Diode Devices Based on Amorphous Selenium Films

phys. stat. sol. (a) 159, 569 (1997)

Subject classification: 73.40.Sx; 73.61.Jc; S5

Diode Devices Based on Amorphous Selenium Films

S. Touihri, G. Safoula, and J. C. Bern�ede

�Equipe Couches Minces et Mat�eriaux nouveaux, Groupe de Physique des Solidespour l' �Electronique, Facult�e des Sciences et des Techniques, 2 rue de la Houssini�ere,BP 92208, F-44322 Nantes C�edex 3, France

(Received July 26, 1996; in revised form October 2, 1996)

Different M1/a-Se/M2 sandwich structures have been studied with M1, M2 � Al, Au, Cr, SnO2.Rectifying contacts have only been obtained when an aluminium underlayer was used. It is shownthat the aluminium/aluminium oxide layer should be in contact with the amorphous selenium (a-Se)film, since there is no rectifying effect when an insulating poly(N-vinyl carbazole) films is intro-duced between the aluminium and the selenium. A forming process, either electrical or thermal, isneeded to obtain the rectifying effect. a-Se/Al samples before and after thermal forming process, i.e.annealing at 325 K for 1 h, have been studied by XPS depth profile. It is shown that a complexaluminium oxide thin films is present at the interface. There is not only alumina but also alumi-nium hydroxyl in this layer. Moreover it is shown that after formation there is interdiffusion be-tween Se and Al. At the interface, the selenium plot follows that of the oxygen present at the inter-face. The OH radicals and/or some oxygen present at the interface can passivate dangling bondspresent in the amorphous selenium. Therefore it is assumed that the interaction between the sele-nium and the aluminium oxides modify the properties of the Al/a-Se contact in such a way thatrectifying contacts are obtained.

1. Introduction

Amorphous selenium films find applications in various electrophotographic devices suchas xerography and xeroradiography [1, 2]. Selenium electrophotographic devices consistof an aluminium sheet which is coated by an amorphous selenium layer.

As such, these thin films have been extensively investigated [3] with regard to theirvarious electrical, optical, thermal and mechanical properties. However, although someinvestigations about the influence of the thin Al2O3 layer present at the interface on thedevice performances have been done [1, 4], a systematic investigation of the dependenceof their electrical properties is still lacking.

The present paper is an effort in this direction in which the role of the interfaciallayer in electrical characteristics is investigated. Electrical and X-ray photoemission meas-urements were made. It is shown that the structure of the aluminium oxide layer iscomplex and that it induces Al/a-Se rectifying contact.

2. Experimental

In order to prevent any crystallization effect during the measurements, selenium powderdoped with 3 at% of arsenic has been used. We have already shown that the crystalliza-tion temperature of As-doped Se layers increases with the arsenic concentration [5]. Sele-nium films deposited from a 3 at% arsenic-doped powder on a substrate heated atTs � 370 K are still amorphous.

S. Touihri et al.: Diode Devices Based on Amorphous Selenium Films 569

Films of thickness �6 mm of 3 at% arsenic-doped selenium (purity 99.999%) were de-posited at room temperature on metal-coated glass substrates by vacuum evaporationof doped powder at a gas pressure �10ÿ4 Pa.

The initial powder was prepared by putting weighed arsenic and selenium powders(3 at% As) in an evacuated silica ampoule, heating them at �1150 K for 24 h, coolingthem to room temperature and finally grinding them.

The substrates were cleaned and then degassed in situ by heating at 400 K for 1 hprior to the deposition of the film.

The metal thin films, with thickness ranging from 100 to 300 nm, were thermally evap-orated onto or below the selenium films. During deposition the substrate holder waswater cooled. The thickness and the evaporation rate of the films were measured in situby the vibrating quartz method. The evaporation rate of the selenium was about2 nm sÿ1. The thickness of the layers was controlled by a stylus. The structural proper-ties of the films were checked by XRD diffraction, by using an analytical X-ray system,type DIFFRACT AT V3.1 Siemens instrument, which uses a graphics program EVA.

