Lateral Etch Rate Research Articles

Effects of O2 addition to SiCl4/SiF4 and the thickness of the capping layer on gate recess etching of GaAs-pseudomorphic high electron mobility transistors

The effects of the addition of a small amount of O2 to SiCl4/SiF4 plasma and the thickness of the GaAs capping layer on gate recess etching of GaAs pseudomorphic high electron mobility transistor (p-HEMT) devices have been studied. During gate recessing of GaAs/AlGaAs p-HEMTs using a selective reactive ion etching (RIE) process in SiCl4/SiF4 plasma [see S. K. Murad, N. I. Cameron, P. D. Wang, S. P. Beaumont, and C. D. W. Wilkinson, Microelectron. Eng. 27, 439 (1995)], it was found that the profile of the gate changes dramatically from undercut to nearly vertical with no lateral etching when the thickness of the capping layer was reduced below 40 nm. This vertical profile puts the gate metal too close to the recess edges in devices with a ≤30 nm capping layer. The addition of a very small amount of O2 to SiCl4/SiF4 plasma was seen to increase GaAs etch rates remarkably, while maintaining the high selectivity over AlGaAs. This increase in the etch rate agrees well with optical emission spectroscopic observations which indicate that the Cl emission has increased by more than an order of magnitude for the addition of only 1.5% of O2 to SiCl4/SiF4 plasma. This selective RIE process of SiCl4/SiF4/O2 was applied to gate recess etching of GaAs/AlGaAs/InGaAs p-HEMTs with various capping layer thicknesses. The profile of the T gates changed from nearly vertical (with no undercut or gate offset) to undercut with a lateral etch rate depending on the thickness of the capping layer. The lateral etching rate also strongly depends on the O2 flow. The p-HEMT devices gate recessed using this process (SiCl4/SiF4/O2) exhibited an ft of 120 GHz, gm of 700 mS/mm and a gain of 9.5 dB at 94 GHz.

Simulation and dry etching of field emitter tips in Si

To increase the efficiency of field emission devices, the design of the emitters should be optimized. A simulation program using the finite element method was developed to investigate the effects of emitter geometry on the emission current and voltage for ungated emitters. It was found that the tip-anode spacing and tip radius have the most significant effects on the emitted current. The simulation predicted that a field emitter consisting of a vertical base and a cone shaped tip on the top has a higher enhancement factor and therefore a lower operating voltage. Emitters were fabricated in Si based on the optimized design. Dry etching in CF4 was used to control the slope of the SiO2 mask. The SiO2 sidewall angle decreased from 80° to 51° as pressure was increased from 40 to 300 mTorr. The SiO2 mask profile was also changed substantially by adding CHF3 or O2 to CF4 during etching. An electron cyclotron resonance source was used to generate a Cl2 plasma to etch Si field emitters. SiO2 mask was eroded during etching to control the Si emitter profile. The slope of the Si etch profile increased from 52° to 78° as the SiO2 slope increased from 17° to 45°. Radio frequency and microwave power had little influence on the Si etch profile, but they were adjusted to achieve the desired etch rate. The lateral etch rate of SiO2 decreased at higher pressure, resulting in more vertical Si profiles. Oxygen was also added to the Cl2 plasma and the Si sidewall angle was changed between 47° and 80°. Using the above results, the Si etch profile was controlled and field emitter tips and arrays with different geometries were etched in Si. To comply with the optimum design, emitter tips with two different sidewall angles of 85° and 70° were fabricated by changing the pressure from 5 to 0.5 mTorr during the etch. Arrays of emitter tips with sidewall angle of 80°, 11 μm height, and 2.2 μm basewidth were also fabricated at 10 mTorr with a packing density of 4×106 tip/cm2.

