ASTM E1329-00
(Practice)Standard Practice for Verification and Use of Control Charts in Spectrochemical Analysis
Standard Practice for Verification and Use of Control Charts in Spectrochemical Analysis
SCOPE
1.1 This practice covers procedures for determining if a spectrochemical analysis is under statistical control.
1.2 Criteria are presented for determining when corrective action is required.
1.3 Control will be effected by using verifiers to test instrument response. It is recommended, although not required, that this be accompanied by the plotting of control charts.
1.4 The preparation of control charts is described.
1.5 Limitations---The procedures which are described do not apply to analyses which requires a calibration each time a set of analyses is run. Reference is made specifically to optical emission spectroscopy, but the practice has a more general application.
1.6 This practice does not apply to validation procedures that monitor the correctness of calibration.
General Information
Relations
Standards Content (Sample)
NOTICE: This standard has either been superseded and replaced by a new version or withdrawn.
Contact ASTM International (www.astm.org) for the latest information
Designation: E 1329 – 00
Standard Practice for
Verification and Use of Control Charts in Spectrochemical
Analysis
This standard is issued under the fixed designation E 1329; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (e) indicates an editorial change since the last revision or reapproval.
1. Scope 3. Terminology
1.1 This practice covers procedures for determining if a 3.1 Definitions—For definitions of terms used in this prac-
spectrochemical analysis is under statistical control. tice, refer to Terminologies E 135 and E 456 and Practice
1.2 Criteria are presented for determining when corrective E 876. Refer also to the glossary of terms and symbols
action is required. appearing in MNL 7.
1.3 Control will be effected by using verifiers to test 3.2 Definitions of Terms Specific to This Standard:
instrumentresponse.Itisrecommended,althoughnotrequired, 3.2.1 control limits—in control charts, the upper and lower
that this be accompanied by the plotting of control charts. limits of a statistic that are not expected to be exceeded,
1.4 The preparation of control charts is described. designated as UCL and LCL respectively in this practice. For
1.5 Limitations—The procedures that are described do not the statistic that is the average of more than one reading or
apply to analyses that require a calibration each time a set of determination, the upper and lower limits will be equidistant
analyses is run. Reference is made specifically to optical from a central line (CL) representing the expected average. For
emission spectroscopy, but the practice has a more general the statistic of either standard deviation or range, the upper
application. limit will be farther from the central line if the lower limit is
1.6 This practice does not apply to validation procedures zero.
that monitor the correctness of calibration. 3.2.2 normalization—a procedure for correcting readings to
a common basis. A special case of normalization is standard-
2. Referenced Documents
ization in which readings are made to conform to an existing
2.1 ASTM Standards:
calibration. Normalization permits gathering data in different
E 135 Terminology Relating to Analytical Chemistry for
periods of time and correcting for drift in a way that may be
Metals, Ores, and Related Materials independent of standardization routines.
E 158 Practice for Fundamental Calculations to Convert
3.2.3 variation—difference in an observed value from a
Intensities into Concentrations in Optical Emission Spec- norm.
trochemical Analysis
3.2.3.1 assignable cause—variation which can be identified
E 305 Practice for Establishing and Controlling Spectro- and corrected. It may be the result of a condition of an
chemical Analytical Curves
instrument or a method of operation. For example, signal
E 456 Terminology Relating to Quality and Statistics intensities may be affected because a spectrometer is not
E 876 Practice for Use of Statistics in the Evaluation of
profiled properly.
Spectrometric Data 3.2.3.2 chance or common cause—random variation which
2.2 Other ASTM Documents:
consistently affects a system, contributing to the imprecision in
MNL 7 Manual on Presentation of Data and Control Chart a predictable way. In the application of control charts, the
Analysis
assumption is made that chance causes of variation are
normally distributed.
4. Significance and Use
This practice is under the jurisdiction of ASTM Committee E01 on Analytical
4.1 Consistency in analysis depends on being aware of a
Chemistry for Metals, Ores, and Related Materials and is the direct responsibility of
significantchangeininstrumentalresponse,suchasthatcaused
Subcommittee E01.22 on Statistics and Quality Control.
Current edition approved May 10, 2000. Published July 2000. Originally
by drift or changes in analytical precision, or both, and taking
published as E 1329 – 90. Last previous edition E 1329 – 96.
corrective action. The usual corrective action for drift is
Annual Book of ASTM Standards, Vol 03.05.
standardization. Standardization, however, when there is no
Annual Book of ASTM Standards, Vol 14.02.