X-ray photoelectron spectroscopy was carried out in a commercial photoelectron spectrom-eter equipped with a twin anode X-ray gun (MgKa and AlKa lines at 1253.6 and 1486.6 eV,respectively).1� Data acquisition and treatment were carried out with a computer.

High-resolution scans with a good signal to noise ratio were obtained with the magne-sium source operating at 10 kV and 10 mA at the selected energy of 50 eV. The quanti-tative studies were based in the determination of the Se3d, As3d, Al2p, Au4f7=2, O1sand C1s peak areas with 0.57, 0.39, 0.12, 2.2, 0.61 and 0.2, respectively, as sensitivityfactors (the sensitivity factors of the spectrometer are given by the manufacturer). Thesubstrates were grounded with silver paste to avoid charge effects. The change in compo-sition with depth through the interface was traced by recording successive XPS spectraobtained after argon ion etching for short periods. As estimated from standard seleniumthin films, the rate of sample erosion was approximatively 10 nm minÿ1. Sputtering wasperformed at pressures of less than 5� 10ÿ4 Pa, a 10 mA emission current and a 5 kVbeam energy using an ion gun. Before sputtering the pressure was below 5� 10ÿ7 Pa.

The decomposition of XPS peaks into different components was made after subtrac-tion of the background. The developed curve-fitting programs permit the variation ofparameters such as Gaussian/Lorentzian ratio, the FWHM, the position and the inten-sity of the contributions. The parameters were optimized by the curve fitting programsin order to obtain the best fit.

The dark current±voltage (I±V ) measurements were carried out by the static methodusing a programmable Keithley electrometer 617 as the current meter.

570 S. Touihri, G. Safoula, and J. C. Bern�ede

1� With a Leybold spectrometer at University of Nantes, CNRS.

Fig. 1. Drawing of the M1/Se/M2 sandwich

The geometry of samples is drawn in Fig. 1. The guard ring was grounded to preventsurface leakage. The devices studied were: glass/Al/a-Se/Au, glass/Au/a-Se/Au, glass/Al/a-Se/Al, glass/SnO2/a-Se/Au, glass/Al/PVK/a-Se/Au, glass/Cr/PVK/a-Se/Au. Asample called M1/a-Se/M2 means that the M1 layer is the metal glass coating. In thecase of SnO2-coated glass the substrates have been provided by ªSolemsº.

3. Experimental Results

The crystalline structure of the sample has been checked before any other measurements.The XRD spectra obtained have shown that the selenium films were systematically amor-phous. The composition has also been checked by XPS analysis it is shown that the compo-sition of the deposited layer is nearly the same as that of the initial powder.

3.1 I±V characteristics

The I±V characteristics obtained with the samples Al/a-Se/Au are reported in Fig. 2. Itcan be seen that there is a ªformingº process of the samples. While the I±V characteris-tics of the first cycle is poorly rectifying, they exhibit a rectifying ratio of about 106 at1 V after the third cycle. The forward polarization is obtained when the upper goldelectrode is positively biased. The same result is immediately obtained if before any I±Vmeasurement the sample is annealed for 1 h at 320 K. The rectifying effect is also ob-tained with Al/Se/Al samples. Usually the vacuum was broken between Al and Se de-positions, however, even when the vacuum is not broken, the Al/Se/Au samples after aforming process exhibit the same rectifying behaviour. As reported in Fig. 3, curve b,

Diode Devices Based on Amorphous Selenium Films 571

Fig. 2. I±V characteristics of a glass/Al/a-Se/Au sample: (a) first, (b) second, (c) third measure-ment

ohmic contacts are obtained with the Au/Se/Au samples. It can be seen in Fig. 4 that,even if there is a small formation process, the sample Al/PVK/Se/Au does not exhibitthe diode characteristics of the Al/a-Se/Au sample (Fig. 3, curve a). In the case ofSnO2/Se/Au samples, the characteristics are reported in Fig. 3, curve c. It can be seenthat here also there is no rectifying effect even if there is some asymmetry of the I±Vcharacteristics.

