Publications

We are developing a method of constructing compact, three-dimensional photonics systems consisting of optical elements, e.g., lenses and mirrors, photo-detectors, and light sources, e.g., VCSELS or circular-grating lasers. These optical components, both active and passive, are mounted on a lithographically prepared silicon substrate. We refer to the substrate as a micro-optical table (MOT) in analogy with the macroscopic version routinely used in optics laboratories. The MOT is a zero-alignment, microscopic optical-system concept. The position of each optical element relative to other optical elements on the MOT is determined in the layout of the MOT photomask. Each optical element fits into a slot etched in the silicon MOT. The slots are etched using a high-aspect-ratio silicon etching (HARSE) process. Additional positioning features in each slot's cross-section and complementary features on each optical element permit accurate placement of that element's aperture relative to the MOT substrate. In this paper we present the results of the first fabrication and micro-assembly experiments of a silicon-wafer based MOT. Based on these experiments, estimates of position accuracy are reported. We also report on progress in fabrication of lens elements in a hybrid sol-gel material (HSGM). Diffractive optical elements have been patterned in a 13-micron thick HSGM layer on a 150-micron thick soda-lime glass substrate. The measured ms surface roughness was 20 nm. Finally, we describe modeling of MOT systems using non-sequential ray tracing (NSRT).

GaN Schottky diodes were exposed to N₂ or H₂ Inductively Coupled Plasmas prior to deposition of the rectifying contact. Subsequent annealing, wet photochemical etching or (NH₄)₂S surface passivation treatments were examined for their effect on diode current- voltage characteristics. We found that either annealing at 750 °C under N₂, or removal of ~500-600 Å of the surface essentially restored the initial I-V characteristics. There was no measurable improvement in the plasma-exposed diode behavior with (NH₄)₂S treatments.

The effect of Inductively Coupled Plasma H{sub 2} or Ar discharges on the breakdown voltage of p-GaN diodes was measured over a range of ion energies and fluxes. The main effect of plasma exposure is a decrease in net acceptor concentration to depths of 400-550{angstrom}. At high ion fluxes or energies there can be type conversion of the initially p-GaN surface. Post etch annealing at 900 C restores the initial conductivity.

Abstract not provided.

Anisotropic, smooth etching of the group-III nitrides has been reported at relatively high rates in high-density plasma etch systems. However, such etch results are often obtained under high de-bias andlor high plasma flux conditions where plasma induced damage can be significant. Despite the fact that the group-III nitrides have higher bonding energies than more conventional III-V compounds, plasma-induced etch damage is still a concern. Attempts to minimize such damage by reducing the ion energy or increasing the chemical activity in the plasma often result in a loss of etch rate or anisotropy which significantly limits critical dimensions and reduces the utility of the process for device applications requiring vertical etch profiles. It is therefore necessary to develop plasma etch processes which couple anisotropy for critical dimension and sidewall profile control and high etch rates with low-damage for optimum device performance. In this study we report changes in sheet resistance and contact resistance for n- and p-type GaN samples exposed to an Ar inductively coupled plasma (ICP). In general, plasma-induced damage was more sensitive to ion bombardment energies as compared to plasma flux. In addition, p-GaN was typically more sensitive to plasma-induced damage as compared to n-GaN.

Four different F{sub 2}-based gases (SF{sub 6}, NF{sub 3}, PF{sub 5}, and BF{sub 3}) were examined for high rate Inductively Coupled Plasma etching of Si. Etch rates up to {approximately}8 {micro}m/min were achieved with pure SF{sub 6} discharges at high source power (1500W) and pressure (35mTorr). A direct comparison of the four feedstock gases under the same plasma conditions showed the Si etch rate to increase in the order BF{sub 3} < NF{sub 3} < PF{sub 5} < SF{sub 6}. This is in good correlation with the average bond energies of the gases, except for NF{sub 3}, which is the least strongly bound. Optical emission spectroscopy showed that the ICP source efficiently dissociated NF{sub 3}, but the etched Si surface morphologies were significantly worse with this gas than with the other 3 gases.

The role of additive noble gases He, Ar and Xe to C&based Inductively Coupled Plasmas for etching of GaN, AIN and InN were examined. The etch rates were a strong function of chlorine concentration, rf chuck power and ICP source power. The highest etch rates for InN were obtained with C12/Xe, while the highest rates for AIN and GaN were obtained with C12/He. Efficient breaking of the 111-nitrogen bond is crucial for attaining high etch rates. The InN etching was dominated by physical sputtering, in contrast to GaN and AIN. In the latter cases, the etch rates were limited by initial breaking of the III-nitrogen bond. Maximum selectivities of -80 for InN to GaN and InN to AIN were obtained.

