Thin Solid Films, 236 (1993) 1-5 1

Properties of transparent conducting oxides deposited at room temperature

Lynn Davis Motorola, Inc., Advanced Manufacturing Technologies, Land Mobile Products Sector, Fort Lauderdale, FL 33322 (USA)

Abstract

A range of indium tin oxide (ITO) films were deposited on glass substrates at room temperature using reactive dc sputtering. The coatings were deposited in an Ar/O2 plasma, and their appearance immediately after deposition ranged from mirror-like to brown to clear. The chemical and physical properties of each of these films were benchmarked with a battery of analytical tools including XPS, Auger, XRD, resistance and thickness measurements. The impact of annealing on the films was also studied. Annealing the films tends to convert all coatings to transparent conductive layers due to oxidation reactions. Film resistance was also found to decrease with annealing. The chemical phenomena responsible for these characteristics are discussed.

1. Introduction

Indium tin oxide (ITO) films are widely used in a variety of optoelectronic applications such as LCDs, energy efficient windows, solid state image sensors, solar cells and CRTs. ITO is essentially an In203 based material that has been doped with Sn to improve electrical conductivity. Tin acts as a cationic dopant in the In203 lattice and substitutes at the indium sites to bind with interstitial oxygen [1]. Tin can exist as either SnO (valence 2) or SnO2 (valence 4). Since indium has valence 3 in In203, the presence of SnO2 would result in n doping of the lattice because the dopant would add electrons to the conduction band. In contrast, the pres- ence of SnO would lower the electron density in the conduction band. During low temperature deposition, tin is present in the ITO films primarily as SnO, result- ing in low carrier densities and high film resistances. However, annealing the films has been found to create SnO2 and results in the formation of an n-type semicon- ductor with high carrier density (about 5 x 1020 cm -3) and low resistance (about 5 x 10-4f~ cm) [1, 2]. Al- though Sn doping of In203 is the most common, other doping schemes are possible. A p-type material could be formed by doping with a metal of valence 2. Alterna- tively, anion doping could be accomplished by replac- ing a portion of the O atoms in In203 with F or C1.

The two main factors controlling film resistance and electrical conductivity are carrier mobility and carrier density [3, 4]. The majority carriers in ITO are elec- trons. For an ITO film deposited from a 90:10In/Sn target, the density of Sn atoms is 3.0 x 1021 cm -3 [5]. Hence, since Sn atoms can donate one electron to the conduction band, the theoretical maximum carrier den-

sity due to Sn doping alone is 3.0 x 10 21 c m -3 . How- ever, since oxygen vacancies also contribute to the electron density, the carrier density can be greater than this theoretical limit. The carrier mobility of ITO films is of the order of 1 -100cmEV -1 s -1, with values of 2 0 - 4 0 c m 2 V -1 s -~ being common for sputtered thin films [ 5]. Unfortunately, modifying the properties of the film to increase carrier density usually lowers electron mobility and vice versa. Hence, obtaining the lowest possible resistance is a trade-off between carrier density and electron mobility. ITO films can be readily de- posited with resistances of 5-1000 t) square -1. Typi- cally, ITO resistances of 100-300 t) square -1 are required for conventional LCD applications such as twisted nematic (TN) or supertwisted nematic (STN) displays.

Indium t in films can be deposited using a variety of methods. Atmospheric deposition can be achieved with a spray pyrolysis technique. Vacuum deposition can be accomplished through evaporation, CVD, PECVD, r.f. sputtering or dc sputtering. The best ITO films are deposited using sputtering methods, although this pro- cedure does not allow the control over the In/Sn ratio in the film that CVD does [1, 5-7]. For reactive dc sputtering, metallic In/Sn targets are used and the film composition is determined by the deposition plasma environment. Metallic In/Sn films are deposited in a pure argon plasma in dc sputtering; ITO films are deposited in a mixed Ar/O 2 plasma. Generally, the oxygen is introduced near the substrate in order to reduce target oxidation, which has an adverse effect on sputter rate [8, 9].

This paper reports findings from studies of ITO films deposited by reactive dc sputter deposition. The films

0040-6090/93/$6.00 © 1993 -- Elsevier Sequoia. All rights reserved

2 L. Davis / Transparent conducting oxides deposited at room temperature

were characterized with a battery Of analytical tech- niques, and the impact of plasma parameters on film performance was studied.