3.2 XPS depth profile

The XPS depth profiles of Al/Se samples, before and after annealing in room atmo-sphere at T � 325 K, are reported in Fig. 5. In that case selenium layers, 40 nm thick,have been deposited on aluminium. It can be seen that there is some increase of theinterdiffusion between Se and Al after annealing. Before annealing, the 40 nm of Se arenearly totally etched after five minutes of argon sputtering, this is in good agreementwith the etching speed of Se which has been measured on a reference film deposited onglass substrate. It can be concluded that there is not a strong interdiffusion between Aland Se. After the mild annealing (325 K, 1 h), the profile is strongly modified. The Alpeaks are visible at the surface of the samples, which means that an interdiffusion occursduring the annealing. Some aluminium appears at the surface, while selenium is detectedin the depth of the aluminium film. In fact, it can be seen that the selenium plot follows


Fig. 3. I±V characteristics of (a) glass/Al/a-Se/Au sample, (b) glass/Au/a-Se/Au sample,(c) glass/SnO2/a-Se/Au sample


Fig. 4. I±V characteristics of a glass Al/PVK/a-Se/Au: (a) first, (b) second, (c) third measure-ment

Fig. 5. XPS depth profiles of a sample glass/Al/a-Se (thickness a-Se� 40 nm) before annealing andafter formation by annealing �T � 325 K, t � 1 h)

39 physica (a) 159/2

that of oxygen which means that probably the interdiffusion takes place in the alumi-nium oxide layer.

In order to discuss the different binding energies the C1s peak (284.8 eV) has beentaken as reference.

The binding energy does not vary strongly with the etching time. When the Al2p lineappears it is split in two parts. One is situated at 72.7 eV and the other, broader one, issituated at about 75.2 eV, with a broad tail on the high-energy side. When the etchingtime increases, the former peak increases while the latter decreases. These two compo-nents of the Al peak, in a first approximation, may be attributed to metal and to alumi-nium oxide, respectively. However, X-ray photoelectron spectroscopy [6] shows that thechemical shift between Al and Al2O3 is only 2 eV, while 2.5 eV are measured here. More-over after decomposition in two peaks (Fig. 6) the oxide peak is quite large. The FWHMis higher than 2.5 eV, while, with our spectrometer, the FWHM of the Al2p line ofaluminium in Al2O3 should be around 2 eV. Moreover there is a good agreement be-tween the experimental and the expected FWHM of the line attributed to metallic Al(1.25 eV). Therefore, since the peak exhibits a broad tail on the high-energy side, abetter fit between experimental and theoretical curves is obtained when the aluminiumpeak is decomposed in three components (Fig. 7). A possible explanation for the highbinding energy state is the existence of hydroxyl groups [7]. This hypothesis is also inline with the observation of the binding energy of the O1s peak, 532.6 eV. This value isnearer to the binding energy of Al(OH)3 than that of Al2O3 [6].

Therefore, at the interface, the Al peak should be attributed to three components: Al,Al2O3, Al2(OH)3. After annealing the selenium is present all over this transition layer,


Fig. 6. XPS spectrum of Al2p, after 2 min etching, decomposed in two components: ÐÐÐ experi-mental curve, ±± � ±± � ±± theoretical curve, ±± ±± ±± ±± different components


Fig. 7. XPS spectrum of Al2p decomposed in three components after a) 2 and b) 15 min etching:ÐÐÐ experimental curve, ±± � ±± � ±± theoretical curve, ±± ±± ±± ±± different components

39*

however, no reaction between Se and the components of this layer has been shown byXPS since the binding energy of Se does not vary from that expected for the Seelement.

4. Discussion

In all the samples electrically characterized, the selenium layer doped with arsenic wasamorphous. However, diode characteristics have only been obtained with an aluminiumunderlayer. This rectifying behaviour is obtained only after a forming process, the form-ing process being either electrical or thermal.

