Creep response analysis of metallic materials is essential for ensuring the long-term reliability of MEMS devices that operate under sustained mechanical stress, elevated temperature, or both. Because micro-scale metals such as electroplated nickel, gold, aluminum, copper, and solder alloys can exhibit significant time-dependent deformation even at modest operating temperatures, accurate creep characterization directly supports design optimization, lifetime prediction, and process qualification. This article outlines the sample types, testing items, analytical methods, and typical applications involved in professional MEMS metallic creep response analysis.
Technical Overview
Creep is defined as the time-dependent, permanent deformation of a material under a sustained load or stress, occurring at levels below the instantaneous yield strength of the material. In bulk engineering alloys, creep is traditionally associated with high homologous temperatures, typically above 0.4 times the absolute melting temperature. In MEMS (Micro-Electro-Mechanical Systems), however, the situation is fundamentally different. Metallic structural films are produced by sputtering, evaporation, or electroplating, resulting in fine grain sizes, high defect densities, texture, impurity incorporation, and a very high surface-to-volume ratio. These microstructural features substantially lower the creep resistance of MEMS metals, allowing measurable creep at room temperature in materials such as electroplated nickel, and pronounced creep in low melting point metals such as solder alloys and gold.
A complete creep response analysis captures the classical three-stage creep curve: the primary (transient) stage where strain rate decreases due to strain hardening, the secondary (steady-state) stage where hardening and recovery balance at a minimum strain rate, and the tertiary stage where damage accumulation leads to acceleration and eventual rupture. At the MEMS scale, the secondary creep rate, the stress exponent, and the activation energy are the key quantities used to identify the governing deformation mechanism, whether dislocation glide and climb, grain boundary sliding, or diffusion-controlled flow such as Nabarro-Herring and Coble creep.
The practical importance of creep in MEMS devices is significant. Sustained deflection of suspended beams and membranes, gradual loss of contact force in RF switches, frequency drift in resonators, hysteresis in MEMS mirrors, offset shift in inertial sensors, and thermomechanical fatigue of solder joints in packaging are all manifestations of metallic creep. Quantifying the creep response early in development allows designers to select materials, adjust geometries, and set derating limits that guarantee stable performance over the intended service life.
Testing Samples
Electroplated Nickel and Nickel Alloys
Electroplated nickel is one of the most widely used structural metals in surface micromachining and LIGA-type processes. Deposited layers typically exhibit grain sizes in the sub-micrometer to several-micrometer range and may contain sulfur, carbon, or organic additive residues from the plating bath. These characteristics make electroplated Ni particularly susceptible to room-temperature creep and time-dependent stress relaxation. Test samples include freestanding dog-bone tensile specimens, fixed-fixed beams, folded flexures, and comb-drive-supported structures released from the sacrificial layer.
Gold and Gold Alloys
Gold is used extensively in RF MEMS switches, contact pads, and wire-bonded interconnects. Because of its low melting temperature and low hardness, gold membranes and switch bridges exhibit measurable creep under the residual tensile stress inherent to deposited films. Samples for creep analysis include clamped membranes, micro-cantilever contact beams, and blanket films on substrate for indentation testing.
Aluminum and Aluminum Alloy Films
Aluminum metallization inherited from CMOS processes is used in CMOS-MEMS structures and mirror platforms. Sputtered Al films contain columnar grains and significant residual stress, and electromigration-assisted creep can occur under combined thermal and current loading. Representative samples include mirror plates, micromachined bridges, and patterned interconnect lines.
Copper Metallization and Through-Silicon Vias
Copper is the dominant conductor in advanced interconnects, through-silicon vias (TSVs), and wafer-level packaging. Electroplated Cu vias experience sustained thermo-mechanical stress after cooling from process temperatures, driving creep and stress relaxation that affect keep-out zones and nearby transistor mobility. Test samples include via chains, blanket Cu films, and freestanding Cu micro-specimens.
Solder and Bonding Layers
Solder alloys such as Sn-Ag-Cu, Sn-Bi, and eutectic Sn-Pb, along with bond layers such as Au-Sn and Au-In, are used for die attach, hermetic sealing, and wafer-level bonding. These low melting point alloys creep readily at operating temperatures. Samples include solder bump arrays, bonded die assemblies, and shear or tensile coupons extracted from bonded stacks.