2. Experimental details

The In-Sn films were deposited in the prechamber of a Kratos XSAM800 XPS, which is shown schematically in Fig. 1. Pumping in the deposition chamber was accomplished with a 1201s -1 turbomolecular pump backed by a rotary pump, and the base pressure was 6.5 × 10 -6 Pa. An Aja circular magnetron with 5 cm targets was used to deposit all films. The experiments described below were typically performed at a constant dc magnetron power of 73-150 W. Films were de- posited from a 90:10 In:Sn metal alloy (purity 99.99%) unless otherwise noted. Argon was introduced into the chamber through a variable leak valve behind the mag- netron, and the argon partial pressure was manually set at 0.4-1.0 Pa. Oxygen was introduced into the chamber through a stainless steel needle, directed at the sub- strate, attached to a variable leak valve. The oxygen partial pressure was manually set at 0-5 Pa. This ar- rangement of Ar and 02 sources is necessary in order to minimize the oxidation of the sputter target during use. If the In:Sn target becomes oxidized, the sputter rate drops substantially, since oxides usually sputter more slowly than metals [8, 9]. Film thickness was measured by a quartz crystal oscillator residing directly above the substrate during deposition. The typical thickness of the films deposited in this study was 50-100 nm. The ITO films were deposited on glass or quartz slides subjected to a cleaning regimen consisting of a methanol rinse followed by 5 min of UV/ozone cleaning.

A variety of analytical techniques was employed to characterize the films. XPS studies were performed using a Kratos XSAM800. Because of the instrument arrangement, the thin films could be deposited in the

Needle Valve

Oxygen l

XPS [-~

Gate Valve

~ 1 1 Viewport [ I I

~ ,~bstrate

[ Magnetron [__

Sputter Source

. , , , i Fast Insertion Probe

Ne~l~e Argon Valve u

Fig. 1. Schematic of deposition chamber used in this work.

XPS prechamber and then immediately analyzed with- out exposing to atmosphere. XPS spectra were gener- ally acquired using Mg K~ radiation unless otherwise noted. Auger and SIMS surface analysis were gathered using a Perkin-Elmer 610 Auger/SIMS spectrometer. All films that were analyzed with either Auger or SIMS were exposed to atmosphere during the transfer be- tween the deposition chamber and the spectrometer. Hence, atmospheric contaminants such as carbon or oxygen were present on these films. Film resistance was checked with a Veeco four point probe.

3. Results

3.1. Characteristics of In-Sn films By varying the O2/Ar mixture in the plasma, at least

four different classes of In:Sn films can be deposited by reactive sputtering from a metallic 90/10 In/Sn target. Using a 100% Ar plasma, the deposited In/Sn films exhibit a dull grey metallic appearance. Introducing a small amount of oxygen (10%-25% at 75W plasma power) results in a film that has a mirror-like finish. Increasing the oxygen content of the plasma to 25%- 40% (at 75 W plasma power) results in a dark brown film. Plasmas containing more than 40% oxygen pro- duce a film that is slightly yellow to clear. Below, these films are arbitrarily classified as Type 1 (metallic film) to Type 4 (maximum oxygen content). Some of the physical properties of these four classes of In:Sn films are given in Table 1. As discussed below, the film type produced from a given plasma composition depends on plasma parameters such as power and voltage.

The resistance of the In-Sn films varied widely, depending on the deposition conditions. The lowest resistance films were the Type 1 coatings, which are metallic. Adding oxygen to the plasma increased film resistance by several orders of magnitude, with the resistance of the Type 4 films generally exceeding the capability of the Veeco four point probe. Although the Type 1 and Type 2 coatings exhibit excellent electrical conductivities as deposited, they are reflective coatings and unsuitable for use in optoelectronic applications. In contrast, the Type 3 and Type 4 coatings exhibit good optical qualities as deposited, but the electrical conduc- tivity of these films is too high. Hence, in the as- deposited state, these In-Sn films are either too resis- tive or lack optical transparency for use in most appli- cations. Consequently, post-deposition processing is required to improve the properties of the films.

The resistance of In:Sn films was generally lower following post-deposition annealing at temperatures greater than 125 °C. Representative resistances of the four types of In:Sn film as a function of annealing temperature are shown in Fig. 2 for 65 nm films. In:Sn

L. Davis / Transparent conducting oxides deposited at room temperature

TABLE 1. Physical properties of four types of In:Sn film

Class Plasma composition Color a Crystallograph3P

Ar Oxygen

Resistance a (fl square-l)

1 100% 0% Dull grey Metallic In 1-6 2 75%-90% 10%-25% Mirror-like Amorphous 100-1500 3 60%-75% 25%-40% Brown Slight ITO 300-7000 4 < 60% > 40% Clear to yellow Amorphous > 400 000

aThese physical properties are for the as-deposited films. As discussed below, significant changes occur during annealing of the films.