When another bottom electrode is used the �I±V � characteristics are ohmic or nearlyohmic, even when a PVK layer is inserted between Al and the selenium layer. Thereforethe use of a bottom aluminium electrode covered with its natural thin oxide layer(3 nm) is not sufficient to obtain diode characteristics, the contact Al/aluminium oxide/a-Se, is necessary.

These results can be compared to those obtained with the polycrystalline selenium-based diodes. Polycrystalline selenium diodes have long been of technical importanceand the nature of the rectification process has been studied by many workers [8 to 10].Selenium diodes consist of a roughened aluminium plate which is covered by a layer ofbismuth and then selenium. A cadmium counter electrode is then applied, and the wholeunit is annealed to obtain hexagonal selenium. Good rectification properties require elec-trical and/or thermal ªformingº. Therefore, there are some striking similarities betweenthe results obtained with the amorphous devices of the present work and the classicalpolycrystalline diodes. It appears that (i) the Al/aluminium oxide under the electrode isnecessary to obtain diode characteristics, (ii) a formation process is needed.

It is clear from these conditions that the forming process is related to the Al/alumi-nium oxide/a-Se interface. It is well known that it is difficult to obtain band bending atthe contact metal/amorphous semiconductor, because of the high density of localizedstates near the Fermi level in amorphous semiconductors. This high density of states isstill small compared to metals, but it is enough to make the amorphous side of thedevice behave more like the metallic side of a Schottky diode rather than like a crystal-line semiconductor [11]. Therefore, even if amorphous selenium is p-type it is difficult toobtain a rectifying contact between metal and a-Se. In the present sample the formationof a rectifying contact is related to the aluminium oxides at the interface Al/a-Se. Manyauthors [12 to 14] have shown that the aluminium oxides are inhomogeneous with thepresence of a high thin resistive layer at the interface Al/Al2O3 and/or an ªn-domainº onthe side nearer to aluminium. These domains are related to non-stoichiometric oxides.

In the present work we have shown this inhomogeneity by XPS depth profiles(Fig. 5). These aluminium oxygen compounds can interact during the formation processwith selenium. There is at least interdiffusion at the interface during formation (Fig. 5),probably this interdiffusion modifies the properties of the amorphous material. Underthe effect of heating and/or electrical strength, the OH of the hydroxyl and/or someoxygen of the alumina can passivate dangling bonds of the amorphous selenium layer atthe interface oxide/a-Se. It has been shown [15], in the case of polycrystalline samples,that non-deoxidized selenium is p-type while deoxidized Se is n-type. In the case ofamorphous selenium Kastner et al. [16] have shown that the stable structural defectsissued from dangling bonds (C0) are either positively, C�3 , or negatively, Cÿ1 , charged.


Oda et al. [17] have shown that when an impurity with large electron affinity such asoxygen or hydroxyl is bonded to a selenium chain it becomes negatively charged, be-cause its electronegativity �c � 3:5� is larger than that of selenium �c � 2:4�. The exist-ence of these negative defects [Dÿ] will affect the neutrality conditions of structuraldefects as follows:

�C�3 � � �Cÿ1 � � �Dÿ�with �Dÿ� � �OHÿ� and/or [Oÿ].

Oda et al. [17] have shown that the [Dÿ] traps introduced by oxygen are shallow trapswhich are temperature dependent. They are effective only at low temperature but not atroom temperature, because at this temperature carriers trapped on these levels can beeasily released. The [C�3 ] and [Cÿ1 ] traps are deeper and not so sensitive to temperature.

By increasing the impurity (Oÿ or OHÿ) content the density of structural defects Cÿ1becomes smaller than that of C�3 . Since Cÿ1 are hole traps and C�3 electron traps, it canbe concluded that the p-type of the selenium will be enhanced by the interaction betweenamorphous selenium and aluminium oxides. Then the amorphous side of the contact willnot behave any more like the metallic side of a Schottky diode with a Fermi level pinnednear the centre of the gap but like a crystalline semiconductor which will modify theband scheme of the interface in such a way that a rectifying contact is obtained. The OHcan also induce microcrystallization of Se at the interface [18]. Therefore under bias, thecontact after formation will behave like a metal/semiconductor contact: the depletedlayer of Se±aluminium hydroxyl will behave like classical p-type semiconductor.

