We describe how a simple qualitative understanding of the interfacial reactions occurring during typical ULSI processes for junction formation, dopant activation, and contact silicidation can be used to eliminate end-of-range interstitial dislocation loops and beneficially impact the diffusion of dopants. Following a brief discussion of the well-documented effects of oxidation and nitridation on extended defects and dopant diffusion, conditions for elimination of implantation-induced defects are specified. Cross-section and plan-view TEM along with angle lapping and chemical etching of implanted and diffused junctions are presented to illustrate the application of point defect engineering to process technology.

We studied formation and annihilation of dislocation loops formed beyond the amorphous/crystalline interface after indium and boron dual implantation and subsequent annealing in the 800–1 100°C temperature range. The residual damage for low (40 keV) and high (200 keV) energy In implants were compared. The depth of the amorphous region in the sample implanted with the higher energy ions was reduced by using anodic oxidation and etching, to equate it with that of the sample implanted by lower energy ions. This enabled the study of the effect of surface proximity on residual disorder upon annealing. The damage was strongly dependent on the energy of In ions. No end-of-range damage was observed for the low energy implant. High energy implantation resulted in end-of-range dislocation loops, stable below 1050°C. The loops kinetics was neither affected by their proximity to the surface, nor by In precipitation. Monte-Carlo full cascade simulation has been used to estimate the depth distribution of interstitials and vacancies produced by In implant.

Shallow junctions were formed in single-crystal Si(100) by implantation of As at energies between 2 and 17.5 keV followed by conventional furnace annealing or by rapid thermal annealing (RTA). Cross-sectional transmission electron microscopy (XTEM) showed that defect-free shallow junctions could be formed at temperatures as low as 700 °C by RTA, with about 60% dopant activation. From a comparison of short-time and long-time annealing, it is proposed that surface image forces are responsible for the efficient removal of end-of-range (EOR) dislocation loops

Recent results indicate that thermal budgets associated with ion implantation induced end of range damage removal is affected by the presence of a free surface. Low energy BF2 implants (6 keV) were done into both single crystal and Ge preamorphized silicon substrates. Rapid thermal processing was used to study the residual end of range defect structure in the temperature range from 700 to 1000°C. 6 keV, 5E14 cm-2 BF2 implantation resulted in formation of continuous amorphous layers approximately 10 nm deep with a mean B penetration of approximately 7 nm. Conventional TEM analysis found the structures to be completely free of any spanning “hairpin” dislocations or stacking faults associated with the BF2 implant for all the annealing temperatures. For anneals between 700 °C and 900°C end of range damage formation resulted, but the size of the dislocation loops remained small. Annealing at 1000°C, 10 seconds showed no evidence of residual end of range damage. Location of the end of range damage region close to the free surface was found to decrease the thermal budget required for the removal of ion implantation induced radiation damage.

Complete removal of end-of-range (EOR) defects in ion implanted silicon has been achieved by the formation and growth of near noble silicides (CoSi 2 and NiSi 2 ) and refractory silicides (MoSi 2 and WSi2). Continued generation of vacancies during the silicide growth was found to be essential to reduce EOR defects. The results are consistent with the suggestion of the presence of a vacancy diffusion barrier near the EOR defects

Extensive experimental investigations which have been reported on the ion-induced motion of the interface between the crystalline and amorphous phases of silicon are shown to fit a model based on a single defect. The model accounts for the temperature and flux dependence of the interface motion. The defects annihilate each other by binary recombination, and have a motion energy of 1.2 eV. The defect is believed to be the dangling bond in the amorphous phase.

Using the new technique of Z-contrast scanning transmission electron microscopy (STEM), we have been able to study the segregation of Sb at an advancing SPE growth interface and the resulting interface breakdown. The first direct information is obtained on Sb diffusion in the amorphous phase, which is many orders of magnitude enhanced over tracer crystalline values. This controls both the dopant incorporation and the stability of the resulting supersaturated alloy. These results are compared to the behavior of the low melting point substitutional diffusers and the interstitial diffusers.

Heavily doped Si:Ga has been prepared by liquid phase epitaxy (LPE) and by ionimplantation with rapid thermal annealing (RTA) or laser annealing (LA). Peak substitutional Ga concentrations obtained by each technique were 1.5, 2.5 and 2.9 ×1020cm-3, respectively. Substitutional fractions (>90%) were similar in the three types of samples, and the conductivity scaled with the total Ga concentration. A lattice expansion per substitutional Ga atom in Si of +0.9 ± 0. l×10-24cm3 /atom was measured by double crystal x-ray diffraction. The average nearest neighbor Si-Ga bond length measured with extended x-ray absorption fine structure (EXAFS) was 0.237 ± 0.004 nm, indistinguishable, to within experimental error, from the intrinsic Si-Si bond lngth, 0.235 nm. Combining these two results the lattice strain per hole in the Si valence band was calculated, +0.4 ± 0.8x10G-cm3. This result complements the lattice contraction per electron in the Si conduction band (-1.8 ± 0.4x10-24cm 3) already reported for Si:As [G. S. Cargill III, J. Angilelloand K. L. Kavanagh, Phys. Rev. Letters 61, 1748 (1988)].

