Transport gap of organic semiconductors in organic modified Schottky contacts

Transcription

1 Applied Surface Science (2003) Transport gap of organic semiconductors in organic modified Schottky contacts Dietrich R.T. Zahn *, Thorsten U. Kampen, Henry Méndez Institut für Physik, TU Chemnitz, D-09107, Germany Abstract Two different organic molecules with similar structure, 3,4,9,10-perylenetetracarboxylic dianhydride (PTCDA) and N,N 0 - dimethyl-3,4,9,10-perylenetetracarboxylic diimide (DiMe-PTCDI), were used for the modification of Ag Schottky contacts on sulphur passivated GaAs(1 0 0) (S-GaAs). Such diodes were investigated recording in situ current voltage (I V) characteristics. As a function of the PTCDA thickness the effective barrier height of Ag/PTCDA/S-GaAs contacts initially increases from 0:59 0:01 to 0:72 0:01 ev, and then decreases to 0:54 0:01 ev, while only a decrease in barrier height from 0:54 0:01 to 0:45 0:01 ev is observed for DiMe-PTCDI interlayers. The initial increase and decrease in effective barrier height for PTCDA and DiMe-PTCDI respectively, is correlated with the energy level alignment of the lowest unoccupied molecular orbital (LUMO) with respect to the conduction band minimum (CBM) of S-GaAs at the organic/inorganic semiconductor interface. Whilst there is an additional barrier for electrons at the PTCDA/S-GaAs interface of about 150 mev, i.e. the LUMO lies above CBM, the LUMO is aligned or below CBM in the DiMe-PTCDI case. The results also shine light on the important issue of the transport gap in organic semiconductors for which an estimation can be obtained. # 2003 Elsevier Science B.V. All rights reserved. Keywords: Transport gap; Organic semiconductor; Schottky contacts 1. Introduction The field of organic electronics is continuously growing, fuelled by the promise of novel devices and applications that can be derived from electronically and optically active organic and inorganic/organic hybrid material systems. In this context one topic of interest is the controlled modification of electronic and transport properties of traditional inorganic metal/semiconductor junctions by means of organic layers at the interface [1,2]. Forrest et al. used two perylene derivatives similar in their structural formula, 3,4,9,10-perylenetetracarboxylic dianhydride * Corresponding author. address: zahn@physik.tu-chemnitz.de (D.R.T. Zahn). (PTCDA) and N,N 0 -dimethyl-3,4,9,10-perylenetetracarboxylic diimide (DiMe-PTCDI), to modify Metal/ GaAs Schottky contacts [3,4]. From their ex situ I V measurements they determined barrier heights which are, in most cases, larger than what can be obtained from unmodified metal/gaas contacts. In this work, we investigated these organic modified interfaces using I V characteristics recorded in situ and additionally ultraviolet photoemission spectroscopy (UPS). The results not only reveal the possibility of controlled tuning of the effective barrier height by the interlayer thickness, but also contribute to the important issue of the transport gap in organic semiconductors. The evolution of the effective barrier height for very thin organic interlayers allows the transport gap to be estimated if the position of the /03/$ see front matter # 2003 Elsevier Science B.V. All rights reserved. doi: /s (03)

2 424 D.R.T. Zahn et al. / Applied Surface Science (2003) highest occupied molecular orbital (HOMO) is known. The latter value is derived from the UPS measurements. 2. Experimental Tellurium doped n-gaas(1 0 0) (Freiberger Compound Materials GmbH) with a concentration of N D ¼ cm 3 was used as a substrate. Wafer pieces are first degreased in consecutive ultrasonic baths of acetone, ethanol and deionised water for 5 min each, then etched in a solution of S 2 Cl 2 :CCl 4 (1:3) for 10 s, followed by rinse in CCl 4, acetone, ethanol and deionised water during 5 s each and finally dried in a N 2 flow. After this procedure the substrate is transferred into an ultra-high vacuum (UHV) chamber and annealed for 30 min at 470 8C and a pressure below mbar. The resulting sulphur passivated GaAs(1 0 0) surfaces, from now on being denoted as S-GaAs, exhibit 2 1 surface reconstructions as judged by low energy electron diffraction. Additional details about the passivation process and surface reconstruction are given elsewhere [5,6]. The organic layer is deposited by organic molecular beam deposition (OMBD) onto the S-GaAs substrate with the thickness monitored by a quartz microbalance, previously calibrated using atomic force microscopy measurements. For I V measurements, samples with organic layer thickness varying from 0 to 60 nm were prepared. For the upper metallic contact in the diodes, silver dots of 2: cm 2 area and 200 nm thickness were evaporated through a shadow mask. Ohmic contact to the backside of the substrate is achieved by inserting metallic Ga In alloy paste between the substrate and the cleaned copper plate and subsequent annealing. Typically 20 diodes were measured in situ for each sample using a picoammeter HP 4140B. UPS spectra (hn ¼ 21:22 ev) were recorded using an ARUPS 10 system. 3. Results and discussion Let us first consider the HOMO position at the organic/s-gaas interface as derived from UPS measurements. The PTCDA/S-GaAs interface was already studied in a previous work [2] revealing that the HOMO, here defined as the maximum (upper edge) of the HOMO peak in the UPS spectra, is located 0:96 0:10 ev below the valence band maximum (VBM) of S-GaAs. The energy position is determined from the intercept of two linear extrapolations, one describing the background and the second one in the low binding energy region of the spectrum, being tangent to the curve in the inflection point. Comparable UPS spectra for the DiMe-PTCDI case are shown in Fig. 1. Here the position of VBM in S- GaAs is found at 1:02 0:10 ev with respect to the Fermi level E F (¼0 binding energy). The evolution of the spectra with increasing DiMe-PTCDI deposition reveals that there is no detectable change in band Fig. 1. Valence band spectra of DiMe-PTCDI deposited on S-GaAs for several thicknesses.

