Rutherford Backscattering Spectrometry (RBS) is a non-destructive, quantitative ion beam analysis technique widely used for compositional and structural characterization of semiconductor materials and thin films. By measuring the energy spectrum of backscattered ions, RBS delivers standards-free quantification of elemental composition, film thickness, areal density, and depth profiles, making it an essential tool for process development, failure analysis, and quality control in the semiconductor industry.
Technical Overview
Rutherford Backscattering Spectrometry (RBS) is an ion beam analysis technique in which a collimated beam of light ions, most commonly 2.0 MeV helium ions (He+ or He2+), is directed at a solid sample inside a high-vacuum chamber. A small fraction of these ions undergo elastic scattering from atomic nuclei near the sample surface. The energy of each backscattered ion carries two layers of information: the kinematic factor, which is determined by the mass of the target atom, identifies the element, while the incremental energy loss of the ion traveling into and out of the sample establishes the depth scale. Because the Rutherford scattering cross sections are precisely known, RBS provides absolute quantitative results without requiring reference standards, a decisive advantage over many complementary techniques such as XPS, AES, or EDS.
- Absolute, standards-free quantification of elemental composition with typical accuracy of 1–5%
- Non-destructive analysis with depth information extending roughly 1–2 micrometers for standard helium beams
- Near-surface depth resolution of approximately 5–10 nm, degrading gradually with depth
- High sensitivity for heavy elements in light matrices, reaching detection limits near 0.1 atomic percent
- Compatible with crystalline channeling measurements for lattice damage and defect evaluation
Principle of Measurement
When an energetic helium ion collides elastically with a stationary target nucleus, the ion retains a fraction of its energy defined by the kinematic factor K, which depends on the mass ratio of the collision partners and the scattering geometry. Heavy target atoms such as hafnium or gold backscatter ions with relatively high energy, whereas light atoms such as oxygen or nitrogen produce signals at significantly lower energies. As ions penetrate the sample, they continuously lose energy through electronic stopping, so atoms located deeper in the film return ions with lower measured energies. A silicon surface barrier detector or PIPS detector positioned at a backscattering angle of typically 165–170 degrees records the energy spectrum, which is subsequently simulated using software such as SIMNRA, RUMP, or DataFurnace to extract thickness, composition, and depth distribution.
Key Performance Characteristics
- Mass resolution: excellent for light and medium-Z elements; neighboring heavy elements may require geometry optimization or heavier probe ions to resolve
- Depth resolution: approximately 5–10 nm at the surface of a silicon substrate under normal incidence
- Analysis depth: up to approximately 1–2 µm for 2 MeV He; deeper profiling is possible with higher-energy beams
- Beam spot: typically 1–2 mm in diameter, providing an averaged large-area measurement
- Environment: ultra-high vacuum chamber to prevent beam scattering from residual gas
Testing Samples
RBS accommodates a broad range of semiconductor materials and device structures, from blanket monitor wafers to patterned fragments. Samples should be reasonably flat and larger than the beam footprint, typically at least 5 x 5 mm, although full wafer sections can often be accommodated depending on chamber geometry.
Silicon-Based Thin Films and Multilayer Stacks
Dielectric layers such as SiO2, Si3N4, and silicon oxynitrides, polycrystalline silicon, amorphous silicon, and oxide-nitride-oxide multilayers are routinely measured for thickness, stoichiometry, and interfacial mixing.
High-k Dielectrics and Gate Stack Materials
Hafnium oxide, zirconium oxide, aluminum oxide, hafnium silicates, and lanthanum-based oxides are analyzed for composition, metallic contamination, and interfacial layer growth, together with metal gate layers such as TiN and TaN.
Silicides and Metallization Layers
Transition metal silicides including TiSi2, CoSi2, NiSi, and NiPtSi are characterized for phase stoichiometry, thickness uniformity, and substrate consumption after rapid thermal processing.
Compound Semiconductors and Optoelectronic Materials
III-V and II-VI materials such as GaN, AlGaN, InGaAs, GaAs, CdTe, and HgCdTe, as well as photovoltaic absorbers like CIGS and CZTS, are measured for alloy composition and depth uniformity.
Ion-Implanted and Processed Wafers
Implanted layers with heavy dopants or species such as Ge, As, Sb, or Sn can be quantified for retained dose and post-anneal redistribution.
Testing Items
RBS analysis delivers a comprehensive set of quantitative results tailored to semiconductor process and materials engineering needs.
- Elemental composition and stoichiometry of films, expressed in atomic percent
- Layer thickness and areal density in atoms per square centimeter
- Depth profiles of constituent and impurity elements
- Retained dose verification of ion-implanted species
- Interfacial reaction, interdiffusion, and mixing assessment
- Density evaluation when thickness is independently known
- Crystalline quality indicators, including minimum yield, defect density depth distribution, and impurity lattice location via channeling
Quantitative Composition and Stoichiometry
Because scattering yields scale directly with areal density and known cross sections, film compositions are obtained absolutely, which is particularly valuable for verifying high-k dielectric stoichiometry or alloy ratios in compound semiconductors.
Thickness and Areal Density
RBS measures areal density directly from the integrated signal of each layer. Combined with independently determined physical thickness from ellipsometry, X-ray reflectometry, or TEM, the technique provides reliable film density values.
