HAL Id: jpa-00209296
https://hal.archives-ouvertes.fr/jpa-00209296
Submitted on 1 Jan 1980
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.
Evidence of a donor-acceptor type transition in CuGaSe2
A. Poure, G. Aguero, G. Masse, J.P. Aicardi
To cite this version:
A. Poure, G. Aguero, G. Masse, J.P. Aicardi. Evidence of a donor-acceptor type transition in CuGaSe2.
Journal de Physique, 1980, 41 (7), pp.707-712. �10.1051/jphys:01980004107070700�. �jpa-00209296�
Evidence of a donor-acceptor type transition in CuGaSe2
A. Poure, G. Aguero, G. Masse and J. P. Aicardi
Laboratoire de Physique du Solide, Université de Perpignan,
avenuede Villeneuve, 66025 Perpignan Cedex, France (Reçu le 3 octobre 1979, révisé le 22 janvier, accepté le 28 février 1980)
Résumé.
2014Nous faisons des mesures de spectroscopie résolue
entemps (T.R.S.) sur
unebande d’émission large
de CuGaSe2, centrée à 1,59 eV de largeur 40 meV (à la température de l’hélium liquide). Le comportement du déclin est mesure à 4 K pour différentes énergies dans la bande d’émission. Nous montrons que les résultats expé-
rimentaux sont
enaccord
avec unmodèle du type donneur-accepteur.
L’étude du déplacement
enT.R.S. permet de déduire par lissage :
2014
la valeur du demi-rayon de Bohr du niveau le moins profond RB
=5,6 A (le centre le moins profond est donc
à 70 meV
aumoins);
2014 la probabilité maximum de transition WMAx = 108 s-1;
2014 la valeur de h03BD~
=1,53 eV.
Les courbes de déclin théorique et expérimentale sont
enbon accord pour
uneconcentration du porteur majo-
ritaire N
=8 1017/cm3.
Abstract.
2014Measurements of spectrum shift
aremade by
meansof time resolved spectroscopy (T.R.S.)
overa
broad emission band of CuGaSe2, peaked at 1.59 eV with
awidth of 40 meV (at liquid helium temperature).
The behaviour of the decay is measured at 4 K for different emission band energies. We will show that the expe- rimental results are in agreement with
adonor-acceptor model.
From the shift of the spectra in T.R.S.
we canfit :
2014 the Bohr-half-radius value RB
=5.6 Å (the shallowest centre is at least at 70 meV) ;
2014 the maximum radiative transition probability WMAX
=108 s-1;
2014 the value hv~
=1.53 eV.
The theoretical and experimental decay
curves arein good agreement for
apreponderant impurity concentration of N = 8 1017/cm3. By using these various parameters
weobserve that theoretical and experimental T.R.S.
spectra
areconsistent.
Classification Physics Abstracts
78 . 60H
1. Introduction.
-CuGaSe2 is a semi-conductor of the I-III-VI2 type which cristallizes in the chalco-
pyrite system (space group 142d) [1]. Through absorption, electro-reflectance and modulated reflec-
tivity measurements, it has been established that
CuGaSe2 is a direct gap semi-conductor [2-5].
The temperature dependence of the gap is given by [6] :
with
Electrical measurements show that the conducti-
vity of this material is hightly influenced by the impu-
rities : we can vary the resistivity of this material from 5 x 10- 2 Q cm to 10 MQ cm.
N. Yamamoto gives the value of the gap according
to the position of the exciton [7] (E,
=1.711 eV at
20 K). There is only one paper available on the lumi-
nescence of this compound [8]. The results remain of
a qualitative order as the authors give no interpre-
tation of the emission observed.
We have studied the cathodoluminescence of syn- thetic crystals provided by L. Mandel (Salford Uni- versity). In this paper, measurements are carried ou
under electron bombardment at liquid helium tempe-
rature.
We use a 25 kV electron gun with an ajustable
electronic density of bombardment (up to 1 mA/mm2).
The electron beam is modulated at the Wehnelt with
an impulsion duration of 1 lis at a frequency of
10 kHz. The rise and fall time is less than 20 ns.
We use a 7 140 type Oriel spectrometer fitted with
a 600 lines/mm grating (Blaze wavelength of 1 Um), a
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01980004107070700
708
Si cathode photomultiplicator cooled by Peltier
effect. The T.R.S. and lifetime measurements are
performed with a P.A.R. 162 type box-car integrator.
The luminescence observed appears like a broad band (40 meV wide) peaked at 1.59 eV at liquid
helium temperature. Experimental methods used for this study are time-resolved spectroscopy (T.R.S.)
and luminescence decay..
2. Theoretical notion about donors-acceptors.
-F. Williams has already studied and developed this
model in the field of semi-conductors and insula- tors [9]. D. G. Thomas and J. J. Hopfield [10] observe donor-acceptor pairs luminescence in GaP and deve-
lop a theory on their kinetic of recombination.