Another possibility is that the selenium during formation modifies the oxide layerwith the formation of a p-type oxide layer which will induce diode characteristics. Thesehypotheses are not so different: a layer is induced at the interface aluminium oxide/amorphous selenium, this layer is either oxide or selenium dominant and induces theapparition of a rectifying contact.

5. Conclusion

Different sandwich structures have been studied such as Al/a-Se/Au, Au/a-Se/Au, Al/a-Se/Al, SnO2/a-Se/Au and also Al/PVK/a-Se/Au or Cr/PVK/a-Se/Au. A rectifyingeffect has been obtained only when the underlayer is an aluminium film and only whenthis film is in contact with the amorphous selenium. Such rectifying contact is obtainedonly after a forming process. This forming process can be electrical (three to four I±Vcycles) of thermal (annealing for 1 h at 325 K). XPS depth profiles of Al/a-Se structureshave shown that after a thermal forming process there is interdiffusion of the layers. Atthe interface, the selenium plot follows the profile of the aluminium oxides, as shown bythe oxygen depth profile. The aluminium oxide layer is composed of alumina and also ofsome aluminium hydroxyl (Al(OH)3). The aluminium oxides (mainly the OH radicals)are supposed to modify the density of trapping states present in the amorphous seleniumlayer in such a way that the contact becomes rectifying.

References

[1] J. H. Dessauer and H. E. Clark, Xerography and Related Processes, The Focal Press, Lon-don 1965.

[2] J. W. Boag, Xeroradiography, Phys. Med. Biol. 18, 3 (1973).


[3] R. A. Zingaro and W. C. Cooper, Selenium, Van Nostrand/Reinhold Co., New York 1974.[4] I. I. Ibragimov, M. I. Shneidman, V. G. Agaev, and Z. M. Abutalybova, Thin Solid

Films 196, L1 (1991).[5] G. Safoula, C. Suresh, K. Napo, K. Alimi, J. C. Bern�ede, and J. Y. Mevelec, J. non-

crystall. Solids 169, 126 (1994).[6] C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder, and G. E. Muilenberg,

Handbook of X-Ray Photoelectron Spectroscopy, Perkin Elmer Corp., Physical ElectronicsDiv., Eden Prairie (Mont.) 1979.

[7] A. Latef and J. C. Bern�ede, phys. stat. sol. (a) 124, 243 (1991).[8] S. Poganski, Z. Phys. 134, 469 (1953).[9] A. E. Sapega and H. P. D. Lanyon, phys. stat. sol. (a) 11, 713 (1972).

[10] C. H. Champness and M. I. El Azab, Solid State Electronics 24, 643 (1981).[11] M. H. Brodsky and G. H. D�ohler, Crit. Rev. Solid State Sci. 5, 591 (1975).[12] K. S. Chari and B. Malhur, Thin Solid Films 75, 157 (1981).[13] H. Birey, J. Appl. Phys. 49, 2898 (1978).[14] G. Delaunay, F. Ropars, and J. Despaujols, Thin Solid Films 87, 127 (1982).[15] G. B. Abdullaev, S. I. Mekhtieva, D. Sh. Abdinov, and G. M. Aliev, phys. stat. sol. 11,

891 (1965).[16] M. Kastner, D. Adler, and H. Fritzsche, Phys. Rev. Letters 37, 1504 (1976).[17] O. Oda, A. Onozuka, and I. Tsuboga, J. non-crystall. Solids 83, 49 (1986).[18] A. Legros, M. Kun Shi, A. Mouton, and A. Selmani, J. Appl. Phys. 78, 3048 (1995).

578 S. Touihri et al.: Diode Devices Based on Amorphous Selenium Films

Documents

Diode Devices Based on Amorphous Selenium Films