HAL Id: jpa-00220742
https://hal.archives-ouvertes.fr/jpa-00220742
Submitted on 1 Jan 1981
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
TIME-RESOLVED LUMINESCENCE MEASUREMENTS ON AMORPHOUS
PHOSPHORUS
G. Fasol, E. Davis
To cite this version:
G. Fasol, E. Davis. TIME-RESOLVED LUMINESCENCE MEASUREMENTS ON AMOR- PHOUS PHOSPHORUS. Journal de Physique Colloques, 1981, 42 (C4), pp.C4-571-C4-574.
�10.1051/jphyscol:19814124�. �jpa-00220742�
Colloque C4, suppZ6ment au nOIO, Tome 42, octobre 1981
TIME-RESOLVED LUMINESCENCE MEASUREMENTS ON AMORPHOUS PHOSPHORUS G. Fasol and E.A. Davis
*
Cavendish Laboratory, MadingZey Road, Cambridge, CB3 OHE, England
" ~ e p a r t m e n t of Physics, University o f Leicester, Leicester LE1 7RH, England
Abstract.- Time resolved luminescence spectra of amorphous phosphorus are reported. An emission band around 1.17 eV decaying with a time constant of 150 ns and a band at 1.41 eV are found. The latter show a fast (20 ns) oomponent, andinaddition a decay with a broad distribution of lifetimes (average lifetime 3.8 ms) attributed to radiative tunnelling. The absence of a Coulomb shift suggests the presence of charged defects. The results are compared with recent ODMR measurements.
Introduction.- This paper presents information about defect states in amorphous phos- phorus, obtained by measuring the time evolution of the photolqinescence spectrum.
Calculations and theoretical discussions (1,2) have predicted defect levels in the gap due to two- and fourfold coordinated atoms. We find two broad emission bands, one centred at 1.41 eV with a fast and a slow component, attributed to radiative tun- neling, and the other at 1.17 eV which decays with a time constant of around 150 ns.
Experimental.- Bulk amorphous phosphorus of semiconductor quality was studied. The samples were produced by zone melting (3) and Alusuisse, Mining and Chemical Prod- ucts (99.9999% purity) and by Hoechst; for whose sample the sum of metallic impurities was 0.1 ppm. The fourth type studied was produced by S. Veprek using a plasma trans- port process. This material has a few ppm metallic impurities and contains 5-7% hy- drogen (4). Luminescence was excited by 6 ns dye laser pulses. The luminescence spectra were extracted from the raw curves by correcting for the system response; all measurements were made in the linear region of response and sample heating was esti- mated to be less than 1 K below 20 K. The resolution of the monochromator is shown in the figures.
Results.- Fig. 1 shows the evolution of the luminescence spectrum at 17K after ex- citation. The salient feature is the presenceof two bands. The excitation energy is 2.03 eV and lies at the peak of the narrow excitation spectrum (5). At zero delay and at delays greater than 5 us the luminescence is a broad band (HE) peaking at
1.41 eV and has a similar shape to the band found with chopped (80 Hz) illumination (Fig. la). At short delay times (Fig. Ib) a new luminescence band (LE) centred ar- ound 1.17 eV dominates the spectrum. As the delay time increases further it becomes progressively weaker than the band centred at 1.41 eV (HE). The solid lines in Fig.
Ib are fits to the measured spectra of the sum of two gaussian curves with peak en- ergy 1.17 eV and FWHM of 0.27 eV for the LE band and 1.41 eV and 0.42 eV for the HE band respectively, and show clearly the presence of two bands. We used the relative heights of the two bands obtained by fitt'ing to establish the time dependence, which is shown in Fig. 2. The LE band decays nearly exponentially with a time constant of around 150 ns. The HE band has a very rapid initial decay (time constant initially'~20 ns), decaying by four orders of magnitude within the first microsecond.
At later times it shows a behaviour characteristic of a distribution of lifetimes.