Temperature Dependence of In:Sn Film Resistance

8000"

=1% > 400,000 @-r <= 1oo C

0 4oo0"

~ 2000'

100 200 300 400 Temperature

Fig. 2. Resistances of the four classes of In-Sn films as a function of annealing temperature. The optical transparency of these films as a function of annealing temperature is shown in Fig. 3.

film Types 1-3 exhibited measurable resistances at room temperature and after each annealing step. The resistance of Type 2 and Type 3 coatings converged to similar values ( < 100 ~ square-1) following annealing at 250 °C. The transparent Type 4 films exhibited high resistances ( > 400 000 Q square-~) as deposited and af- ter the 100 °C anneal. Only after annealing to tempera- tures greater than 150 °C were measurable resistances recorded for the Type 4 films. Subsequent anneals at 250°C and 350 °C produced resistivities less than 1000 fZ square- 1.

The impact of post-deposition annealing on optical transparency was also monitored for these films; the results are shown in Fig. 3. The effect of post-deposi- tion annealing was tracked by monitoring the percent- age transmission of the 550 nm line, which lies in the visible spectrum. The percentage transmission of the Type 3 and Type 4 films improved slightly on anneal- ing, since these films were highly transparent as de- posited. In contrast, the Type 1 and Type 2 films were converted from reflective coatings as deposited to trans- parent conductors after annealing. The Type 2 film began to turn transparent after annealing to 150 °C for 30 min. Subsequent anneals up to 350 °C produced a film with an optical transparency similar to the an- nealed Type 3 and 4 films. The Type 1 film became

Optical Transmission of In-Sn Films

100

80-

w~ • T ~ T y p e 2 / @ 80- C

~ 40- . / . .; ~ ~ . T y ~ E 20- i I-

0 30C 150 C 250C 350C

Temperature

Fig. 3. Comparison of optical adsorbance at 550 nm for the four classes of In-Sn film shown in Fig. 2.

transparent only after annealing at 300 °C. Even after several hours at 350 °C, the optical transparency of the Type 1 film was not as good as the Types 2-4, but may be acceptable for some applications.

3.2. Composition and structure of films 3.2.1. Surface composition The composition of the as-deposited films was exam-

ined with XPS, immediately after deposition and with- out exposing the film to atmosphere. As expected, films deposited in a pure Ar plasma contained very little oxygen. The oxygen that was present in these films was probably due to background water in the deposition chamber. The addition of oxygen to the plasma results in significant oxygen incorporation into the film. Even at oxygen concentrations of 15% in the plasma, the stoichiometry of the deposited film is close to that of the metal oxide. Higher oxygen concentrations in the plasma produce only minor changes in surface compo- sition. Thus, even though film Types 2-4 had different appearances, the surface compositions are remarkably similar. Representative surface compositions deter- mined in this study are given in Table 2.

The chemical distribution of In, Sn and O in these films was studied with Auger and SIMS depth profiling after removing the films from the deposition chamber.

4 L. Davis / Transparent conducting oxides deposited at room temperature

TABLE 2. Representative surface composition of In:Sn films (surface atomic concentration percentages as determined by XPS using stan- dard sensitivity factors)

Film type Indium Tin Oxygen

1 69.9 12.8 17.4 2 36,9 6.8 56.3 3 32.5 3.8 63.8 4 31.0 3.0 66.0 Theoretical a 35.3 3.9 60.8

aTheoretical values for ITO film reactively sputtered from a 90:10 In/Sn metal target.

Type 1 films contained a small amount of oxygen at the surface of the film and at the interface between the sputtered film and the glass slide. Between these two extremes, the oxygen content dropped rapidly. The high oxygen content of the surface of these films demon- strates that In/Sn oxidation is strongly favored thermo- dynamically and occurs readily on exposure to atmosphere. The increase in oxygen content near the glass substrate is due to scavenging of oxygen from the SiO2 in the substrate. In contrast, the composition of Types 2-4 films was uniform throughout. Thus, the stoichiometry of these films is expected to be consistent throughout.

3.2.2. Crystallography of deposited films X-Ray diffraction was used to study the crystallogra-

phy of the sputter deposited films. For the as-deposited coatings, each type displayed a distinctive crystallogra- phy. The as-deposited metallic Type 1 films exhibited a crystal structure analogous to that of metallic In. In contrast, the mirror-like Type 2 films deposited at room temperature exhibited an amorphous crystallography characterized by a broad featureless diffraction peak centered at 33 ° . This amorphous structure probably limits carrier mobility in these films and results in an increase in resistance, even though the Type 2 films have a carrier density comparable with the Type 1 coatings. Similarly, as deposited at room temperature, the Type 3 and Type 4 films were primarily amorphous, although each contained a small amount of polycrys- talline ITO. After annealing at 350 °C, the four types of In-Sn coating exhibited the diffraction pattern of In203, providing clear evidence of the conversion of the coatings to ITO.

