Search

Measurements of electron temperatures (Te) and electron energy distribution functions (EEDFs) in a dual frequency capacitively coupled etcher were performed by using trace rare gas optical emission spectroscopy (TRG-OES). The parallel plate etcher was powered by a high frequency (60 MHz) “source” top electrode and a low frequency (13.56 MHz) “substrate” bottom electrode. Te first increased with pressure up to ∼20 mTorr and then decreased at higher pressures. Increasing the bottom rf power resulted in higher electron temperatures. Electron temperatures in 90% CF4+10% O2 plasmas were similar to those in 80% CF4+20% O2 plasmas. EEDF exhibited bi-Maxwellian characteristics with enhanced high energy tail, especially at pressures >20 mTorr.

In plasma etching processes, the reactor wall conditions can change over time due to a number of intentional and unintentional reasons, leading to a variability in the radical number densities in the plasma, caused by changes in the probabilities for reactions such as recombination at the walls. This leads to loss of reproducibility in the etching process. Here the authors isolated one such effect in which the feed gas was changed in the absence of a substrate. The transient surface composition of an anodized aluminum surface was determined for inductively coupled plasmas as the gas was switched from Cl2 to O2 and vice versa. The study was carried out with the spinning wall method and Auger electron spectroscopy. When the surface was first conditioned in an O2 plasma and then exposed to Cl2 plasmas, a rapid uptake of Cl was found in the first tens of seconds, followed by a slow approach to a steady-state value within ∼5min of plasma exposure. Conversely, when the surface was exposed to a Cl2 plasma for a ...

The authors have investigated the effects of elevated substrate temperature (Ts) on cleaning of boron residues from silicon substrates in 1%H2–Ar plasmas, following etching of HfO2 in BCl3 plasmas. Vacuum-transfer x-ray photoelectron spectroscopy (XPS) provided a measure of total B removal rates, as well as information on individual BClxOy moities. B cleaning rates increased with Ts in an Arrhenius manner, with an apparent activation energy of 1.7kcal∕mol. Conversely, the Si etching rate decreased with increasing substrate temperature with an apparent activation energy of −0.8kcal∕mol. Therefore, when considering selectivity with respect to Si etching, it is advantageous to remove B at higher Ts. For example, at Ts=235°C, ∼90% of B is cleaned from Si in 10s, while

The authors have investigated plasma etching of HfO2, a high dielectric constant material, and poly-Si in BCl3 plasmas. Etching rates were measured as a function of substrate temperature (Ts) at several source powers. Activation energies range from 0.2to1.0kcal∕mol for HfO2 and from 0.8to1.8kcal∕mol for Si, with little or no dependence on source power (20–200W). These low activation energies suggest that product removal is limited by chemical sputtering of the chemisorbed Hf or Si-containing layer, with a higher Ts only modestly increasing the chemical sputtering rate. The slightly lower activation energy for HfO2 results in a small improvement in selectivity over Si at low temperature. The surface layers formed on HfO2 and Si after etching in BCl3 plasmas were also investigated by vacuum-transfer x-ray photoelectron spectroscopy. A thin boron-containing layer was observed on partially etched HfO2 and on poly-Si after etching through HfO2 films. For HfO2, a single B(1s) feature at 194eV was ascribed to a ...

The authors have investigated the influence of plasma exposure time (t) on the Langmuir-Hinshelwood (i.e., delayed) recombination of O atoms on electropolished stainless steel surfaces using the spinning-wall method. They found a recombination probability (γO) of 0.13±0.01 after about 60min of plasma exposure. γO decreased to 0.09±0.01 for t⩾12h and was independent of the O flux impinging onto the surface. These recombination probabilities are much lower than those obtained in plasma chambers exclusively made of stainless steel, but similar to values recorded in stainless steel reactors with large silica surfaces exposed to the plasma. Near real-time elemental analysis by in situ Auger electron spectroscopy showed that the stainless steel surface became rapidly coated with a Si-oxide-based layer (Fe:[Si+Al]:O≈2:1:9 for t=60min and 1:2:9 for t=12h), due to the slow erosion of the silica discharge tube and anodized Al chamber walls. Thus, the recombination probability of oxygen atoms on stainless steel in plasma reactors with large amounts of exposed silica is largely determined by the amount of sputtered silica coating the chamber walls.

The authors report for the first time Auger electron spectroscopy (AES) of a surface while it is exposed to a high pressure, reactive environment: a 5mTorr inductively coupled plasma. An anodized aluminum cylindrical substrate (a common plasma reactor coating) was rotated within the reactor wall. Differential pumping allowed the substrate to be exposed to the plasma, and then AES as little as 1ms thereafter. Electron-beam-induced charging, a severe problem for conventional Auger analysis of insulators, is remediated in this experiment because the plasma maintains the surface at a constant floating potential. Chlorine, oxygen, and nitrogen plasmas were investigated. O2 plasmas are effective in removing Cl from Cl2 plasma-conditioned surfaces; N2 plasmas are not. During Cl2 plasma exposure, Cl coverage does not decrease with increasing delay time between plasma exposure and Auger analysis, varied by varying the substrate rotation frequency. This is contrary to desorption of Cl2 (detected by line-of-sight ma...

