semi合集-English.pdf - 第675页
SEMI E89-1104 E © SEMI 1999, 2004 2 3.2 This gui de does not address those as pects of measurement uncertainty associ ated with change in the object being measured, either spatially o r temporally. 3.3 This guide does no…
SEMI E89-1104
E
© SEMI 1999, 2004 1
SEMI E89-1104
E
GUIDE FOR MEASUREMENT SYSTEM ANALYSIS (MSA)
This guide was technically approved by the Global Metrics Committee and is the direct responsibility of the
North American Metrics Committee. Current edition approved by the North American Regional Standards
Committee on August 16, 2004. Initially available at www.semi.org September 2004; to be published
November 2004. Originally published September 1999.
E
This standard was editorially modified in October 2004 to correct an editorial error in the title.
NOTICE: This document was completely rewritten in
2004.
1 Purpose
1.1 The purpose of this guide is to provide a consistent
set of terminology and describe a simplified, but
constructive, experimental approach to planning and
performing a measurement system analysis (MSA).
1.2 The goal of an MSA is to characterize the
performance capability of the measurement system
(MS) as it is intended to be used in a manufacturing or
laboratory setting.
1.2.1 Accurately identifying the MS bias and the size
and nature of all sources of variability allows one to
determine whether the MS is capable of performing its
intended function. Moreover, a well-designed MSA
can be used to identify and quantify areas that need the
most improvement.
2 Scope
2.1 This guide covers procedures for determining
specific measures of MS capability including:
measurement variability (i.e., reproducibility)
under a variety of conditions, including
effects of repeatability,
load-unload, and
time, and
bias, including bias-related
linearity,
stability, and
matching tolerance.
2.2 This guide also covers secondary metrics such as
precision-over-tolerance (P/T) ratio and signal-to-noise
ratio (SNR).
2.3 The primary focus of this guide is on determining
measurement capability of automated wafer MSs under
normal operating conditions, but the definitions and
methodologies are extendible to many other measure-
ment situations involving automated measurements on
units such as processed dice, packaged devices, flat
panel displays, piece parts, etc.
2.4 While there is no universally accepted correct way
to conduct an MSA, the approach described in this
document is supported in the technical literature (see
Section 11) and congruent with practices advocated in
ISO 5725-2. The procedures given in this guide
represent an approach to the conduct of an MSA and
provide basic reference methods that should serve for a
variety of applications. Other methods may be
appropriate in certain circumstances.
2.5 The procedures in this guide that are intended to
separate the various sources of nonsystematic (i.e.,
random) errors are based on the use of factorial
experiments and analysis of variance (ANOVA).
Because the primary focus of this guide is on evaluation
of automated MSs, the variability introduced by
different operators is expected to be minimal.
NOTE 1: Information on measurement uncertainty
calculations is provided in Related Information 1. Information
on testing measurement distributions for normality and equal
repeatability is provided in Related Information 2.
NOTICE: This standard does not purport to address
safety issues, if any, associated with its use. It is the
responsibility of the users of this standard to establish
appropriate safety and health practices and determine
the applicability of regulatory or other limitations prior
to use.
3 Limitations
3.1 Determination of MS capability is meaningless
unless the MS is in control. Methodology for
establishing and maintaining MS control is beyond the
scope of this guide. Such methodology should be a part
of a quality management system, such as that mandated
by ISO 9000 or similar standards. Additional guidance
for laboratories without established procedures may be
found in the ASTM Manual on Presentation of Data
and Control Chart Analysis.
1
1 Manual on Presentation of Data and Control Chart Analysis, 6th
edition, MNL 7 (ASTM International, West Conshohocken, PA,
1991)
SEMI E89-1104
E
© SEMI 1999, 2004 2
3.2 This guide does not address those aspects of
measurement uncertainty associated with change in the
object being measured, either spatially or temporally.
3.3 This guide does not address determination of
measurement capability in the case of destructive
measurements on samples, or when the MS alters the
object being measured as a result of making the
measurement.
3.4 This guide does not apply to inter-laboratory
experiments designed to measure inter-laboratory
precision of test methods.
4 Referenced Standards
4.1 ISO Standards
2
ANSI/ISO Z540-2 — Guide to the Expression of
Uncertainty in Measurement
ANSI/ISO/ASQC A3534-1 — Statistics – Vocabulary
and Symbols – Part 1: Probability and General
Statistical Terms.
ISO 3534-3 — Statistics – Vocabulary and Symbols –
Part 3: Design of Experiments.
ISO 5725-2 — Accuracy (trueness and precision) of
measurement methods and results – Part 2: Basic
method for the determination of repeatability and
reproducibility of a standard measurement method.
ISO 9000 — Quality management systems –
Fundamentals and vocabulary, 2000.
NOTICE: Unless otherwise indicated, all documents
cited shall be the latest published versions.
5 Terminology
5.1 Terminology in this section that is not directly used
in this guide, is likely to be encountered while
conducting an MSA.
5.2 Definitions of many other terms related to
metrology and statistics can be found in VIM,
3
ANSI/ISO/ASQC A3534-1, and ISO 3534-3.
5.3 Definitions
5.3.1 accuracy — closeness of agreement between a
test result or the mean of a group of test results made on
an object and its true value.
3
2 ISO Central Secretariat, 1, rue de Varembé, Case postale 56, CH-
1211 Genève 20, Switzerland, web site:
www.iso.ch; ISO standards
are available in the United States through the American National
Standards Institute, web site:
www.ansi.org, and in most other
countries through the ISO member body.
3 International Vocabulary of Basic and General Terms in
Metrology, Second Edition [VIM] (ISO, Genève, 1993).