ASTM Manual Series, ASTM, 6th edition, 1990. real need, can only broaden the spread of subsequent analyses.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.
E1329–00
obtained before and after a set of calibrants has been run, an instrument
One purpose of this practice is to set guidelines that will avoid
problem may have to be investigated and corrected or the operational
“overstandardization.”
environment improved. Reliable calibration data can be obtained only if
4.2 To control manufacturing processes, there must be
an instrument shows a stable operation. Practice E 876 describes ways to
confidencethataconsistentmaterialisbeingproducedandthat
test for drift.
the analysis of the material is reliable. For assurance that the
6.1.1.1 Unless a curve fitting routine is being used that
material meets specification, a purchaser may require the
requires “standardizing” before running a set of reference
supporting record of control charts to assess that proper
materials it is recommended that no normalization be done
analytical control has been maintained.
until all calibration data has been recorded. Strictly speaking
4.3 Ideally, variations in analytical results may be held to
standardization, as defined in Terminology E 135, only can be
chance causes. The concept of a confidence interval or limits
doneafteracalibrationhasbeenestablished.Ifanormalization
on a control chart is based on what can be expected when all
to some prescribed set of readings is done as if it were a
normal precautions are exercised. When results appear to go
standardization before each time a set of reference materials is
out of control, the analyst should consider and correct what
run, the resulting record of readings can be treated as if no
might be an assignable cause. As experience is accumulated,
standardization had been done.
however, it may not seem unusual for readings to drift with
time as optics degrade, detector response changes, or excita- 6.1.2 Choose one set of averages of before and after
tions modify as, for example, when deposits build up on a readings of 6.1.1 as the norm. A grand overall average of the
counter electrode (a correctable assignable cause), or the sets may be used if that seems like a reasonable median of all
longer range effects as an X-ray tube deteriorates. sets.Excludeanyreadingsforaelementinareferencematerial
that does not show comparable repeatability to what was
5. Problems in Applying Control
observed for that element in other materials. For higher level
readings, the comparison should be made to observed relative
5.1 A complication in effecting verification control or in
repeatabilities.
using control charts with spectrochemical analyses is that the
6.1.2.1 For an ideal normalization of readings, determine
measurements being taken are not absolute. Determinations
the regression fit of a set of observed readings, x, to expected
depend upon comparisons of one measurement to another: the
readings, y. This linear regression, which is also supported by
relativeintensityofananalyticallinetotherelativeintensityof
Practice E 305, commonly is done on electronic calculators or
an internal standard line in optical emission spectroscopy; the
computers by the following equations to determine a slope, m,
interrelationship of counts in X-ray spectroscopy under some
and a constant, k, which can be used to correct observed
specified condition of maintaining a fixed intensity from an
readings to an established norm:
irradiating source and holding to a consistent response from a
detector with or without pulse height analyzers and with or
n(xy – (x~(y!
m 5 (1)
without an external monitor; the nonlinear relationship of 2 2
n(~x !– ~(x!
emulsion blackening to radiation in photographic measure-
and
ments; and the relative response in integrating for fixed times
with ostensibly constant radiation sources. Added to these is k 5 ~(y –m(x!/n (2)
the complication of background signal in all techniques.
where the summations of functions of x and y are as fol-
5.2 Itisimportanttorecognizethatthereareseveralsources
lows:
of random variation, including variations from the measuring
x = the observed average readings of an element in a
method as well as inhomogeneity in the specimens.The device
calibration set,
being used to test analytical response is the analytical system
y = the expected normal readings for that element, and
itself.Thisdiffersfromnormalstatisticalprocesscontrolwhere
n = the number of pairs of x and y readings.
an independent and usually more accurate measuring device is
6.1.2.2 Apply the appropriate m and k corrections to the
used to verify the process variability.
averages of the verifiers and standardants, as well as to the
calibrants in each calibration set, as follows:
6. Verifiers
R 5mR 1k (3)
N O
6.1 It is recommended that readings for all potential verifi-
ers as well as standardants be established by measuring them
where:
along with the calibrants.
R = normalized reading, and
N
6.1.1 Ideally, the full set of potential standardants and
R = observed average reading.
