Achieving high mobility ZnO : Al at very high growth rates by dc filtered cathodic arc deposition

Achieving a high growth rate is paramount for making large-area transparent conducting oxide coatings at a low cost. Unfortunately, the quality of thin films grown by most techniques degrades as the growth rate increases. Filtered dc cathodic arc is a lesser known technique which produces a stream of highly ionized plasma, in stark contrast to the neutral atoms produced by standard sputter sources. Ions bring a large amount of potential energy to the growing surface which is in the form of heat, not momentum. By minimizing the distance from cathode to substrate, the high ion flux gives a very high effective growth temperature near the film surface without causing damage from bombardment. The high surface temperature is a direct consequence of the high growth rate and allows for high-quality crystal growth. Using this technique, 500–1300 nm thick and highly transparent ZnO : Al films were grown on glass at rates exceeding 250 nm min−1 while maintaining resistivity below 5 × 10−4 cm with electron mobility as high as 60 cm2 V−1 s−1. (Some figures in this article are in colour only in the electronic version) Currently, indium supply can meet the demand, even in the United States where no indium is mined and little is actually recycled. However, building energy efficiency must be improved world wide and smart, multi-functional windows requiring transparent conductive oxide (TCO) coatings will play an important role. About 108 m2 year−1 of TCO coated glass is already required for the flat panel display and solar cell industry. Smart windows will require TCO coatings over areas of the same order of magnitude, and the photovoltaic market is also expected to grow. This will require a very substantial increase in the indium supply if indium tin oxide (ITO) and other In based materials remain the TCO of choice. Such an increase in the demand of ITO will have a profound impact on the indium price which is already subject to large swings. Furthermore, indium has very unique properties which make it useful for many other important applications which will inevitably suffer. For example, indium is used for coatings for aircraft parts, for cryogenic and vacuum applications, in optoelectronic devices for fibre-optic communications, etc. Aluminium-doped ZnO (AZO) is one of the leading candidates to replace ITO but several obstacles must first be overcome. Aside from AZO being less resilient to moisture and acids, the electrical properties of AZO deposited onto glass are not as good as with ITO. Epitaxial AZO grown on sapphire by pulsed laser deposition (PLD) [1] has shown mobility as high as 70 cm2 V−1 s−1 but mobility is 40–50 cm2 V−1 s−1 when deposited on glass [2, 3]. The resistivity was a record low 0.8 × 10−4 cm in the study by Agura et al [2] but PLD has a very low growth rate and is not well suited for large-area deposition. 0022-3727/11/232003+05$33.00 1 © 2011 IOP Publishing Ltd Printed in the UK & the USA J. Phys. D: Appl. Phys. 44 (2011) 232003 Fast Track Communication Magnetron sputtering is the current standard for large-area glass coating by physical vapour deposition. It can produce AZO/glass with resistivity as low as 5 × 10−4 cm, but typically the mobility is limited to about 30–40 cm2 V−1 s−1 [4]. This means a high carrier concentration is present, which limits transmission of the solar infrared that could be used to heat a building in cold climates or help power a solar cell. There is one publication where an exceptionally high mobility of 53 cm2 V−1 s−1 was reported for as-deposited rf sputtered AZO which was 780–900 nm thick [5]. However, the growth rate was relatively slow at 15 nm min−1. Reactive dc sputtered AZO has achieved a mobility of 46 cm2 V−1 s−1, but no information on the growth rate was reported [6]. Considering that the films were only 300–400 nm thick and that the rate was not even mentioned, it may be reasonable to assume that the growth rate was not considerably higher than that of rf sputtering. There are a few reports of substantially higher growth rates, reaching 580 nm min−1, for AZO grown by reactive magnetron sputtering but the mobility is limited to 20 cm2 V−1 s−1 at best [7]. For AZO, ZnO, and many other crystalline materials, film quality judged by several metrics seems to be inversely proportional to the growth rate of the technique. Atomic layer deposition can produce high quality crystals (of a handful of materials) at very low rates, followed by MBE, and PLD. Then comes RF sputtering, then