Fabrication of Si field emitters by dry etching and mask erosion

A Cl2 plasma generated by an electron cyclotron resonance source was used to etch field emitter arrays in Si. Compared to wet etching, emitter arrays with sharper emitter tips and higher packing density can be formed by dry etching. Mask erosion was used to control the etch profile in Si. SiO2 masks with different profiles were patterned by wet and dry etching, and the effects of the initial SiO2 masks slope and etch conditions on the resultant Si profile were investigated. The lateral etch rate of SiO2 decreased from 363 to 150 nm/min as the SiO2 slope increased from 17° to 45°. As a result, the slope of the Si etch profile increased from 52° to 78° after dry etching to a depth of 1.8 μm. The ion flux and energy, controlled through coupled microwave and rf power, were used to obtain the desired etch rate and basewidth of the emitters. By increasing the pressure during etching, the lateral etch rate of SiO2 was reduced and more vertical Si profiles were developed. As pressure was increased from 0.5 to 10 mTorr, the lateral etch rate of SiO2 reduced from 414 to 144 nm/min while the vertical Si etch rate did not change significantly. This caused the slope of the Si etch profile to increase from 42° to 69°. Using the above technique, arrays of sharp emitter tips in Si with 2.2 μm basewidth and 11 μm height were fabricated and packing densities up to 1×107 tip/cm2 were achieved.

Development of Polymers in O2 Plasmas: Temperature Effects and Transition to Imperfect Anisotropy

A parametric study of the etching of a photoresist is performed in an O2 microwave multipolar plasma using a trilevel resist system. The etch rate and the anisotropy evolutions are reported as a function of ion energy, oxygen pressure and substrate temperature, which are important parameters likely to affect anisotropy. A transition from isotropic to imperfect anisotropic etching occurs when the ion energy is increased. However, residual lateral etching always subsists at room temperature, even at the lowest oxygen concentrations. Above room temperature, in contrast with ion-induced etching, the spontaneous lateral etch rate increases with temperature, suggesting a thermally activated reaction process. The phenomena are interpreted in terms of oxygen coverage on the polymer surface, with a threshold coverage being required for purely spontaneous chemical etching to occur, and on the assumption of a photon-induced desorption of volatile products proportional to the oxygen coverage.

Ultrathin single-crystalline silicon on quartz (SOQ) by 150 °C wafer bonding

Single-crystalline silicon films with thicknesses as thin as 2000 Å have been prepared on thermally mismatched quartz substrates by a simple wafer-bonding approach. Initial bonding at ≈ 80 °C, storage at room temperature for more than 100 h and multi-temperature (maximum 150 °C) consecutive annealing with a 1 °C min −1 ramping rate have been adopted to strengthen the bond and to prevent debonding at the edge of the bonded pairs during annealing and etching, where thermal shearing and peeling stresses are maximum. Final etching by EDP (ethylenediamine—pyrocatechol—water) effectively reduces the peeling failure of the highly stressed thinned silicon layer, mainly due to a reduced lateral oxide etching rate along the interface. The high carrier mobility in the single-crystalline silicon layer and the transparent and insulating quartz substrate provides a new dimension of freedom in applications.

Damage free and selective RIE of GaAs/AlGaAs in SiCl 4/SiF 4 plasma for MESFET and pseudomorphic HEMT's gate recess etching

An apparently damage free and selective Reactive Ion Etching (RIE) process has been developed using a mixture of SiCl 4/SiF 4 plasma for etching GaAs/AlGaAs. Using a rf power of 9–12 W and hence a low dc bias of ≤35 V, with a flow rate ratio of (2–3):9 sccm of SiCl 4:SiF 4, the selectivities we obtained are extraordinarily high >5000:1. No surface induced damage by this process can be detected. Raman scattering measurements made on a heavily doped GaAs layer showed that the etched surfaces improved on etching. Hall effect measurements on pseudomorphic HEMT material structure showed no detectable modification of the sheet carrier concentration even after 10 minute of continuous bombardment of the active layer. This SRIE has been applied to the recessing of T-shaped gates and the lateral etching rate of these gates could be suppressed to <10 nm/min, resulting in a precise control of the gate off-set and hence effective gate length.