The energy of the photon emitted in a model of the
donor-acceptor-pair type is given by the relation :
Eg is the width of the forbidden band, EA and ED
the acceptor and donor level depths, e2 j Kg r is the
Coulombic interaction energy. The function f(r) is
a correction to the Coulombic interaction ; since it only comes into effect for very close-set pairs, we
may overlook it. From formula (1) we see that as the
distance between pairs, r, decreases, the energy of the photon emitted increases. As the donor and acceptor ought to be situated at precise points of
the lattice, only certain discrete values of r are allowed and for the smallest of these a sharp line spectrum should appear; for greater values a broad band is to be seen. The transition probability W(r) between
the donor and the acceptor, in units of time is for- mulated proportionally to the square of the matrix element which brings into effect the overlap integral
of the hydrogenous envelopes FD(r) and FA(r) of the
donors and acceptors :
As the interpair distance increases, the transition probability ought to decrease. Moreover, when we
assume that the electron is loosely bound to the donor, the latter is totally represented by a quasi- hydrogen model, the radiative transition probability W(r) between associated donor-acceptor pairs dis-
tant by r is given by :
where W MAX is a constant and RB the Bohr-half-radius of the shallowest state.
This equation is correct if we restrict ourselves to the effective mass approximation. However, it may be applied to deeper levels with a fictitious RB where
this approximation is no longer exact.
From the above, this kind of model ought to show
a spectrum of lines in the high energy field. Yet no
observation of such structures has ever been reported
in direct-gap semi-conductors. In this case the emis- sion takes the form of a broad band composed of
non-resolved lines stemming from different pairs at
different distances.
Time resolved spectroscopy enables us to observe the spectrum at different times after the end of exci- tation. In the donor-acceptor case we observe a shift
of the emission band towards low energies when t
increases (t-shift). Indeed, most of the close-set pairs,
that is to say those in which the transition probability
is maximal, no longer participate in this lumines-
cence. The t-shift is the first criterium used in the
donor-acceptor transition. The second criterium used is that of spectrum variation according to excitation intensity ( j-shift). This is expressed by a shift of the spectrum towards high energies as the excitation
intensity increases.
The study of light decay at different wavelengths
shows a t -" type law. The rate of decay becomes
more rapid when the energy in the band increases.
3. Experimental results.
-3. 1 TIME RESOLVED, SPECTROSCOPY.
-Figure 1 represents the lumines-
cence spectrum at t
=0. Two emission bands are
observed : one peaked at 1.63 eV, the other at 1.59 eV.
We choose to study the band peaked at 1.59 eV.
At 1.63 eV, the emission has a short lifetime and
disappears rapidly in the T.R.S. spectra.
Fig. 1.
-Emission spectrum at t = 0 (liquid helium temperature).
Figure 2 represents the time resolved spectra at liquid helium temperature. We observe that the spectrum shifts towards lower energies as the time
after the excitation increases. There is a distortion of the spectrum on the high energy side and a slight
decrease of the half-width. The total shift of the maximum position is about 30 meV (between the
time zero and 100 Us).
Fig. 2.
-Time resolved emission spectra :
a --> t =0 ;
b -+ t = 800 ns; c -+ t = 2 J1S; d -+ t = 8 J1S; e -+ t = 100 J1S.
3.2 LUMINESCENCE DECAY IN FUNCTION OF THE POSITION IN ENERGY.
-We observe the luminescence
decay at different energies inside the emission band.
Except at the early time these curves obey a law of
the form I
=t - ", n values are compiled in table I.
Table I.
-Variation versus E deduced from I= t -
law.
3.3 DISCUSSION. - The t-shift can be considered
as the characteristic criterium of the pair recombina-
tion mechanism.
The total variation (30 meV) of the band position
after the end of the excitation can be attributed in the first approximation to the difference between the coulombic energy (e2/Ks r) for the close-set pairs
and the pairs furthest apart.
In the case of CuGaSe2, the lattice constant is
equal to a
=4.614 A. The static dielectric constant is A% = 9. We obtain a limit of the coulombic energy for the close-set pairs equal to 300 meV.
The observed variation indicates a radius of 40 À
for the close-set pairs.
For such a value of the intra-pairs distance, f(r)
can be neglected in equation (1) [11].
For the study of the emission peaked at 1.59 eV
we shall use the formula :
We did not notice any shift as a function of exci- tation. Actually, the excitation range was not suffi- cient to show any shift.
In conclusion we favour the existence of a donor- acceptor model. The study of the luminescence decay
enables us to confirm this interpretation : the lifetime increases as the energy decreases.
In the first part, we show the total behaviour of
luminescence. In the second part, we confirm our
model, by measurement of the fundamental values
(level depth, Bohr radius, maximum transition pro-
bability).
4. Determnnation of fondamental parameters and application to the stady of total decay of hnminescence and sepctra shapes. 4 , 1 DETERMINATION OF THE
and spectra shapes. - 4. 1 DETERMINATION OF THE
MAXIMUM TRANSITION PROBABILITY.