The energy distribution and flux of the fast neutrals and residual ions extracted from a neutral beam source were measured. Positive ions generated in an inductively coupled argon plasma were extracted through a metal grid with high aspect ratio holes. Ions suffered grazing angle collisions with the inside surface of the grid holes, turning into fast neutrals. The neutral energy distribution was always shifted to lower energies compared to the corresponding residual ion energy distribution. The neutralization efficiency increased with power, decreased with boundary voltage and, for thin neutralization grids, was almost independent of plasma gas pressure. The residual ion flux decreased with increasing hole diameter and hole aspect ratio. The fast neutral flux first increased and then dropped as the hole diameter was increased. Results were explained based on plasma molding inside the grid holes.

BCl3-containing plasmas are used to etch HfO2, a high dielectric constant (“high-k”) material. We have investigated several plasmas for their effectiveness in cleaning boron from the underlying Si surface after BCl3 plasma etching of HfO2, while removing a minimum amount of Si. X-ray photoelectron spectroscopy with vacuum sample transfer was used for surface analysis. B cleaning was optimum in dilute H2 plasmas (in Ar) compared to pure H2 plasmas. Dilute H2 plasmas slowed B cleaning process to a controllable time. In a 1% H2–Ar plasma, 20s was required to clean ∼90% B from the surface after a 60s overetch of HfO2 in BCl3 plasmas. The Si substrate was etched

The passivation step used in the “Bosch” process (alternating etching and deposition steps) to perform deep anisotropic silicon etching has been examined in detail. The effect of pressure, inductively coupled plasma power, temperature, flow rate, and bias power on both deposition rate and film composition has been explored over a relatively wide range. Deposition rate was found to vary significantly as a function of temperature, power, and pressure. In contrast, only two film composition regimes were observed: high fluorine-to-carbon ratio (F:C) films (∼1.6) at low pressure∕high power versus low F:C films (∼1.2) at high pressure∕low power. Optical emission spectroscopy of the deposition plasmas also show only two regimes: C2, C3, and F emission dominated (high F:C films) and CF2 emission dominated (low F:C films). A two-step deposition mechanism is assumed: carbon deposition followed by fluorination. Low F concentration and deposition from large fluorine-deficient CxFy species in the CF2-rich plasmas resu...

Recent advances in the interpretation of optical emission spectra from plasmas have made it possible to measure parameters such as electron temperature (Te), relative electron density, and gas temperature (Tg) with this nonintrusive technique. Here we discuss the application of trace rare gas optical emission spectroscopy (TRG-OES), optical actinometry, and N2 rotational spectroscopy to determine Te, relative electron density, fluorine atom concentration, and Tg for fluorocarbon/Ar plasmas in an inductively coupled reactor. Various etch processes, containing mixtures of a carrier gas, C2F6, and C4F8, were evaluated as a function of pressure and flowrate. Ar, Kr, and Ne were used individually or were mixed to comprise the carrier gas. In the case of TRG-OES and optical emission actinometry, a mixture containing equal parts of He, Ne, Ar, Kr, and Xe (∼1% ea.) was added. A method for correcting excitation cross sections is introduced for cases when radiation trapping affects the emission of a rare gas (Ar) t...

Trace rare gases optical emission spectroscopy has been used to measure the electron temperature, Te, in a high-density inductively coupled Cl2–Ar plasma at 18 mTorr as function of the 13.56 MHz radio frequency power and Ar fraction. Only the Kr and Xe emission lines were used to determine Te, because of evidence of radiation trapping when the Ar emission lines were also used for larger Ar fractions. At 600 W (10.6 W cm−2), Te increases from ∼4.0±0.5 eV to ∼6.0±2.0 eV as the Ar fraction increases from 1% to 96%. In the H (inductive, bright) mode, Te, for a “neat” chlorine plasma (including 1% of each He/Ne/Ar/Kr/Xe) increases only slightly from ∼3.8 to 4.0 eV as power increases from 450 to 750 W. This increase is much larger for larger Ar fractions, such as from ∼4.0 to 7.3 eV for 78% Ar. Most of these effects can be understood using the fundamental particle balance equation.

Chemical etching of silicon by fluorine atoms in the absence of ion bombardment is reviewed. Controversies on the identity of etching products and reaction probabilities are discussed. Attempts are made to explain the apparent presence of SiF2 as a primary product in many studies, dating back to 1980, but not in others, including those of Harold Winters from as early as 1979. Reported estimates of reaction probabilities (here defined as the probability of removing a Si atom from the substrate per incident F atom) vary by a factor of 2000. When these values, with some corrections and reasonable adjustments, are plotted as a function of F atom flux, most of them fall on a “universal curve” that reveals a large (∼30-fold) decrease in the reaction probability with increasing F flux, from 0.03 at a F flux 1012 cm−2 s−1 to 0.001 at a flux of 1020 cm−2 s−1. These values were extracted from beam experiments with F atoms generated from cracking of F2, including those by Harold Winters, from isotropic etching in pl...

