Silicon-on-Insulator structures will be an important technological advance used in future VLSI, VHSIC and threedimensional integrated circuits. The most mature SOI technology other than silicon-on-sapphire is SIMOX, or Separation by Implanted Oxygen. High energy oxygen ions are implanted into single crystal silicon until a stoichiometric buried silicon dioxide layer is formed. After implantation, the material is annealed at high temperature to remove implantation induced defects. The structure is completed by the growth of a thin epitaxial silicon layer. Devices and complex circuits have been successfully fabricated by several research groups. This paper reviews the development of this buried oxide SOI technology from 1973 to 1983. The five major sections discuss the advantages of SOI, the basics of buried oxide formation, the literature published between 1973 and 1983, key issues that must be solved before large scale implementation takes place and, finally, predictions of future developments.

Single crystal silicon was implanted with 80, 120, 160 and 240 keV oxygen ions. Rutherford backscattering (RBS) analysis was used to obtain the implanted oxygen profile and the oxygen to silicon ratio in the implanted layer for doses in the range 1016 to 1.5 × 1018 O2+ cm−2 for room temperature implants. The depth and the thickness of the buried oxide layer has been measured as a function of implantation energy and oxygen dose. Chemical formation of stoichiometric SiO2 was confirmed by infra-red (IR) spectroscopy. Both RBS and IR indicate that once a surface oxide layer is formed for very high dose levels, the layer thickness decreases with increasing implanted dose beyond a critical dose level.

SOI structures have been formed in (100) silicon by implanting 400 keV molecular oxygen to a dose of l.8×l018 O atoms cm−2. These samples were annealed at 1150°C for 2 hours with a SILOX cap. Oxygen depth profiles have been determined by SIMS and wafers implanted at about 500°C have been characterized by studying the regrowth kinetics, As drive in and oxidation rate in the top silicon overlay.

Silicon wafers have been implanted with H+ (90 keV) to doses of 5.0E15/ cm2 and 2.OE16/cm2. The wafers were annealed in nitrogen at temperatures between 450 and 700°C for times between 10 and 60 min. The electrically active carrier profiles were measured by capacitance voltage and spreading resistance techniques. The residual damage was measured by TEM and RBS. The electrical measurements were essentially the same in both FZ and CZ silicon implying that oxygen is not playing a role in the donor formation which was observed. The donor concentration peaks near the projected range of the hydrogen after annealing at temperatures between 450–500°C. As reported previously 1000 H+ ions generate 1 donor in the implant peak. In addition, the donor concentration between the surface and Rp has increased more than a factor of 10 above the background concentration after a 450°C 10 min anneal. Anneals of 550°C for 30 min or more annihilates essentially all of the donors. The RBS results show small amounts of damage for the 5.0E15/cm2 implant dose but considerable crystal damage with a dose of 2.0E16/cm2, even after a 500°C, 30 min anneal. Cross-sectional TEM analysis of 500°C annealed samples showed a large number of small loops at depths corresonding to the depth of the peak electrical carrier concentration. The donors are directly correlated to the implant damage and resultant defects. SIMS data shows little diffusion for anneals of 500°C or less but after 550°C, 30 min the peak H concentration decreases by approximately a factor of 10.

Transient, greatly enhanced diffusion has been observed on annealing solid-phase-epitaxial (SPE) grown Si-Sb alloys. This is shown to be due to a high concentration of interstitials being trapped during SPE regrowth. The migration enthalpy, for diffusion of Sb by an interstitialcy mechanism was measured as 1.8 ± 0.2 eV. The interstitials eventually condensed into loops, marking the end of the transient. In a SPE grown Si-Bi alloy a similar transient enhanced diffusion was observed, with an activation energy of 2.0 ± 0.2 eV, but no loops formed.

This paper deals with the application of spectroscopic ellipsometry (SE) and cross-section transmission electron microscopy (XTEM), to the characterization of damaged surface layers in ion implanted Si single crystal. Si samples of 2–6Ω·cm resistivity and <100> orientation were implanted with 28Si+ ions in the dose range of 1.0 × 1016–1.5 × 1016 ions/cm2 using ion energies of 100 and 200 keV. Ion current densities were varied from 6 to 200 μA/cm2. Depth profiles of the implanted samples were evaluated from the spectroscopic ellipsometry data. These calculated profiles were compared with the TEM micrographs of the cross sections of the samples. Excellent agreement is obtained between the two characterization techniques. The characteristics of the depth profiles of the samples, as established by the two techniques, is shown to be the result of annealing occuring during implantation.

The red shift and asymmetric broadening of the LO phonon mode of ion-implanted GaAs are both described quantitatively by a spatial correlation model based on a damage-induced relaxation of the momentum selection rule previously used by Richter, Wang, and Ley to describe similar effects in microcrystalline Si. The success of the model for a qualitatively different disorder microstructure suggests it may be possible to evaluate average sizes of crystallographically perfect regions in semiconductors from the phonon lineshapes of their Raman spectra.

