ELECTRON TRANSPORT IN HYDROGENATED AMORPHOUS SILICON SCHOTTKY BARRIERS AND DEEP LOCALIZED STATES KINETICS

Transcription

1 ELECTRON TRANSPORT IN HYDROGENATED AMORPHOUS SILICON SCHOTTKY BARRIERS AND DEEP LOCALIZED STATES KINETICS S. Deleonibus, D. Jousse To cite this version: S. Deleonibus, D. Jousse. ELECTRON TRANSPORT IN HYDROGENATED AMORPHOUS SILI- CON SCHOTTKY BARRIERS AND DEEP LOCALIZED STATES KINETICS. Journal de Physique Colloques, 98, 42 (C4), pp.c4-487-c <0.05/jphyscol:98403>. <jpa > HAL Id: jpa Submitted on Jan 98 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published 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.

2 CoZZoque C4, supptdment au nolo, Tome 42, octobre 98 ELECTRON TRANSPORT IN HYDROGENATED AMORPHOUS SILICON * S. Deleonibus and D. Jousse (b) Laboratoire de Physique des Composants d Semiconducteurs, ERA CNRS 659, ENSER, 23 rue des Martyrs, 3803 GrenobZe, France * G qo de Energia Institute de Fisica, Universidade EstaduaZ de Ccmzpinas, Campinas, Brazi Abstract : We present a method for analysing transport in Schottky,diodes on hydrogenated amorphous silicon (asih). Experimental results show a correlation between the reduction of carrier col lection velocity, the reduction of barrier height for reduced space-charge regions. In the same way, the kinetics of "midgap" localized states is slowered as bulk states located near the Pt-aSiH contact contribute to capacitance. Introduction. - It is often assumed that transport in bchottky diodes on amorphous materials 3s limited by a diffusion process in the space-charge region. The mean reason put foward is an account for the low carrier mobility in these types of materials. Nevertheless, the thermionic theory has often been supposed to hold to measure the density of states (D.O.S.) in the gap of hydrogenated amorphous silicon (asih) by the C(V) technique [l],c2, C3. In order to avoid the doubt about transport, several authors C4, i5, C6 carry out capacitance measurements at zero bias versus frequency and (or) temperature. In this paper, we present a method of analysing results of current-v0 tage I(V), zero bias low frequency capacitance C(w) characteristics versus temperature which must lead to a better understanding of the actual nature of extended state electron transport through Schottky barriers on asih. We put in evidence the influence of the extension of the space-charge region on the collection velocity of carriers at the metal-asih interface. An analogous treatment is used for the analysis of the kinetics of localized states around the Fermi level with experimentally determined parameters derived from C(w) measurements. The samples investigated were obtained by glow discharge decomposition of silane at three substrate temperatures : 350, 400 and 450 C. The back contact is a 0. ~ ~ thick n' layer (PH3/SiH,,=5%) deposited on a tantal um-alumina substrate. The Schottky contact is obtained by sputtering a 50R-00R thick Pt layer on top of the intrinsic (~lpm thick) asih. The diodes achieved on the 400 C material exhibit more than 2% power efficiency C5. Electrical characterization. - Current-voltage I(V) characteristics are performed in the range of temperature C-9O0C, +8O0C; tially (fig.2) for forward bias lower than the built-in potential Vso (fig.) following the relation :. C5, C7. They first behave exponen- (a) Work supported by COMES (b) On leave from Laboratoire de Physique des Composants a Semiconducteurs ERA CYRS 659. * Samples achieved by 6. BOURDON, CRCGE ilarcoussis, France Article published online by EDP Sciences and available at

3 I(V) = where 0 = 00 exp(- '@B) (2) ideal ity factor of thfdiode. *) - J () i give the barrier height +B (fig-l) I ), B is the b4-l O w t Fig. : Band diagram ofschottky diode on asih Fdiodes on materials p-repared at : a)350 C, b) 400 C, c) 450 C At higher voltage, we obtain a linear dependence of current with voltage. The slope of the characteristics allows us to determine a mean value of the bulk conductance GB of the material. Its activation energy will give the mean electron quasi-fermi level distance (Ec-EF) from the conduction band (Fig.). Thus we have : S GB - UB with ag = a. exp [-!C$] where ag is the mean bulk conductivity, S is the surface of the Pt contact, t the thickness of the intrinsic asih layer, a, is given by : GO = q Ncii (3) Nc being the effective density of states at Ec. I.I the electron nobility, q the electron charge. In the same range of temperature we perform low-frequency C(w) measurements ClHz, lohzl (see fig.3) for Hz capacitance). For temperatures higher than Tresp the capacitance starts to rise : the states around the Fermi level at the neutral bulk boundary-w (fig. l) begin to respond to the modulation. In the same way, when the capacitance saturates at a temperature Tsat the bulk states around EF near the Pt-aSiti interface contribute to capacitance (neglecting the response of surface states). The saturation value of low frequency capacitance (typically Hz fig.3) Csat allows us to deduce the D.O.S. around the midgap in the case of a uniform distribution N : N = - (Table I). Thus, the equilibrium value of the space-charge width will be' : W = L g ( ) (Table I) The electrical potential V(x) having an exponentialshape : V(x) = Vs0 exp - $ 'U LD being the Debye length associated to the uniform 3.O.S. around the midgap : Lo, =(&l ~abie L Results (electrical characterization) E -E a D W " ( 0 ~ )'9 (ev;"(e~) C ~(ev)"~~ (V) 6 0 [~cm-2) (n-lcm-': (~m-~ev-l)(pm) ~0'. 3 ~ 0 ~ ~ 5 ~ ' ~ Y.3x05 3x02 xol6 0.67, P x04.9x02 2.5~0" 0.5 Electron collection velocity. - In a general way, the dark free electron current I,,mframr to the metal can be expessed by : -,. I~-+,, = tic us exp(*) exp N

