Validating UV/Visible Spectrophotometers

Validating UV/Visible Spectrophotometers
T e c h n i c a l
UV/Vis Spectroscopy
Validating UV/Visible
UV/Visible spectrophotometers are widely used by many laboratories – including
those found in academia and research as well as industrial quality assurance. The
technique is mainly used quantitatively (although some qualitative analysis can
also be performed).
For any type of critical determination, whether it be clinical, pharmaceutical or
industrial QC, environmental analysis or research, it is essential that the instrument
is performing according to specification. In some of these applications, it is
important that the instrument performance is monitored regularly and that
there is documentary evidence that this is the case.
PerkinElmer, Inc.
Seer Green, UK
This will apply to any measurements to do with human or animal health (clinical or
pharmaceutical/nutraceutical), food or labs offering an accredited measurement
service (e.g. in accordance with ISO 17025). Validation is also a key requirement
of Good Laboratory Practice (GLP).
Testing frequency
One common question from laboratories relates to the frequency of testing.
There are no regulations, as such, regarding this but for any human health or
nutrition related analysis, a testing frequency of at least monthly should be
considered. If the instrument is in a high sample throughput environment, then
some additional weekly testing (e.g. absorbance and wavelength check) should
be performed.
Steve Upstone
Principal UV/Vis/NIR
Spectroscopy Specialist
N o t e
Testing wavelength accuracy
Wavelength accuracy is normally assessed by using either
a sample containing a series of very sharp peaks such as a
solution of holmium perchlorate or a holmium oxide and/or
didymium doped glass filter or by measuring the emission
from a lamp. If the instrument is equipped with a deuterium
(D2) lamp as the UV source, this can be used. An external
mercury (Hg) lamp can also be used. This is less convenient
than using the previously mentioned methods but methodology
exists (e.g. in the Ph. Eur. tests) for its use as an alternative
to a glass or liquid standard. The advantage of emission
lines is that they are inviolate (i.e. the emission wavelengths
don’t change over time).
Using the deuterium lamp
If the instrument uses a deuterium lamp, it is possible to
measure the very sharp emission line at 656.1 nm by switching
off the visible lamp and measuring the energy (single beam
mode). There is also a weaker emission line at 486.0 nm
that can be used. This is not as rigorous as using other
wavelength standards (as fewer lines are available) but it
does serve as a very convenient and accurate way to assess
that the instrument’s optics are aligned correctly. Some
PerkinElmer instruments (such as the LAMBDA™ 25/35/45)
use the 656.1 nm emission line of D2 upon switch-on to
calibrate the monochromator. Others (such as the LAMBDA
800/900 and LAMBDA X50 series) can perform this calibration
on demand (and store the values in the instrument firmware).
UV WinLab™ v5 and v6 have methods to check this and
produce a report). These methods use the rear beam for the
procedure (just in case a sample has been left in the front
position by mistake). The 486.0 nm emission line (but not
the 656.1 nm line) is one of the possible test points in the
Ph. Eur. 5.2 (2005).
253.7 nm 302.25 nm 313.16 nm 334.15 nm 365.48 nm
404.66 nm 435.83 nm 547.07 nm 576.96 nm 579.07 nm
Figure 2. Wavelengths for mercury emission lamp.
Glass and liquid filters
The most commonly used glass (or quartz) filter is holmium
oxide. This provides a quick and easy way to check wavelength accuracy. This standard gives a series of lines in the
visible region. It does not provide strong absorption peaks
into the UV region and so a solution of holmium oxide in
perchloric acid (holmium perchlorate) is recommended by
some pharmacopoeias but for general non-pharmacopoeia
instrument testing, holmium oxide glass is widely used.
Another material that is used is “didymium” – a mixture of
praseodymium and neodymium. This is recommended by
the U.S. Pharmacopoeia (USP 24) as well as others such as
the JP (Japanese Pharmacopoeia). As with holmium oxide,
it is also available in solution form. Other wavelength standard
solutions commercially available include samarium perchlorate,
which gives a range of wavelengths in the UV and visible
regions between 230 and 560 nm and “rare earth sulphate”
which gives far UV wavelengths between 201 nm and 253 nm.
