Etching of silicon carbide for device fabrication and through via-hole formation

We have investigated the etching of SiC using inductively-coupled-plasma reactive ion etching with SF₆-based and Cl₂-based gas mixtures. Etch rates have been investigated as functions of bias voltage, ICP coil power, and chamber pressure. It will be shown, for the first time, that SiC surfaces etched in Cl₂-based plasmas yield better surface electrical characteristics than those etched in SF₆-based plasmas. We have also achieved SiC etch rates in excess of 1 μm/min which are suitable for micro-machining and via-hole applications. Through via-holes obtained in 140 μm thick SiC at an effective etch rate of 824 nm/min have been achieved. To the best of our knowledge, to date, this is the highest effective etch rate for a through via-hole etched with a masking process compatible with microelectronic fabrication.     

Plasma chemistries for high density plasma etching of SiC

A variety of different plasma chemistries, including SF₆, Cl₂, ICl, 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 4500?-min⁻¹ were obtained for SF₆ plasmas, while much lower rates (≤800?·min⁻¹) 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 ≥10 μm in the SiC. A micromachining process (sequential etch/deposition steps) designed for Si produces relatively low etch rates (<2,000?-min⁻¹) for SiC.     

Inductively coupled plasma reactive ion etching of SiC single crystals using NF<Subscript>3</Subscript>-based gas mixtures

Inductively coupled plasma reactive ion etching of SiC single crystals using NF₃-based gas mixtures was investigated. Mesas with smooth surfaces and vertical sidewalls were obtained, with a maximum etch rate of about 400 nm/min. Effects of CH₄ and O₂ addition to the NF₃ gas and the crystalline quality of substrates were studied during the SiC dry etching using various masks. Selectivity of the photoresist (PR) mask improved from about 0.2 to about 0.4 by the addition of 30% CH₄ during the RIE, although the etch rate decreased by 50–70%. Results also indicated that the substrate quality does not significantly affect the etch results.     

Remote plasma etching of titanium nitride using NF3/argon and chlorine mixtures for chamber clean applications

To avoid plasma induced erosion of chamber hardware, the application of remote plasma sources to activate the etch gases was introduced. We present results on the etch behaviour of titanium nitride (TiN) using mixtures of NF₃, Cl₂ and argon. The gas mixture was excited in a remote plasma source and then routed through a reaction chamber to study the etch behaviour of TiN samples which simulate the situation at the chamber walls. The dependency of the TiN etch rate on temperature, gas flow, composition and pressure was examined. While the temperature (studied in the range 25-300 °C) turned out to be the most sensitive parameter, the general etch rate was mainly dependent on the availability of atomic fluorine. Etch products and NF₃/Cl₂ dissociation have been monitored by quadrupole mass spectrometry and infrared spectroscopy. While NF₃ showed a high decomposition up to 96%, chlorine decomposition was not observed. However the addition of chlorine increased the etch rates up to 260% in the low pressure/low temperature regime. Surface effects of chlorine addition are indicated by X-Ray Photoelectron Spectrometry and REM surface analysis.     

Etching of <Superscript>10</Superscript>Boron with SF<Subscript>6</Subscript>-based Electron Cyclotron Resonance Plasmas for Pillar-Structured Thermal Neutron Detectors

Isotopically enriched ¹⁰boron for use in pillar-structured neutron detectors was successfully etched in an electron cyclotron resonance (ECR) plasma using SF₆-based plasmas. The effects of radio frequency (RF) power, ECR power, gas flow rate, H₂ and O₂ incorporation into the plasma, and gas mixture ratios were examined. Etch rates up to approximately 1.35 μm/min were realized. In addition, etch morphology was examined, and the final shape of ¹⁰boron-coated pillars could be controlled through the etch gas mixture utilized. Selectivity to the underlying Si structure was apparent from scanning electron microscopy (SEM) micrographs of completed etches.     