Comparison of advanced plasma sources for etching applications. I. Etching rate, uniformity, and profile control in a helicon and a multiple electron cyclotron resonance source

We have studied the etching performance of two commercially available low pressure, high density plasma sources and their application for the etching of 0.35 μm features in polysilicon films. The two sources are a rf-inductively coupled helicon made by Lucas Labs of Sunnyvale, CA and a multipole electron cyclotron resonance (ECR) source made by Wavemat of Plymouth, MI. The sources are mounted on a dual chamber etching platform to remove platform dependent effects. Performance metrics consist of measuring the polysilicon etching rate, etching rate uniformity, and profile control in HBr gas-phase chemistry. The effect of applied source power, applied rf-bias power, and reactor pressure on the etching rate and uniformity is examined using a response surface experiment. Profile control is determined by examining nested and isolated lines and trenches using oxide mask/polysilicon/oxide structures. In both sources, high uniformity and vertical profiles are obtained at low reactor pressure, high applied source power, and applied rf-bias powers between 50 and 60 W. To decrease the lateral etching rate and increase the anisotropy of the etching process, approximately 3% of O2 is added to the feed-gas. For the helicon, the operating point for best uniformity is at 2.0 mTorr, 2500 W applied source power, and 57 W applied rf-bias power resulting in a measured etching rate of 2340 Å/min and uniformity of ±3.3%(2σ). For the ECR, the operating point for best uniformity is at 2.8 mTorr, 1370 W applied source power, and 60 W applied rf-bias power resulting in a measured etching rate of 2580 Å/min and uniformity of ±1.4%(2σ). Since both sources exhibit remarkably similar performance for the etching of polysilicon films, other factors such as ease of operation, plasma stability, and plasma ignition sequence become relatively more important when deciding which source to use for a particular application.

Low temperature electron cyclotron resonance plasma etching of GaAs, AlGaAs, and GaSb in Cl2/Ar

Sidewall etching of GaAs, AlGaAs, and GaSb in electron cyclotron resonance Cl2/Ar discharges is found to be completely suppressed by cooling the semiconductor sample to −30 °C during the process. Vertical etch rates of ≳1500 Å min−1 at 1 mTorr and −50 V dc bias are obtained for all three materials under conditions where the lateral etch rates are negligible. Ex situ chemical analysis of the sidewall shows substantially increased Cl-containing residue on low temperature etched samples, which can be removed by a 5 min H2 plasma clean-up step. The exploitation of temperature to control undercutting enables use of simpler gas chemistries because there is no need to form a sidewall polymer.

Dry processed, through-wafer via holes for GaAs power devices

The fabrication of through-wafer via holes in GaAs substrates by plasma etching and laser drilling is reported. Using a low pressure (15–20 mTorr), low dc bias (−150 V) Cl2/BCl3 discharge with Cl2-to-BCl3 ratios ≤0.2, we are able to produce narrow (≤30 μm) via holes. This enables the use of a higher density of vias in closer proximity to the active GaAs power devices than can be achieved with normal diameter (100–150 μm) holes. Microwave enhancement of the plasma density using an electron cyclotron resonance source increases the GaAs vertical and lateral etch rates and requires use of low Cl2-to-BCl3 ratios in order to retain the anisotropic nature of the vias. Multiple pass (∼100 per hole) drilling of vias with a Q-switched Nd-YAG frequency doubled (532 nm), 30 mW laser has also successfully produced through-wafer connections. This provides a maskless, versatile method for producting vias customized for a particular wafer, but is less developed than plasma etching.

Electron cyclotron resonance plasma etching of photoresist at cryogenic temperatures

An electron cyclotron resonance oxygen plasma discharge was used to anisotropically etch photoresist at a low substrate temperature (−100 °C). The results of using a lower temperature are seen in a reduction in lateral etch rate, with concomitant improvement in anisotropy. Langmuir probe and flux analysis at the substrate with a quadropole mass spectrometer was used to characterize the plasma stream as a function of the operating conditions. The plasma stream flux was composed of approximately 10 times as many reactive oxygen neutrals as O2+ ions. Etch rate was found to be strongly affected by the ion power density impinging on the substrate. Four mechanisms were identified that may contribute to lateral etching. Lateral etching was observed to decrease by the combined application of rf substrate bias and low substrate temperature.