Time-dependent behavior of chlorine inductively coupled plasmas is presented for Si etching, following NF3-Ar plasma cleaning of a chamber coated with Y2O3. Optical emission intensities were recorded throughout the processes for Cl, O, F, Si, SiClx=1-3, SiF, and N2, as well as from added trace rare gases Xe and Ar for determination of number densities for selected species by actinometry. Time-dependent Langmuir probe measurements of ion and electron number densities and electron energy distributions were also carried out. Ex situ x-ray photoelectron spectroscopy measurements of the surface composition of Y2O3 coupon pieces after different etching and clean processes were also performed. Initially fluorinated yttria surfaces are shown to have a relatively high probability for loss (“recombination”) of Cl through formation of both Cl2 and SiClx. As etching proceeds, SiClx abstracts F from the surface and deposits Si and Cl, lowering of the heterogeneous recombination of Cl. The initially high recombination ...

As new, advanced high-k dielectrics are being developed to replace SiO2 in future generations of microelectronics devices, understanding their etch characteristics becomes vital for integration into the manufacturing process. We report on the etch rates and possible mechanisms for one such dielectric, Zr1−xAlxOy (x≈0.2), in plasmas containing a mixture of Cl2 and BCl3, as a function of gas composition and ion impact energy. Higher concentrations of BCl3 enhance the etch rate as well as selectivity of Zr1−xAlxOy etching as compared to the etching of α-Si, whereas increasing ion energy increases the etching rates but decreases selectivity. In a high density helical resonator plasma, etching rates on the order of 700 A/min and 1:1 selectivity are typical. Angle-resolved x-ray photoelectron spectroscopy was used to study the composition of the upper ∼30 A of the film, before and at the end of the etching process. We found that the etching rate of Zr1−xAlxOy does not change with time for the range of Cl2/BCl3 ...

The surface during the etching of Si in a Cl2 inductively-coupled plasma (ICP) was analyzed by laser desorption (LD), followed by detection of the desorbed species by monitoring the transient changes by plasma-induced emission (PIE). Optical emission from Si, SiCl, SiCl2, and possibly other species was detected in situ using this LD-PIE method as a function of rf power, substrate bias, and pressure. The surface coverage of chlorine was determined by normalizing the LD-PIE signal by either of two ways: by the electron density, as measured by microwave interferometry, or by using the background PIE signal. Little change in surface coverage was observed as the ion density was changed by varying the rf power supplied to the ICP, confirming the observation made using laser-induced fluorescence (LIF) detection (LD-LIF). The LD-PIE signal is related to the density of species desorbed from the surface and subsequently excited by electrons. LD-PIE analysis of the surface is more versatile than LD-LIF—sometimes it can detect the desorbed species while LD-LIF cannot. Still, it requires calibration to account for the electron collision-induced excitation of the laser-desorbed species.

Using trace rare gases-optical emission spectroscopy (TRG-OES) and Langmuir probe measurements, electron temperatures (Te) were obtained in Cl2/BCl3/N2 plasmas in an inductively coupled plasma system, under typical processing conditions for metal etching. A small amount (1.7% each) of the five rare gases was added to the plasma and emission spectra were recorded. TRG-OES Tes corresponding to the high-energy tail of the electron energy distribution function were derived from the best match between the observed and computed rare gas emission intensities. Te was determined as a function of total pressure, source power, fraction of BCl3 added to Cl2 and substrate material (SiO2, Al, and photoresist). Positive ion densities and relative electron densities were also measured for some of these conditions. At source and bias powers of 1000 and 100 W, TRG-OES Tes in Cl2/BCl3/N2/rare gas plasmas increased from 1.4 eV at 40 mTorr to 2.3 eV at 3 mTorr, about 15% lower than values computed from a global model and ∼1.4...

Nonideal feature profile anomalies such as undercut, tapered, or bowed sidewalls and microtrenches at the base of trench sidewalls are often observed after etching masked silicon (Si) in chlorine (Cl2) plasmas. Off-normal impact with subsequent scattering and/or focusing of ions is believed to be the primary cause of these anomalies. Localized buildup of negative charge on the insulating mask sidewalls is one possible source of the ion deflection. Here we show that nearly identically shaped sidewalls and microtrenches were formed when Si features were etched in several Cl2 plasmas (the rf source and substrate bias powers and pressure were varied) regardless of whether the mask was insulating (SiO2) or conducting (Si or tungsten). Therefore, differential charging of the insulating mask material is not the fundamental cause of these profile characteristics. Ion angular distributions for each experimental plasma condition were estimated and the angle through which an ion would be deflected by a charged mask ...

We have studied the effects of source and bias powers, pressure, and feed gas composition on the shapes of SiO2-masked crystalline silicon features etched in a transformer-coupled high density plasma system. Higher etching rates were obtained at higher source and bias powers, and higher pressure. The etching rates of isolated and nested trenches, isolated lines, and holes were nearly the same, indicating a negligible pattern density dependence. We did, however, observe a very weak decrease in etch rates with increasing aspect ratio at 2 mTorr in a pure Cl2 plasma. At 10 mTorr, no aspect ratio dependence was observed, except at the highest source and bias powers. Microtrenching was observed under certain plasma conditions and could be reduced by using higher bias powers. At 10 mTorr in a pure chlorine plasma, we observed a slight taper at the bottoms of the etched features and the formation of narrow microtrenches near feature corners. At 2 mTorr, the microtrenches were broader and overlapped near the cent...