We have investigated a phenomena first reported by Pfiester and Griffin, that the presence of implanted Ge in Si can substantially reduce excess self-interstitial concentrations [ J. R. Pfiester and P. B. Griffin, Appl. Phys. Lett., 52, 471 (1988) ]. By studying the effects of Ge implantation on P diffusion, we are able to deduce that residual implantation damage can act as an efficient sink for self-interstitials. This effect can also be produced by Si self-implantation, demonstrating that there is nothing unique about the chemical indentity of Ge in reducing self-interstitial concentrations. Our experiments provide solid evidence that there is no unexpectedly strong interaction of Ge with self-interstitials, a situation that would undermine the validity of previous Ge diffusion experiments aimed at studying Si self-diffusion. Our experimental results show that the effect of Ge implantation on P diffusion is a complicated function of implantation conditions. Diffusion is affected by the order of thl P and Ge implants as well as by changes in implant energies and doses.

Rapid thermal annealed tail regions of shallow junction arsenic implants into silicon have been investigated. Tail profiles have been roduced by an anodic oxidation and stripping technique after implantation to fluences of 1014 to 1016 cm−2 and by implanting through a layer of silicon dioxide. Electrical activation and diffusion have been achieved by rapid thermal annealing in the temperature range of 800 to 1100 °C. Electrically active defects remain after annealing. Spreading resistance and deep level transient spectroscopy results are presented. The diffusion of the arsenic tail is discussed and compared with currently accepted models.

In device fabrication, dopants are frequently implanted into silicon through silicon dioxide masks. A consequence of this technique is the co-implantation of recoiled oxygen into the substrate. This study investigates the effect of recoiled oxygen on the widely observed transient enhanced boron diffusion. Comparison of the spreading resistance profiles of annealed through-oxide and directly implanted samples reveals that transient enhanced diffusion of boron can be suppressed by the former process. Continued annealing of the through-oxide implanted silicon recovers the enhanced diffusion of boron. This behavior is believed to be due to precipitation of recoiled oxygen. The mechanisms leading to the above observations are discussed and transmission electron microscopy support presented.

We have performed a systematic investigation of boron implantation at 30 keV into <100> n-type silicon in the 77 –300 K temperature range and mostly at 9×1015 cm−2 fluence. The analyses have been performed with ion channeling and cross sectional transmission electron microscopy both in as-implanted samples and in samples annealed in vacuum furnace at 500 °C and 850 °C for 30 min. We confirm the impossibility of amorphization at room temperature and the presence of residual damage mainly located at the boron projected range. On the contrary, a continuous amorphous layer can be obtained for implants at 77 K and 193 K; the thickness of the implanted layer is increased by lowering the temperature, at the same time the amorphous-crystalline interface becomes sharper. Sheet resistance measurements performed after isochronal annealing shows an apparent reverse annealing of the dopant only in the sample implanted at 273 K. The striking differences between light and heavy ions observed at room temperature implantation disappears at 77 K and full recovery with no residual damage of the amorphous layer is observed.

We have utilized plasma immersion ion implantation (PIII) to demonstrate effective gettering of metallic impurities in silicon wafers. Metallic impurities such as Ni, Cu or Au were intentionally diffused into Si as marker impurities. The Ar or Ne atoms were ionized in an electron cyclotron resonance (ECR) plasma chamber. The ions were accelerated by a negative voltage applied to the wafer and implanted into the wafer. The as-implanted saturation dose can be as high as 5×1016cm−2. After an annealing step at 1000°C for 1 hour in a N2 ambient, the retained doses and the amount of gettered impurities were measured with Rutherford backscattering spectrometry (RBS). With a retained Ar dose in 1015cm−2 range after annealing, the gettered Ni, Cu and Au were 3.0×1014cm−2, 3.0×1014cm−2 and 4.4×1013cm−2 respectively.

We have observed strong gold gettering by implanted carbon in silicon. It was found that the gettering agents in carbon implanted layers are point defects associated with singular carbon atoms. The positions of the gettered Au atoms were found to be distorted substitutional sites. A pointdefect gettering model is proposed to explain our findings.