3.3. Impact of plasma parameters 3.3.1. Impact of plasma composition on deposition

rate and film structure In reactive sputtering, the properties of the deposited

film are influenced by plasma settings and composition. In order to identify these effects, a series of experiments

DeposlUon Rate of ITO Films ~5

10- 2 8 % ~

Type

o

o o . o a

60 80 100 120 140 160 Power ON)

Fig. 4. Deposition rate of ITO films as a function of plasma power and plasma composition. The percentage oxygen in the plasma is indicated. Each data point is an average of at least four experiments, and the error bars represent the standard deviation.

were performed monitoring the deposition rate of In/Sn as a function of both plasma power and plasma compo- sition. The results are shown in Fig. 4.

In-Sn films deposited in a 28% 02 plasma were Type 3 coatings at plasma powers below 100W. Above 100 W, the deposits were Type 2. This transition could easily be followed by the physical appearance of the films and by X-ray diffraction, as discussed below. In contrast, the 43% 02 plasma resulted in the deposition of Type 4 films at all powers up to 150 W. The change in the type of film deposited with the 28% 02 plasma suggests that the In/Sn metal is being deposited so fast at the higher powers that the films become oxygen deficient. A similar phenomenon would be likely to occur for the 43% oxygen plasma at sufficiently high In/Sn deposition rates (i.e. high plasma power). There- fore, in order to deposit a given film type, it is critical that the plasma composition and plasma power are matched to appropriate values. These values are likely to be different for each deposition system due to differ- ent geometries and Ar/O2 mixing.

3.3.2. Influence of plasma settings on film resistance A series of 50 nm films were deposited at various

plasma powers for the 28% and 43% 02 plasmas in order to determine the impact of plasma power on the films' resistivity. In general, the resistance of films de- posited in the 43% 02 plasma was outside the range (i.e. >500 000 f~ square -1) of the Veeco resistance meter, prior to annealing, as discussed above. In contrast, the resistivity of films deposited in a 28% 02 plasma varied from 5.3 × 10-4~ ') cm to 0.35 f~ cm prior to annealing. For comparison, the resistance of bulk metallic Cu is 1.7 x 1 0 - 6 ~'2 cm (i.e. 17 l~O cm). Annealing the films at 400 °C for 30 min lowered the resistances, and exposed the dependence of plasma composition and plasma

L. Davis / Transparent conducting oxides deposited at room temperature 5

TABLE 3. Dependence of film resistivity on plasma power and plasma composition: values represent an average of at least four films after 400 °C anneal; standard deviations are given in parentheses

Power Resistivity (IJ~ cm -1) Watts

28% 0 2 Plasma 43% 0 2 Plasma

73 572 (215) 2470 (1050) 98 366 (119) 2870 (632)

123 1170 (245) 2250 (663)

power on film resistivity. These results are given in Table 3.

Clearly, the resistivity of films deposited in the 43% 02 plasma is higher at all values of plasma power than those deposited in lower oxygen-containing plasmas. In addition, increasing the plasma power does not neces- sarily improve film performance, as demonstrated by films deposited in a 28% 02 plasma. The resistivities of these films decreased with an increase in plasma power up to 100W. Above 100 W, the resistivity increased. This is probably due to the fact that deposition at powers greater than 100 W results in a Type 2 deposit, whereas deposition below 100 W produces Type 3 films, as described above. This finding illustrates the impor- tance of controlling plasma variables in order to obtain the desired ITO films. If the optimum parameters are not met, then a different ITO film, with different physi- cal properties, may be deposited. Post-processing of the film may recapture the desired performance, but is likely to cost more than depositing the film correctly.

4. Conclusions

At least four different types of In-Sn film can be deposited merely by varying the argon/oxygen mixtures in the deposition plasma. The films range in appearance from metallic coatings (deposited in a pure Ar environ- ment) to transparent indium tin oxide films (deposited in a 50:50 Ar/O2 plasma). The properties of the as- deposited films depend not only on plasma composi- tion, but also on parameters such as plasma power. Although four distinct films are deposited at room temperature, the physical properties of these coatings converged after annealing at temperatures greater than 150 °C. Annealing produced films with lower resistance due to increased film crystallization and liberation of high valency tin dopes. Annealing also improved the optical transmission of the films.

References

1 G. Frank and H. Kostlin, Appl. Phys., 27,4 (1982) 197. 2 C. H. L. Weijtens, J. Electrochem. Soc., 138 (1991) 3432. 3 C. Kittel, Introduction to Solid State Physics, Wiley, New York, 1976. 4 N. W. Ashcroft and N. D. Mermin, Solid State Physics, Saunders,

Philadelphia, PA, 1976. 5 T. J. Coutts, Transparent Conducting Oxides, American Vacuum

Society Short Course Notes. 6 V. Vasu and A. Subrahmanyam, Thin Solid Films, 193/194 (1990)

696. 7 T. Maruyama and K. Fukui, J. Appl. Phys., 70 (1991) 3848. 8 J.L. Vossen and W. Kern, Thin Film Processes, Academic, New York,

1978. 9 J. L. Vossen and W. Kern, Thin Film Processes II, Academic, New

York, 1991.