3 D.R.T. Zahn et al. / Applied Surface Science (2003) bending. After 10 nm of DiMe-PTCDI the HOMO position is 2:09 0:10 ev relative to E F. Therefore the HOMO is located 1:07 0:10 ev below VBM. Moreover, the formation of an interface dipole can be derived from the shift of the secondary electron cutoff at high binding energy (see left panel of Fig. 1). Its value of D ¼ (0:27 0:10) ev and the HOMO position are important ingredients for the energy level alignment diagram presented in Fig. 4. The HOMO position and its offset with VBM are in particular important when holes have to be considered in the electrical transport. Considering electrons, however, the barrier at the interface is determined by the offset between lowest unoccupied molecular orbital (LUMO) and conduction band minimum (CBM) and the LUMO is the relevant transport level. Other techniques have to be employed to evaluate the LUMO position which is defined in analogy to the HOMO as the minimum (lower edge) of the LUMO peak as can be measured, e.g. by inverse photoemission when taking care of the experimental broadening involved [7,8]. We have recently shown that the electron affinity EA, i.e. the distance between LUMO and the vacuum level, can be derived for PTCDA by a systematic variation of the interface dipole between PTCDA and GaAs [9]. Using EA ¼ 4:12 0:10 ev for PTCDA the related transport gap is in the range of ev. It is then clear (see also Fig. 4) that electrons experience an additional barrier at the PTCDA/S-GaAs interface of around 150 mev. The consequence of this barrier is directly evident from the I V measurements which reveal an increase in barrier height when a very thin PTCDA interlayer is inserted (see Fig. 3, upper panel) in excellent agreement with the photoemission results. The subsequent decrease for thicker layers is explained in terms of an increase in image force lowering resulting from the low dielectric constant of PTCDA (e PTCDA ¼ 2, this value is for the direction perpendicular to the molecular plane which is the one relevant for transport since the molecules lie flat on the substrate surface, with a distance of around 0.32 nm between layers [10]) compared to GaAs (e GaAs ¼ 13:1). Therefore I V measurements can be extremely helpful to at least provide a coarse estimate of the LUMO position with respect to CBM. I V characteristics for various DiMe-PTCDI interlayer thickness in Ag/S-GaAs diodes are displayed in Fig. 2. I V characteristics for S-GaAs/DiMe-PTCDI/Ag diodes with different interlayer thicknesses. Fig. 2. Clearly the current is larger for all diodes that contain DiMe-PTCDI interlayers than for the one without. Therefore the effective barrier height is lowered for all interlayer thicknesses as can also be seen in the plot of effective barrier heights versus interlayer thickness in Fig. 3 (lower panel). Here the barriers were derived from the saturation current extrapolated to zero bias employing thermionic emission theory. This immediately indicates that in contrast to the PTCDA case there is no additional barrier at the interface. Consequently it follows as a coarse estimate that the LUMO at most can be aligned with CBM or lies somewhat below CBM. The uncertainty in the Fig. 3. Effective barrier height F eff, as a function of interlayer thickness.