-4 .1.1 Theory.
-1n the non-compensated case, the basic hypothesis
is that depopulation of the centres is homogeneous
and the time decay of the zero-phonon line may be
written in the form :
where K is a constant and Q(t) > is the average
decay of the neutral donors (acceptors) and Ec the
coulombic energy.
In the compensated case, formula (3) is still avai- lable in the hypothesis that donor-acceptor tunneling
is always slower than the redistribution of the carriers
on the shallowest centres. Therefore the population
of these centres remains homogeneous as in the previous case.
If we derive (3) we obtain the Coulomb energy variation law for the emission peak maximum at a given time t [12].
From this we deduoe :
4.1.2 Experimental study.
-For the pairs fur-
thest apart, the coulombic energy is approximately
zero. The emission is peaked at :
We determine this value on the long time T.R.S.
spectrum in the low energy wing. We calculate
Table II. - T.R.S. results.
710
Table II gives T.R.S. results and enables us to calculate Ee MAX for the emission peak maximum.
Formula (4) can be written in the form :
with x
=EcMAX’
By fitting our results to the last formula we find :
In the hydrogeneous approximation the shallowest level depth would then be : 73 meV.
The results of this fitting are given in table III.
Table III.
-Peak emission energies versus time (two parameters fitting).
Without estimating hl’oo as previously, we now fit
our results with the three parameters RB, hi,. and WMAX.
Formula (4) becomes :
where X
=ho (peak emission energy).
By fitting to this formula we find :
The new values of hi,
=f(t) calculated with these parameters are compiled in table IV.
We deduce for the shallowest level depth : 71 meV
and the value of hl’oo
=1.53 eV which is quite similar
to the estimation we first made from the low energy
wing in the long time T.R.S. spectrum.
So we will now use :
Table IV.
-Peak emission energies versus time (three
parameters fitting).
4. 2 STUDY OF TOTAL LIGHT DECAY. - 4.2 .1
Theory [10].
-Non compensated case (lst case) NA > ND,
Q(t) > is the average probability that the electron is on the donor
where n is the concentration of the preponderant impurity.
The light intensity I(t) is given by :
-
Compensated case ND
=NA.
In this case, it is more difficult to calculate the
decay. We find that Q(t) )comp is given by :
with
So
and we deduce
with the results obtained in the non compensated
case, we can calculate I(T)n.comp x 6(r) >n.comp and
the corresponding t of the compensated case by for-
mula
We calculated formulas (6) and (8) on a Mitra 15 CII calculator.
We show the total decay for different values of the
impurity concentration and for the calculated values
RB and WmAx in the non compensated case (Fig. 3).
Fig. 3.
-Theoretical total-light-decay
curvesfor the values of
WMAX
=108 s-1 and RB
=5.6 A for several concentrations.
The shape of the curves is strongly influenced by
the product nRB. Although the parameter WMAX
affects the position of the curves, it does not affect
their shape.
In our case, as the function Q(t) > remains cons-
tant (Q(t) N 1) over the time range involved, the
curves of the non compensated case are similar with the compensated ones.
, 4.2.2 Experimental results. - Previous calcula- tions are only correct if all the pairs involved in the broad band have been initially saturated. In our
experiments this condition has been fulfilled. Figure 4 represents the luminescence decay in a time range between 10-’ s and 10-4 s.
The comparison between previously calculated
Fig. 4.
-Experimental total-light-decay
curve(points). Theore-
tical total-light-decay
curvefor n = 8
x1017/CM3 (solid).
curves and the total decay curve obtained by experi-
ment shows a good agreement in the case of a dopage
n = 8 x 1017/cm3 .
This value would appear to be compatible with the impurity concentration of that compound.
4.3 STUDY OF SPECTRA EVOLUTION IN FUNCTION OF TIME.
-The shape of the spectrum at different times, in function of energy is given by formula (3).
From this expression, we can see that the concen-
tration only intervenes in Q(t) >.
We calculated formula (3) at different times.
Results are given on figure 5.
wWe observe that in the case of close-set pairs (at early times) theoretical spectra follow the experi-
mental curves on the high energy side.
When time increases, the spectral shift on the high
energy side is characteristic of donor-acceptor beha-
viour. On the low energy side the variation between calculated and experimental curves is important,
even at early times.
For long times, experiment doesn’t fit at all with
theory, the emission seems govemed by vibration
effects. It would be interesting to correct the theore- tical spectra given by equation (3) with the additional
assumption that the pairs radiate with a spectral shape given by the long time curve (figure 2 spec- trum e) and to compare them with the experimental
curves [10].
712
Fig. 5.
-Theoretical
curvefor the spectra assuming that
apair
of separation
rradiates only at the photon energy given by eq. (1) (solid).
a -->t
=800 ns ; b - t
=2 uS;
c -t
=8 ps; d --> t
=20 ps;
e - t =