- Email: [email protected]
Ion implantation of sulfur and silicon in GaAs
290
World Abstracts on Microelectronics and Reliability
Res. Devel. 24 (5), 537 (September 1980). Tile proximity effect in electron-beam lithography, which is due to electron scattering in the resist and wafer, results in oonuniform exposure and development for patterns in which the incident doses of all the shapes are the same. Correction for this effect has been accomplished in tile past primarily by varying the incident doses of all the shapes in order to achieve an equal average resultant dose per unit area for all shapes. We show that in the case of dense circuits with linewidths of about 1 lam or smaller, two enhancements to the proximity correction technique can be easily implemented. One of these is a simple approach to shape breakup (partitioning) to enable dose correction to be applied nonuniformly within the original design shapes. The other technique is a new type of algorithm for forming subsets of the design to perform self-consistent dose correction. These two enhancements are applied to LSI chip data for dense circuits and are shown to permit fabrication of circuits which would be morc difficult to process using the proximity correction techniques described previously, due to the particular geometries present in these circuit designs. We also show the application of step and repeat pattern recognition algorithms to compact the resulting data, and consequently to reduce the amount of data by an amount which is greater than the increase in the number of shapes caused by partitioning.
Noise in phosphorus-implanted buried channel MOS transistors. S. T. LIu, O. N. TUFTE, A. VAN OER ZtEL, S. Y. PAl and W. LARSON. Solid-St. Electron. 23, 1195 (1980). Noise measurements on phosphorus-implanted buried channel MOS transistors are reported. The data are interpreted as two superimposed generation-recombination (,q-r) spectra with different time constants plus a white spectrum due to thermal-like noise of the conducting channel.
Reactive ion etching in chlorinated plasmas. GERALDINE C. SCHWARTZ and PAUL M. SCHAIBLE. Solid St. Technol. p. 85 (November 1980). Single crystal silicon and films of aluminum and a l u m i n u m - 4 % copper can be reactively ion etched in chlorinated plasmas to produce non-undercut. vertical-walled patterns. CCI, and CI 2 have been used: the relative advantages and disadvantages of each are discussed. Results are given for the dependence of etch rates on input rf power, gas pressure, concentration and flow rate of reactant gas, batch size, and temperature. AICu films must be heated during RIE to etch them completely; at the higher temperature, preferential lateral chemical etching becomes more pronounced. When silicon is heavily doped with arsenic ( >/102°/cm 3) undercutting can occur: the extent depends on the choice of reactant, and the concentrations of arsenic and reactant; cooling the wafer prevents undercutting.
An approach to high speed laser trimming of thick film resistors. J. R. SIMS, P. G. CRETER,J. W. SoucY and A. HowE. Solid St. Technol. p. 135 (October 1980). A study of the effects of varying laser trimming speed indicates faster speeds can be used and still result in stable resistors conforming to military specifications. A set-up procedure is detailed and a short term measurement parameter is correlated with long term stability characteristics.
Bonding-failure analysis with electron beam and nuclear scattering techniques. A. CHmSTOU, G. NELSON and H. M. DAY. IEEE Trans. Reliah. R-29 (3), 250 (August 19801. Failure mechanisms related to gold bonding were determined using the scanning electron microscope, Auger electron spectroscopy, X-ray energy dispersive spectroscopy, electron microprobe analysis, and deuteron probe analysis. Power transistors from different lots were analyzed; there were four types of bond defects. Type I defect is darkened carbon inclusions in the bond area. Type 11 defect is a carbon buildup region on the posts probably in the form of graphite particles which adhered to the gold plating during processing.
Type 111 defects are adsorbed carbon surface films. Type IV defects are gross discontinuities m the gold plating. Auger electron spectroscopy analysis and electron probe analysis proved that defects 1, It, and II1 are subsurface and that further exposure to time-temperature and stress will result in carbide precipitates and hence cracks at the interface. Deuteron probe (DP) analysis of the same bond area has shown the presence of subsurface carbon (the predominant contaminant) thus verifying the Auger results. Contaminants such as manganese, iron boron, cobalt, nickel, chlorine, and sulfur were detected by energy dispersive X-ray analysis in concentrations of 100ppM or more. These contaminants result in microcracks and voids which are formed by a time-temperature process. All bond-pull data had a bimodal distribution similar to that reported by Horsting. The bimodal distribution of pull-strength, carbon subsurface inclusions, and excessively high levels of contamination (from the plating bathl, all indicate that a contamination-related failure mode exists.
