"Study of Indium Tin Oxide (ITO) for Novel Optoelectronic Devices"
Ph.D. thesis by Shabbir A Bashar

5.2 Patterning ITO

Patterning ITO is necessary not only for appraising the quality of the film (using TLM method for example) but also for fabricating useful micro-contacts to devices. As for metals and semiconductors, there are two potential methods of patterning ITO:

  1. lift-off lithography
  2. etch lithography

5.2.1 ITO Patterning by Lift-off Lithography

Due to the high velocity particles involved in a reactive r.f. sputtering system, the liftoff technique was not satisfactory. Both thin (< 1 mm) and thick (> 3 mm) photoresists (PR) were used in the experiments for ITO deposition by liftoff technique; whereas the thin PR was not removable in acetone, by flood exposure under UV lamp or in heated KOH solution, the thicker PR was less difficult to remove. However, the walls of the thick PR were rigidly stuck to the substrate. Microscopic examination of the thin PR showed numerous cracks on its surface most probably caused by PR hardening and baking as a result of penetration of incident particles through its entire depth. In case of the thick PR, the side walls are most probably damaged by bombardment of high velocity plasma particles reflected from the deposited ITO as shown in Figure 5.9:

Figure 5.9: Photoresist wall hardening by reflected plasma particles

In another experiment, the substrate was patterned with PR as usual and placed on the Nordiko substrate table at a 45 degree angle to reduce damage to the PR. However, the PR was still difficult to remove while the ITO deposition rate dropped below satisfactory levels (< 1 /sec).

Therefore, ITO patterning was carried out by etching methods only. These include wet chemical etching and ion milling or dry etching as presented in the following sub-sections.

5.2.2 Wet Chemical Etching

It was found that ITO readily etches in HF solution (HF:H2O2:10H2O) and HCl solutions of various strengths. However, ITO cannot be etched in ammonia solutions (8NH3:3H2O2:400H2O). The etch rates in HF were very high (between 100/sec to 150/sec) and often uncontrollable. Subsequently HCl solution was used for wet etching ITO in majority of the cases.

Figure 5.10: Wet Chemical Etch Rate of ITO as a function of HCl concentration at room temperature and controlled humidity

The undiluted etch solution contained 36% HCl corresponding to a molar solution. Figure 5.10 shows the etch rate of ITO deposited on S.I. GaAs substrates as a function of HCl concentration in the etchant. These reactions were carried out in the cleanroom under controlled conditions where the temperature was kept below 20C and the humidity below 60%. Although not documented here, excess temperature and humidity resulted in increased etch rates. Therefore, to avoid undesirable results, the above ambient conditions were maintained for subsequent ITO etching.

As mentioned earlier, the wet chemical etch process is highly dependent on a number of ambient conditions. In addition, we noticed that a certain amount of ITO - varying from 0.5mm to 1.5mm was often etched away from underneath the photoresist. In severe cases, thin traces of the photoresist remained on the substrate.

Figure 5.11 is a photo-micrograph of a TLM pattern chemically etched using HCl to produce ITO mesas on S.I. GaAs substrate. The mesas are 110mm wide; the amount of ITO removed completely due to under etching was 1.5mm. This is an extreme example of lack of control of wet chemical etching technique.

Figure 5.11: Photo Micrograph of an ITO mesa, etched using HCl solution. The figure shows characteristic coarse edges due to solvents creeping under the resist; blemishes and traces of thin photoresist can also be seen

For relatively large dimensions (> 100 mm), wet chemical etching using the above solution was both adequate, repeatable and relatively easy to accomplish. During the course of this investigation, the most frequently used chemical for wet etching ITO, where deemed suitable, was a 1HCl : 1H2O solution corresponding to 18% HCl by content. The corresponding etch rate was 8/sec.

5.2.3 Effect of Adding Zn Dust to HCl Solution

The addition of zinc dust to the ITO surface is said to enhance wet chemical etching in HCl solutions. A GaAs sample with ITO was first left in a mixture of water and Zn powder. It was then dried by leaving it on a "blotting" surface (i.e. cleanroom wipe) which left fine granules of Zn powder on the sample surface. The sample was then immersed in the HCl solution.

The ITO removal did not seem to be affected by the immersion in the Zn:H2O suspension; also the etch on the surface was non-uniform. The residue Zn was removed from the sample by placing it in a 3H2O:1HNO3 bath. But this latter solution seemed to etch the GaAs substrate. Therefore, this method of ITO patterning was deemed to be unsuitable for controlled and uniform micro patterning of ITO as required in this project.

5.2.4 Selective RIE of ITO in Argon Plasma

The commonly used solution of HCl for ITO definition by wet chemical etching results in removals of approximately 0.5 mm of ITO underneath the photoresist mask. Thus where near micron structures need to be reproduced, wet chemical etching techniques are often unsuitable. Although CH4/H2 mixtures can be used for reactive ion etching of ITO [148], this is a potentially explosive gas mixture which is unsuitable for use in environments without relatively expensive exhaust set-ups operating round the clock to remove any build-ups. Dry etching of ITO using a gas mixture of acetone, argon and oxygen - essentially a hydrocarbon etch with acetone being the source of reactive organic radicals in the plasma discharge - has also been reported [149]. However, this technique needs cumbersome optimisation of the various constituent gas partial pressures to ensure that no carbon debris is left on the surface.