4 426 D.R.T. Zahn et al. / Applied Surface Science (2003) Fig. 4. Energy level alignment at the interface between S-GaAs(1 0 0) and PTCDA (left) and DiMe-PTCDI (right). LUMO position is reflected by the grey bar used in the right hand part of Fig. 4. The difference in LUMO-CBM alignment between PTCDA and DiMe-PTCDI is corroborated taking two facts into account. First there is also a slight difference in HOMO VBM alignment, namely the DiMe-PTCDI HOMO to GaAs VBM offset is by around 100 mev larger than in the PTCDA case. Secondly it is worth having a look at the optical gaps determined from the position of the first peak in optical absorption. The values are 2.22 and 2.14 ev for PTCDA and DiMe- PTCDI, respectively [11,12]. If the transport gap shows the same trend in size as the optical one, then this would provide another 80 mev or so for lowering the LUMO relative to CBM in the DiMe-PTCDI case. Both effects together amount to approximately 180 mev well in agreement with what is needed to find the LUMO aligned or slightly below CBM. We consider finally the evolution of effective barrier height with DiMe-PTCDI interlayer thickness. There is a decrease to a minimum barrier of 0.45 ev at around 30 nm followed by a slight increase. The decrease seems to be compatible with the image force lowering picture applied for the PTCDA. This is not surprising when considering that both systems under study are very similar, i.e. they are basically the same except for the tiny difference in molecular structure. However, we recently found that this tiny difference has a quite pronounced effect on molecular ordering [13]. While PTCDA molecules are lying flat on the substrate surface, the molecular plane of DiMe- PTCDI molecules is tilted in the range of with respect to the substrate plane. It thus seems that the molecular orientation has a dramatic influence on electrical transport through the structures studied. One possibility is an increased contribution of holes to the transport as they were reported to have a higher mobility parallel to the plane of the molecules [14]. Indeed our own preliminary charge deep level transient measurements gave first hints that hole injection and/or transport have a significant influence in the DiMe-PTCDI case. Still further experimental and theoretical effort is required to clarify this issue. The discrepancy between the results presented here and previous results by Forrest et al. [3,4] also deserve a final comment. The difference clearly lies in the fact that measurements in this work were performed in situ, i.e. in UHV. The previous results obtained ex situ are strongly affected by the interaction of ambient gas molecules with the organic layer. Indeed our own ex situ measurements are quite comparable to Forrest s results. 4. Summary and conclusions UPS measurements were employed to determine the HOMO VBM offset for the DiMe-PTCDI/S- GaAS case to be 1:07 0:10 ev and the interface dipole to be 0:27 0:10 ev. The comparison of results from I V measurements for PTCDA and DiMe-PTCDI interlayers reveal that there is no additional barrier for electrons at DiMe-PTCDI/S-GaAs interface thus establishing that the LUMO position is aligned or slightly below the CBM of S-GaAs for the DiMe-PTCDI case. The DiMe-PTCDI transport gap is therefore likely to be slightly lower than of PTCDA, i.e. near 2:42 0:10 ev.

5 D.R.T. Zahn et al. / Applied Surface Science (2003) Acknowledgements Financial support by the Bundesministerium für Bildung and Forschung (BMBF contract No. 05 KS1OCA/1) and the EU DIODE network (HPRN- CT ) is gratefully acknowledged. The authors would like to thank Antoine Kahn for fruitful discussions. References [1] A. Vilan, A. Shanzer, D. Cahen, Nature 404 (2000) 166. [2] T.U. Kampen, S. Park, D.R.T. Zahn, Appl. Surf. Sci. 190 (2002) 461. [3] S.R. Forrest, M.L. Kaplan, P.H. Schmidt, J.M. Parsey Jr., J. Appl. Phys. 58 (1985) 867. [4] S.R. Forrest, M.L. Kaplan, P.H. Schmidt, J. Appl. Phys. 55 (1984) [5] D.R.T. Zahn, T.U. Kampen, S. Honecker, W. Braun, Vacuum 57 (2000) 139. [6] C. Gonzáles, I. Benito, J. Ortega, L. Jurczyszyn, J.M. Blanco, R. Pérez, F. Flores, T.U. Kampen, D.R.T. Zahn, W. Braun, Phys. Rev. B, to be submitted. [7] I.G. Hill, A. Kahn, Z.G. Soos, R.A. Pascal Jr., Chem. Phys. Lett. 327 (2000) 181. [8] A. Kahn, private communications. [9] S. Park, T.U. Kampen, R.T. Zahn, Appl. Phys. Lett 79 (2001) [10] S.R. Forrest, Chem. Rev. 97 (1997) [11] R. Kaiser, M. Friedrich, T. Schmitz-Hübsch, F. Sellam, T.U. Kampen, K. Leo, D.R.T. Zahn, Fresenius J. Anal. Chem. 363 (1999) 189. [12] V. Bulovic, P.E. Burrows, S.R. Forrest, J.A. Cronin, M.E. Thompson, Chem. Phys. 210 (1996) [13] T.U. Kampen, G. Salvan, A. Paraian, C. Himcinschi, A.Y. Kobitski, M. Friedrich, D.R.T. Zahn, Appl. Surf. Sci., in press. [14] S.R. Forrest, M.L. Kaplan, P.H. Schmidt, J. Appl. Phys. 60 (1986) 2406.