Laser desorption laser-induced fluorescence (LD-LIF) detection of GeCl was used to determine in situ the surface coverage of chlorine during the etching of germanium by Cl2 in an inductively coupled plasma (ICP) reactor. The ICP operated in the dim mode for radio frequency (rf) power ≲350 W and in the bright mode for higher powers. The etch rate was 3.5 μm/min with 540 W rf power and −40 V substrate bias. The chlorine surface coverage was about 2× that with chlorine flow only and the plasma off, both with dim- and bright-mode operation, and was independent of rf power within each mode for laser repetition rates of 0.2, 5, and 15 Hz. Similarly, the chlorination of the adlayer did not change when the ion energy was increased from 16 to 116 eV by increasing the substrate bias voltage, both with dim- and bright-mode operation. This was confirmed by x-ray photoelectron spectroscopy measurements in a similar high density reactor, where it was found that the surface density of chlorine was ∼2.6×1015 Cl/cm2. As t...

Thin SiO2 layers were subjected to short exposures (10–40 s) to HBr/O2 and Cl2 high-density plasmas, simulating the over-etching process encountered when polycrystalline Si gate electrodes are etched down to the gate oxide layer. Following this treatment, the samples were transferred under vacuum to an x-ray photoelectron spectrometer and spectra were recorded as a function of the take-off angle between the sample surface plane and the photoelectron collection lens. These angle-resolved measurements were inverted, using a maximum entropy approach, to obtain depth profiles. After etching in Cl2 or HBr plasmas at an ion energy of ∼40 eV (obtained with a grounded stage and a plasma potential of 40 V), surface layers were formed with halogen areal densities of ∼2×1015 cm−2, distributed over a half-depth of 10–20 A. These results (both absolute areal densities and depth distributions) are similar to those found previously for etching of Si under the same conditions. For SiO2, buildup of Cl or Br near the surf...

The etching of silicon by a chlorine inductively coupled plasma (ICP) was studied using laser desorption laser-induced fluorescence (LD-LIF) analysis to determine the surface coverage of chlorine during steady-state etching. Laser interferometry was used to measure etch rates, and optical emission actinometry and Langmuir probe analysis were used to characterize the plasma. The ICP operated in the dim mode for radio frequency (rf) powers ≲ 350 W and in the bright mode for higher powers. Under typical operating conditions in the ICP bright mode, the ion density was about 4× 1011/cm3, the electron temperature was 2.8 eV, and about 90% of the Cl2 was dissociated. The chlorine surface coverage in the dim and bright modes was ∼2.0× that with chlorine flow and the plasma off, and increased slowly with power. This coverage ratio monotonically increased from ∼1.6× to 2.5× that with the plasma off as the ion energy was increased from 16 to 116 eV by increasing the rf substrate bias voltage during bright mode opera...

Spectroscopic ellipsometry was applied to study the damage induced by chlorine plasma etching of crystalline silicon (100). Two etching modes (with different dc bias voltages) were investigated; (1) reactive ion etching with radiofrequency (rf) power applied only to the wafer stage and (2) a high density helicon plasma with and without rf power on the stage. Bias voltages from 0 to −350 V were employed. A two layer model was used to interpret the ellipsometry data. The interfacial layer was modeled as a mixture of amorphous silicon (a-Si), crystalline silicon (c-Si), and chlorinated silicon (SiClx), and the top surface was modeled as a mixture of a-Si and SiClx. The evolution of the damaged surface layer was investigated as a function of plasma exposure time and dc bias. Real-time measurements show that the damaged surface layer reaches a saturated thickness and composition after about 4 s of plasma exposure. The thickness of the surface layer increases with mean ion energy (i.e., negative dc bias voltage...

A method is described for determining the electron temperature of a low pressure plasma of the type used in microelectronics materials processing. A small amount of an equal mixture of He, Ne, Ar, Kr, and Xe is added to the process gas (in this example Cl2) and the intensities of optical emission lines from the Paschen 2p levels of the rare gases are recorded. The observed emission intensities are compared with those computed from a model that includes electron impact excitation from the ground state, as well as two-step electron impact excitation through intermediate metastable levels. This latter route is shown to be the dominant one for nearly half of the levels. Using adjusted, published electron impact excitation cross sections and assuming a Maxwellian electron energy distribution, the electron temperature (Te), the only adjustable parameter, was determined from the best match between the observed and computed intensities. For a high density, helical resonator Cl2 plasma at 10 mTorr, Te=2.2±0.5 eV w...

Optical emission was used to determine absolute Cl2 number densities in a Cl2 high‐density, helical resonator plasma that was used to etch Si. Emission from Cl2 at 306 nm was monitored, and rare‐gas actinometry was used to correct for changes in the electron energy distribution. This emission band of Cl2 is ascribed in this study to a state at 8.4 eV, or possibly to one at 9.2 eV. A state of Xe at 9.8 eV, emitting at 823.2 nm, was chosen as a suitable match to the energy of the emitting state of Cl2. Absolute number density measurements were obtained by extending the measurement to extremely low powers where little dissociation occurs. This simple self‐calibration method has advantages over other more complicated techniques such as laser induced fluorescence that rely on external calibrations. It is also much easier to carry out than mass spectrometry and is more sensitive than optical absorption. It is applicable to feed gases with optical emission from states with energies above about 8 eV. In this stud...