Plasma and wet chemical etching of In0.5Ga0.5P

Dry and wet chemical etching of epitaxial In_{0, 5}Ga_0.5P layers grown on GaAs substrates by gas-source molecular beam epitaxy have been investigated. For chlorine-based dry etch mixtures (PCl₃/Ar or CC1₂F₂/Ar) the etching rate of InGaP increases linearly with dc self-bias on the sample, whereas CH₄/H₂-based mixtures produce slower etch rates. Selectivities of ≥500 for etching GaAs over InGaP are obtained under low bias conditions with PCl₃/Ar, but the surface morphologies of InGaP are rough. Both CC1₂F₂/Ar and CH₄/H₂/Ar mixtures produce smooth surface morphologies and good (≥10) selectivities for etching GaAs over InGaP. The wet chemical etching rates of InGaP in H₃PO₄:HC1:H₂O mixtures has been systemically measured as a function of etch formulation and are most rapid (∼1 μn · min⁻¹) for high HCl compositions. The etch rate,R, in a 1:1:1 mixture is thermally activated of the formR ∝ , whereE _a = 11.25 kCal · mole⁻¹. This is consistent with the etching being reaction-limited at the surface. This etch mixture is selective for InGaP over GaAs.     

Efficient Outdiffusion of Hydrogen from Mg-Doped Nitrides by NF<Subscript>3</Subscript> Annealing

High concentration (more than 1 × 10¹⁸ cm⁻³) of hydrogen atoms remaining in Mg-doped GaN epitaxial layers grown by metalorganic chemical vapor deposition even after conventional annealing in N₂ ambient could induce degradation in GaN-based devices containing Mg-doped layers. In this study, by annealing Mg-doped nitrides in NF₃ ambient, we successfully reduced residual hydrogen below mid-10¹⁷ cm⁻³, which is much smaller than by N₂ annealing. NF₃ annealing enhances outdiffusion of hydrogen from the bulk, which is possibly because the nitrogen and fluorine radicals decomposed from NF₃ accelerate desorption of hydrogen adatoms from the surface. The proposed method for Mg activation would improve the reliability of GaN-based light-emitting diodes and laser diodes.     

Formation technology of through metallized holes to sources of high-power GaN/SiC high electron mobility transistors

The technology of through metallized holes to sources of high-power GaN/SiC high electron mobility transistors is studied. The dependences of the reactive ion etch rate of SiC in the inductively coupled plasma discharge on the pressure of the SF₆/O₂/Ar gas mixture (5–40 mTorr), the high-frequency power applied to the bottom electrode (200–300 W), the working gas flow ratio (5 : 1 : (0–10)), and the bottom electrode temperatures (5–50°C) are studied. Based on these dependences, the hole etching process on 76-mm-diameter SiC substrates 50 and 100 μm thick is developed. The process features smooth etched-surface morphology, a high rate (1 μm/min), and low high-frequency power deposited into the inductively coupled plasma discharge (1000 W). The developed process of hole etching in SiC substrates is characterized by the selectivity coefficient S = 12 and the anisotropy coefficient A = 13. Films based on NiB are recommended as masks for etching through holes into SiC substrates. The processes of through-hole metallization by the electrochemical deposition of Ni and Au layers are developed.     

Surface contamination and damage from CF4 and SF6 reactive ion etching of silicon oxide on gallium arsenide

Two reactive ion etchants, CF₄ and SF₆, have been compared in terms of plasma characteristics, silicon oxide etch characteristics, extent of RIE damage, and formation of barrier layers on a GaAs surface after oxide etch. It was found that higher etch rates with lower plasma-induced dc bias can be achieved with SF₆ plasma relative to CF₄ plasma and that this correlates with higher atomic fluorine concentration in SF₆ plasma. RIE damage, measured by loss of sheet conductance in a thin highly-doped GaAs layer, could be modelled as a region of deep acceptors at a high concentration in the conductive layer. By relating the sheet conductance change to the modelled damaged layer thickness, it was found that the RIE-damaged thickness from both CF₄ and SF₆ plasmas had the same linear relation to plasma dc bias. Barriers to subsequent GaAs RIE were created during oxide overetch at the GaAs surface. The barriers were identified by XPS as ∼20 A of GaF₃ for CF₄ plasma and ∼30 A of GaF₃ on top of As_xS_y for SF₆ plasma. Ellipsometry was used to routinely determine the presence or absence of the barriers which could be removed in dilute ammonia.     