Fig. 3 shows the shift of the peak of the emission spectrum as a function of delay
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19814124
JOURNAL DE PHYSIQUE
Fig.] : Evolution of the lumdnescence spectrum after excitation with 6 ns pulses of a dye laser at E = 2.03 eV. The spectra are on a linear scale and displaced for cla- rity and normalized to the same height. The solid lines in ( b j are least squares fits of the sum of two gaussian bands. The dashed curves give the two gaussian bands used to fit the decay curve at 800 ns delay.
time. For times larger than 900 ns the HE band dominates. A major result of this investigation is that any shift of the HE band at times up to 100 us is smaller than the experimental resolution of 30 meV. For times larger than 100 us the HE band shifts gradually to lower energies, the total detected shift being around 60 meV.
Measurements on polished, ground and freshly broken surfaces showed the same lumin- escence behaviour as did samples from the three industrial sources. The material prepared by plasma transport containing 5-7% hydrogen has about 50% lower quantum efficiency. In that material the LE band had a shorter lifetime compared to the other types. The spectral shape of the two bands is temperature independent within the spectral resolution, but the LE peak decreases faster with increasing temperature than HE.
Discussion.- The main result of this investigation is the discovery of two separate emission bands (LE and HE). Competing radiative recombination processes are known to occur in a-As2Sj (6). The previously measured temperature dependence and the ex- citation spectrum of amorphous phosphorus (5) and the results of this paper, show many similarities between amorphous phosphorus and As2S3. The component of the HE band dominating at delay times greater than 1 us is best explained by radiative tun- nelling. The apparent absence of a Coulomb shift seems to indicate a process without Coulmb interaction. Our results, of couse, do not exclude Coulomb interaction, but they do put a lower limit on the value of the Bohr radius of the larger of the car- riers. The solid lines in Fig. 3 show the shift expected, assuming the arguments of Ref.(7) and (8). The dielectric constant of phosphorus is ~ = 6 . 1 (9). Thus R, must be at least 15g to explain the results. This limit becomes considerably higher, if
the zero delay peak is included in this reasoning. At long times such a weak localisation is thought to be unlikely. The explanation in terms of
(squares) as a function of delay time for a 6 ns excitation pulse, as determined from the fits shown in Fig. lb. The solid curve shows an exponential decay with time constant 152 ns, the broken curve shows a convolution of exponential decay curves with a distribution of lifetimes G(T)
shown in the insert.
red P h o s p h o ~ s l = 17K ;E 2 . 0 3 e L
Fig. 3: Shift of the peak of the lumines- censce spectrum as a function of delay time after a 6 ns pulse. The error bars show the uncertainty of determining the peak of a broad band. The dashed lines show the Coulomb shift (Ref. 8) .E' would be expected if there is attraction between the recombining partners before recombina- tion and none after recombination and E"
for the opposite case. R, is the Bohr ra- dius of the larger carriers.
radiative tunnellingis confirmed by analysis of the decay curves. The solid line in Fig. 2 shows that the long tlme component of HE can be fitted well by a de- cay curve as expected for independent decay of centres with a distribution of life- times. Such an analysis has been described in Ref. 10 for a-Si. The distribution function of lifetimes G(T) for which the best fit was achieved is shown in the insert of Fig. 2, where TG(T) is plotted to take account of the logarithmic scale. The mean lifetime is < T > = 3.8 ms. Since a short pulse emphasizes the short delay times, this determinafiion has obviously some uncertainty for longer decay times.
The fact that HE dominates the spectrum at zero delay is at first puzzling.
