Rutherford Backscattering Spectrometry (RBS) Testing of Alloy Thin Films: Principles, Methods, and A

Rutherford Backscattering Spectrometry (RBS) is a non-destructive ion beam analysis technique that provides standardless, quantitative measurement of composition, thickness, and depth distribution in alloy thin films. By energy-analyzing MeV helium ions elastically scattered from the sample, RBS delivers absolute areal densities with 1–3% accuracy and nanometer-scale depth resolution, making it indispensable for metallization qualification, diffusion barrier evaluation, and multilayer process control. This article explains the physical principles, measurable parameters, test methods, and industrial applications of RBS testing for alloy thin films.

Technical Overview

Rutherford Backscattering Spectrometry is a quantitative surface and thin film analysis technique based on the elastic scattering of light, high-energy ions—most commonly 1.5–3.0 MeV helium ions—from the nuclei of atoms in a solid target. When a collimated He+ or He2+ beam strikes an alloy film, a small fraction of the ions undergo Coulomb (Rutherford) scattering at large angles and are detected by a charged-particle detector positioned near the backscattering direction, typically at 160–170 degrees. The measured energy spectrum of the backscattered ions encodes both the masses of the atoms present in the film and their depth distribution.

Physical Principle

Each backscattered ion loses energy in two distinct processes: a discrete energy transfer during the elastic nuclear collision, governed by the kinematic factor K, and continuous electronic energy loss along the inbound and outbound paths, described by stopping powers. The kinematic factor increases monotonically with target atomic mass; for scattering at 180 degrees, K is approximately equal to the square of (M2 − M1)/(M2 + M1), where M1 is the projectile mass and M2 is the target nucleus mass. Consequently, the energy axis of a spectrum maps directly onto atomic mass, while the additional energy loss suffered by ions emerging from greater depth maps onto a depth scale. Because the Rutherford scattering cross section is a well-known analytic function proportional to (Z1Z2/E)², quantitative elemental areal densities can be obtained without matrix-matched reference standards.

Information Content of the Spectrum

• Elemental identification from the kinematic edge positions of individual spectral contributions
• Areal density of each element in atoms per square centimeter, determined absolutely from the integrated spectral yield and the measured beam dose
• Atomic composition and stoichiometry of single or multilayer alloy films
• Film thickness derived from spectral peak widths when the layer density is known or assumed
• Depth profiles of composition, interfacial mixing, and diffusion following thermal or environmental exposure

Method Performance Characteristics

• Depth resolution: typically 8–20 nm at the surface for normal incidence, degrading with depth due to energy straggling
• Analysis depth: approximately 1–2 µm for heavy-alloy films, extendable to roughly 10 µm for low-density or light matrices
• Quantitative accuracy: 1–3% relative for major elements in simple binary and ternary alloys
• Detection sensitivity: down to about 0.01–0.1 atomic percent for heavy elements in light matrices; reduced sensitivity for light elements (O, N, C) in heavy matrices, and hydrogen is invisible to conventional RBS
• Non-destructive nature: beam-induced damage is negligible for metallic films
• Mass resolution limitation: adjacent heavy elements with close atomic numbers can produce overlapping signals that must be deconvoluted by spectral simulation

Testing Samples

RBS testing is applicable to any laterally uniform, smooth, vacuum-compatible alloy film or multilayer stack prepared on a solid substrate. Typical sample categories submitted to accredited third-party laboratories and research facilities include the following.

Metallic Alloy Thin Films

Binary and ternary metallization alloys such as Ni–Cr, Cu–Ni, Al–Cu, Ag–Al, and sputtered stainless steel (Fe–Cr–Ni) films are routinely characterized to verify deposit stoichiometry, thickness, and batch-to-batch consistency.

Semiconductor Contact and Silicide Films

As-deposited and annealed Ti, Ni, Co, Pt, and Ni(Pt) layers on silicon, as well as TiSi2, NiSi, and CoSi2 phases, are analyzed for phase stoichiometry, residual unreacted metal, and interfacial reaction depth.