Simulation of profile evolution in silicon reactive ion etching with re-emission and surface diffusion

This article describes a model that simulates etching profiles in reactive ion etching. In particular, models are developed to explain the significant lateral etch rate that is observed in many etch profiles. The total etch rate is considered to consist of two superimposed components: an ion-assisted rate and a purely ‘‘chemical’’ etch rate, the latter rate being due to etching by radicals in the absence of ion bombardment. The transport of radicals to the evolving interface is studied for two different transport mechanisms: re-emission from the surface and diffusion along the surface. For the case of transport by surface re-emission, a reactive sticking coefficient is defined for the radicals, and a formulation is developed to simulate etching for any value (between zero and unity) that this sticking coefficient may assume. When the sticking coefficient approaches either zero or unity, the method of characteristics is shown to be useful for profile simulation. Transport of radicals by surface diffusion is also investigated, and it is shown that the important dimensionless parameter governing profile evolution is the Damkohler number. The two models are compared to experiments performed on the etching of silicon in a SF6 plasma, and the surface re-emission model is shown to accurately predict the development of etching profiles.

Low-Temperature Microwave Plasma Etching of Crystalline Silicon

Low-temperature microwave plasma etching of crystalline silicons is described. Vertical and lateral etch rates of Si and the selectivities of Si to photoresist are measured as a function of water temperature within a range of -150 to +30° for SF6, and (3) the selectivities become high at low temperatures (e.g., >40 at -90°C and 2.3 Pa). This etching enables highly anisotropic Si etching at a high etch rate and high selectivity with fluoride gases. Less-polymerizing-type gases can provide high etch rates. An etching model of the ion-bombarded surfaces in discussed. The model implies that separate control of the side wall reaction and the horizontal surface reaction is archived by the low-temperature etching.

The Influence of Native Oxide Layers of InP on the Shape of Etching Profiles at Resist Edges

Profile etching experiments made on in solutions of revealed that the profile shape is strongly dependent on the resist orientation and on whether or photoresist is used as a mask. It was found that the local etch rate of the (111)In crystallographic planes can be significantly enhanced near the resist edges. In the extreme case the dissolution rate was found to be controlled by diffusion in solution which resulted in rounded profiles. A model is proposed which can account for this apparently anomalous etching behavior. The anisotropic etching of thin surface layers is essential in this model. Experimental results can be explained if it is assumed that native oxide layers covering the monocrystalline substrates play this role. The local etching kinetics of crystallographic facets and, consequently, the profile shape are strongly determined by the lateral etch rate of these oxide layers.

Profile evolution in the multi-level technique

The oxygen reactive ion etching of sub-half micron features in organic layers produces a variety of profile effects depending on both system parameters and pattern arrangements. Vertical and lateral etch rates of the bottom resist in a tri-level resist system were investigated with particular attention to the trench aspect ratio. In order to properly study the etch behaviour and to evaluate their various facets, a new version of the simulation program ADEPT (Advanced Simulation of Dry-Etching Process Technology) was developed. This now permits inclusion of physical and chemical descriptions of the etch process. Examples of profile phenomena observed in the tri-level technique are compared with corresponding output from the simulation program.

The etching mechanism of titanium polycide in a mixture of SF6 and O2

Reactive ion etching of titanium disilicide films on top of undoped polycrystalline silicon has been investigated in order to determine the etching mechanism of the silicide in a fluorine plasma. Mixtures of SF6 and O2 and of CF4 and O2 were used. Vertical and lateral etch rates have been determined as a function of pressure and the amount of oxygen addition. The density of fluorine atoms and positive ions in the plasma has been determined from optical emission spectroscopy and electric probe measurements, respectively. From these results, the fluxes of fluorine to and away from the sample and the ion flux towards the sample have been calculated. With x-ray photoelectron and Auger electron spectroscopy, the surface constitution after etching has been studied. The TiSi2 layer is etched anisotropically, independent of pressure and the addition of oxygen. A relation between the etch rate and the density of fluorine atoms in the plasma is not found, in contrast with results obtained elsewhere [K. C. Cadien, S. Sivaram, and C. D. Reintsema, J. Vac. Sci. Technol. A 4, 739 (1986)]. From our results it is shown that the etch rate of TiSi2 in a low-pressure fluorine plasma is not limited by the supply of reactive fluorine to the surface. It is limited by ion-induced desorption or ion-induced formation of volatile titanium fluoride compounds. The magnitude of the observed vertical and lateral etch rate of silicon are described by means of the reaction probability of fluorine at the silicon surface and the density of fluorine atoms in the plasma. Anisotropic etching of the polysilicon layer in a SF6/He/O2 plasma occurs when more than 10% oxygen is added. The effect of oxygen on etch rates and anisotropy is due to a surface reaction and not to gas-phase reactions. It is argued that an oxyfluoride layer is formed on the silicon surface that hinders the formation of saturated silicon fluoride species.