5.3.1.1 Discussion — Accuracy depends on both the
precision and bias of the measurement process. Since
random components of error (resulting in imprecision)
and systematic components of error (resulting in bias)
cannot be completely separated in routine use, the
reported accuracy must be interpreted as a combination
of these two elements.
5.3.2 bias — difference between the population mean
of the test results from a measurement process and the
true (accepted reference) value of the property being
measured.
5.3.2.1 Discussion — Bias is a systematic component
of measurement uncertainty. One or more systematic
error components may contribute to the bias. The true
value and the population mean are both unknown. The
true value may be estimated with the use of a consensus
value. If sufficient measurements are made to
adequately mitigate the effects of measurement
variability, the population mean may be estimated from
the sample mean
n
i
i
x
n
x
1
1
(1)
where:
x = sample mean,
n = number of measurements, and
x
i
= i
th
measurement value.
5.3.3 calibration — set of operations that establish the
relationship between values of quantities indicated by a
measurement system (MS) and the corresponding
values assigned to reference materials.
5.3.3.1 Discussion — The purpose of calibration is to
reduce or eliminate bias in the MS.
5.3.4 certified reference material (CRM) — reference
material, one or more of whose property values are
certified by a technically valid procedure, accompanied
by or traceable to a certificate or other documentation
issued by a certifying body.
5.3.5 coefficient of variation (CV) — population
standard deviation expressed as a percentage of the
mean value.
5.3.5.1 Discussion — CV can be estimated from the
sample standard deviation, s, and the sample mean,
,x of a distribution as follows:
100
x
s
CV (2)
CV is an appropriate measure of variability only when
the sample standard deviation is proportional to the
mean; otherwise it varies with the value of the
measurand. If the sample standard deviation is
SEMI E89-1104
E
© SEMI 1999, 2004 3
independent of the value of the measurand, it is more
appropriate to use it directly rather than CV.
5.3.6 effect change in the expected value of a given
response due to the change of a given factor from one
level to another. It is a measure of influence that a
particular variable level has on the output variable.
5.3.6.1 fixed effect — variable for which estimates of
the mean are obtained for each level.
5.3.6.2 random effect — variable for which estimates of
the mean are not obtained for each level; rather the
variable is treated as a variance component.
5.3.7 factor — predictor variable whose level is
changed with the intent of assessing its effect on the
response variable (in a designed experiment) [adapted
from ISO 3534-3].
5.3.7.1 crossed factor(s) — two factors are crossed
when every level of one factor appears with every level
of the second factor.
5.3.7.2 fixed factor — factor that has either all of its
levels represented in an experiment or levels selected
by a nonrandom process.
5.3.7.3 nested factor(s) — factor that has a different set
of levels appearing within each level of a second factor.
Factor B is nested in factor A when randomization of
the levels of factor B is restricted to specific levels of
factor A.
5.3.7.4 random factor — factor that has randomly
sampled levels from a population of levels.
5.3.8 gage — alternate spelling of gauge.
5.3.9 gauge — instrument used to assign a value to a
quantitative or qualitative characteristic of a physical
entity or phenomenon.
5.3.10 interaction — effect for which the apparent
influence of one factor on the response variable
depends upon one or more other factors [ISO 3534-3].
5.3.11 level value of a factor (in a designed
experiment) [adapted from ISO 3534-3]. Also called
“setting of a variable.”
5.3.12 linearity — absence of changes in variability or
bias as measurements are made at different points
within the measurement range.
5.3.12.1 Discussion — Traditional definitions of
linearity ignore the fact that variability can change over
the measurement range, as well as bias. The
assumption of constant variability over the
measurement range should be verified during the MS
analysis.
5.3.13 lower specification limit (LSL) — value of an
attribute below which a product is said to be
nonconforming.
5.3.14
matching tolerance (
m
) — difference in bias for
any two measurement systems (MSs) of the same kind
made under the conditions of reproducibility.
5.3.15 measurand — particular attribute of a phenome-
non, body or substance subject to measurement. [VIM]
5.3.16 measurement resolution, of a gauge — smallest
difference in measurand that can be meaningfully
distinguished by the gauge.
5.3.17 measurement subsystem any set of entities,
processes, or conditions that share a common purpose
in the measurement.
5.3.17.1 Discussion — A measurement subsystem may
contain one or more of its own subsystems. For
example, a wafer handling mechanism may be further
composed of wafer loading and wafer positioning
subsystems.
5.3.18 measurement system (MS) — all entities,
procedures, and conditions that can influence the test
result obtained with a given measurement process.
5.3.18.1 Discussion — The MS may include, but is not
limited to, the gauge, operators, setup mechanics,
wafers, locations on a wafer, environmental conditions,
software used by the gauge, measurement method, etc.
The MS may be comprised of measurement
subsystems.
5.3.19 measurement system analysis (MSA) —
procedure in which relevant sources of bias and
variability associated with a measurement system (MS)
are estimated.
NOTE 2: MSA is also sometimes called gauge (or gage)
repeatability and reproducibility (GRR or GR&R).
5.3.20 nested design — experimental design in which
different levels of one factor appear in each level of a
second factor.
5.3.21 population standard deviation (
) — square root
of the population variance.
5.3.22 population variance (
2
) — measure of
dispersion associated with a population distribution.
5.3.22.1 Discussion — For continuous distributions, the
population variance is the second central moment.
5.3.23 precision — general estimator of the variability
of a measurement process about the mean value of the
test results obtained.
5.3.23.1 Discussion — Precision is a random
component of measurement uncertainty. Unless the