Holmium oxide quartz
279.2 nm
453.7 nm
637.7 nm
360.9 nm
536.5 nm
Holmium perchlorate solution 241.15 nm 287.15 nm
361.5 nm
536.3 nm
Didymium glass
440.4 nm
513.4 nm
879.3 nm
481.0 nm
684.5 nm
Figure 3. Wavelengths for holmium oxide glass, holmium perchlorate solution
and didymium glass.
Figure 1. Deuterium peak showing peak close to 656.1 nm (smaller peak is
due to the heavier tritium).
Using a mercury emission source
The use of a mercury (Hg) emission lamp is another way to
check wavelengths. This lamp offers more lines compared to
deuterium – particularly in the UV region. The practicalities
of using such a lamp are a little more difficult. The mercury
lamp needs to be placed in the instrument’s lamp compartment.
The principal mercury emission lines are shown in Figure 2.
Figure 4. Spectrum of holmium perchlorate solution for wavelength accuracy
Photometric (absorbance) accuracy testing
As with wavelength accuracy testing, either solutions or
glass/quartz filters can be used. For general applications
neutral density grey glass filters (such as the NIST® 930D
set) are a convenient way to test the photometric accuracy
of instruments. These offer a number of wavelengths in the
visible region. For testing in the UV either “metal on quartz”
glass filters or solutions must be used. Metal on quartz filters
have a reflective coating applied to a quartz plate. This can
cause issues due to the strong back reflection sent back
through the instrument’s optics so they may not be suitable
for all types of instruments. The coating needs to be treated
with care as any damage will alter the optical properties.
The most commonly used solution for checking absorbance
accuracy is potassium dichromate. The original 1988 Ph. Eur.
method tests absorbance at four wavelengths – 235, 257,
313 and 350 nm using between 57.0 and 63.0 mg of
potassium dichromate in 0.005 M sulphuric acid diluted to
1000 mL. Since 2005 a second solution has been added to
provide an additional test point at 430 nm. This uses the
same amount of potassium dichromate but is made up to
100 mL (i.e. it is ten times more concentrated than the
original solution). In both cases the A 1%/1 cm value is
recorded and checked against the target range.
Nictonic acid (0.1 M hydrocholoric acid) in a range between
6 and 24 mg/L) can be used to assess photometric accuracy
in the far UV region (at 213 nm and 261 nm). This is not an
official procedure but reference materials are commercially
available for this test.
Care has to be taken when preparing the solutions as
potassium dichromate is hygroscopic and so it is important
that this is dried thoroughly otherwise the weight will be
incorrect and this could result in an unnecessary failure.
Double aperture correction for photometric
No instrument has a perfectly linear detector response and
so the observed photometric accuracy will have a greater
deviance from the certified values as a result. It is possible to
correct for this systematic error by using a technique called
double aperture correction. This procedure is only available
for instruments at the upper performance range. To correct
for this, it is necessary to purchase a double aperture
accessory. This accessory consists of a series of shutters –
either manual or motorized – which are able to measure the
top and bottom half of the beam. The mechanism must be
highly precise such that there is no overlap between the two
measurements. It is then possible to calculate a correction factor
that can be used for subsequent measurements. Instrument
manufacturers will quote photometric accuracy both with
and without double aperture correction as part of their
specifications. It will also be used by some national standards
laboratories as part of their determinations but it is very rare
to find it used elsewhere due to its complexity and offering
a precision that is unnecessary for most measurements.
Figure 5. Double aperture accessory.
Figure 6. Double aperture correction for detector non-linearity.