4 where s is the mean collection velocity of carriers at the interface. If the current is diffusion limited then US is the drift velocity vds at the interface : us =!JFS = vds where Fs is the field at the interface given by : F = (5) (Table ). D If the thermionic theory h0 ds, then us is the metal collection velocity and will have the thermal velocity vth as a maxinun value C8. Thus, we will have according to relation (2) : 00 = q NCuS. (5). From relations (3) and (5) we deduce : IOQ -.-A US Thus from relation (4) : 00 u 00. m=&. The ratio us/vds can then be evaluated from experimentally derived parameters (Table ). Dynamic behaviour of midgap states. - When the states located at W start contributin~ to capacitance their response time ~ r is ~ given s by ~ S.R. statistics : E -EF Tresp = to^ exp k+. Where TOF =. (6). When the capacibsnee saturates we expectfa;fh;%e:l::~~s~ocated at a distance ~ from the conduction band to contribute plainly to capacitance C5 with a response time : Tsat = TGB exp.-.- ~ Q B where ~Tsat '@B.= *@B Vth2 N EC S (7). In (6) and (7) : OF an6 are the electronmc capture cr~ss)::cttons of the states around the Fermi level respectively at W and at the interface, vthl and vth2 are the thermal velocities at Tresp and Tsat,!{(Ec) is the density of states at Ec. We thus consider that the bulk states at the don't necessarilly obey the same kinetics as the states at (Ec-EF) even if the energy difference qvs0 is small as it is the case for 350 C and 450 C samples (Table I). For a ~iven worki?~ frequency vw must have a response temperature and a saturation tenperature and : v ' = Tresp = 'sat (see fig.3 for Hz capacitance). In practice, we observe a l inear 6ependence of Log v as a function of (Tres )-l (fis.4). In figure 4 the activation energy given experimentally is (E~-EF) : trat shows a posteriori that the bulk Fermi level determined from the linear part of I(V) characteristics is correct. From the pre-exponential factor of the curves in fig.4 we can determine QF (Table ). For a given material, if the response kinetics of the states around EF were the same at W and at the interface then the quantity AT = (Ts - Tresp) should be proportional to Vs0 accor"ing to S.R.H statistics. For sm?f values of VSo, AT is too large (Table ). For the sane samples, the rea at difference between TQF and TQ@B shows that the kinetics is slowered as the bulk states contribute to capacitance in materials with a thin space-charge region. In the sane way, the ratio us/vds gets smaller (Table ) and the barrier decreases as W decreases. Possible interpretation. - We are in right to imagine that the static and dynamic b e h a v i o u r y diode have the same origin. The possibility of carrier collection by intermediate trapping in localized band-tail states in the spacecharge region must not be excluded. Thus, the ratio us/vds decreases as W decreases.

5 V( l!:) loo o IOO T(oE) : C versus T for materials defined in fig.2 at Hz. Figure 4 : Log v versus (Tresp) -l Classically, we would have expected the contrary if transport were diffusion limited. As a matter of fact, the effect is nore pronounced as the band-bending at the Pt-aSiH contact is more important : (case of 350 and 450 C materials) : us is smaller than vds by a factor of 0'. The dynamic behaviour follows the static behaviour. (Table ). The communication between deep states at interface and the conduction band is probably established via intermediate trapping of the electrons in band-tail states. Thus ro$a can be higher than r o by ~ a factor of 02. Conclusion. - We have presented a method of analysis of I(V), C(w) characteristics versus temperature that can lead to a better understanding of transport in Pt-aSiH Schottky diodes as well as the kinetics of deep localized gap states in asih. We have proposed a possible interpretation for the reduction of collection velocity of extended states electrons at the metal-asih Schottky contact. This phenomena is accompanied by an enhancement of thermal release time of the deep bulk states at the surface Fermi level and a reduction of the barrier height. That reflects also the indirect influence of the D.O.S. around the midgap on the entended states electron transport. References C A.J. Snell, K.D. Mackenzie, P.G. Le Conber and W.E. Spear J. don Cryst. Solids 35 & 36 (980). C2 P. Viktorovitch, D. Jousse, P,. Chenevas-Paule and L. Vieux-Rochas Rev. Phys. Appl.- 4 (579). C3 P. Viktorovitch and D. Jousse J. igon Cryst. Solids, 35 S 35 (980). C4 J. Seichler, W. Fuhs, H. Flell and H.M. Kelsch J. ilon Cryst. Solids, (980). CS R. Basset, D. Jousse, S. Deleonibus, F. Ducatel and 6. Eourdon Proceedings of 3rd E.C. Photovoltaic Solar Energy Conference, Cannes (980). C6 P. Viktorovitch, G. Moddel, Appl. Phys. 3 (9) (Sept. 980). C7 S. Deleonibus, D. Jousse, R. Basset, Presented at ESSDERC'80 York (Ga) C8 S.M. Sze, Physics of Semiconductor devices, Wiley,!.dew-York (969).