Ion implantation of sulfur and silicon in GaAs. S. G. LIu, E. C. DOUGLAS. C. P. Wu, C. W. MAGI-I!, S. Y. NARAYAN, S. T. JOLLY, F. KOLONDRA and S. JAIN. RCA Rer. 41, 227 (June 1980). This paper describes implantation of 28Si and a-'S into semi-insulating GaAs to generate high mobility n layers, Implantation of 2uSi over an energy range of 40keV to 1.2 MeV and fluence ranging from 10 ~2 to l0 t ~'c m - 2 was studied. Implanted layers were characterized by secondary ion mass spectrometry (SIMS), differential Van der Pauw measurements, and differential CV measurements. A capless thermal annealing technique under arsenic overpressure that results in high activation efficiency and excellent surface morphology was developed. Implanted n-layers as deep as 1 ~.tm have been realized. The measured depth distribution of electron concentration is broader for S-implanted than for St-implanted GaAs as a result of thermal diffusion during annealing. Tile diffusion coefficients at 825°C for S and Si in GaAs were deduced from electrondensity profilesto be 2-5 x 10- ~*and ~< 10- ~s cm2/s, respectively. The redistribution of the implanted Si following thermal or laser annealing was studied using SIMS. Preliminary results on redistribution of Cr in the semi-insulating GaAs due to implantation and annealing are also presented.
Annealing of Si-SiO2 interface states using Ar-ion-implantdamage-gettering. B. GOLJA and A. G. NASSmIAN. Solid-St. Electron. 23, 1249 (19801. The effectivenes~ of Ar-ion-implant damage gettering on the Si-SiO 2 interface states has been investigated using MOS techniques and Rutherford backscattering. Silicon wafers of (1001 orientation were used in the study. Some wafers were intentionally contaminated with Au and then Ar-ion-implant was performed on the back surface and the damaged layer subsequently annealed at 1050°C for times of 15 and 60rain in a nitrogen ambient. The quasi-static I-V technique of Kuhn was used to obtain experimental curves which were correlated with tile theory to extract the interface state density and distribution. The effectiveness of Ar-ion-implant in removing Au from the interface is clearly demonstrated by the structure of the quasi-static 1 V curves. Rutherford backscattering of t'~N + ions was carried out on the wafers indicating that the removal of gold from the interfacial traps is associated with its removal from the bulk.
Registration mark detection for electron-beam lithography-ELI system. DONALD E. DAVIS. IBM d. Res, Develop. 24 (5), 545 (September 1980). In electron-beam lithography for the direct exposure of wafers for integrated circuit manufacturing, accurate registration is necessary to achieve the required pattern overlay. This paper examines elements that should be considered to optimize the registration mark detection process in an automatic registration system for an e-beam lithography tool such as IBM's ELI. Included is a section on
World Abstracts on Microelectronics and Reliability
Res. Devel. 24 (5), 537 (September 1980). Tile proximity effect in electron-beam lithography, which is due to electron scattering in the resist and wafer, results in oonuniform exposure and development for patterns in which the incident doses of all the shapes are the same. Correction for this effect has been accomplished in tile past primarily by varying the incident doses of all the shapes in order to achieve an equal average resultant dose per unit area for all shapes. We show that in the case of dense circuits with linewidths of about 1 lam or smaller, two enhancements to the proximity correction technique can be easily implemented. One of these is a simple approach to shape breakup (partitioning) to enable dose correction to be applied nonuniformly within the original design shapes. The other technique is a new type of algorithm for forming subsets of the design to perform self-consistent dose correction. These two enhancements are applied to LSI chip data for dense circuits and are shown to permit fabrication of circuits which would be morc difficult to process using the proximity correction techniques described previously, due to the particular geometries present in these circuit designs. We also show the application of step and repeat pattern recognition algorithms to compact the resulting data, and consequently to reduce the amount of data by an amount which is greater than the increase in the number of shapes caused by partitioning.
Noise in phosphorus-implanted buried channel MOS transistors. S. T. LIu, O. N. TUFTE, A. VAN OER ZtEL, S. Y. PAl and W. LARSON. Solid-St. Electron. 23, 1195 (1980). Noise measurements on phosphorus-implanted buried channel MOS transistors are reported. The data are interpreted as two superimposed generation-recombination (,q-r) spectra with different time constants plus a white spectrum due to thermal-like noise of the conducting channel.
Reactive ion etching in chlorinated plasmas. GERALDINE C. SCHWARTZ and PAUL M. SCHAIBLE. Solid St. Technol. p. 85 (November 1980). Single crystal silicon and films of aluminum and a l u m i n u m - 4 % copper can be reactively ion etched in chlorinated plasmas to produce non-undercut. vertical-walled patterns. CCI, and CI 2 have been used: the relative advantages and disadvantages of each are discussed. Results are given for the dependence of etch rates on input rf power, gas pressure, concentration and flow rate of reactant gas, batch size, and temperature. AICu films must be heated during RIE to etch them completely; at the higher temperature, preferential lateral chemical etching becomes more pronounced. When silicon is heavily doped with arsenic ( >/102°/cm 3) undercutting can occur: the extent depends on the choice of reactant, and the concentrations of arsenic and reactant; cooling the wafer prevents undercutting.