Amorphous Si is nucleated heterogeneously at grain boundaries during irradiation of polycrystalline Si by 1.5 MeV Xe+ ions for temperatures of 150–225°C. Following formation at grain boundaries, the amorphous Si layer grows at a rate comparable to the growth rate of a pre-existing amorphous-crystal interface, resulting in a decrease in average grain size and a marked change in the grain size distribution. The heterogeneous nucleation kinetics of amorphous Si are strongly dependent on grain boundary structure. A simple atomistic model for amorphous phase formation, which suggests that the nucleation kinetics are dependent on the point defect mobilities and grain boundary structure, is related to the experimental results.

lons of boron, phosphorous, titanium and neon were implanted into (0001) oriented 6H SiC crystals at 300 K. Implantation energies and fluences were chosen to produce equal (calculated) displacements per atom at similar peak damage depths and a randomized (metaminct or amorphous) zone extending to the front surface. RBS/channeling was used to test the amorphization criteria. Dopant effects on regrowth kinetics and microhardness have been determined by isochronal annealing.

Damage introduced during ion implantation of semiconductor materials coalesce at a certain critical dose, for a particular energy and ion species. After this threshold dose rapid changes occur in the reflectance. This may be used for studying the amorphization process, or it may be applied as a non--destructive dosimeter and uniformity tool.

An automated reflectometer was developed for studying reflectance during ion implantation of semiconductors with various ion species. Results on argon and phosphorous implants into silicon at energies ranging from 50 to 240 keV are presented.

Focused ion beam induced deposition is already in use commercially for the repair of clear defects in photomasks, where missing absorber is added. Research is being carried out to extend this technique to the repair of x-ray lithography masks and to the restructuring and repair of integrated circuits, particularly in the prototype phase. In this technique a local gas ambient is created, for example, by aiming a small nozzle at the surface. The gas molecules are thought to adsorb on the surface and to be broken up by the scanned focused ion beam. A deposit is formed with linewidth equal to the beam diameter which can be below 0.1 Ό m. At small beam diameters and low currents (50–100 pA) the time to deposit 1Όm3 is in the vicinity of 10–20 sec. If the gas is a hydrocarbon, the deposit is largely carbon, which is useful for photomask repair. On the other hand, if the gas is a metal halide or a metal organic, the deposit is metallic. The deposits have substantial concentrations of impurities due to the atoms in the organometallic, to the ion species used, or to the ambient in the vacuum chamber. Thus the resistivities of the "metal" films deposited typically range from 150 to 1000 ΌΏcm which is usable for some repairs. (Pure metals have resistivities in the range 2.5 to 12 pQcm.) We have deposited gold from dimethyl gold hexafluoro acetylacetonate and have achieved linewidths down to 0.1 Όm, patches of 1 Όm thickness with steep side walls and in some cases, resistivities approaching the bulk value. Other workers have reported deposits of Al, W, Ta, and Cr. We will review previous work in the field and present some of our own current results.

Cross-sectional transmission electron microscopy observations have been carried out to clarify two-dimensional depth distributions of lattice defects generated in high-dose (5 x 1015 ions/cm2 ), P, As, BF2 and B implanted, annealed submicron Si areas as a function of implantation areas. Monte Carlo simulation is also adapted for ion-implantation into submicron Si through fine mask patterns to predict the effect of mask size on spatial damage and impurity profiles. Simulation results predict that the above profiles have a strong mask size dependence for regions below the critical size, where the dopant concentration decreases and damage depth moves toward the surface-side with a reduced implantation area. Some experimental results support simulation results, although most defects, mainly in the P and As implantation, are confined within the original amorphized layers, independent of mask size. However, in BF2 and B implantation, unexpected defect behavior such as variations in defect distribution from one implanted layer to another is found to occur in submicron regions doped by implantation.

mechanistic model of the ion beam induced deposition (IBID) process is proposed, and an experimental system for measuring IBID is described. In IBID, the kinetic energy of the incident ion is deposited in the substrate surface resulting in the ejection of atoms (sputtering), and the decomposition of adsorbed molecules. Net deposition occurs when the number of atoms added through decomposition exceeds the number removed by sputtering.

Gold films were deposited by decomposing dimethyl gold hexafluoroacetylacetonate (C7H7F6O2Au) molecules with a 5 keV argon ion beam. The beam current density was 5 μA/cm2. These films were deposited on quartz crystal microbalances (QCM). The QCM's provide an in-situ measurement of the deposition rate as a function of ion dose, dose rate, gas pressure, and substrate temperature. The mass of the depositing film increases with increasing ion dose. The deposition yield, or mass deposited per incident ion, is independent of the power input of the ion beam and therefore, the decomposition is not due to macroscopic heating. The deposition yield is shown to depend on the sputter yield of the substrate. The yield increases with increasing gas pressure and decreasing substrate temperature. This increase is proportional to the adsorption of the organometallic gas.