Depth Profiling
Energy loss of the probing ion converts the spectrum into a depth scale, enabling non-destructive depth profiles of layers up to roughly one micrometer, including buried layers and interfacial reaction zones.
Crystallinity via Channeling RBS
Aligning the ion beam with a major crystallographic axis reduces backscattering yield from a well-ordered lattice. The minimum yield quantifies crystal perfection, while the depth dependence of dechanneling reveals damage profiles in implanted or processed layers.
Testing Methods
Analyses are performed on a megavolt tandem or single-ended accelerator equipped with a scattering chamber, precision goniometer, and semiconductor particle detectors. Standard operating conditions use a 2.0 MeV helium beam at a backscattering angle near 170 degrees.
Standard Random RBS
The sample is oriented to avoid channeling directions, and the energy spectrum is collected to determine composition, thickness, and depth profiles. Simulation software fits the spectrum using known stopping powers and scattering cross sections, with mass and depth resolution optimized through detector geometry selection.
RBS Channeling Analysis
A goniometer aligns the beam to a crystal axis or plane of single-crystal samples. Comparing aligned and random spectra yields the minimum channeling yield, quantifies lattice disorder versus depth, and can locate impurity atoms on substitutional or interstitial sites.
Elastic Recoil Detection Analysis (ERDA)
For hydrogen and other light elements that standard RBS cannot detect, a heavier ion beam such as silicon or chlorine recoils light nuclei toward a forward detector, often with a range foil to stop scattered primaries. ERDA extends coverage to hydrogen content and moisture-related issues in dielectrics.
Complementary Ion Beam Techniques
Depending on the analytical question, RBS can be combined with nuclear reaction analysis for light element quantification or particle-induced X-ray emission for trace heavy element screening, providing a complete ion beam characterization workflow.
Sample Preparation and Handling
- Clean the surface of organic contamination; avoid thick native oxides if they interfere with surface-region analysis
- Provide samples at least 5 x 5 mm in size; larger pieces or wafer sections are acceptable
- Insulating samples are analyzed with electron flood compensation to control charging
- Mark the region of interest, since the beam integrates over a 1–2 mm footprint
Applications
RBS is deployed across research, development, and manufacturing stages of semiconductor technology wherever absolute composition and depth information are required.
- Verification of high-k gate dielectric composition and interfacial layer growth in advanced CMOS gate stacks
- Silicide process window optimization, including phase identification, thickness control, and substrate consumption measurement
- Ion implantation qualification through retained dose measurement and post-anneal damage recovery assessment
- Barrier layer integrity studies for copper metallization, including TaN and TiN stability against diffusion
- Stoichiometry control of compound semiconductor epilayers for LED, laser, RF, and power device applications
- Thin film photovoltaic absorber development, such as composition grading in CIGS and CZTS structures
- Ferroelectric and piezoelectric film qualification, including PZT and AlScN composition verification
- Failure analysis of interfacial reactions, oxidation, and intermixing in multilayer device stacks
FAQ
What elements can RBS detect?
RBS readily detects all elements heavier than the probing ion, which in practice means from carbon and nitrogen upward when helium beams are used. Sensitivity is highest for heavy elements in light matrices and lowest for light elements in heavy matrices. Hydrogen cannot be detected by conventional RBS and requires ERDA.
Is RBS a destructive technique?
No. The helium ion dose used for a spectrum is low enough that sputtering, lattice damage, and heating are negligible for most semiconductor materials. Samples remain suitable for subsequent analyses such as TEM or electrical testing.
What are the typical detection limits?
For heavy elements in a light matrix, detection limits approach 0.1 atomic percent, corresponding to areal densities on the order of 10^13–10^14 atoms per square centimeter. Light elements embedded in heavy matrices may exhibit limits of several atomic percent.
How deep can RBS profile?
With a standard 2 MeV helium beam, useful depth information extends roughly 1–2 micrometers, with the best depth resolution of 5–10 nm preserved near the surface and degrading with depth due to energy straggling.
How does RBS compare with SIMS, XPS, and TEM-EDS?
RBS offers absolute quantification without standards and non-destructive depth profiling, whereas SIMS provides superior trace sensitivity with dynamic range down to ppm levels but requires calibration standards. XPS offers surface sensitivity and chemical state information but probes only the top few nanometers without sputtering. TEM-EDS provides nanoscale spatial resolution but semi-quantitative accuracy at typical operating conditions. The techniques are complementary and often used together.
Can RBS determine crystal quality?
Yes. Channeling RBS measures the minimum backscattering yield along a crystal axis; high-quality silicon typically shows minimum yield below 2 percent. Depth-resolved dechanneling analysis quantifies implantation damage, and angular scans can determine whether impurity atoms occupy substitutional or interstitial lattice sites.
What accuracy can be expected for composition and thickness?
With proper calibration and simulation, absolute compositions are typically accurate to 1–5 percent, and areal densities to approximately 3–5 percent. Depth accuracy depends on the stopping power database and is generally better than 5 percent for common semiconductor materials.
What sample sizes and forms are accepted?
Standard measurements require pieces of at least 5 x 5 mm with a reasonably flat surface. Wafer fragments, cleaved chips, and coated witness coupons are all suitable. Insulating samples can be measured with charge compensation, and patterned areas can be analyzed provided the region of interest exceeds the beam footprint.