reactive sputtering and then ultra-high rate techniques like atmospheric plasma torches which are best suited for amorphous materials and thick porous films. A frequently overlooked growth technique capable of high quality crystal formation is cathodic arc deposition. It is typically disregarded in major review papers on TCOs and has a low profile at most large conferences without a hard coating theme. Pulsed filtered cathodic arcs have produced AZO [8] with mobility above 40 cm2 V−1 s−1, but the growth rates are comparable to PLD. However, unlike PLD, cathodic arcs can be operated in a dc mode, giving much higher growth rates. AZO has been produced by dc cathodic arc in the past [9–11], but the quality was inferior to the films produced by reactive magnetron sputtering. This fact is likely why effort into dc arc growth of AZO and other TCOs seems to have diminished. However, cathodic arc has the inherent advantage of producing a highly ionized plasma of cathode material. Energetic condensation from a plasma is known to give dense films due to both the kinetic and potential energy of the arriving ions [12]. Substrate biasing can control the kinetic energy of the arriving ions, but kinetic energy also brings momentum which can damage the growing film. On the other hand, neutralization of the ion at the film surface releases the potential energy of the ion as heat, which can locally anneal and densify the films leading to better crystal quality. Thus, it is surprising that dc arc-grown AZO did not previously outperform the sputter-deposited films which condense from neutral atoms with very few ions in the arriving flux of particles. As we will show, the limiting factor for previous work on dc arc grown AZO is the relatively large path the plasma followed from cathode to substrate. A long path is typically needed to make room for the magnetic plasma filter necessary to produce particulate-free coatings [13], but it significantly reduces the ion flux. In this work, very high quality AZO with mobility as high as 60 cm2 V−1 s−1 was produced at rates exceeding 250 nm min−1 by minimizing the distance the plasma travelled. The high quality of these films compared with previous reports is attributed to the relatively high ion flux reaching the surface in the compact arrangement. The high flux locally raises the temperature at the film surface and allows for unusually high-quality crystal growth at high rates. Film deposition was done in a vacuum chamber which reaches a base pressure of 1 × 10−5 Torr when the substrate is at elevated temperatures. Oxygen gas was then fed in at a rate of 20–50 sccm with the total pressure in the range 1–5 mTorr. Optimum oxygen pressure depends strongly on the arc current and these ranges were found to give the best balance for low resistivity and high visible transparency. Readily available, very-low-cost Zn containing about 4 at% Al was used for the cathode (40$ kg−1). Arc currents typically used in this system are a relatively modest 25–70 A dc. The arc plasma was filtered using a helical quarter-torus open coil carrying 400 A dc and generating a magnetic field estimated to be 50 mT. Anode–filter separation as well as filter– substrate distance was minimized, with a 300 mm cathode– substrate separation. Borosilicate microscope glass slides (25 mm × 75 mm, 1 mm thick) were used as substrates. They were cleaned with Liquinox, a widely used glass detergent containing ethylene diamine tetra acetate, designed to yield completely residue free surfaces. The samples were thoroughly rinsed with tap water and then the substrates were quickly dried using dry nitrogen, leaving a streak-free surface with no visible particulates or residues. Samples were pre-heated to an initial substrate temperature (TS) of 425 ◦C at most, but excellent material could be deposited at TS = 200 ◦C, and reasonably well performing material could be deposited on room temperature (RT) substrates. Before deposition the substrates were exposed to a 150 W oxygen plasma from a constricted plasma source [14] for 2 min, as a critical last step in the environmentally friendly cleaning procedure. Films were well adherent, even when grown up to several μm thick. Based on our personal experience in the lab growing more than 100 samples, this surprisingly simple cleaning procedure gave an initial glass surface equal to or better than those cleaned with solvents (ethanol and acetone) as judged by spurious residues on the surface. Growth time was usually 2 min and limited to 5 min, sometimes by overheating of the substrate if the plasma flux