Therefore we investigated the suitability of a number of gases for dry etching ITO: freon, oxygen and argon. Dry etching using Ar was found to be the most suitable in terms of speed and selectivity. Figure 5.12 and Figure 5.13 show the results of this investigation.

Figure 5.12: ITO Etched Thickness in argon ambient versus time under various r.f. powers

A 4300 ITO layer was deposited on a Si substrate by reactive r.f. sputtering, the sample was then patterned using standard photolithography process to enable thickness measurements to be carried out following dry etching. Thick photoresist (> 3 mm) was used as the mask for these experiments which were performed at three suitable r.f. powers of 100W, 150W, and 200W. The dry etching chamber was pumped to pressures better than 1e-4 torr; Ar was then flowed in at a constant rate to maintain a chamber pressure of 11 mtorr prior to exciting the plasma.

The last point on the 150W power line in Figure 5.12 (corresponding to 40 minutes and 6300) was not used for the least square fit as the Si substrate was found to have been etched also. Etch rates of 35/min, 111/min and 131/min were obtained for 100W, 150W and 200W respectively.

Figure 5.13: ITO Etch Rate in Argon ambient vs. r.f. power of RIE

Figure 5.13 shows that the etch rate is non-linearly dependent on the r.f. power. Also pin-holes were visible on substrates etched at higher powers. Both these effect suggest that the process produces substrate heating and that the process of ITO removal in an Argon plasma is that of a physical knocking of the ITO species rather than by a chemical reaction [150]. Based on these findings, the r.f. power of 100W was decided to be the most controllable and suitable for further dry etching work.

As even the thick photoresist was damaged during the RIE process and was therefore difficult to remove, an alternative material for the mask was necessary. A metal which has a sufficiently lower etch rate than ITO in the Ar plasma but nevertheless can be subsequently removed chemically in preference to ITO would be the ideal substitute. Aluminium was the preferred candidate on both counts.

A resistively evaporated layer of Al was therefore used as a mask. ITO is selectively dry etched over Al in argon plasma. Figure 5.14 shows that while the ITO etch rate at 100W r.f. power and 11 mtorr pressure, is 35/min, the corresponding figure for Al is only 8/min making Al a more than adequate candidate to be used as a mask for dry etching ITO.

Figure 5.14: Selective RIE of ITO over Al in argon plasma at a r.f. power of 100W

Following dry etching, the Al is selectively removed by placing the sample in a super saturated luke warm solution ( 30C) of KOH for app. 2 minutes. A 1000 layer of Al was found to be adequate. Thicker layers of Al were difficult to remove chemically; although further heating the KOH solution accelerated Al removal, this also caused partial etching of the ITO. Hence, all subsequent dry etching of delicate ITO patterns were masked using 1000 of thermally evaporated Al.

In comparison to the ITO dry etching techniques of Adesida et al [148] using CH4/H2 mixtures, these etch rates are an order lower. This difference most likely arises from a variation in the etch mechanisms involved in the two cases: whereas the carbon in the methane/hydrogen mixture is said to react with the oxygen in ITO, there are no similar reactive species in the Ar plasma. Therefore, it is postulated that the etching of ITO in Ar plasma is essentially an ion milling process.

The etch rates obtained in this investigation for the RIE of ITO is, however, comparable to those reported by Saia et al [149]: 22/min in C3H6O/Ar/O2 or 60/min in hydrogen chloride gas.

Figure 5.15 shows a photo micrograph of a TLM mesa pattern etched on a test ITO sample using the dry etching technique discussed thus far. In comparison with the results obtained for a similar test sample etched using HCl solution, these results show near perfect edges replicated from the mask. There is no apparent lateral etching or residue of debris on the substrate.

Figure 5.15: Photo micrograph of an ITO mesa pattern, etched using RIE. The figure shows near perfect edges while the substrate appears remarkably clean

As discussed in Chapter 6, using this dry etching technique with Al as the mask, a 2mm FET gate structure was successfully produced in the fabrication of the first transparent gate HEMT. Therefore, this technique was used in all subsequent fabrication ITO structures below 50mm and proved to be a reliable and very convenient method for patterning delicate ITO structures.

1998: Shabbir A. Bashar (in accordance with paragraph 8.2d, University of London Regulations for the Degrees of M.Phil. and Ph.D., October 1997). The Copyright of this thesis rests with the author, and no quotation from it or information derived from it may be published without the prior written consent of the author.
Title Dedication Acknowledgements Abstract Symbols Contents

This page is hosted by
Dr. Shabbir A. Bashar, e-mail: shabbir@betelco.com