Etching of 4H-SiC using a NF<Subscript>3</Subscript> inductively coupled plasma

Etching of silicon carbide (SiC) was conducted in a NF₃/CH₄ inductively coupled plasma (ICP) at low pressure. The etch responses examined include the etch rate, surface roughness, and profile angle. For the variations in the source power, the direct-current (DC) bias strongly affected the etch rate. The profile angle varied inconsistently with the bias power. It was commonly observed without regard to the pressure level that, at lower gas ratios, the surface roughness was inversely related to the DC bias. At higher gas ratios, the surface roughness seemed to be dominated by surface reactions. In estimating etch mechanisms, the DC bias played an important role in qualitatively separating chemical and physical effects.     

Boron diffusion into 6H-SiC through graphite mask

Experimental results are reported of selective diffusion of boron in 6H−SiC. Photoluminescence spectroscopy scanning electron microscopy cathodoluminescence imaging, secondary-ion mass spectroscopy optical microscopy, and stain-groove technique were used to characterize the selectively doped regions fabricated by diffusion from the vapor phase through a graphite mask. Local p-doped regions of dimensions down to ∼20 μm in diameter were formed on an n-type substrate using the graphite mask. Maximum concentration of boron atoms at the surface, obtained by SIMS, varied from 3×10¹⁹ cm⁻³ to 6×10¹⁹ cm⁻³, depending upon the temperature of diffusion, while the p-n junction depth measured by the stain-groove technique varied from 0.5 μm to 1.2 μm. Planar p-n junction diodes fabricated on the diffused regions exhibited good rectification characteristics with a breakdown voltage of about 1000 V.     

Deep dry etching of borosilicate glass using SF6 and SF6/Ar inductively coupled plasmas

Deep reactive ion etching (DRIE) of borosilicate glass was carried out using SF₆ and SF₆/Ar plasmas in an inductively coupled plasma (ICP) reactor. Electroplated Ni on Cu (≅50 nm)/Cr (≅100 nm)/glass structure using patterned SU-8 photoresist mask with a line spacing of 12-15 μm was used as a hard-mask for plasma etching. Plasma etching of borosilicate glass was performed by varying the various process parameters such as the gas chemistry, the gas flow ratio, the top electrode power, and the dc self-bias voltage (V_dc). In the case of using SF₆ gas only, the profiles of the etched channel showed the undercut below the Ni hard-mask due to a chemical etching and the microtrenching at the bottom of the etched channel. An optimized process using the SF₆ plasmas showed the glass etch rate of ≅750 nm/min. The addition of the Ar gas to the SF₆ gas removed the undercut and microtrenching but decreased the etch rate to ≅540 nm/min. The increasing and decreasing time-dependent etch rates with the etch depth in the SF₆ (200 sccm) and SF₆(60%)/Ar(40%) plasmas, respectively, were ascribed to the different ion-to-neutral flux ratios leading to the different etch process regime.     

Reactive ion etching of SiC using C2F6/O2 inductively coupled plasma

The inductively coupled plasma-reactive ion etching (ICP-RIE) of SiC single crystals using the C₂F₆/O₂ gas mixture was investigated. It was observed that the etch rate increased as the ICP power and bias power increased. With increasing sample-coil distance, O₂ concentration, and chamber pressure, the etch rate initially increased, reached a maximum, and then decreased. Mesas with smooth surfaces (roughness ≤1 nm) and vertical sidewalls (∼85°) were obtained at low bias conditions with a reasonable etch rate of about 100 nm/min. A maximum etch rate of 300 nm/min could be obtained by etching at high bias conditions (≥300 V), in which case rough surfaces and the trenched sidewall base were observed. The trenching effect could be suppressed by etching the samples on anodized Al plates, although mesas with sloped (60–70°) sidewalls were obtained. Results of various surface characterization indicated little contamination and damage on the etched SiC surfaces.     