O
verifiers should be run before and after a series of calibrants to
The grand averages of the normalized readings of the
permit normalizing all calibration data to a common basis. To
standardants and verifiers will become the values used for
achieve the best normalization of data, readings should be
standardizing.
recorded for all elements of interest on every standardant and
6.1.3 If the analytical system only can support the early
verifier, even if there is no knowledge of expected concentra-
convention of “two-point” standardization, and if the only
tions. Unless there is a marked change in the before and after
permissible normalization is a quasi-standardization, before
measurements, the averages of a set of before and after
collecting calibration data it is still advisable to record all
readings will be used for normalization.
readings for all elements in all reference materials to establish
NOTE 1—If there appears to be a drift between readings of standardants a full record of what can be expected for all the reference
E1329–00
materials (see 8.6). The initial set of “normal” readings are 6.3.1 If a deliberate change is made in the slope of a
reasonable starting points. Neither the preferred method of calibration curve after the collection of data, such as might be
using a regression fit nor the recommendation of waiting until done in the transformation in 6.1.4, the effective standard
all data have been logged before assigning normal values are deviation of the reading will be the previous observed standard
infallible. Modification of these values always should be an deviation divided by the factor used to change the slope of the
option as more experience is gained. It is expected, however, curve. Thus, if a standard deviation has been calculated as
thatthepreferredmethodswillarriveattheidealnormalvalues being 0.6 when a curve slope (change of concentration divided
earlier. by change in reading) at some point was 0.4, it would become
0.3ifthecurvewasmadetwiceassteep,thatis,whentheslope
6.1.3.1 If the operating system is based on two-point stan-
dardization, Eq 3 still would be used to normalize or standard- at the same point was changed to 0.8.
ize readings. The generation of slope and constant corrections,
7. Use of Confidence Interval to Control Spectrochemical
however, would be as follows:
Analysis
m 5 ~H –L !/~H –L ! (4)
R R O O
7.1 Practice E 876 uses Student’s t-table to establish the
and
range of reading or concentration around an average that will
k 5H –m H (5) include the true reading or concentration at some confidence
~ !
R O
level. The calculation includes the standard deviation of the
where:
measurement.To be effective, the standard deviation should be
H = Reference or normal reading of the high standardant,
R
estimated with at least 16 df. The interval straddling the
L = Reference or normal reading of the low standardant,
R
average will be 6ts/ n , where t is a factor from the t-table
=
H = Observed reading of the high standardant, and
O
for some probability level, s is the estimate of standard
L = Observed reading of the low standardant.
O
deviation, and n is the number of readings taken for one
6.1.4 Ifdataarelatertransformedbyaslopeandinterceptto
observation. If control of a method depends upon observing an
give a different scaling for the calibration, the same transfor-
intensity reading, the confidence interval may be in terms of an
mation must be applied to the readings of standardants and
intensity reading. If a method uses a computer to display
verifiers.
concentration, the confidence interval should be in terms of
6.2 If a verifier (or a new standardant) is established after a
concentration.
calibration has been defined, the expected reading can be
7.1.1 If the confidence interval is used to judge when drift
established as follows:
has occurred, it will be appropriate to use a confidence level of
6.2.1 Shortly after a standardization, run the verifier in
95 % to anticipate when control may be in jeopardy.
replicate and keep a record of its average reading. Average
7.1.2 Itwillbesatisfactorytouse2.0asanapproximationof
about ten such observations made after new standardizations to
a t-factor for the 95 % confidence. When it is observed that a
obtain a good representation of the expected reading.
verifier has exceeded this range, often referred to as a
6.2.1.1 Normalization coefficients are determined by mak-
“2-sigma” limit, a standardization should be done. This can be
ing a linear regression fit of normal readings as a function of
a half correction if the verifier reads less than three times the
observed readings, such as is done in Practice E 305 in
standard deviation, the “3-sigma” limit. Practice E 305 de-
establishing a straight line relationship by the method of least
scribes procedures for half corrections. See also 8.5.3.
squares. The “normal” set of readings can be either overall
7.1.3 Any verifier which exceeds the “3-sigma” limit will
averages or a set that appears to be a median of all sets. The
require a full standardization correction unless there is an
“slope” of this regression becomes the proportional factor, m,
assignable cause for the divergence. For an assignable cause, a
and the “intercept” the constant, k.
second verification can be made after correcting the problem to
6.2.2 If a verifier has to b
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.