Specimen Geometry and Preparation
Regardless of material, specimens must be carefully prepared with controlled release etching to avoid surface contamination and stress modification. Critical documented parameters include film thickness, grain size and orientation, residual stress measured by wafer curvature or micro-Raman techniques, hardness, and elastic modulus determined by instrumented indentation. These baseline properties are prerequisites for meaningful interpretation of creep data.
Testing Items
A comprehensive creep response analysis typically quantifies the following parameters:
- Full creep strain versus time curves at controlled stress and temperature, segmented into primary, secondary, and tertiary stages.
- Steady-state (minimum) creep rate as a function of applied stress for stress exponent determination.
- Stress exponent n from the power-law relationship between strain rate and stress, used to identify the rate-controlling mechanism.
- Apparent activation energy Q derived from temperature-dependent creep rate measurements using Arrhenius analysis.
- Threshold stress behavior and the transition from power-law to power-law breakdown at high stresses.
- Primary creep parameters describing the transient response and anelastic recovery upon unloading.
- Stress relaxation curves, quantifying the decay of stress or contact force under fixed displacement, essential for switch contact evaluation.
- Time-to-rupture and strain-to-failure for ductile failure assessment where applicable.
- Monkman-Grant relationship constants linking steady-state creep rate with rupture life for lifetime extrapolation.
- Grain size dependence of creep rate to distinguish grain-boundary-controlled from lattice-controlled mechanisms.
- Extrapolated long-term deformation and deflection predictions for service conditions spanning ten or more years.
Testing Methods
Instrumented Nanoindentation Creep Testing
Nanoindentation is the most widely used method for creep characterization of MEMS metallic films because it requires only a small, flat sample area and provides precise load and displacement resolution. A typical protocol applies a rapid load ramp to a target load, followed by a prolonged hold at constant load during which penetration depth is continuously recorded. The depth-time data are converted to creep strain and strain rate using the expanding cavity or constant hardness analysis approaches. Load-level series and strain-rate-jump tests at multiple temperatures allow extraction of the stress exponent and activation energy. Strict thermal drift correction and a stable temperature stage, often ranging from room temperature to 300 degrees Celsius or higher, are mandatory because creep displacements can be small relative to instrumental drift. Testing generally follows the principles of ISO 14577 for instrumented indentation.
On-Chip MEMS Tensile and Beam Testing
Freestanding micro-specimens patterned directly on the wafer capture the true size effect and as-deposited microstructure of MEMS metals. Electrostatic comb drives, thermal actuators, or external micro-probe and grip systems apply controlled tensile loads to dog-bone specimens while displacement is measured by integrated markers, capacitive sensing, or digital image correlation. Sustained loading at fixed force yields direct creep strain curves; sustained fixed displacement yields stress relaxation data. Cantilever and fixed-fixed beam bend tests, in which a known load is applied and deflection is tracked over hours or days, provide an economical alternative for process monitoring.
Bulge Testing of Membranes
In bulge testing, a freestanding membrane is pressurized with a controlled gas or liquid pressure and the dome height is monitored optically or interferometrically over time. Holding a constant pressure produces a biaxial creep condition representative of MEMS membranes and sealed cavities, from which plane-strain creep properties are extracted using analytical membrane relations.
Stress Relaxation Testing
Many MEMS failure modes involve fixed strain rather than fixed load, for example a pre-deflected switch bridge or a compressed solder layer. Stress relaxation tests impose a fixed displacement and record the decay of reaction force or inferred stress. The relaxation curve is mathematically linked to the underlying creep compliance through the viscoelastic correspondence principle, allowing creep constants to be identified from relaxation data and vice versa.
Accelerated Temperature and Stress Conditioning
Because service life in the field may exceed ten years, laboratory tests apply elevated temperature and elevated stress to accelerate the kinetics. Testing at several temperature and stress combinations, followed by Arrhenius and power-law extrapolation using time-temperature parameters such as the Larson-Miller parameter, enables prediction of deformation at use conditions. Validation of the extrapolation against the identified physical mechanism is essential to avoid non-conservative lifetime estimates.