Various mechanisms have been proposed to explain photo-assisted etching (PAE) of Si, including photogenerated carrier-mediated etching, photon-stimulated desorption, and photon-induced damage (breaking of Si–Si bonds) caused mainly by vacuum ultraviolet photons irradiating the substrate. In this study, the authors use optical emission spectroscopy to gain an insight into possible in-plasma PAE mechanisms. Emissions from Cl, Si, SiCl, and Ar were recorded as a function of power while etching p-Si in Cl2/Ar Faraday-shielded inductively coupled plasmas at a pressure of 60 mTorr with no substrate bias. Under these conditions, ion-assisted etching was negligible and PAE was dominant. The Si:Ar optical emission intensity ratio, ISi/IAr (proportional to the etching rate of Si), increased substantially with power. Accounting for the contribution to this signal from the dissociation of SiClx (x = 1–4) etch products, the residual increase in the emission indicated that the PAE rate also increased with power. Time r...

Using x‐ray photoelectron spectroscopy (XPS) and real‐time, laser‐induced thermal desorption–laser‐induced fluorescence (LD–LIF), we have determined the coverage of Br and Cl on Si(100) surfaces that are etched in mixed HBr/Cl2 plasmas. Halogen coverages measured by XPS after etching are directly proportional to the fraction of the respective halogen in the feed gas. LD–LIF was detected from SiCl(g) and SiBr(g) products with intensities that are a semiquantitative measure of instantaneous Cl and Br coverages. Saturated coverages during etching in Cl2 and HBr plasmas are 1.0×1015 Cl/cm2 and 6.0×1014 Br/cm2, respectively. Etch rates at these two extremes are 2170 and 1330 A/min, and therefore are proportional to the respective halogen coverages. It therefore appears that the rate of formation of volatile Si–halides, stimulated by ion bombardment, is lower for HBr mainly because less halogen is available on the surface at saturated coverage. Langmuir probe measurements indicate that the ion flux is 17% lower...

Silicon nitride (SiN, where Si:N ≠ 1:1) films low pressure-chemical vapor deposited on Si substrates, Si films on Ge on Si substrates, and p-Si samples were exposed to plasma beams emanating from CH3F/O2 or CH3F/CO2 inductively coupled plasmas. Conditions within the plasma beam source were maintained at power of 300 W (1.9 W/cm3), pressure of 10 mTorr, and total gas flow rate of 10 sccm. X-ray photoelectron spectroscopy was used to determine the thicknesses of Si/Ge in addition to hydrofluorocarbon polymer films formed at low %O2 or %CO2 addition on p-Si and SiN. Polymer film thickness decreased sharply as a function of increasing %O2 or %CO2 addition and dropped to monolayer thickness above the transition point (∼48% O2 or ∼75% CO2) at which the polymer etchants (O and F) number densities in the plasma increased abruptly. The C(1s) spectra for the polymer films deposited on p-Si substrates appeared similar to those on SiN. Spectroscopic ellipsometry was used to measure the thickness of SiN films etched u...

We have used laser‐induced thermal desorption, combined with laser‐induced fluorescence of SiCl(g) to study, in real time, the Si‐chloride (SiClx(ads)) layer that is present on the surface during Si etching in a high‐plasma density, low pressure Cl2 helical resonator plasma. The SiClx(ads) layer that builds up during etching contains about twice as much Cl as the saturated layer that forms when Si is exposed to Cl2 gas. By varying the laser repetition rate we determined that the surface is chlorinated with an apparent first‐order time constant of ∼6 ms at 1.0 mTorr, and 20 ms at 0.3 mTorr. Therefore in the plasma at pressures above ∼0.5 mTorr, the SiClx(ads) layer reaches saturated coverage on a time scale that is short compared to the time required to etch one monolayer (40 ms). From the weak dependence of the SiClx(ads) layer coverage on discharge power (0.2–1 W/cm3), substrate bias voltage (from 0 to −50 V dc), and pressure (0.5–10 mTorr), we conclude that ion flux, and not neutral etchant flux (i.e., ...

Interferometric thermometry is a noncontact method of directly measuring the temperature of a semiconductor wafer. Periodic modulation in the intensity of reflected laser light is detected as the temperature‐dependent change in the optical pathlength causes transitions from constructive to destructive interference between the front and back surfaces of the wafer. In its simplest implementation, only a temperature change is detected; heating and cooling are indistinguishable. In this article, a simple method for determining the direction of temperature change is described and demonstrated, based on sensing phase differences in periodic signal intensity detected from two regions of slightly different thickness. A statistical analysis reveals that this method will have an extremely high probability of correctly determining the direction of temperature change in almost all instances.

Cu is receiving increasing attention as a replacement for Al in future Si ultra‐large‐scale integrated circuits due to its lower resistivity and potentially better reliability in terms of electromigration resistance and stress‐induced voiding. Metalorganic chemical vapor deposition (MOCVD) of Cu offers the advantages of conformal coverage and selective growth. Whatever the means of deposition, a diffusion barrier and adhesion layer such as TiN is required. To understand the nucleation mechanisms of Cu film growth on TiN during MOCVD, we have studied the thermal decomposition of two metalorganic precursors, hexafluoroacetylacetonate Cu(I) vinyltrimethylsilane [CuI(hfac)(vtms)], and bis (hexafluoroacetylacetonate) Cu(II) [CuII(hfac)2] by x‐ray photoelectron spectroscopy and temperature programmed desorption mass spectrometry in an ultrahigh vacuum (UHV) system.Experiments were carried out on polycrystalline air‐exposed (i.e., oxidized) TiN and N2 ion beam sputter‐cleaned TiN. These surfaces are representati...

