Rutherford Backscattering Spectrometry (RBS) is a quantitative, non-destructive ion beam analysis technique that serves as the gold standard for characterizing oxygen implantation processes in semiconductor manufacturing. This article provides a comprehensive technical analysis of how RBS measures implanted oxygen dose, depth distribution, and layer quality, covering sample requirements, testing protocols, and key applications in SOI wafer production and ion implantation process control.
Technical Overview
Rutherford Backscattering Spectrometry (RBS) is a quantitative ion beam analysis technique that determines the elemental composition and depth distribution of materials in the near-surface region with depth resolution on the order of 10–20 nanometers. In a standard RBS measurement, a collimated beam of light ions—most commonly 2.0 MeV helium ions—is directed onto the sample surface at normal incidence. A small fraction of these ions undergoes elastic nuclear scattering with atoms inside the sample and is backscattered toward a silicon surface barrier detector positioned at a scattering angle of approximately 165 to 170 degrees. Because the energy retained by each backscattered ion depends both on the mass of the target atom and on the depth at which the scattering event occurred, the recorded energy spectrum simultaneously encodes compositional and depth information without requiring external calibration standards.
Oxygen implantation is a critical process step in advanced semiconductor manufacturing, most prominently in the fabrication of Separation by IMplantation of OXygen (SIMOX) silicon-on-insulator (SOI) substrates. In this process, high doses of oxygen ions, typically in the range of 1×10^17 to 2×10^18 ions per square centimeter, are implanted into single-crystal silicon at energies between 50 and 200 keV. A subsequent high-temperature anneal above 1300 degrees Celsius drives the implanted oxygen to react with the silicon matrix, forming a stoichiometric buried silicon dioxide layer while preserving a single-crystal silicon film at the surface. The precise verification of implanted dose, projected range, and depth profile directly determines the final buried oxide thickness, the top silicon film thickness, and the overall electrical performance of devices built on the substrate.
RBS is uniquely suited to analyzing oxygen implantation processes because it provides absolute dose quantification, non-destructive depth profiling, and complementary crystallographic information through ion channeling. In channeling mode, the incident beam is aligned with a major crystallographic axis of the substrate, which suppresses the backscattering yield from the crystalline lattice and highlights the yield from displaced, interstitial, or amorphized atoms. This allows engineers to simultaneously quantify the implanted oxygen distribution and assess the implantation-induced lattice damage, providing a complete picture of the process window before and after annealing.
Testing Samples
SIMOX and SOI Wafers
Fully and partially processed SIMOX wafers represent the primary sample category for RBS oxygen implantation analysis. Both pre-anneal implanted wafers and post-anneal finished SOI structures can be characterized to verify dose retention, buried oxide formation, and silicon overlayer quality.
Oxygen-Implanted Silicon Substrates
Bare single-crystal silicon wafers subjected to oxygen ion implantation at various energies and doses are routinely analyzed for dose calibration, range verification, and process qualification during implanter setup and process transfer.
Compound Semiconductor Materials
Materials such as silicon carbide, gallium nitride, and gallium arsenide implanted with oxygen for defect engineering, semi-insulating layer formation, or optical property modification can be analyzed, provided the oxygen signal can be resolved from the host matrix signal.
Oxide and High-k Dielectric Films
Thin dielectric films incorporating implanted oxygen, including silicon dioxide, hafnium oxide, and mixed oxide stacks, can be evaluated for stoichiometry, thickness, and interfacial mixing.
Sample Requirements
- Flat, smooth surface with roughness preferably below a few nanometers to preserve depth resolution.
- Minimum sample area of approximately 5 × 5 millimeters; full wafers up to 300 millimeters can typically be accommodated with appropriate handling fixtures.
- Conductive samples are preferred; insulating samples may require electron flood gun compensation or thin conductive coating considerations to mitigate charging.
- Samples must be vacuum compatible and free of loose particulates, oils, or organic contamination.
Testing Items
RBS analysis of oxygen-implanted materials delivers a set of quantitative process metrics that directly map onto implantation process parameters.
- Retained oxygen dose in atoms per square centimeter, measured with a typical absolute accuracy of 3 to 5 percent.
- Depth distribution of implanted oxygen, including the projected range and straggle of the implanted profile.
- Peak oxygen concentration at the implantation maximum, expressed in atomic percent.
- Buried oxide layer thickness and effective stoichiometry in annealed SIMOX structures.
- Interface abruptness between the silicon overlayer, buried oxide, and substrate.
- Surface native oxide thickness, which contributes to the total oxygen signal and must be deconvolved from the implanted profile.
- Crystalline quality assessment through channeling measurements, including the minimum yield of the substrate and the damage peak intensity in the implanted region.
- Interdiffusion or mixing assessment after thermal processing.
