Etch rate repeatability testing is a core process qualification and statistical process control activity that verifies how consistently dry plasma or wet chemical etch tools remove material from wafer surfaces across consecutive runs. Performed by accredited third-party laboratories and in-fab quality teams in line with precision frameworks such as ISO 5725 and measurement system analysis guidance such as SEMI E89, the test quantifies run-to-run etch rate stability, within-wafer uniformity, and wafer-to-wafer variation. This article explains the technical principles, sample types, test items, measurement and statistical methods, industry applications, and frequently asked questions related to etch rate repeatability evaluation.
Technical Overview
Etch rate is defined as the thickness of material removed per unit of time, typically expressed in nanometers per minute (nm/min) or Angstroms per minute, and is calculated by measuring film thickness or step height before and after a fixed-duration etch process. Etch rate repeatability describes how consistently the same etch tool, recipe, and operator can reproduce this value over a series of consecutive runs performed under nominally identical conditions. In statistical terms, repeatability follows the definitions established in ISO 3534-1 and ISO 5725-1, which characterize it as the closeness of agreement between independent test results obtained under repeatability conditions: same operator, same equipment, same method, and a short interval of time.
Repeatability is typically quantified using descriptive statistics across repeated runs, including the mean etch rate, standard deviation, coefficient of variation (CV), and total range. For most production etch processes in advanced semiconductor manufacturing, a run-to-run etch rate variation of less than 1 to 3 percent (CV) is expected, while qualification-grade equipment often demonstrates variation below 1 percent. Because film thickness budgets at advanced nodes are extremely tight, even small drift in etch rate directly translates into critical dimension (CD) variation, profile distortion, residue defects, or loss of underlying layers.
Etch rate repeatability testing should be clearly distinguished from related but separate concepts. Within-wafer uniformity describes spatial variation across a single wafer surface, wafer-to-wafer repeatability describes variation between substrates in the same lot, and reproducibility, as defined in ISO 5725, describes variation when conditions such as operator, day, or chamber are changed. A rigorous test program evaluates all of these components separately, and applies measurement system analysis (MSA) practices consistent with SEMI E89 to ensure that observed variation is caused by the etch process rather than by the metrology tool. Accredited third-party laboratories performing these evaluations typically operate under quality management systems aligned with ISO/IEC 17025.
Testing Samples
Etch rate repeatability studies are most commonly performed on blanket (unpatterned) films deposited on 100 mm to 300 mm silicon wafers or on wafer pieces, because blanket layers allow direct thickness metrology at identical sites before and after etching. Patterned monitor wafers with open-area test structures are also used when selectivity or pattern transfer behavior must be assessed in parallel.
Silicon Dioxide and Silicon Nitride Dielectric Films
Thermal silicon dioxide, PECVD and TEOS-based oxides, LPCVD and PECVD silicon nitride, silicon oxynitride, and ONO (oxide-nitride-oxide) stacks are the most widely used qualification vehicles. These films are stable, well characterized, and directly representative of the dielectric etch steps used in gate oxide removal, spacer formation, contact opening, and pad oxide release processes.
Polysilicon and Conductor Films
Doped and undoped polysilicon, titanium nitride, tantalum nitride, tungsten, molybdenum, and aluminum films are tested to evaluate conductor etch modules used for gate electrodes, wordlines, bitlines, and interconnect levels. Metal film etch rate repeatability is particularly sensitive to plasma power stability and chamber wall conditioning, making these samples valuable stress vehicles.
Compound Semiconductor and Wide-Bandgap Materials
Silicon carbide, gallium nitride, sapphire, gallium arsenide, and indium phosphide substrates and epilayers are submitted for repeatability testing by power device, RF component, and photonic device manufacturers. These materials etch slowly and require aggressive chlorine- or fluorine-based plasma chemistries, so stable, repeatable etch rates are essential for controlling trench depth in SiC MOSFETs and GaN HEMT device structures.
Masking, Sacrificial, and Low-k Dielectric Layers
Photoresist, bottom anti-reflective coatings, spin-on carbon hard masks, amorphous carbon, organic sacrificial layers, and porous low-k dielectrics such as SiCOH are tested to characterize mask erosion rates and selectivity margins. Repeatability data for these layers determine whether a process window can sustain long over-etch steps without mask failure or underlayer damage.
MEMS, SOI, and Specialty Substrates
Silicon-on-insulator device layers, quartz and glass substrates, ceramic coupons, and bulk silicon pieces are tested for MEMS release, deep reactive ion etching (DRIE), and microstructure fabrication processes, where vertical etch depth repeatability directly controls mechanical device performance such as resonator frequency and sensor gap dimensions.
Testing Items
A comprehensive etch rate repeatability evaluation typically reports the following items, each derived from pre-etch and post-etch metrology data collected across a defined run matrix:
- Mean etch rate per run, expressed in nm/min or Angstroms per minute
- Etch rate standard deviation and coefficient of variation across consecutive runs (run-to-run repeatability)
- Within-wafer etch rate non-uniformity, calculated as sigma divided by mean across multiple measurement sites
- Wafer-to-wafer etch rate variation within a single lot
- Lot-to-lot etch rate stability over an extended monitoring period
- Absolute etch depth and etch depth accuracy relative to target
- Selectivity of the etched film relative to the masking layer and underlying stop layer
- Endpoint time or time-to-clear stability for patterned or open-area test structures
- Sidewall angle, etch bias, and anisotropy repeatability for patterned monitor wafers
- Measurement system gauge repeatability, confirming that metrology noise is a small fraction of process variation
Testing Methods
Sample Preparation and Baseline Metrology
Test films are deposited under controlled conditions, and baseline thickness is mapped before etching using spectroscopic ellipsometry, spectral reflectometry, or white-light interferometry at a fixed array of sites, commonly 9, 25, or 49 points per wafer. When thickness-optical methods are unsuitable, such as for opaque metal films, masked windows are created by photolithography so that a physical step height can be measured with a stylus profilometer or atomic force microscope after etching. Accurate, site-matched baseline data is the foundation of any meaningful repeatability calculation.
