Conductive polymers offer charge dissipation in GaN and ZnO sample processing
Conducting polymers such as polythiophenes promise to replace metallic films for charge dissipation in semiconductor processing. This switch should provide ease of use and deliver results of unprecedented quality, say Rafal Dylewicz and Faiz Rahman from the University of Glasgow, UK.
Creating structures in semiconductor materials is fundamental to making all semiconductor devices. A wide variety of process techniques are employed to craft various types of surface features. Several of these rely on the use of charged particles – electrons or ions – for their operation. Such processes include ion-based etching, electron-beam lithography and focused ion-beam techniques. Both electron-beam lithography and FIB techniques can be used for patterning semiconductor surfaces and are widely used in both research and industry.
A direct side effect of any such technique is charge accumulation, which results in sample charging. While conductive substrates tend to lose accumulated charge quickly, less conductive substrates tend to keep their charge and the resulting electric field causes problems during sample processing.
It is customary to coat vulnerable substrates with a thin layer of a suitable metal in order to obtain a high conductivity surface layer capable of effective charge dissipation. Both the deposition and the subsequent removal of such metallic charge dissipation films is an added complication. This is especially true when samples need to be processed further and the metallic film needs to be removed. This happens, for instance, when samples are merely observed using scanning electron microscopes (SEMs) before undergoing further processing.
Our group at the University of Glasgow, UK, has been exploring the use of conducting organic polymers for replacing metallic charge dissipation layers. Such materials offer convenience in both application and subsequent removal of electrically conducting thin films. Examples of potential polymers include polyaniline, polythiophenes and polyfluorenes. The ability of thin polythiophene layers, in particular, to dissipate accumulated charge in electron-beam lithography process on wide bandgap semiconductors has been the subject of much of our recent investigations. Wide bandgap semiconductors such as SiC, ZnO and GaN are important members of the compound semiconductor family, so their processing is of vital importance to both researchers and industrial engineers.
Our work has focused on ZnO and GaN, typical wide bandgap semiconductors that display low surface conductivity. The aim of this work was to create dense periodic nano-patterns in hydrogen silsesquioxane (HSQ) negative type e-beam resist, so that passive photonic devices could be fabricated in these semiconductors by a subsequent dry etch process. We used a commercially available 2.5 percent water-based dispersion of poly(2,3- dihydrothieno-1,4-dioxin)-poly(styrenesulfonate), i.e. PEDOT:PSS from Sigma-Aldrich. The high electrical conductivity and good oxidation resistance of these polymer films make them suitable for electromagnetic shielding and noise suppression applications.
The optical transmission spectrum of the polythiophene film deposited on a float glass substrate reveals a featureless transmission curve. The polymer film, therefore, possesses high transparency throughout the visible spectrum and even into the near-IR and near-UV regions. In addition, values of extinction coefficient for a thin polythiophene film are negligible in a wide range of wavelengths, including the visible light spectrum, as experimentally determined with the use of a rotating analyzer ellipsometer. The optical transparency of polythiophene charge dissipation layers makes it easy to see the sample surface and perform any alignment operations required to align patterns relative to pre-existing device features.
Our process is simple, inexpensive and can be used with any wide bandgap semiconductor or dielectric material. However, it is described here in the context of electronbeam lithography exposure of dense and high-resolution patterns in hydrogen silsesquioxane (HSQ) negative type resist deposited on bulk ZnO and GaN/AlN-on sapphire substrates.
Both scanning electron microscope inspection and electron-beam lithography patterning of ZnO and GaN face difficulties because these materials are not able to efficiently dissipate the charge that accumulates during such processes. Consequently it is common practice to perform e-beam lithography of wide bandgap semiconductors with a thin conducting metal layer, usually aluminium or gold, deposited on top of the e-beam resist.
Furthermore, the processing of ZnO is difficult due to the fact that it is an amphoteric oxide, and is thus easily attacked by both acids and bases – typically used for the removal of metal films.