Stray light testing
Stray light can be defined as the amount of unwanted
wavelengths that are present in the sample beam. In a
conventional “dispersive” instrument, the monochromator is
before the sample and the sample receives one wavelength
at a time. After the sample, there is only the detector. This
means that any stray light that gets past the monochromator
will be seen by the detector. Stray light has a strong impact
on the instrument’s ability to record accurately at higher
absorbances and is a key specification when assessing the
quality of a spectrophotometer. For a single monochromator
instrument (as found in most chemical and biochemical
laboratories) the best achievable stray light value (using
ASTM® methodology) will be around 0.01%T. This equates
to one in every 10,000 photons being due to stray light. This
may sound like a small amount but it has to be remembered
that absorbance is a logarithmic scale and so 3AU (equivalent
to 0.1%T) equates to only one in a thousand photons actually
passing through the sample and reaching the detector. If the
stray radiation is mainly “far” stray light (i.e. wavelengths
some distance from the actual analytical wavelength) then it
is unlikely to be absorbed by the sample and thus will cause
the linearity to become worse at higher absorbances as the
absorbance becomes swamped by the [constant] stray light.
If linear performance greater than 3A is required, systems
with more than one monochromator must be considered –
either premonochromator (using a smaller auxiliary
monochromator) or a full double monochromator (two
equivalent monochromators). It should, however, be taken
into consideration that for most liquid samples, they can
be diluted or measured in short pathlength (e.g. 5 mm)
cuvettes and so measurement at such a high absorbance
value shouldn’t be necessary – multiple monochromator
systems are usually only necessary for materials and optics
analysis where it is not possible to adjust the sample.
All tests should be carried out using a 1 cm (10 mm) quartz
Figure 8. ASTM® stray light test at 220 nm using a 10 g/L sodium iodide
(NaI) solution showing stray light to be less than 0.01% T.
Stray light testing for pharmacopoeia
Figure 7. Effect of instrumental stray light on absorbance linearity.
Stray light tests should be carried out on the basic instrument
and in a configuration specified by the manufacturer for
that particular test. If any accessories are fitted (e.g. sippers,
integrating spheres, cell changers, etc.) these can have a
detrimental effect on the stray light value due to additional
light leakage, etc. Clear plastic tubing entering the sample
chamber from outside can act as a fiber optic and transfer
light from outside to the inside of the instrument. It is
recommended to either use black tubing or to shroud the tubing
(an outer black tubing sheath works well) for about 5 cm
prior to entering the sample area in order to minimize this.
The European Pharmacopoeia (Ph. Eur.) performs the test
in absorbance (unlike ASTM® where transmittance is used
as the ordinate mode) and the recorded value of a 12 g/L
aqueous potassium chloride solution at 198 nm should
be 2A or greater. This test was revised in 2005 as the test
wavelength was moved from 200 nm (which had been used
since 1988) to 198 nm. The original 200 nm wavelength
was influenced somewhat by the sample temperature and
whether or not the solution had been degassed (sealed
cuvettes were less prone to this) and so moving the test
point down to 198 nm made the test more robust.
Stray light testing to ASTM® methodology
ASTM® stray light testing involves the use of two solutions
– 10 g/L sodium iodide (NaI) for 220 nm and 50 g/L sodium
nitrite (NaNO2) for 340 nm and 370 nm. These solutions
have a very sharp cut-off in the UV region. This means that
if light is detected at a wavelength where the sample is
supposed to be completely blocking, this must be due to
stray light passing through the transparent part of the sample
spectrum. As a further confirmation, it is possible to block
the beam (the metal frame of our glass filters inserted “sideways on” into the instrument makes an ideal beam blocker)
and compare the measurement of the stray light solution to
the blocked beam – the difference between the two giving
the stray light value. A similar test can be performed at
650 nm using a 0.1 g/L methylene blue solution or measuring
around the cut-off of pure acetone between 250 and 320 nm.
Figure 9. Stray light testing according to Ph. Eur. using 12 g/L KCl solution
showing the difference in stray light in a premonochromator instrument
(LAMBDA 45) compared to a single monochromator instrument (LAMBDA
25/35). It also shows the improvement in the result when using 198 nm as the
test point instead of 200 nm.
