HAL Id: jpa-00220741
https://hal.archives-ouvertes.fr/jpa-00220741
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.
PRIMARY AND SECONDARY PHOTOCURRENTS IN n-TYPE AND p-TYPE a-Si:H FILMS
H. Welsch, W. Fuhs, K. Greeb, H. Mell
To cite this version:
H. Welsch, W. Fuhs, K. Greeb, H. Mell. PRIMARY AND SECONDARY PHOTOCURRENTS IN
n-TYPE AND p-TYPE a-Si:H FILMS. Journal de Physique Colloques, 1981, 42 (C4), pp.C4-567-C4-
570. �10.1051/jphyscol:19814123�. �jpa-00220741�
CoZZoque C4, supple'ment au nOIO, Tome 42, octobre 1981
PRIMARY AND SECONDARY P H O T O C U R R E N T S I N n - T Y P E AND p - T Y P E a - S i : H FILMS
H.M. Welsch, W. Fuhs, K.H. Greeb and H. Me11 Fachbereich Physik, U n i v e r s i t a t Marbzrrg, F. R. G.
Abstract.- The spectral dependence of the primary (reverse) and secondary (forward) photocurrents, Jr and Jf, in Schottky barrier diodes made from glow discharge deposited a-Si:H is studied in the energy range 0.8-3.1 eV using a two beam experimental set-up. Below 1.5 eV significant differences are observed between the spectra of Jf and Jr and also between diodes made from n-type and p-type material. They are interpreted as arising from differences between the generation rates of mobile electrons and mobile holes caused by the asymmetry of the density of gap states.
Introduction.- Photoconductivity spectra are commonly used to obtain information on sub-bandgap absorption in thin films of a-Si, where direct optical transmission ex- periments fail. Many of these spectra exhibit a shoulder between 1.2 and 1.4 eV which obviously depends on the history of the sample112. The origin of this struc- ture is still a matter of debate. It is often interpreted as arising from optical transitions into or from localized gap states and thus associated with an absorption constant a in the range 10-100 cm-'. On the other hand most of these measurements use secondary photocurrents3 and it is not clear whether under the experimental con- ditions the photoconductive gain and hence the UT-product is independent of energy.
Recently crandal14 suggested, that the reverse current Jr in a Schottky barrier diode is a more accurate measure of a, since Jr is a prlmary photocurrent3. On the other hand, one has to consider that, for sub-bandgap exitation, the generation rates of mobile majority and minority carriers might be largely different and that Jr then is determined by the smaller one of these rates which would underestimate the total optical absorption. In order to clarify this problem we have studied the spectral dependences of the forward and reverse bias currents of a-Si:H Schottky diodes prepared by using both n- and p-type material.
Experimental details.- The a-Si:H films were prepared by the decomposition of si- lane in a capacitively coupled glow discharge system using the following paramet-ers:
substrate temperature 280°c, pressure during deposition % 0.5 mbar, flow rate 5 sccm, rf-power h 2 W. The Schottky diodes had the familiar structure. For n-type material: stainless steel/nf-a-~i:H/undoped a-Si:H/high work function metal; for p-type material: stainless steel/p+-a-si:~/slightly doped a-Si:H (up to 300 vppm B2H6)/low work function metal. The Schottky contacts were prepared by evaporation of the various metals in a cryo-pumped vacuum system.
The spectral dependences of the photocurrents in the diodes were measured in the energy range 0.8-3.1 eV using a two beam experimental set-up: A uniformly ab- sorbed continuous bias illumination and chopped monochromatic light of low inten- sity. The resulting ac-photocurrent was detected by lock-in-technique. It is impor- tant to note that there were no significant phase shifts, indicating that the re- sponse time of the photocurrent was smaller than the reciprocal chopper frequency
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19814123
JOURNAL DE PHYSIQUE
f
n-type(12.5 Hz). In our samples we only found enhancement of the photoconduc- tivity. No quenching effects were ob- served as recently has been reported by pearsons5. The forward currents were measured with a bias voltage of
1 V and the reverse currents with - 3 V in the case of n-type and
-
1 V on p-type material.Results and Discussion.- The photo- currents with reverse and forward bias, Jr, and Jf
,
,, of the n-type Schottky diodes show the behaviour typical for a Schottky barrier (Fig.l).Jrrn saturates, when at high pho- ton energy the light is absorbed pre- dominantly within the collection width. At low photon energies (below
1.7 eV) we observe, like other au- thors6, a pronounced influence of the contact material, which is attributed to internal photoemission. This be-
hv (eV1 comes evident from a Fowler plot
which gives straight lines and extra- polates to barrier heights consistent Fig. 1: Spectral dependence of the forward with other measurements, namely
and reverse bias currents, Jf,, and 1.05 eV for Pt, 0.95 eV for Ni and Jr,n. Dashed line: variation of Jrvn 0.88 eV for Ag-electrodes. Photoemis- expected after correcting for photo- sion thus has to be taken into ac- emission. Dotted line: Absorption count in case of the reverse blas constant u from optical transmission. photocurrents in the energy region be-
low 1.7 eV. Subtracting this contri- bution one finds the steeper edge of Jrtn indicated by the dashed line in Fig. 1.