Experimental conditions for uniform anisotropic etching of silicon with a microwave electron cyclotron resonance plasma system

The etch rate uniformity of a SF6 +Ar plasma is studied for various discharge geometries using an electron cyclotron resonant microwave plasma disk reactor operating at 2.45 GHz. At a distance of 10 mm below the 9-cm-diam plasma source 2-in. (5-cm) wafers can be etched uniformly in a free-fall diffusion dominated discharge. The effect of the ratio of SF6 to Ar on vertical and lateral etch rates was investigated and correlated with the 704-nm F* emission line intensity and positive ion flux to the surface. Surface contamination and impurity implantation in etched silicon are documented using Auger electron spectroscopy.

Ion energy and anisotropy in microwave plasma etching of polymers

A fresh approach has been applied to study the effect of ion energy on the anisotropy of polymer etching. A combination of microwave plasma excitation and RF biasing of an electrode on which etching substrates were placed, was used to independently control plasma density and the kinetic energy with which ions from the plasma strike the substrate. An empirical relationship is demonstrated which relates the rate of etching for polyimide films to the plasma density and the square root of the ion energy. Observations of lateral and vertical etch rates in a multilayer photoresist system have shown that in O 2 plasmas there is a significant ion assisted component to the etching, and that this component is important even when ions have energies as low as 26eV.

Etch behaviour and lateral etch rates of trilevel resist systems in an rie-system using SOG and Ti intermediate layers

Trilevel resist systems using several spin-on-glasses (SOG) as intermediate layers are compares to systems with SiO or Ti intermediate layers referring to O2-RIE etch behaviour. While the SOG shows different surface damages after the O2 plasma, the SiO and Ti layers are nearly smooth. The lateral etch rates depend on the baking temperature and baking time. Because of the higher resistance of Ti and SiO the ideal profile conditions shift to higher power/pressure ratios.

Application of Emission Spectroscopy for Profile Control during Oxygen RIE of Thick Photoresist

Oxygen reactive ion etching of the planarizing photoresist layer in a three‐layer resist system is described. The vertical and lateral etch rates of the thick resist layer, as well as the degree of isotropy of the resist profile, are reported as functions of RF power and oxygen pressure. The optical emission spectra are recorded and interpreted for oxygen plasmas used in etching, revealing the presence of oxygen atoms (O) and ionized oxygen molecules . It is demonstrated that trends in the degree of isotropy which occur with changes in RF power and oxygen pressure can be predicted by monitoring the emission intensity ratio.

Undercut in a CF4-Based High-Pressure Poly-Si Plasma Etch

The behaviour of the plasma etching of polycrystalline Si (P-doped and undoped) in a CF4-based high-pressure glow discharge is studied as a function of the power density, of the exposed poly-Si surface area (the so called loading effect), of the flow rate of CF3Cl and of the flow rate of N2O and O2 as additive gases. Partially in contrast with the current results reported in the literature, the experimental data here presented show that P-doped poly-Si is etched slower than undoped poly-Si in the mixture CF4+CF3Cl+N2O as well as in the mixture CF4+CF3Cl+O2, probably as a result of the different interaction mechanisms between substrate and gaseous reactants, which could be established at such high pressure and flowrate as in our experiment. Furthermore, our data reveal that the lateral etch rate (or undercut) is critically determined by the effective concentration of atomic F. Finally, we observed that CF3Cl has a clear anisotropic mechanism of etch.