Some pharmaceutical companies use their own stray light
test solutions such as a saturated solution of lithium carbonate
(at 227 nm) or sodium chloride (at 205 nm) as part of their
in-house procedures but these are not officially recognized
by the major pharmacopoeias.
Resolution testing
The European Pharmacopoeia introduced a resolution test
as part of the original 1988 methodology. This is based on
measuring a solution of 0.02% w/v solution of toluene in
hexane, measuring the ratio of the peak (at 269 nm) and
the trough (at 266 nm) and measuring the ratio. This ratio
should be 1.5 or greater unless prescribed in the monograph. This means that, although it is highly desirable for
an instrument to pass this test, it is not essential but most
manufacturers regard this to be a part of making an instrument “pharmaceutical compliant” and some carry out the
measurement as part of their final test procedure. In practice, an instrument with a bandpass (slit width) of around
1.5 nm should pass this test. For a fixed bandpass instrument, the measurement should give fairly constant results as
there is no mechanical movement involved. If the instrument
has variable bandpass – particularly if the slits can be varied
continuously using a stepper motor “jaw” mechanism, test
results may vary more and so the test should be carried out
more frequently.
Additional testing
It may be useful to perform additional tests on the instrument
using methodology not contained in pharmacopoeias but
using a subset of the manufacturers’ own in-house service
procedures. These can give an indication of things such as
lamp deterioration. Some examples are:
Absorbance stability (drift) – measuring how an absorbance
changes (usually in absorbance per hour) under stipulated
Baseline flatness – the instrument is autozeroed with nothing
in the sample compartment across its working range and
this is then measured. Some manufacturers specify flatness
over a restricted range (e.g. 200-800 nm) which may not be
the full working range of the instrument. This is normally
the case with photomultiplier based instruments where the
photomultiplier suffers a fairly major loss of sensitivity above
800 nm.
Noise – this is measured at a particular wavelength under
prescribed conditions. It may be measured at zero absorbance
or at a specified absorbance value (e.g. 1A).
Reproducibility (both in terms of wavelength and photometric
Testing reflectance
Some laboratories wish to test the accuracy in reflectance.
These tend to be the more highly specified instruments and
they must be equipped with a suitable reflectance attachment.
This is usually an integrating sphere for diffuse and total
reflectance and either a relative or absolute attachment for
specular reflectance.
Figure 10. Resolution test using 0.02% w/v toluene in hexane (in this example
ratio is approx 1.9).
In 2005 the European Pharmacopoeia introduced a second
method involving a solution of toluene in methanol and
based on second derivative (D2) spectroscopy. At the time of
writing, this test is not widely used.
For diffuse reflectance, it is possible to purchase diffuse
reflectance standards. These are either grey standards where
differing amounts of carbon black are added to the white
material or color standards where the white material is
dyed. For perfectly diffuse materials, Spectralon® is probably
the best material. This is a proprietary white sintered PTFE
material which is resistant to yellowing and can be cleaned
easily. When measuring these standards, it is best to use as
large a beam as possible. The standards aren’t completely
homogenous and so, for very small beams, it is possible
to pick up individual “smuts” of carbon which will give
an inconsistent result. By measuring the %RSD (Relative
Standard Deviation (Figure 16 – Page 9) of these standards,
the expected tolerance can be assessed.
For measuring specular reflectance, the most commonly
used material is a front surface coated aluminum mirror as
this reflects across the UV/Vis/NIR range. These mirrors are
less reliable in the UV range as any surface scratches will
contribute to higher scatter in the UV region due to the
physics of scatter phenomena. One other issue with these
mirrors is that they are often required to be traceable as
part of a laboratory gaining accreditation (e.g. ISO 17025).
Traceable mirrors often have a poor specification – such
as ±1% reflectivity and this may not be good enough for
some applications (such as the measurement of laser mirrors
where accuracy of 0.1%R is highly desirable). The reason for
this wide tolerance is that national standards labs compare
their standards with other laboratories and so, the more
laboratories are involved, the higher the statistical spread of
data which increases the specified uncertainty.