Kinetic Studies of the Interfacial Reaction of the Ba2YCu3O6+x Superconductor with a CeO2 Buffer

Interfacial reactions between the Ba₂YCu₃O_6+x superconductor and the CeO₂ buffer layers employed in coated conductors have been modeled experimentally by investigating the kinetics of the reaction between Ba₂YCu₃O_6+x films and CeO₂ substrates. At 810°C, the Ba₂YCu₃O_6+x -CeO₂ join within the BaO-Y₂O₃-CeO₂-CuO _x quaternary system is nonbinary, thereby establishing the phase diagram topology that governs the Ba₂YCu₃O_6+x /CeO₂ reaction. At a mole ratio of Ba₂YCu₃O_6+x :CeO₂ of 40:60, a phase boundary was found to separate two four-phase regions. On the Ba₂YCu₃O_6+x -rich side of the join, the four-phase region consists of Ba₂YCu₃O_{6 +x} , Ba(Ce_1−z Y _z )O_3−x , BaY₂CuO₅, and CuO _x ; on the CeO₂ rich side, the four phases were determined to be Ba(Ce_1−z Y _z ) O_3−x , BaY₂CuO₅, CuO _x and CeO₂. The Ba₂YCu₃O_6+x /CeO₂ reaction is limited by solid-state diffusion, and the reaction kinetics obey the parabolic rule, x = Kt ^1/2, where x = thickness of the reaction layer, t = time, and K = a constant related to the rate constant; K was determined to be 1.6 × 10⁻³ μm/s^1/2 at 790°C and 4.7 × 10⁻³ μm/s^1/2 at 830°C. The activation energy for the reaction was determined to be E _act = 2.67 × 10⁵ J/mol using the Arrhenius equation.     

Effects of nitrogen trifluoride on the growth and properties of plasma-enhanced chemical-vapor-deposited diamond-like carbon films

The effects of nitrogen trifluoride (NF₃) on the growth and properties of plasma-enhanced chemical-vapor-deposited diamond-like carbon (DLC) films were investigated. The addition of NF₃ increases the deposition rate of DLC film due presumably to the removal of activated hydrogen species by the fluorine radical (F⁻). Diamond-like carbon films deposited in a methane/NF₃ mixture have a higher refractive index, a lower bulk resistivity, and a lower optical bandgap compared to films deposited in pure methane due to a lower hydrogen content in the films. Moreover, the bulk resistivity of methane/NF₃ DLC films remains constant for annealing temperatures below 400°C. Thus, DLC films deposited with NF₃ are more stable than DLC films deposited without NF₃.     

Characteristics of germanium dry etching using inductively coupled SF6 plasma

Inductively coupled SF₆ plasma etching of germanium (Ge) was investigated at different inductively coupled plasma (ICP) power levels, the SF₆ flow rate, and the working pressure. The etch rate of Ge increases from 1007 to 2447 nm/min as the SF₆ flow rate increases from 10 to 60 sccm. Also, the etch rate of Ge increases from 265 to 1007 nm/min as the ICP power level increases from 100 to 400 W whereas the etch rate of Ge decreases from 552 to 295 nm/min as the working pressure increases from 5 to 20 mTorr. The etch profile is isotropic. As SF₆ flow, ICP power and working pressure decrease the surface roughness decreases. Optical emission spectroscopy was used to examine the gas phase species in the plasma, and emission from excited atomic S and F has been identified. Composition of the surface due to SF₆ plasmas has been obtained using X-ray photoelectron spectroscopy. Reaction layers on germanium due to inductively coupled SF₆ plasma etching are found to be a thin, layer with of G–-S, Ge–F and Ge–O bonded species.     

Microstructure coarsening during static annealing of 60Sn40Pb solder joints: III intermetallic compound growth kientics

The growth of the total (Cu₃Sn+Cu₆Sn) intermetallic compound layer in Cu-60Sn40Pb solder joints during static annealing at 50°C to 150°C was described by the equation hi=h_o+A_o exp(−Q_a/RT)t^p with h_o=0–0.3 μm, p=0.38–0.70, A_o=1.9×10⁻⁴–3.4×10⁻⁴ m/s^p, and Q_a=25.5–30.9 kJ/mole. These constants are within the range of those obtained by others and give values of D_o and Q which are in reasonable accord with those for the diffusion coefficients in Cu₃Sn and Cu₆Sn₅ determined in diffusion couples. The deviation of the values of the time exponent p from the ideal of 0.5 for diffusion growth may be due to inaccuracies or errors pertaining to the measured thickness (especially h_o) and the complex nature of the diffusion process.     