High resolution transmission electron microscopy (HRTEM) has been applied to the study of amorphization and recrystallization mechanisms in Se+ implanted (100) GaAs. An Se+ dose of 1 × 1014 cm−2 at 450 keV (projected range 1550Å) produced an amorphous band in the depth range 250 to 2150Å below the surface. Annealing at 400°C resulted in the epitaxial regrowth of the upper and lower transition region (0–250Å and 2150 – 2500Å, respectively). Regrowth of the amorphous layer was found to proceed by the nucleation and propagation of the dense network of stacking faults bundles. These bundles disappeared on higher ≥600°C) temperature annealing. Amorphization and recrystallization mechanisms in Se+ implanted GaAs are discussed in light of these HRTEM results.

We have studied the formation of heavily doped n-type layers in LPE GaInAs using ion implantation. 400 keV selenium ions have been implanted in dose ranges of 5 × 1013 to 1 × 1015 cm−2 at room temperature. For the high dose implants we have reproducibly achieved activities of 20–40% and sheet Hall mobilities of 700–1000 cm−2 V−1 s−1 and peak carrier concentrations of about 1019 cm−3. TEM and RBS results indicate that for long time anneals residual damage persists in the implanted layers, however, anneals at 800°C for 30 seconds perfectly recrystallize the implanted layers.

Implantation damage in single crystal of CdS produced by 60 keV Bi+ and 45 keV Ne+ at 50 K and at 300 K has been studied. Measurements of Cd disorder and dechanneling behaviour have been made by means of RBS/channeling for He ions ranging in incident energy from 1.0 to 2.8 MeV either along <0001> or <1120> axial channeling directions. The amount of disorder measured were two orders of magnitude lower than the calculated Cd disorder. Damage when analysed along the <0001> axis is larger than when analysed along the <1120> axis.Xmim values for the implanted crystals decreases as the EO increases, when analysed along <0001> direction. TEM observations of Bi implanted samples show that the dislocation loops of b = 1/3 <1120> are produced. Attempts have been made to correlate the RBS/channeling results with the defect strictures observed in microscopy.

Annealing in the 600–900°C temperature range, using either a halogen lamp for periods of seconds or a furnace for periods of minutes, has been applied to activate Be implanted InP samples. The Be was ion-implanted at room temperature into InP substrates. The substrates were uncapped and in close contact with another smooth surface during annealing.

It was found that in the 1017–1019/cm3 range, the acceptor concentration increased with the temperature of anneal. It was also found that redistribution effects decreased with decreasing anneal time. For short anneals, the optimum condition for an 80–90% activation in the 1017–1018/cm3 range was estimated to be 10 seconds at 950°C. It was also found that after rapid anneal, the carrier concentration profile closely approximated the asimplanted Be profile. In the case of 1018–1019/cm3 implants, which were rapidly annealed, there was a low concentration component (1016–1017/cm3) to the redistribution; this ancillary component to the main active distribution was detected by SIMS and was electrically inactive. For long term annealing this electrically inactive Be component was partly converted to substitutional Be and became electrically active. Rapid thermal annealing eliminated this conversion.

Proton and α-particle channeling were used to study the radiation damage caused by the implantation of 160 keV Mg ions in InSb. The implantations took place at various substrate temperatures ranging from room temperature to temperatures just below the melting point and doses ranging from 5.1013 to 1.1016 Mg+ cm−2. The isochronal annealing of the room temperature implanted crystals started at 200°C and damage could not be completely removed even at temperatures just below the melting point. For crystals implanted at elevated substrate temperatures no annealing effects during implantation occured up to 400°C. Above 400°C a sharp reduction of damage indicates that the rate of formation of more complex defect configurations during the implantation process becomes smaller than the annihilation rate of the vacancy-interstitial pairs. A non-linear dependence exists between the degree of radiation damage in the InSb lattice and the implanted dose.

The dependence of the implanted layer composition on total dose, dose rate and target chamber environment for Cr+ implanted Si have been studied by means of Rutherford Back Scattering (RBS) and Auger Electron Spectroscopy (AES). Implantation of Cr+ for doses up to 2 × 1018 ions/cm2 and a fixed dose rate and energy were carried out in an ultra high vacuum (UHV) system as well as in a diffusion pumped vacuum (DPV) system. For the former, the maximum Cr concentration was about 42%. On the other hand, implantation of Cr in a DPV system resulted in a much higher peak concentration (86%) and retention.

Both the RBS and AES results positively demonstrate the existence of extensive surface carbon for a Si-rich surface and a chromium oxide layer for the Cr-rich surface. This result suggests that the interaction of oxygen or carbon occurs preferentially and depends on the surface composition.