The effect of Cl2 pressure and substrate temperature (Ts) on the absolute, steady‐state etch rates of InP(100) has been measured with ∼1% accuracy from ∼0.02 to 1000 A/sec by infrared laser interferometry. An unexpected finding was that the etch rate could exceed the calculated evaporation rate of InCl3 by a factor of 50, indicating a weaker binding energy of InCl3 to InP relative to itself. At sufficiently low Ts, the measured slow etch rate is only a factor of ∼3.5 above that predicted from the evaporation rate of InCl3. The transition between the fast and slow etch rates occurs as a first‐order phase transition at a substrate temperature, T*s, which depends on the Cl2 pressure. If InP is etched at T*s, the surface roughens while at other Ts the surface remains smooth. The explanation for the surface roughening is that at T*s, InCl3 nucleates to form InCl3 islands. At T*s the etch rate is approximately fifteen times slower where the InP surface is covered by InCl3 compared to the bare InP surface. The l...

This article reports the application of interferometric thermometry to the measurement of silicon wafer temperatures during exposure to low‐pressure (1 mTorr), high‐density (ne= 1 × 1011 cm3) helical resonator plasmas. This technique measures the temperature‐dependent optical pathlength from the front to the back of a wafer polished on both sides. The measurement does not perturb the plasma, nor does it suffer from interference from the plasma, problems which often plague other techniques. Nonetching N2 plasmas heat the wafer to 140 °C in the first 3 min, and 200 °C after 10 min when the wafer is simply placed on the platen with no provisions for heat dissipation. When the wafer is clamped to an o‐ring seal on the platen and a small gap between the wafer and platen is pressurized with helium (so‐called helium backside cooling), the wafer temperature rise is reduced. With a helium pressure of 10 Torr between the wafer and platen, the temperature was 40 °C after 2 min and 70 °C after 10 min. When silicon wafers are etched in a Cl2 plasma, the interferometric signal contains contributions from both heating and etching, which respectively increase and decrease the optical pathlength within the wafer. A deconvolution of this signal using the known etch rate indicates that plasma‐induced heating in a Cl2 plasma is approximately the same as in a N2 plasma of about the same power, indicating that any contributions from heating due to chemical reactions in the etching process are negligible compared to heating from the plasma.

Products of pulsed laser‐induced decomposition of triethylgallium (TEGa) and trimethylgallium (TMGa) chemisorbed on GaAs(100) were determined and compared to those observed in thermal desorption. The desorbing products, C2H5, C2H4, and Ga(C2H5)x (x=2 and/or 3) for TEGa decomposition, and CH3 and Ga(CH3)x (x=2 and/or 3) for TMGa, are the same as those produced in thermal desorption. However, rapid substrate heating with the pulsed excimer laser causes large changes in the relative yields. With laser induced heating, Ga‐alkyl desorption is almost completely suppressed in favor of further decomposition to yield hydrocarbon products and leave Ga on the surface, while in thermal desorption the Ga‐alkyl is a major product. Furthermore, for laser‐induced thermal decomposition of TEGa, C2H5 production is enhanced with respect to C2H4. These effects are ascribed to decomposition of chemisorbed GaR2(ads) (R=CH3, or C2H5), leading to formation of adsorbed GaR(ads) and desorption of R(g). This decomposition reaction ...

We report studies of the kinetics of thermal decomposition of triethylgallium (TEGa), trimethylgallium (TMGa), and trimethylindium (TMIn) adsorbed on GaAs(100) in ultrahigh vacuum. The adsorbed layers were prepared by dosing GaAs(100) at room temperature, to either saturated coverage or coverages below saturation. The relative coverage of carbon was monitored by x‐ray photoelectron spectroscopy (XPS) as the substrate temperature was slowly increased (0.6–3.2 °C/min). Products were detected at faster heating rates (0.7–6 °C/s) with a differentially pumped quadrupole mass spectrometer. The substrate temperature was measured by infrared laser interferometric thermometry. The kinetic analysis also makes use of XPS and mass spectrometric data on laser‐induced, rapid thermal decomposition (heating rates of ∼1011 °C/s ). TEGa dissociatively chemisorbs on GaAs(100) at room temperature. Heating the substrate from room temperature to ∼500 °C results in desorption of a Ga–alkyl at low temperature, ascribed mostly to...