Synthesis of metastable carbon-silicon-nitrogen compounds by ion implantation

The feasibility of carbon-silicon nitride formation (β−Si_1.5C_1.5N₄, the homologue of equilibrium β−Si₃N₄ or hypothetical β−C₃N₄) has been investigated by high dose N⁺ implantation into polycrystalline β−SiC (cubic phase). Thin films were formed using 100 keV implantations with varying ion doses and target temperatures. X-ray diffraction with a position-sensitive detector and cross-sectional transmission electron microscopy revealed that the as-implanted surfaces contained ∼0.1 μm thick buried amorphous layers. Rutherford backscattering spectroscopy showed that the peak concentration of nitrogen saturated up to approximately 54 at.% with increasing doses, suggesting formation of a new phase.     

In-situ plasma cleaning of stainless steel III-V MOCVD growth systems

A plasma enhanced, in-situ, dry etching process for the cleaning of stainless steel III-V Metal Organic Chemical Vapor Deposition growth systems was investigated as a function of etchant gas, flow rate, electrode configuration, power density and plasma frequency. The plasma enhanced etching process was investigated using Ar, CH₄ (5% in H₂), CCl₂F₂ (Freon 12)/Ar and Cl₂/Ar plasmas with flows varying from 5 to 25 seem. The plasma was excited using three electrode configurations, and two radio frequency generators (90–460 KHz and 13.56 MHz), singly and in combination. The plasma power was varied over the range from 200 to 700 Watts (∼0.2W/cm² – 0.7W/cm²). The etching rates of GaAs, InP, As, and Mo were measured using a weight difference method. The Cl₂/Ar plasmas exhibited etching rates typically 5 to 10 times greater than that of CCl₂F₂ plasmas, which in turn is several times greater than that of the other etchant gases investigated. At 400 W, elemental As etch rates, as high as ∼180μm/hr and ∼20μm/hr were achieved using Cl₂ and CCl₂F₂ plasmas, respectively. InP/GaAs etch rates using Cl₂ were ∼30μm/hr and using CCl₂F₂ were ∼7μm/hr. Plasma characteristics and etch rate measurements are reported. The in-situ process investigated is a safe, cost effective and an efficient method for increasing reactor uptime.     

Assessment of Tight Pixel Geometry Influence on Surface Chemistry of Hg1? X Cd X Te Focal Plane Array by Time of Flight–Secondary Ion Mass Spectroscopy: Novel Analytical Methods

Factors such as lateral control of surface chemistry, reproducible etch depths and widths, and elimination of cross-contamination are vital to achieving world-class Hg_1-X Cd _X Te focal plane infrared detector arrays. Raytheon Vision Systems (RVS) has made significant progress toward a greater understanding of these factors. The current investigation applies time of flight–secondary ion mass spectroscopy (TOF-SIMS) to assess the manner in which Hg_1-X Cd _X Te surface chemistry is influenced by three different critical physical-chemical factors: (1) process chemistry, including baseline and four alternatives; (2) patterned photoresist format; and (3) etch geometry, including aspect ratio, trench depth, trench width, and unit cell spacing. The first of two patterned photoresist formats consisted of an array having 15-μm unit cells with 5-μm-wide and either 3.5-μm- or 6-μm-deep trenches. The second format consisted of a special diagnostic array of parallel dry-etched stripes having various permutations of trench depths (6 μm and 10 μm), trench widths (3 μm and 5 μm), and trench-to-trench separations (3 μm, 5 μm, and 20 μm). The surface chemistry results relative to etch-depth/width ratios, exposed Hg_1−X Cd _X Te area (etch widths), and trench separation distances show that these parameters have a measurable influence on cross-contaminant abundance and type and on the relative ranking of the process cleaning efficacies. Novel analytical methods for using TOF-SIMS data to qualitatively assess cleaning efficacy, geometry-dependent surface species distributions, the polar/hydrophilic and nonpolar/hydrophobic nature of the processed surface, and the dependence of cleaning efficacy on surface chemistry are also discussed. These methods are intended for use in a variety of studies.