There are two possible explanations: (A)In addition to the tunnellingprocess a sepa- rate radiative channel exists with a high generation probability, a lifetime of aro- und 20 ns and emits the same HE band. (B) The radiation at zero delay is due to the same process as HE at long delays. This explanation would demand a larger density of cemtres than (A) and would mean that the rapid initial decay is caused by non- radiative transitions which would affect HE much more than LE. In view of the weaker temperature dependence of HE explanation (A) is thought to be the better one. To in- terpret the LE band the optically detected magnetic resonance measurements by S. De- pinna and B.C. Cavenett (to be published) must be considered. Apart from an enhanc- ing resonance g = 2 with a spectral resonance like our HE band they find another en- hancing resonance with a spectral response peaking around 1.1 eV showing characteris tics of a state with triplet multiplicity. If it is assumed that this resonance corresponds to the same process as our LE band with a lifetime of around 150 ns,it is unlikely to be due to a (S = +;, s = +;) pair, since such a transition should be slow. In view of the fact that ODMR uses continuous excitation while we use 6 ns pul- ses, it is not impossible that the triplet resonance and our LE band derive from different processes although we do not think this is likely. In addition it was found that if the resonance corresponding to the HE band is excited, the low energy luminescence decreases, but not vice versa, proving that a transfer of carriers be- tween the two recombination processes is important.
Assuming that the Stokes shift between the excitation and luminescence spectra is due to lattice relaxation at a trap, the thermal trapping depths would be around 0.29 eV for HE and 0.41 eV for LE. It has been proposed, that two-fold (P;) and four fold coordinated (P;) atoms could be stable in charged states (1,ll). These centres could trap one or two carriers. The HE band could be explained as in a-AspSg, by
JOURNAL DE PHYSIQUE
the recombination of a hole trapped on PT with an electron localized in the band edge or by the equivalent process on P i with reversed charges. But recombination of an exciton bound to such a defect or to an IVAP (as suggested by S. Depinna and B.C. Cavenett) would be expected to be non-radiative (12). Uncharged dangling bonds
(P?) were also proposed to be stable (11). They could trap single carriers, leading to tunnelling recombination, or they could trap excitons. In this case the Coulomb interaction would contribute to the transition energy. It would then be necessary to explain why the Coulomb interaction term is smaller than our experimental reso- lution. Band to band transitions (6) would be possible candidates to explain the components with short lifetimes; but explanation of the tunnellingprocess by band to band transitions as in a-Si (10) would again demand a Coulomb shift and would give difficulties in explaining the cross-feeding effect found in the ODMR experi- ments.
We have found that two bands with a great variety of lifetimes contribute to the luminescence in amorphous phosphorus. Recent ODMR measurements confirm our results and provide additional information. We found evidence fortunnelling proces- ses, for which we determine the mean life-time, and we found an indication of ther- malisation, which could give an estimate on the range of the band edge localized
states. A combination of existing models should be able to explain the present re- sults, although we do not have a consistent picture yet.
Acknowledgements.- The authors are very grateful to Professor N.F. Mott for dis- cussions and to Dr. A.D. Yoffe for encouraging this work and for many helpful dis- cusslons to S. Depinna and Dr. Cavenett for discussions and the communication of results prior to publication and to the Science Research Council for support. One of us (G.F.) is very grateful to E.P. OIReilly and R.T. Phillips for many discus- sions, to the British Council for a Research Fellowship and Hoechst and Alusuisse
for samples.
References
l . ELLIOTT, S.R.,DAVIS, E.A., J. Phys. C
12
(1979) 2577.2. O'REILLP, E.P. and KELLY, M.J., J. Phys. C., to be published.
3. CREMER, J. andKRIBBE
,
H., Chemie-Ingenieur-Technik,36
(1964) 957.4. BRUNNER, J.,THULER, M.,VEPREK, S.,WILD, R., J. Phys. Chem. Solids
40
(1979) 967.
5. KIRBY, P.B. andDAVIS, E.A., J. Non-Cryst. Solids.
35
&36
(1980) 945.6. SHAH, J., Phys. Rev. B.
2,
(198.0) 4751.7. WILLIAMS, F.E., J. Phys. Chem. Solids 12, (1960), 265.
8. STREET, R.A., Sol. State. Commun. 34, n.980) 157.
9. LANNIN, J.S. and SHAWNABROOF, ~.d.,Tolid State Commun. 28 (1978) 497.
10. TSANG, C. andSTREET, R.A., Phys. Rev. B 19 (1979) 3027,
1 1 . GREAVES, G.N.,ELLIOTT, S.R. andDAVIS, E.A., Adv. in Physics, -
2
(1979) 49.12. STREET, R.A., Phys. Rev. B