Testing Methods
Measurement Geometry and Beam Conditions
The standard configuration uses a 2.0 MeV helium ion beam in Cornell or IBM geometry, with the detector at a scattering angle near 170 degrees to maximize mass resolution and depth sensitivity. The accumulated beam charge, measured through current integration on the sample holder, normalizes the spectrum yield so that dose values are obtained absolutely. Typical beam currents range from 10 to 50 nanoamperes, and total charge per spectrum is commonly on the order of 10 to 40 microcoulombs depending on the required statistics.
Random Spectrum Acquisition
A random spectrum is collected with the sample rotated and tilted several degrees away from any channeling direction so that all lattice atoms contribute equally to the scattering yield. This spectrum contains the oxygen depth profile: because oxygen is lighter than silicon, the oxygen signal appears as a distinct step superimposed on the low-energy edge of the silicon substrate signal, with the step height proportional to local oxygen concentration and its width reflecting the implanted layer thickness.
Channeling Spectrum Acquisition
A second spectrum is acquired with the beam aligned along a major crystal axis, such as the silicon (100) or (110) direction. The aligned spectrum reveals the dechanneling and direct scattering from displaced atoms, allowing quantitative evaluation of implantation damage and recovery after annealing.
Spectrum Simulation and Data Reduction
- Spectra are fitted using simulation codes such as SIMNRA or RUMP, which model elastic scattering kinematics, stopping powers, and energy straggling.
- A layered model is constructed containing the surface oxide, silicon overlayer, oxygen-rich buried layer, and crystalline substrate.
- Simulation parameters, including layer thicknesses, oxygen concentrations, and areal densities, are iteratively adjusted until the simulated spectrum matches the experimental data.
- Stopping power uncertainties of roughly 3 percent dominate the absolute dose error budget.
Quality Assurance
Reference measurements on standard materials, such as thermally grown SiO2 films of certified thickness, are performed periodically to validate energy calibration, detector resolution, and charge collection accuracy, ensuring long-term reproducibility of dose measurements across measurement campaigns.
Applications
RBS oxygen implantation analysis supports a broad range of industrial and research activities across the semiconductor value chain.
- SIMOX SOI wafer manufacturing, where buried oxide thickness and dose uniformity directly govern wafer yield and device isolation performance.
- Ion implanter qualification and dose calibration, providing traceable absolute dose verification that complements internal dosimetry.
- Process development for low-dose and modified SIMOX variants, including internal thermal oxidation steps and dose-energy matrix optimization.
- High-temperature anneal recipe development, tracking oxygen redistribution and layer sharpening as a function of anneal conditions.
- Compound semiconductor defect engineering, where oxygen implantation creates semi-insulating regions for device isolation in GaN and SiC technologies.
- Academic and materials research on ion-solid interactions, defect accumulation, and implantation-enhanced oxidation mechanisms.
- Failure analysis and incoming material inspection for SOI substrates used in RF, power, and radiation-hardened device production.
FAQ
What is the detection limit for oxygen in silicon by RBS?
Because oxygen scatters helium ions more weakly than the heavier silicon matrix, the oxygen signal appears as a small feature superimposed on the silicon edge. Practical detection limits are approximately 0.5 to 1 atomic percent for oxygen in silicon, which is fully sufficient for implanted doses used in SIMOX and isolation processes but may be insufficient for trace-level contamination studies.
Is RBS a destructive technique?
For inorganic semiconductor samples, RBS is essentially non-destructive. The helium beam deposits negligible heat and causes no measurable material loss or chemical alteration, so samples can be returned to production or subjected to subsequent process steps after analysis.
What depth range can RBS probe?
With a 2.0 MeV helium beam, the analyzable depth in silicon extends to roughly 1 to 2 micrometers, comfortably covering the projected ranges of oxygen implants at energies up to several hundred keV. Higher beam energies can be used for deeper profiles when required.
How accurate is the implanted dose measurement?
Dose accuracy is typically 3 to 5 percent, limited primarily by stopping power knowledge and detector solid angle calibration. Precision between repeated measurements on the same sample is considerably better, often within 1 to 2 percent, making RBS highly effective for comparative process studies.
How does RBS compare with Secondary Ion Mass Spectrometry for oxygen profiling?
SIM offers lower detection limits and better depth resolution, but it is inherently destructive and requires calibration standards. RBS provides absolute quantification without standards and adds channeling-based crystal quality information, so the two techniques are complementary rather than competing; many process qualification programs specify both.
What sample sizes and formats can be accepted?
Pieces as small as 5 × 5 millimeters and full wafers up to 300 millimeters in diameter are commonly handled. Angled cleaved pieces are acceptable provided the analysis region is flat and the surface normal is well defined.
Can RBS distinguish implanted oxygen from native surface oxide?
Yes. The surface oxide produces an oxygen signal at the highest energies, separated from the deeper implanted peak by the silicon overlayer thickness. Simulation software deconvolves the two contributions, reporting surface oxide and implanted dose independently.
What turnaround time should be expected?
Standard single-point measurements with simulation and reporting are typically completed within three to five working days, while wafer mapping or multi-condition process matrices may require additional scheduling depending on the number of measurement points.