The authors have studied the interaction of HBr/Cl2/O2 inductively coupled plasmas with reactor chamber wall deposits, with and without Si etching, using the “spinning wall” technique. The spinning wall is part of the reactor chamber walls, allowing near-real-time analysis of the composition of surface layers via Auger electron spectrometry and determination of species desorbing off the walls by mass spectrometry. In HBr plasmas with no bias voltage on the Si substrate, and hence no Si etching, HBr is ∼30% dissociated, and H2 and Br2 form in the plasma. Layers deposited on the reactor chamber contained little if any Br under these conditions. Adding O2 to an HBr plasma leads to formation of H2O and increased Br2 (compared to a pure HBr plasma) products that desorb from the spinning wall. H2O has a very long residence time on the surface. With bias voltage applied to the Si substrate in an HBr plasma, mass spectrometer signals are prominent for SiBr and SiBr3, and weaker for SiBr2, SiBr4, Si2Br4, Si2Br5, and Si2OBr5. Under these conditions, a SiOxBry layer deposits on the spinning wall. Adding 20% O2 to HBr stops etching and eliminates Br from the surface layer, indicating that Br on the reactor walls is a result of SiBrx impingement and not from bromination by impinging Br. With HBr/Cl2 plasmas and no bias on the stage, a SiOxCly layer deposits, and no Br is detected. HCl, BrCl, and Br2 were detected in the line-of-sight leak, around the spinning wall, of a HBr/Cl2 (1:1) gas mixture in the absence of a plasma. Residence time analysis of species in the chamber and a change in the product distribution with a change in the composition of the layer deposited on the chamber wall suggest that reactions forming these products in the absence of a plasma occur on the reactor walls. With a plasma and bias on the Si substrate, both Br and Cl incorporate in the layer, and a rich spectrum with numerous SiClxBry peaks was observed up to at least 500 amu. The spectrum does not change with the addition of 6% O2. Adding 20% O2 suppresses Br adsorption, but Cl still adsorbs. In 40% O2/HBr/Cl2 plasmas with stage bias, Cl adsorption also ceases and no etching products are observed in the mass spectrum.

A CH3F/CO2 inductively coupled plasma (ICP), sustained in a compact plasma reactor, was investigated as a function of power (5–400 W) and feed gas composition, at a pressure of 10 mTorr, using optical emission spectroscopy and rare gas actinometry. Number densities of H, F, and O increased rapidly between 74% and 80% CO2, ascribed to the transition from polymer-covered to polymer-free reactor walls, similar to that found previously in CH3F/O2 ICPs at 48% O2. Below 40% O2 or CO2, relative emission intensity ratios were almost identical for most key species in CH3F/O2 and CH3F/CO2 ICPs except for higher OH/Xe (a qualitative measure of OH and H2O densities) over the full range of CH3F/O2 composition. The number density of H, F, and O increased with power in CH3F/CO2 (20%/80%) plasmas (polymer-free walls), reaching 4.0, 0.34, and 1.6 × 1013/cm3, respectively, at 300 W. The CO number density increased with power and was estimated, based on self-actinometry, to be 8.8 × 1013/cm3 at 300 W. The CO2 number density...

As the tolerances in gate dimensions in integrated circuit manufacturing become ever more stringent, plasma process conditions must be very tightly controlled. The reactor chamber wall contamination is one of the major causes of process drifts and is therefore of prime importance. Here, the authors report a study of the role of Ti contamination on an oxidized silicon surface in affecting the heterogeneous recombination coefficient of O in an O2 inductively coupled plasma reactor. Recombination coefficients were measured, using the spinning wall method, with in-situ Auger electron spectroscopy (AES) for surface analysis during plasma operation. The O-atom recombination coefficient on a Ti-free surface was found to be 0.034. After using an evaporation source to deposit a small amount of Ti on the spinning wall (17% of the atomic composition obtained from AES), the O recombination coefficient decreased to 0.022. A possible mechanism is proposed in which Ti reacts with ≡Si-O• sites that are active in recombining O, forming ≡Ti-O• sites that are less efficient for O recombination.

The field of plasma etching is reviewed. Plasma etching, a revolutionary extension of the technique of physical sputtering, was introduced to integrated circuit manufacturing as early as the mid 1960s and more widely in the early 1970s, in an effort to reduce liquid waste disposal in manufacturing and achieve selectivities that were difficult to obtain with wet chemistry. Quickly, the ability to anisotropically etch silicon, aluminum, and silicon dioxide in plasmas became the breakthrough that allowed the features in integrated circuits to continue to shrink over the next 40 years. Some of this early history is reviewed, and a discussion of the evolution in plasma reactor design is included. Some basic principles related to plasma etching such as evaporation rates and Langmuir–Hinshelwood adsorption are introduced. Etching mechanisms of selected materials, silicon, silicon dioxide, and low dielectric-constant materials are discussed in detail. A detailed treatment is presented of applications in current silicon integrated circuit fabrication. Finally, some predictions are offered for future needs and advances in plasma etching for silicon and nonsilicon-based devices.

Selectivity of etching between physical vapor-deposited TiN and TaN was studied in chlorine-containing plasmas, under isotropic etching conditions. Etching rates for blanket films were measured in-situ using optical emission of the N2 (C3Πu →B3Πg) bandhead at 337 nm to determine the etching time, and transmission electron microscopy to determine the starting film thickness. The etching selectivity in Cl2/He or HCl/He plasmas was poor (

Chlorine atom recombination coefficients were measured on silicon oxy-chloride surfaces deposited in a chlorine inductively coupled plasma (ICP) with varying oxygen concentrations, using the spinning wall technique. A small cylinder embedded in the walls of the plasma reactor chamber was rapidly rotated, repetitively exposing its surface to the plasma chamber and a differentially pumped analysis chamber housing a quadruple mass spectrometer for line-of-sight desorbing species detection, or an Auger electron spectrometer for in situ surface analysis. The spinning wall frequency was varied from 800 to 30 000 rpm resulting in a detection time, t (the time a point on the surface takes to rotate from plasma chamber to the position facing the mass or Auger spectrometer), of ∼1–40 ms. Desorbing Cl2, due to Langmuir–Hinshelwood (LH) Cl atom recombination on the reactor wall surfaces, was detected by the mass spectrometer and also by a pressure rise in one of the differentially pumped chambers. LH Cl recombination...