Diffusion Barrier and Protective Coatings

Refractory nitride barriers such as Ta–N, Ti–N, and W–N, aluminide diffusion coatings, and MCrAlY-type overlay alloys are tested for nitrogen content, layer thickness, and elemental interdiffusion after thermal cycling.

Magnetic and Spintronic Multilayers

Stacked structures containing CoFeB, Co–Pt, Fe–Pt, PtMn, IrMn, Ru, and Ta layers are measured layer-by-layer to confirm individual thicknesses, composition, and interface sharpness before and after annealing.

Precious Metal and High-Entropy Alloy Films

Au–Ag, Au–Ni, Au–Pd, and Pt–Ir deposits for electronics and decorative applications, as well as multi-principal-element (high-entropy) alloy films with four or more cations, are verified for target composition.

Sample Requirements

• Flat, smooth surface with roughness preferably below the nominal depth resolution
• Lateral homogeneity across the beam spot, typically 0.5–2 mm in diameter
• Sample dimensions commonly from a few square millimeters up to full wafer sections, depending on chamber design
• Vacuum compatibility; insulating samples may require a thin conductive coating or charge compensation

Testing Items

RBS testing of alloy thin films typically reports the following measurable quantities:

• Elemental areal density (atoms/cm²) for each constituent, obtained as an absolute, standardless result
• Atomic fractions (at.%) and film stoichiometry of single layers and individual layers within multilayer stacks
• Layer thickness (nm), calculated from areal density using the known or assumed layer density
• Composition–depth profiles, including gradual grading such as Ga gradients in photovoltaic absorber alloys
• Interface width, intermixing depth, and diffusion profiles after annealing, oxidation, or service exposure
• Heavy-element impurity and contaminant screening at levels down to the 0.01–0.1 at.% range
• Retained dose and sputtered areal density in ion-implanted or plasma-treated alloy films
• Hydrogen content by elastic recoil detection (ERDA) where light-element quantification is required
• Crystalline quality and lattice disorder in single-crystal or epitaxial alloy films using RBS in combination with ion channeling

Testing Methods

Sample Preparation and Mounting

Specimens are cleaned of particulate and organic contamination, mounted on a manipulator stage with conductive fixation, and aligned to the beam axis with a precision of better than one degree. Reference positions are recorded so that repeated measurements, for example before and after annealing, probe the identical location.

Instrumentation and Measurement Geometry

A tandem electrostatic accelerator produces a mass-analyzed beam of 4He ions, typically at 1.5–3.0 MeV, with currents of a few to a few tens of nanoamperes. Backscattered particles are registered by a silicon surface-barrier or PIPS detector placed at a scattering angle of 160–170 degrees with a solid angle on the order of 5–15 msr. The collected charge, measured with a current integrator on an electrically biased sample, normalizes the spectrum for absolute yield determination. Standard fluences of 5–20 µC keep acquisition times in the range of 10–60 minutes while avoiding any measurable beam damage.

Calibration and Energy Reference

The energy calibration of the detection chain is verified using scattering from reference standards such as a heavy metal film (Au or Bi) on a light substrate, whose kinematic edges appear at precisely calculable energies. Beam energy and detector geometry are confirmed before each measurement campaign to keep systematic uncertainties within 1%.

Spectrum Simulation and Data Reduction

Quantitative results are extracted by simulating the measured spectrum with dedicated software that models the single-scattering approximation, uses evaluated or semi-empirical stopping power databases for the inbound and outbound paths, and applies detector resolution and energy straggling functions. Layer models containing assumed thicknesses, densities, and compositions are iteratively refined until the calculated spectrum matches the measured spectrum over the full energy range. Reported uncertainties combine counting statistics, stopping power uncertainties, and geometric contributions.