Controlled Etch Execution
The etch is executed under a frozen recipe in which all critical parameters are fixed and logged, including RF or microwave power, chamber pressure, gas flow rates, electrode temperature, chuck helium backside pressure, and etch duration. Chamber seasoning and warm-up dummy runs are performed before data collection so that measurements reflect steady-state process behavior rather than transient wall conditions. A typical test matrix consists of at least three consecutive runs with three to five monitor wafers per run, although full equipment qualification studies may extend to ten or more runs. For wet etch processes, bath temperature, chemical concentration or bath age, and agitation are controlled and recorded in the same manner.
Post-Etch Metrology
After etching, the identical wafer sites are re-measured using the same instruments and recipes employed for the baseline. Etch rate is then calculated as the difference between pre-etch and post-etch thickness divided by etch time, for example ER = (T0 - T1) / t, where T0 is pre-etch thickness, T1 is post-etch thickness, and t is etch time in minutes. Cross-section scanning electron microscopy is used for patterned samples where vertical and lateral etch dimensions must both be characterized, and in-situ optical emission or laser interferometry endpoint traces can be retained as supplementary evidence of process stability.
Statistical Evaluation of Repeatability
Etch rate data are aggregated to compute the run mean, standard deviation, and coefficient of variation, where CV (%) = (standard deviation / mean etch rate) x 100. Within-wafer uniformity is calculated site-by-site for each wafer, while run-to-run and wafer-to-wafer components are separated through nested analysis. Consistent with the precision framework of ISO 5725, the repeatability standard deviation can be used to derive a repeatability limit, beyond which two single results under repeatability conditions should be considered suspect. Results are also plotted on statistical process control charts, such as X-bar and R or individual and moving range charts, to reveal drift, shift, or cyclical behavior, and process capability indices such as Cp and Cpk are calculated when specification limits exist.
Measurement System Verification
Before attributing variation to the etch process, the metrology itself is verified by repeatedly measuring unetched reference sites to establish gauge repeatability, following the measurement system analysis concepts outlined in SEMI E89. Studies involving multiple laboratories or multiple metrology tools can additionally follow precision-evaluation practices in the style of ASTM E691. Independent evaluations conducted by accredited third-party laboratories under ISO/IEC 17025 management systems provide unbiased confirmation of in-fab results and are widely accepted for vendor acceptance and process transfer documentation.
Applications
Etch rate repeatability testing supports nearly every stage of the semiconductor and microfabrication equipment lifecycle, from initial process development through high-volume manufacturing control:
- New equipment acceptance and qualification for reactive ion etch, ICP, CCP, downstream, and wet bench etch systems
- Routine daily, weekly, or per-shift process monitoring and statistical process control in production fabs
- Chamber-to-chamber matching and fleet-wide tool matching across multiple etch platforms
- Process transfer between development fabs and production fabs, or between different equipment suppliers
- Post-maintenance recovery verification, including chamber cleaning, part replacement, and seasoning validation
- New material and new chemistry introduction, such as SiC trench etching, GaN gate recess, and low-k integration
- Photomask blank etching, MEMS deep silicon etching, and packaging-related wafer thinning or release processes
- Research and development programs in universities and institutes that require documented process stability for publication and reproducibility
FAQ
How many runs and measurement points are required for a valid repeatability test?
A common minimum configuration is three consecutive runs with three monitor wafers per run and 9 to 49 measurement points per wafer, which yields enough data to compute meaningful standard deviations. Full equipment qualification or chamber matching studies typically use ten or more runs spread over several days to capture longer-term stability effects.
What level of etch rate repeatability is considered acceptable?
Acceptance criteria depend on the device node and application. A run-to-run coefficient of variation below 1 percent is considered excellent and is often required for advanced logic and memory etch steps. Values between 1 and 3 percent are typical for many production processes, while values above 5 percent generally trigger investigation of RF power stability, gas flow calibration, chamber wall condition, or endpoint timing.
What is the difference between repeatability and uniformity?
Uniformity describes how etch rate varies spatially across a single wafer within one etch run, whereas repeatability describes how the average etch rate varies between repeated etch runs over time. A process can be highly uniform within each wafer yet still exhibit poor run-to-run repeatability, and both characteristics must be qualified independently.
Which metrology technique should be used?
Spectroscopic ellipsometry and spectral reflectometry are preferred for transparent dielectric films on silicon. Stylus profilometry is effective whenever a masked step is available, particularly for opaque metal films. Atomic force microscopy provides high-precision step measurements, and cross-section scanning electron microscopy is used when patterned features must be profiled. The key requirement is that the same tool, recipe, and sites are used for both pre-etch and post-etch measurements.
What are the most common causes of poor etch rate repeatability?
Frequent root causes include RF generator power drift, mass flow controller calibration error, chamber wall polymer buildup or insufficient seasoning, unstable substrate temperature, endpoint detection jitter, photoresist loading effects, and, for wet processes, bath chemistry depletion or temperature fluctuation. A disciplined repeatability test isolates these factors by holding all parameters constant and correlating variation with logged tool signals.
How often should etch rate repeatability testing be performed?
Repeatability verification is performed at new tool installation, after every preventive maintenance event involving chamber exposure, after recipe or hardware changes, and on a periodic monitoring schedule defined by each fab's statistical process control plan. Accredited third-party laboratories are frequently engaged for vendor acceptance testing, process transfer audits, and independent arbitration when in-fab data are disputed.