Sputter-deposited W-based contacts on p-GaN (N{sub A} {approximately} 10{sup 18} cm{sup {minus}3}) display non-ohmic behavior independent of annealing temperature when measured at 25 C. The transition to ohmic behavior occurs above {approximately} 250 C as more of the acceptors become ionized. The optimum annealing temperature is {approximately} 700 C under these conditions. These contacts are much more thermally stable than the conventional Ni/Au metallization, which shows a severely degraded morphology even at 700 C. W-based contacts may be ohmic as-deposited on very heavily doped n-GaN, and the specific contact resistance improves with annealing up to {approximately} 900 C.

GaN implanted with donor(Si, S, Se, Te) or acceptor (Be, Mg, C) species was annealed at 900-1500°C using AlN encapsulation. No redistribution was measured by SIMS for any of the dopants and effective diffusion coefficients are ≤2×10-13 cm2 · s-1 at 1400°C, except Be, which displays damage-enhanced diffusion at 900°C and is immobile once the point defect concentration is removed. Activation efficiency of ∼90% is obtained for Si at 1400°C. TEM of the implanted material shows a strong reduction in lattice disorder at 1400-1500°C compared to previous results at 1100°C. There is minimal interaction of the sputtered AlN with GaN under our conditions, and it is readily removed selectively with KOH.

A comparison of KOH, NaOH and AZ400K solutions for UV photo-assisted etching of undoped and n+ GaN is discussed. The etching is diffusion-limited (Ea < 6kCal·mol-1) under all conditions and is significantly faster with bias applied to the sample during light exposure. No etching of InN was observed, due to the very high n-type background doping (> 1020cm-3) in the material.

Proton implantation in GaN is found to reduce the free carrier density through two mechanisms - first, by creating electron and hole traps at around Ec-0.8eV and Ev+0.9eV that lead to compensation in both n- and p-type material, and second, by leading to formation of (AH)O complexes, where A is any acceptor (Mg, Ca, Zn, Be, Cd). The former mechanism is usefid in creating high resistivity regions for device isolation, whereas the latter produces unintentional acceptor passivation that is detrimental to device performance. The strong affinity of hydrogen for acceptors leads to markedly different redistribution behavior for implanted in n- and p-GaN due to the chemical reaction to form neutral complexes in the latter. The acceptors may be reactivated by simple annealing at 2600{degrees}C, or by electron injection at 25-150{degrees}C that produces debonding of the (AH) centers. Implanted hydrogen is also strongly attracted to regions of strain in heterostructure samples during annealing, leading to pile-up at epi-epi and epi-substrate interfaces. II? spectroscopy shows that implanted hydrogen also decorates VG, defects in undoped and n-GaN.

The role of the inert gas additive (He, Ar, Xe) to C12 Inductively Coupled Plasmas for dry etching of GaAs and GaSb was examined through the effect on etch rate, surface roughness and near-surface stoichiometry. The etch rates for both materials go through a maximum with Clz 0/0 in each type of discharge (C12/'He, C12/Ar, C12/Xc), reflecting the need to have efficient ion-assisted resorption of the etch products. Etch yields initially increase strongly with source power as the chlorine neutral density increases, but decrease again at high powers as the etching becomes reactant-limited. The etched surfaces are generally smoother with Ax or Xe addition, and maintain their stoichiometry.

The effects of the additive noble gases He, Ar and Xe on chlorine-based Inductively Coupled Plasma etching of InP, InSb, InGaP and InGaAs were studied as a function of source power, chuck power and discharge composition. The etch rates of all materials with C12/He and C12/Xe are greater than with C12/Ar. Etch rates in excess of 4.8 pndmin for InP and InSb with C12/He or C12/Xe, 0.9 pndmin for InGaP with C12/Xe, and 3.8 prdmin for InGaAs with Clz/Xe were obtained at 750 W ICP power, 250 W rf power, - 1570 C12 and 5 mTorr. All three plasma chemistries produced smooth morphologies for the etched InGaP surfaces, while the etched surface of InP showed rough morphology under all conditions.