Complementary Ion Beam Techniques

• Nuclear Reaction Analysis (NRA) for sensitive quantification of low-Z elements such as nitrogen, oxygen, carbon, and boron in heavy-alloy matrices
• Elastic Recoil Detection Analysis (ERDA) for hydrogen and deuterium depth profiling
• Particle-Induced X-ray Emission (PIXE) for trace-element sensitivity complementing the major-element quantification of RBS
• RBS/channeling on single-crystal films to evaluate epitaxial quality and defect densities

Quality Assurance and Reporting

Testing performed in accredited third-party laboratories is conducted under quality management systems consistent with ISO/IEC 17025 requirements. Test reports typically include the experimental conditions, the measured and simulated spectra, areal densities, atomic compositions, derived thicknesses with stated density assumptions, and a documented uncertainty budget.

Applications

RBS is widely applied wherever absolute, depth-resolved composition data of alloy films are required for development, qualification, or failure analysis:

• Microelectronics metallization: verification of Al–Cu interconnect alloys, Cu seed and cap layers, W, Ti, and TiN stack thicknesses, and silicide contact stoichiometry and reaction depth
• Diffusion barrier engineering: confirmation of Ta–N and Ti–N composition windows and quantitative assessment of barrier breakdown and intermixing after thermal stress
• Magnetic recording and spintronics: layer-by-layer metrology of CoFeB, Co–Pt, Fe–Pt, and exchange-bias stacks, including interface broadening after field-annealing treatments
• Hard and tribological coatings: nitrogen content and thickness of TiAlN, CrN, and AlCrN films, together with crater depth correlation for wear studies
• Corrosion and high-temperature oxidation: growth kinetics and interdiffusion profiles of aluminide, chromia-forming, and MCrAlY protective coatings
• Photovoltaics: composition and Ga grading profiles in Cu(In,Ga)Se2 absorbers and stoichiometry control of contact alloys
• Optical and decorative coatings: thickness and composition of Ag-alloy mirror layers, anti-tarnish caps, and gold-colored nitride finishes
• Advanced materials research: composition confirmation of sputtered high-entropy alloy films, thin film solid-state reaction studies, and radiation damage investigations

FAQ

Is RBS a destructive test?

No. The helium dose deposited is small, and metallic alloy films are unaffected by the analysis. Insulating or temperature-sensitive materials may require reduced beam current, but for metals the measurement is effectively non-destructive and the same spot can be re-measured repeatedly.

What is the detectable thickness range for alloy films?

Reliable quantitative analysis covers approximately 5 nm to 1–2 µm for dense heavy-alloy films. Very thin layers are quantified through their areal density even when the thickness itself cannot be resolved, while the practical upper limit in light matrices can reach about 10 µm with degraded depth resolution.

How sensitive is RBS to impurities?

Heavy elements in a light matrix can be detected down to roughly 0.01–0.1 atomic percent. Sensitivity drops significantly for light elements in heavy matrices, and hydrogen cannot be measured by conventional RBS; NRA and ERDA are recommended in those cases.

Can RBS distinguish neighboring elements such as iron and nickel?

Adjacent heavy elements produce partially overlapping signals. For elements differing by only one or two mass numbers, the kinematic edges are close, and accurate results rely on full spectral simulation and, where possible, on complementary information such as known layer structure or PIXE data.

What accuracy can be expected?

Major-element compositions are typically accurate to 1–3% relative, and derived thicknesses to 3–5%, with the main uncertainty contributions coming from stopping power values and any assumed layer density.

How does RBS compare with EDS, XPS, and SIMS?

Unlike EDS, which probes micrometers deep and requires standards for best accuracy, RBS is surface-sensitive and intrinsically quantitative without reference materials. Compared with XPS and SIMS, RBS offers superior quantitative depth profiling of the top micrometer but lower trace sensitivity, so the techniques are frequently combined in comprehensive characterization programs.

What sample sizes and forms are accepted?

Samples ranging from small coupons of a few square millimeters to wafer sections can be accommodated depending on the sample chamber. The essential requirements are flatness, lateral homogeneity on the millimeter scale, and compatibility with high vacuum.

Can multilayer alloy stacks be analyzed layer by layer?

Yes. As long as adjacent layers differ sufficiently in mass, thickness, or composition, RBS resolves individual layer signals and reports separate areal densities and compositions. Interfaces broader than roughly 10 nm are directly visible in the spectrum as gradual spectral slopes.