Modeling and Simulation
Experimental constants are incorporated into constitutive models such as the Norton power law, the Garofalo hyperbolic sine formulation, or unified viscoplastic models including primary creep and back-stress evolution. These models are implemented in finite element analysis of full device geometries to simulate long-term deflection, contact force loss, frequency shift, and packaging stress evolution. Model predictions are verified against device-level accelerated aging data before release to production.
Applications
- RF MEMS switches and relays, where creep of gold or nickel bridges governs the long-term degradation of contact force, insertion loss, and switching threshold voltage.
- MEMS resonators and oscillators, where stress relaxation in metallic electrodes and structural layers causes resonance frequency drift that must be bounded for frequency stability specifications.
- Optical MEMS mirrors and scanners, where hinge and platform creep leads to pointing error, set-point drift, and hysteresis in laser projection and telecommunications systems.
- Inertial sensors such as accelerometers and gyroscopes, where creep of metallic proof-mass attachments and spring structures contributes to bias instability over temperature and time.
- Thermal actuators and micro-heaters, in which sustained high-temperature operation demands creep-resistant structural metals and validated deformation budgets.
- Wafer-level packaging, hermetic sealing rings, solder bump interconnects, and copper TSV structures, where creep and stress relaxation control joint integrity, fatigue life, and stress state in adjacent silicon devices.
- Material and process qualification for new plating chemistries, annealing conditions, and deposition parameters, using creep response as a sensitive indicator of microstructural quality.
- Failure analysis and root-cause investigation, comparing the creep signature of field-returned or aged devices against pristine baseline material.
FAQ
Why do MEMS metals creep at room temperature when bulk alloys do not?
MEMS metallic films have very fine grains, high defect and impurity content, and large surface-to-volume ratios. Grain boundary mediated mechanisms such as Coble creep and grain boundary sliding scale strongly with inverse grain size, while homologous temperatures for low melting point metals like solder and gold are already high at room temperature. Together these factors shift measurable creep to much lower temperatures than in well-annealed bulk alloys.
How is creep measured when the specimens are only a few micrometers thick?
Instrumented nanoindentation is the standard approach, since it locally loads the film with micron-scale resolution and records sub-nanometer depth changes during a load hold. For structural verification, on-chip tensile and beam specimens patterned with the product process flow are tested using electrostatic or thermal actuation, ensuring that the measured behavior reflects the actual device material.
What is the difference between creep and stress relaxation testing?
Creep testing holds stress constant and measures the growth of strain over time, while stress relaxation testing holds strain or displacement constant and measures the decay of stress over time. Both probe the same underlying time-dependent deformation, and the choice depends on the device boundary condition: suspended masses under gravity or preload resemble creep, whereas clamped bridges and compressed bonds resemble relaxation.
How are short-term laboratory tests used to predict ten-year device behavior?
Tests are performed at multiple elevated temperatures and stress levels, from which the stress exponent and activation energy are extracted. Using Arrhenius extrapolation, power-law scaling, and time-temperature parameters such as the Larson-Miller parameter, the creep rate at service conditions is calculated. The extrapolation must be anchored to a verified deformation mechanism to remain physically valid, and results are confirmed with device-level accelerated aging.
What stress exponent indicates which creep mechanism?
As a general guideline for MEMS metals, a stress exponent near 1 suggests diffusion-controlled creep or grain boundary sliding, values around 2 to 3 are often associated with grain boundary sliding accommodated by diffusion in fine-grained electroplated films, and values above about 5 indicate dislocation climb-controlled creep. Mechanism identification should combine the stress exponent with activation energy and grain size dependence.
How long does a typical creep characterization program take?
Individual indentation hold tests typically run from several minutes to a few hours depending on the material responsiveness, while wafer-level relaxation and on-chip beam tests may run from days to weeks to capture the full transient. A complete program covering multiple temperatures and stresses, constitutive model fitting, and finite element life prediction generally spans four to eight weeks.