An approach to high speed laser trimming of thick film resistors. J. R. SIMS, P. G. CRETER,J. W. SoucY and A. HowE. Solid St. Technol. p. 135 (October 1980). A study of the effects of varying laser trimming speed indicates faster speeds can be used and still result in stable resistors conforming to military specifications. A set-up procedure is detailed and a short term measurement parameter is correlated with long term stability characteristics.
Bonding-failure analysis with electron beam and nuclear scattering techniques. A. CHmSTOU, G. NELSON and H. M. DAY. IEEE Trans. Reliah. R-29 (3), 250 (August 19801. Failure mechanisms related to gold bonding were determined using the scanning electron microscope, Auger electron spectroscopy, X-ray energy dispersive spectroscopy, electron microprobe analysis, and deuteron probe analysis. Power transistors from different lots were analyzed; there were four types of bond defects. Type I defect is darkened carbon inclusions in the bond area. Type 11 defect is a carbon buildup region on the posts probably in the form of graphite particles which adhered to the gold plating during processing.
Type 111 defects are adsorbed carbon surface films. Type IV defects are gross discontinuities m the gold plating. Auger electron spectroscopy analysis and electron probe analysis proved that defects 1, It, and II1 are subsurface and that further exposure to time-temperature and stress will result in carbide precipitates and hence cracks at the interface. Deuteron probe (DP) analysis of the same bond area has shown the presence of subsurface carbon (the predominant contaminant) thus verifying the Auger results. Contaminants such as manganese, iron boron, cobalt, nickel, chlorine, and sulfur were detected by energy dispersive X-ray analysis in concentrations of 100ppM or more. These contaminants result in microcracks and voids which are formed by a time-temperature process. All bond-pull data had a bimodal distribution similar to that reported by Horsting. The bimodal distribution of pull-strength, carbon subsurface inclusions, and excessively high levels of contamination (from the plating bathl, all indicate that a contamination-related failure mode exists.
Ion implantation of sulfur and silicon in GaAs. S. G. LIu, E. C. DOUGLAS. C. P. Wu, C. W. MAGI-I!, S. Y. NARAYAN, S. T. JOLLY, F. KOLONDRA and S. JAIN. RCA Rer. 41, 227 (June 1980). This paper describes implantation of 28Si and a-'S into semi-insulating GaAs to generate high mobility n layers, Implantation of 2uSi over an energy range of 40keV to 1.2 MeV and fluence ranging from 10 ~2 to l0 t ~'c m - 2 was studied. Implanted layers were characterized by secondary ion mass spectrometry (SIMS), differential Van der Pauw measurements, and differential CV measurements. A capless thermal annealing technique under arsenic overpressure that results in high activation efficiency and excellent surface morphology was developed. Implanted n-layers as deep as 1 ~.tm have been realized. The measured depth distribution of electron concentration is broader for S-implanted than for St-implanted GaAs as a result of thermal diffusion during annealing. Tile diffusion coefficients at 825°C for S and Si in GaAs were deduced from electrondensity profilesto be 2-5 x 10- ~*and ~< 10- ~s cm2/s, respectively. The redistribution of the implanted Si following thermal or laser annealing was studied using SIMS. Preliminary results on redistribution of Cr in the semi-insulating GaAs due to implantation and annealing are also presented.
Annealing of Si-SiO2 interface states using Ar-ion-implantdamage-gettering. B. GOLJA and A. G. NASSmIAN. Solid-St. Electron. 23, 1249 (19801. The effectivenes~ of Ar-ion-implant damage gettering on the Si-SiO 2 interface states has been investigated using MOS techniques and Rutherford backscattering. Silicon wafers of (1001 orientation were used in the study. Some wafers were intentionally contaminated with Au and then Ar-ion-implant was performed on the back surface and the damaged layer subsequently annealed at 1050°C for times of 15 and 60rain in a nitrogen ambient. The quasi-static I-V technique of Kuhn was used to obtain experimental curves which were correlated with tile theory to extract the interface state density and distribution. The effectiveness of Ar-ion-implant in removing Au from the interface is clearly demonstrated by the structure of the quasi-static 1 V curves. Rutherford backscattering of t'~N + ions was carried out on the wafers indicating that the removal of gold from the interfacial traps is associated with its removal from the bulk.
Registration mark detection for electron-beam lithography-ELI system. DONALD E. DAVIS. IBM d. Res, Develop. 24 (5), 545 (September 1980). In electron-beam lithography for the direct exposure of wafers for integrated circuit manufacturing, accurate registration is necessary to achieve the required pattern overlay. This paper examines elements that should be considered to optimize the registration mark detection process in an automatic registration system for an e-beam lithography tool such as IBM's ELI. Included is a section on