.11 H. Masuda, A. Fujishima and K. Honda, Chem. Lett., (1980) 1153.12 Y. Nakato, A. Tsumura and H. Tsubomura, in A. J. Nozik (ed.), Photoeffects at
Semiconductor-Electrolyte Interfaces, ACS Symp. Ser., American Chemical Society,
Washington, DC, 1981, p. 145.13 S. Menezes and B. Miller, J. Electrochem. Sot., 130 (1983) 517.
14 F. Decker, B. Pettinger and H. Gerischer, J . Electrochem. Sot., 130 (1983) 1335.
15 F. Decker, M. Abramovich and P. Motisuke, J. Electrochem. Sot., 131 (1984) 1173.
16 P. M. Smiley, R. N. Biagioni and A. B. Ellis, J. Electrochem. Sot., 131 (1984) 1068.
19 G. Jones and J. Woods, J. Lumin., 9 (1974) 389.
20 S. Satoh and K. Igaki, Jpn. J. Appl. Phys., 22 (1983) 68.
21 G. Petermann, H. Tributsch and R. Bogomolni, J. Chem. Phys., 57 (1972) 1026.22 A. B. Ellis, B. R. Karas and H. H. Streckert, J. Chem. Sot., Faraday Discuss., 70
(1980) 165.
23 A. Fujishima, Y. Maeda, S. Suzuki and K. Honda, Chem. Lett., (1982) 179.
24 B. R. Karas and A. B. Ellis, J. Am. Chem. Sot., 102 (1980) 968.
25 K. Ando, A. Yamamoto and M. Yamaguchi, J. Appl. Phys., 51 (1980) 6432.
26 K. Ando, A. Yamamoto and M. Yamaguchi, Jpn. J. Appl . Phy s., 20 (1981) 679.
27 R. E. HoIIingsworth and J. R. Sites, J. Appl. Phys., 53 (1982) 5357.
28 W. S. Hobson and A. B. Ellis, J. Appl. Phys., 54 (1983) 5956.
29 R. Garuthara, M. Tomkiewicz and R. P. Silberstein, J. Appl. Phys., 54 (1983) 6787.
30 A. Etcheberry, J. Gautron, R. Parsons and J. L. Sculfort, 35th Meeting of Znt erna-
tional Societ y of Electrochemist ry, Berkeley, CA, August 5 - 10, 1984, AnstractA3-31.
31 I. Jimenez, F. Decker and P. Salvador, 35th Meeting of International Society of
Electrochemist ry, Berkeley, CA, August 5 - 10, 1984, Abstract 112-3.
32 Y. Nakato, A. Tsumura and H. Tsubomura, Chem. Phys . Lett ., 85 (1982) 387.
33 M. S. Wrighton, D. S. Ginley and D. I. Morse, J. Phy s. Chem., 78 (1974) 2229.
34 B. R. Karas, D. J. Morano, D. K. Bilich and A. B. Ellis, J. Electrochem. Sot ., 127
(1980) 1144.
35 B. R. Karas, H. H. Streckert, R. Schreiner and A. B. Ellis, J. Am. Chem. Sot., IO3
(1981) 1648.
36 H. Gerischer and I. Mattes, 2. Phys. Chem. N.F., 49 (1966) 112.
37 H. Gerischer, R. McIntyre and B. Smandek, 35th Meeting of International Society
of Electrochemist ry, Berkeley, CA, August 5 - 10, 1984, Abstract A3-1.
38 M. Tomkiewicz and R. Garuthara, 35th M eeting of International Society of Electro-
chemistry , Berkeley, CA, August 5 - 10, 1984, Abstract A3-2, in J. Appl. Phys., 58
(1985) 1662.
39 N. Miiller, M. Abramovich, F. Decker, F. Iikawa and P. Motisuke, J. Electrochem.