Whereas between 1.5 and 2 eV Jf,n and Jrln are almost parallel, they deviate in the high energy region because Jf,, is sensitive to the inhomogeneous illumination. Be- low 1.5 eV Jf,, exhibits a pronounced shoulder. The height of this shoulder depends on preparation: it is enlarged by doping, electron or ion bombardment. Such a shoul- der is often reported in the literature and has been explained by quite different assumptions. The important question is whether this structure arises from absorption and therefore reflects a structure in
the densitv of states or from a varia- tion of the photoconductive gain and
hence of the VT-product with photon en-
N (E)
ergy. Our measurements were carried out using modulated light and an intensive continuous bias illumination. With this procedure the quasi Fermi levels in the specimen are fixed. Thus the gain should be kept constant. The spectrum of Jf,n then reflects the variation of the gene- ration rate of majority carriers (elec-
trons). Jrfn, on the other hand, deter- I
I I
mined by the smaller one of the gene-
I I
ration rates of minority and majority I I
carriers, does not exhibit a shoulder. 1
Ev Ec
To explain this difference we
refer to the field effect density of Fig. 2: Field effect density of states.
states7 sketched in Fig. 2. At high Arrows indicate transitions photon energies absorption leads to which create mobile carriers.
both mobile electrons and mobile holes
the transitions g, and gh. The asymmetry of the density of gap states suggests a smaller generation rate for holes than for electrons, i.e. gh << ge. In case of the reverse current both types of mobile carriers are needed to establish a stationary photocurrent. Jr,, is therefore limited by the generation rate gh of mobile minority carriers. Thus taking Jrnn as a measure for a would neglect the processes leading to free electrons and, therefore, would underestimate the optical absorption. The for- ward current, on the other hand, is determined by the mobile majority carriers only, i.e.-by ge, the stronger one of the generation rates. We argue, therefore, that Jfrn rather than Jr should be taken as a measure for the absorption constant a. Fittlng a (hv)
,
obtaineh from transmission measurements (dotted curve in Fig. 1 ),
to Jf,, we find a2
10 cm-' at the photoconductivity shoulder near 1.4 eV.In order to check this interpretation we have performed the same measurements on Schottky Barriers prepared by using boron doped a-Si:H films. These Schottky di- odes have surprisingly good properties if low work function metals like Yb or Sm are used. The barrier height obtained with Yb on a-Si:H films moderately doped with bo- ron amounted to l eV. According to the I-V characteristic and to the photoresponse the majority carriers are holes. In Figs. 3 and 4 the spectral dependences of p- type and n-type barriers are shown in comparison. It is obvious, that in the Jf
spectra, whith measure the generation rates of the corresponding majority carriers, the low energy shoulder is considerably lower in case of the boron doped films (Fig.
3a). In view of the asymmetry in the field effect density of states this trend is to be expected. The Jr spectra (Fig. 3b) on the other hand, show a more similar be- haviour. Saturation in the high energy range is expected to occur when absorption occurs predominantly inside the collection width. According to these curves the col- lection width is practically the same in n- and p-type barriers. Indeed a more de- tailed analysis yields for the collection width a value of 0.27 'km at - 1 V in the p-type diodes. No major differences are to be expected in the low energy range he- cause both currents are determined by the smaller one of the generation rates ge and gh which, by the above arguments, is gh. Indeed, if one accounts for the photoemis- sion contribution in the n-type diodes, the curves are practically pacallel in the whole energy range investigated. It is worth noting, that in case of p-type material
Fig. 3: Spectral dependences of the forward and reverse-bias currents of n-type (Jf,n, Jr,,) and p-type (Jf,p, JrPp) Schottky diodes. Details see text
JOURNAL DE PHYSIQUE
Fig. 4: Spectral dependences of the forward and reverse bias currents of n-type (Jf,n, JrRn) and p-type (Jflp, Jrtp) Schottky diodes. Details see text.
the contributions of internal photoemission are considerably smaller than in the n- type diodes. The reason for this is not yet clear. A parallel behaviour is also ex- pected for Jf and Jrrn because both mpasure the hole generation rate gh. Fig. 4a shows that this is the case after subtraction of the photoemission contribution to !p Jrrn (dotted line). In Fig. 4b finally Jf,, and Jrrn are compared. Jf,n arises from processes generating mobile electrons, g,, and thus has a shoulder at low energy.
Jr,p 1s determined by the smaller one of the generation rates g, an6 gh which, due to the asymmetry of the density of gap states, is gh at photon energies below 1.4 eV.
After all, neither Jf nor Jr is a correct measure for the absorption constant a. In any case only absorption processes are recorded which lead to mobile carriers which in case of Jf are the majority carriers. Jr being carried by equal numbers of electrons and holes is limited by the smaller one of the generation rates ge and gh.
Therefore, Jr appreciably underestimates the optical absorption in the low energy region. We believe that the most accurate measure of a is the forward current Jf in n-type, i.e. undoped and P-doped, diodes. The photocurrents in p-type diodes yield values for the absorption constant which are too low, since absorption processes which generate electrons are ignored.
Acknowledgement.- Financial support by the Bundesminister fiir Forschung und Techno- Sogie (BMFT) is gratefully acknowledged.
ReEerences
1. LOVELAND R.J., SPEAR W.E., AL-SHARBATY A., J.Non-Cryst.Solids
13
(1973/74) 55 2. ANDERSON D.A., MODDEL G . , PAUL W., J.NOn-CrySt-Solids 35 & 36 (1980) 345 3. BUBE R.H., Photoconductivity of Solids, Wiley & Sons, New York 1967 4. CRANDALL R.S., Phys.Rev.44
(1980) 7495. PEARSONS P.D., Solid State Comm.
36
(1980) 8516. WRONSKI C.R., ABELES B., CODY G.D., TIEDJE T., Appl.Phys.Lett.
2
(1980) 96 7. MADAN A., LeCOMBER P.G., Proc. 7th Intern. Conf. on Amorphous and Liquid Semi-conductors, Edinburgh 1977, 377