Use of conducting organic polymers in place of metallic films does not involve any special resist preparation steps. Processing involves spin-coating of a conductive polymer (PSS:PEDOT) on top of a HSQ-coated sample, electronbeam writing of dense patterns in the resist, removal of the PSS:PEDOT layer and, finally, development of the exposed HSQ e-beam resist. A layer of PEDOT:PSS polymer, spin-coated on a glass cover-slip, was investigated with a Hitachi S4700 scanning electron microscope at a moderate accelerating voltage of 5 kV, without any additional sputter-coated conductive layer. It is seen in Figure 1a that PSS:PEDOT appears to form a porous film, which contributes to the ease of its removal with water. A schematic diagram showing the use of commercially available PEDOT:PSS conductive polymer to dissipate charge in electron-beam lithography, for both bulk ZnO and epitaxial GaN/AlN/sapphire samples, is shown in Figure 1b.
Fig. 1. PSS:PEDOT conductive polymer used for an advanced micro- and
nano-fabrication processes: a) scanning electron microscope investigation
of a polymer surface, after spin-coating deposition; b) schematic
presentation of ZnO and GaN samples used in the experiment, where
HSQ e-beam resist was further patterned by electron-beam lithography
Comparison of the experimental results for the ZnO sample appears in Figure 2. The fabricated nano-patterns included a 50 μm x 10 μm area of W1 (one row of holes removed) and W3 (three rows of holes removed) photonic crystal (PhC) waveguides with a triangular lattice of holes (periodicity of 550 nm, designed hole diameter of 440 nm). Numerical data in Figure 2a include three different cases, where no electron dissipation layer, a 40-nm-thick aluminium layer and a 100-nm-thick conductive polymer layer were used on top of the HSQ resist. Thus, conventional aluminium and the proposed polymer approach were compared and good agreement between these results is reported, whilst the new method considerably simplifies sample processing.
Fig. 2. Results of electron-beam lithography on bulk ZnO samples: a) linear approximation of measured hole diameter as a function of exposure dose, for three investigated cases; b) SEM micrographs of a photonic crystal lattice fabricated in HSQ resist on bulk ZnO, for an exposure without any electron dissipation layer (top) and an exposure with 100-nm-thick PSS:PEDOT layer used (bottom).
Scanning electron microscope observations of the resulting photonic crystal patterns, exposed with the same dose of 474 μC/cm2 are shown in Figure 2b at two different magnifications of 2k and 70k. For pure HSQ resist without a charge dissipation layer (Figure 2b, top) severe overexposure of the PhC pattern was observed. Despite properly defined holes on the edges of an array, the middle part of the PhC lattice exhibited signs of a strong proximity effect, which is indicated by a decrease in SEM observation contrast. When a conductive polymer layer was used sharply defined holes were obtained within highly uniform photonic crystal lattices, as additionally indicated by high contrast SEM micrographs (Figure 2b, bottom).
After sample processing, the spin-coated conducting polymer may be easily removed due to its solubility in water, which makes it a perfect solution for the processing of amphoteric oxide samples, such as ZnO. GaN processing also benefits from the use of a polymer dissipation layer due to the extended exposure range and the avoidance of dense pattern overexposure in HSQ. The use of PEDOT:PSS coating, therefore, makes e-beam processing of ZnO and GaN sample much simpler, quicker, and less expensive.
This benefit is, of course, not limited to just these two semiconductors but may be extended to the processing of other semiconductors and dielectric materials. Although not described here, our work shows that FIB-based processing can also similarly benefit from the application of conducting polymers.
The authors would like to thank the technical staff of the James Watt Nanofabrication Centre and the Kelvin Nanocharacterisation Centre at the University of Glasgow, United Kingdom. They would also like to thank Szymon Lis (Wroclaw University of Technology, Poland) for ellipsometer measurements.
© 2011 Angel Business Communications. Permission required.
R. Dylewicz et al. Electron. Lett. 46 1025 (2010)
R. Dylewicz et al. J. Vac. Sci. Technol. B 28 817 (2010)