The interplay between chlorine inductively coupled plasmas (ICP) and reactor walls coated with silicon etching products has been studied in situ by Auger electron spectroscopy and line-of-sight mass spectrometry using the spinning wall method. A bare silicon wafer mounted on a radio frequency powered electrode (−108 V dc self-bias) was etched in a 13.56 MHz, 400 W ICP. Etching products, along with some oxygen due to erosion of the discharge tube, deposit a Si-oxychloride layer on the plasma reactor walls, including the rotating substrate surface. Without Si-substrate bias, the layer that was previously deposited on the walls with Si-substrate bias reacts with Cl-atoms in the chlorine plasma, forming products that desorb, fragment in the plasma, stick on the spinning wall and sometimes react, and then desorb and are detected by the mass spectrometer. In addition to mass-to-charge (m/e) signals at 63, 98, 133, and 168, corresponding to SiClx (x = 1 – 4), many Si-oxychloride fragments with m/e = 107, 177, 19...

Ion energy distributions (IEDs) were measured near the edge of Faraday-shielded, inductively coupled pulsed plasmas in Ar, Kr, or Xe gas, while applying a synchronous dc bias on a boundary electrode, late in the afterglow. The magnitudes of the full width at half maximum of the IEDs were Xe > Kr > Ar, following the order of the corresponding electron temperatures in the afterglow, Te(Xe) > Te(Kr) > Te(Ar). The measured decays of Te with time in the afterglow were in excellent agreement with predictions from a global model. Measured time-resolved electron and positive ion densities near the plasma edge did not decay appreciably, even in the 80 μs long afterglow. This was attributed to transport of ions and electrons from the higher density central region of the plasma to the edge region, balancing the loss of plasma due to diffusion. This provides a convenient means of maintaining a relatively constant plasma density in the afterglow during processing using pulsed plasmas.

The authors report a new, important phenomenon: photo-assisted etching of p-type Si in chlorine-containing plasmas. This mechanism was discovered in mostly Ar plasmas with a few percent added Cl2, but was found to be even more important in pure Cl2 plasmas. Nearly monoenergetic ion energy distributions (IEDs) were obtained by applying a synchronous dc bias on a “boundary electrode” during the afterglow of a pulsed, inductively coupled, Faraday-shielded plasma. Such precisely controlled IEDs allowed the study of silicon etching as a function of ion energy, at near-threshold energies. Etching rates increased with the square root of the ion energy above the observed threshold of 16 eV, in agreement with published data. Surprisingly, a substantial etching rate was observed, independent of ion energy, when the ion energy was below the ion-assisted etching threshold. Experiments ruled out chemical etching by Cl atoms, etching assisted by Ar metastables, and etching mediated by holes and/or low energy electrons ...

This article reviews methods for studying reactions of atoms and small molecules on substrates and chamber walls that are immersed in a plasma, a relatively unexplored, yet very important area of plasma science and technology. Emphasis is placed on the “spinning wall” technique. With this method, a cylindrical section of the wall of the plasma reactor is rotated, and the surface is periodically exposed to the plasma and then to a differentially pumped mass spectrometer, to an Auger electron spectrometer, and, optionally, to a beam of additional reactants or surface coatings. Reactants impinging on the surface can stick and react over time scales that are comparable to the substrate rotation period, which can be varied from ∼0.5 to 40 ms. Langmuir–Hinshelwood reaction probabilities can be derived from a measurement of the absolute desorption product yields as a function of the substrate rotation frequency. Auger electron spectroscopy allows the plasma-immersed surface to be monitored during plasma operatio...

InP thin films have been deposited on several types of substrates, via 193 nm excimer laser induced photochemical decomposition of In(CH3)3 and P(CH3)3 gas‐phase precursors. Deposition results from gas‐phase photochemistry, while surface irradiation stimulates removal of carbon and crystallization. InP films deposited on (100)InP substrates have been studied by scanning electron microscopy, transmission electron microscopy, and Rutherford backscattering spectroscopy (RBS). Films range from amorphous to epitaxial, depending upon conditions (most notably fluence incident on the substrate). The best film deposited at ∼0.1 J/cm2 and at a steady‐state temperature of only ∼320 ° C, had an RBS spectrum indistinguishable from that of the single crystal substrate. To our knowledge, this is the first report of a 100% photochemically induced deposition of an epitaxial, III–V compound semiconductor thin film (i.e., no growth in the absence of light), and also the first report of an epitaxial InP film deposited from o...

Plasma etching techniques for III‐V compound semiconductors are reviewed, emphasizing design considerations in the choice of gases, discharge parameters, and substrate temperature. Mechanisms are proposed for anisotropic, isotropic, and crystallographic chemical etching. Applications of plasma etching in device fabrication are given.