Patterning the group-IH nitrides has been challenging due to their strong bond energies and relatively inert chemical nature as compared to other compound semiconductors. Plasma etch processes have been used almost exclusively to pattern these films. The use of high-density plasma etch systems, including inductively coupled plasmas (ICP), has resulted in relatively high etch rates (often greater than 1.0 pmhnin) with anisotropic profiles and smooth etch morphologies. However, the etch mechanism is often dominated by high ion bombardment energies which can minimize etch selectivity. The use of an ICP-generated BCl~/C12 pkyma has yielded a highly versatile GaN etch process with rates ranging from 100 to 8000 A/rnin making this plasma chemistry a prime candidate for optimization of etch selectivity. In this study, we will report ICP etch rates and selectivities for GaN, AIN, and InN as a function of BCl~/Clz flow ratios, cathode rf-power, and ICP-source power. GaN:InN and GaN:AIN etch selectivities were typically less than 7:1 and showed the strongest dependence on flow ratio. This trend maybe attributed to faster GaN etch rates observed at higher concentrations of atomic Cl which was monitored using optical emission spectroscopy (OES). ~E~~~~f:~ INTRODUCTION DEC j 4898 Etch selectivi

BC13, with addition of Nz, Ar or Hz, is found to provide smooth anisotropic pattern transfer in GaAs, GaN, GaP, GaSb and AIGriAs under Inductively Coupled Plasma conditions, Maxima in the etch rates for these materials are observed at 33% N2 or 87$'40 Hz (by flow) addition to BC13, whereas Ar addition does not show this behavior. Maximum etch rates are typically much higher for GaAs, Gap, GaSb and AIGaAs (-1,2 @rein) than for GaN (-0.3 ymu'min) due to the higher bond energies of the iatter. The rates decrease at higher pressure, saturate with source power (ion flux) and tend to show maxima with chuck power (ion energy). The etched surfaces remain stoichiometric over abroad range of plasma conditions.

A parametric study of etch rates and surface morphologies of In-containing compound semiconductors (InP, InGaAs, InGaAsP, InAs and AlInAs) obtained by BClj-based Inductively Coupled Plasmas is reported. Etch rates in the range 1,500-3,000 &min. are obtained for all the materials at moderate source powers (500 W), with the rates being a strong function of discharge composition, rf chuck power and pressure. Typical root-mean-square surface roughness of-5 nm were obtained for InP, which is worse than the values obtained for Ga-based materials under the same conditions (-1 run). The near surface of etched samples is typically slightly deficient in the group V element, but the depth of this deficiency is small (a few tens of angstroms).

A parametric study of the etch characteristics of GaN, AIN and InN has been earned out with IC1/Ar and IBr/Ar chemistries in an Inductively Coupled Plasma discharge. The etch rates of InN and AIN were relatively independent of plasma composition, while GaN showed increased etch rates with interhalogen concentration. Etch rates for all materials increased with increasing rf chuck power, indicating that higher ion bombardment energies are more efficient in enhancing sputter resorption of etch products. The etch rates increased for source powers up to 500 W and remained relatively thereafter for all materials, while GaN and InN showed maximum etch rates with increasing pressure. The etched GaN showed extremely smooth surfaces, which were somewhat better with IBr/Ar than with IC1/Ar. Maximum selectivities of- 14 for InN over GaN and >25 for InN over AIN were obtained with both chemistries.

High density plasma etching of GaAs, GaSb and AIGaAs was performed in IC1/Ar and lBr/Ar chemistries using an Inductively Coupled Plasma (ICP) source. GaSb and AlGaAs showed maxima in their etch rates for both plasma chemistries as a function of interhalogen percentage, while GaAs showed increased etch rates with plasma composition in both chemistries. Etch rates of all materials increased substantially with increasing rf chuck power, but rapidly decreased with chamber pressure. Selectivities > 10 for GaAs and GaSb over AlGaAs were obtained in both chemistries. The etched surfaces of GaAs showed smooth morphology, which were somewhat better with IC1/Ar than with IBr/& discharge. Auger Electron Spectroscopy analysis revealed equi-rate of removal of group III and V components or the corresponding etch products, maintaining the stoichiometry of the etched surface.

A parametric study of Inductively Coupled Plasma etching of InP, InSb, InGaP and InGaAs has been carried out in IC1/Ar and IBr/Ar chemistries. Etch rates in excess of 3.1 prrdmin for InP, 3.6 prnh-nin for InSb, 2.3 pm/min for InGaP and 2.2 ~rrdmin for InGaAs were obtained in IBr/Ar plasmas. The ICP etching of In-based materials showed a general tendency: the etch rates increased substantially with increasing the ICP source power and rf chuck power in both chemistries, while they decreased with increasing chamber pressure. The IBr/Ar chemistry typically showed higher etch rates than IC1/Ar, but the etched surface mophologies were fairly poor for both chemistries.