No surface compositional variation could be observed by the RBS experiments for Cr implanted in a UHV system for different dose rates. In contrast, for implantation in a DPV system, higher concentrations can be achieved for lower dose rates.

We report the preliminary results of a study to determine the dependence of the near surface composition and structure on total dose, dose rate, vacuum condition and substrate orientation for Al implantation into Si (111) and Si (100) with doses up to 2 × 10l8 ions/cm2. Our studies include the results of Rutherford Back Scattering (RBS), Auger Electron Spectroscopy (AES) and x-ray diffraction measurements on samples implanted with a 100 keV energy in a diffusion pumped vacuum (DPV) system (10−6 Torr) with and without a LN2 trap and in an ultra high vacuum (UHV) system (2–4) x 10−8 Torr.

Results of high dose rate (50 μA/cm2 ) implantation into Si (111) in an untrapped DPV system indicate that Al segregates with a preferred (111) orientation. For a dose of 1 × 1018 ions/cm2 the surface is Al-rich to a depth of 2500Å while for lower doses the surface is silicon-rich. A carbon build-yp occurred for samples prepared by low dose rate (5 μA/cm2 ) implantation. However, no Al segregation could be observed for doses of less than 1018 ions/cm2 . A similar behavior has been observed for Si (100) except that Al segregation occurs with a polycrystalline structure. Moreover, the segregated Al is present at depths greater than the projected range.

When implantation was carried out in a DPV system with a LN2 trap, no carbon peaks could be observed by RBS regardless of the dose rate. For these conditions, as well as for the implantation of Al in an UHV system, we find Al segregation with a polycrystalline structure independent of the dose rates and target orientations we used. Al is observed at a depth greater by a factor of two than the expected value from the Rpcalculations. The Al depth penetration increases with the dose of implantation.

Investigation has been made on the effects of double implant of BF2+−As+, BF2+−P+, B+−As+ and B+−P+ on the thermal annealing behaviors of silicon. For specimens with As+ or P+ as the major dopants, the annealing behaviors were similar to those singly implanted with As+ or P+. For samples with BF2+ or B+ as the major impurities, drastic changes in annealing characteristics were found in comparison with BF2+ or B+ single implant specimens. High density twins, including primary twins on all four {111{ planes and twelve types of secondary twins were observed in BF2+−As+ and BF2+−P+ double implant samples.

Crystals of float-zone, Czochralski, and Web Si having widely differing oxygen concentrations were imrplanted at 300°K with large fluences of hydrogen. Experiments using RBS/channeling, profilometry and microscopy are reported. For room temperature implants the step-height h, the distribution of displaced atoms ND(x) and the total number of displaced atoms/cm2 (aD) are all essentially independent of the oxygen convent of the Si.

Arsenic and argon implantation damage is characterized by Rutherford backscattering in GaAs undoped VPE buffer layers grown on Cr-O doped semi-insulating substrates and capless annealed in a H2 −As4 atmosphere provided by AsH3. The damage detected in the RBS channeled spectra varies as a function of the ion mass, the implant depth and the annealing temperature of the stress-free controlled atmosphere technique. This damage is discussed in terms of the stoichiometric disturbances introduced by the implantation process. The as-implanted and annealed damage characteristics of the Ar and As implants are correlated to the electrical activation characteristics of Si and Se implants in GaAs, respectively.

To measure the sputtered ions during implantation a specially designed UHV-target chamber with a SIMS apparatus was set up. Quantitative analysis are possible with an Auger spectrometer. Disturbances in the stoichiometry in InP are measured during implantation of Sn. The enrichment of the doped surface of InP with the lighter component phoshorus will be discussed in consideration of preferential sputtering and recoil effects during implantation. Measured depth profiles of Sn in InP will be compared with calculated distributions on condition that sputtering takes place. The sputtering yield of InP bombarded by 120 keV Sn+ is 17±5.

Samples of n+-GaAs implanted with 300 keV protons have been examined using high resolution electron microscopy, capacitance-voltage profilometry, and infrared reflectance. In contrast to previously reported results, electron microscopic examination of the as-implanted samples revealed the presence of dislocation loops and/or precipitates both near the wafer surface and at the bottom of the implanted layer. These results are corroborated by electrical and optical measurements.

Alterations to the structure and properties of ceramics are complex due to the range of bonding types encountered and the necessity for maintaining local charge balance. Ion damage can occur as a result of ionizing effects as well as displacement collisions. Ion species, implantation temperature, implantation energy, and the specific bonding characteristics of the host are important parameters in determining the structure and properties of implanted ceramics. Some of these effects will be illustrated for Al2O3 implanted with chromium or zirconium and silicon carbide implanted with chromium.