The etch rate of GaN under W-assisted photoelectrochemical conditions in KOH solutions is found to be a strong function of illumination intensity, solution molarity, sample bias and material doping level. At low e-h pair generation rates, grain boundaries are selectively etched, while at higher illumination intensities etch rates for unintentionally doped (n - 3x 10^12Gcm-3) GaN are 2 1000 .min-l. The etching is diffusion limited under our conditions with an activation energy of - 0.8kCal.mol-1. The etched surfaces are rough, but retain their stoichiometry. PEC etching is found to selectively reveal grain boundaries in GaN under low light illumination conditions. At high lamp powers the rates increase with sample temperature and the application of bias to the PEC cell, while they go through a maximum with KOH solution molarity. The etching is diffusion-limited, producing rough surface morphologies that are suitable in a limited number of device fabrication steps. The surfaces however appear to remain relatively close to their stoichiometric composition.

A variety of different plasma chemistries, including SF6, Cl2, IC1 and IBr, have been examined for dry etching of 6H-SiC in high ion density plasma tools (Inductively Coupled Plasma and Electron Cyclotron Resonance). Rates up to 4,500~"min-1 were obtained for SF6 plasmas, while much lower rates (S800~.min-') were achieved with Cl2, ICl and IBr. The F2- based chemistries have poor selectivity for SiC over photoresist masks (typically 0.4-0.5), but Ni masks are more robust, and allow etch depths 210pm in the SiC. A micromachining process (sequential etch/deposition (<2,000Angstrom min-1) for SiC steps) designed for Si produces relatively low etch rates.

A systematic study of the etch characteristics of GaN, AlN and InN has been performed with boron halides- (BI{sub 3} and BBr{sub 3}) and interhalogen- (ICl and IBr) based Inductively Coupled Plasmas. Maximum etch selectivities of -100:1 were achieved for InN over both GaN and AlN in the BI{sub 3} mixtures due to the relatively high volatility of the InN etch products and the lower bond strength of InN. Maximum selectivies of- 14 for InN over GaN and >25 for InN over AlN were obtained with ICl and IBr chemistries. The etched surface morphologies of GaN in these four mixtures are similar or better than those of the control sample.

The role of extended and point defects, and key impurities such as C, O and H, on the electrical and optical properties of GaN is reviewed. Recent progress in the development of high reliability contacts, thermal processing, dry and wet etching techniques, implantation doping and isolation and gate insulator technology is detailed. Finally, the performance of GaN-based electronic and photonic devices such as field effect transistors, UV detectors, laser diodes and light-emitting diodes is covered, along with the influence of process-induced or grown-in defects and impurities on the device physics.

Current and future generations of sophisticated compound semiconductor devices require the ability for submicron scale patterning. The situation is being complicated since some of the new devices are based on a wider diversity of materials to be etched. Conventional IUE (Reactive Ion Etching) has been prevalent across the industry so far, but has limitations for materials with high bond strengths or multiple elements. IrI this paper, we suggest high density plasmas such as ECR (Electron Cyclotron Resonance) and ICP (Inductively Coupled Plasma), for the etching of ternary compound semiconductors (InGaP, AIInP, AlGaP) which are employed for electronic devices like heterojunction bipolar transistors (HBTs) or high electron mobility transistors (HEMTs), and photonic devices such as light-emitting diodes (LEDs) and lasers. High density plasma sources, opeiating at lower pressure, are expected to meet target goals determined in terms of etch rate, surface morphology, surface stoichiometry, selectivity, etc. The etching mechanisms, which are described in this paper, can also be applied to other III-V (GaAs-based, InP-based) as well as III-Nitride since the InGaAIP system shares many of the same properties.

Inductively coupled plasma (ICP) etching characteristics of GaSb and AIGaAsSb have been investigated in BC13/Ar and Clz/Ar plasmas. The etch rates and selectivity between GaSb and AIGaAsSb are reported as functions of plasma chemistry, ICP power, RF self-bias, and chamber pressure. It is found that physical sputtering resorption of the etch products plays a dominant role in BC13/Ar ICP etching, while in Clz/Ar plasma, the chemical reaction dominates the etching. GaSb etch rates exceeding 2 ~rnhnin are achieved in Clz/Ar plasmas with smooth surfaces and anisotropic profiles. In BC13/Ar plasmas, etch rates of 5100 Mmin and 4200 Mmin are obtained for GaSb and AIGaAsSb, respectively. The surfaces of both GaSb and AIGaAsSb etched in BC13/Ar plasmas remain smooth and stoichiometric over the entire range of plasma conditions investigated. This result is attributed to effective removal of etch products by physical sputtering. For a wide range of plasma conditions, the selectivity between GaSb and AIGaAsSb is close to unity, which is desirable for fabricating etched mirrors and gratings for Sb-based mid-IR laser diodes.