When testing reflectance, there can be some debate about
the “correct” result for a particular sample. Most standards
are fairly ideal in nature so, for example, Spectralon® is
almost perfectly diffuse and an aluminum mirror is almost
perfectly specular. Actual samples may lie somewhere in
between – displaying a mixture of both types of reflectance.
In order to address this issue, some industries (e.g. paint
manufacturers) sometimes use calibrated ceramic tiles as
they have a higher specular component.
• Photometric accuracy using grey glass standards. Some
grey glass standards have test points in the NIR region in
addition to the UV/Visible range.
• Reflectance – both Spectralon® and aluminum mirrors can
be used to check reflectance in the NIR region. Spectralon®
cuts off above 2500 nm.
Validating PerkinElmer UV/Vis and UV/Vis/NIR
Spectrophotometers Using UV WinLab software
UV WinLab software (v4 and higher) contains features to aid
with the calibration of spectrophotometers.
For medium performance instruments (LAMBDA 20, 25/35,
40/40P and 45), there is the Instrument Performance
Verification (IPV) module. This module contains over twenty
tests that the user can perform to check performance. Some
tests require a standard or an external mercury lamp whereas
others do not.
When measuring reflectance (unless using an absolute
accessory), it is necessary to zero the instrument and accessory
using a standard whose reflectance is known. It is then
necessary to perform a mathematical correction to remove
the contribution of this standard and convert the relative
measurement into an absolute one. If this is not done
(i.e. the relative value is used) for a set of absolute standards,
the results will be invalid.
Testing in the Near Infrared (NIR) region
Some spectrophotometers also have near infrared capability
and can measure in excess of 900 nm and so it is necessary
to assess performance in this region. There are some test
procedures that are applicable in this region. These include:
• Wavelength accuracy of the NIR grating using the second
order deuterium emission line at 1312.2 nm or 972 nm
(i.e. two times 656.1 nm and 486 nm. respectively).
• Wavelength accuracy using a holmium oxide which gives
a peak at 1938 nm.
• Wavelength accuracy using a McCrone standard. This is a
composite standard which give additional lines in the NIR
from 481 nm to 2326.8 nm (available from the U.K. NPL).
• Stray light using water at 1420 nm chloroform at 2365 nm.
Figure 11. IPV module – methods list.
Each test can be populated with standard serial number and
calibration date information (where applicable) and a testing
frequency for each test can also be established. In the ES
(Enhanced Security) version of the software, it is possible
to set the permissions such that an analyst cannot use the
instrument if it has failed the calibration or the calibration is
overdue (this can be over-ridden by a system administrator,
if required). If any tests are required that are not part of the
IPV module, these can be created using standard UV WinLab
Test Standards Required
Wavelength accuracy – third party lamp Third party lamp (e.g. mercury)
Wavelength accuracy glass Calibrated standards set (B0507805). Traceable to German PTB
Wavelength accuracy solution Pharmacopoeia standards set (B2500100) using holmium perchlorate (Ph. Eur. test)
Wavelength accuracy – D2 No standard required – uses instrument’s deuterium lamp (486 nm and 656.1 nm)
Wavelength repeatability – D2 No standard required – uses instrument’s deuterium lamp (486 nm and 656.1 nm)
Resolution – benzene vapor Sealed Benzene cell (toluene acceptable substitute)
Resolution – toluene Pharmacopoeia standards set (B2500100) using toluene in methanol (Ph. Eur. test)
Photometric accuracy glass Calibrated standards set (B0507805). Traceable to German PTB
Photometric accuracy – K2Cr207 Pharmacopoeia standards set (B2500100) using K2Cr207 solution (Ph. Eur. test)
Photometric accuracy – KNO3 solution KNO3 solution
Photometric repeatability glass Calibrated standards set (B0507805). Traceable to German PTB
Baseline stability (flatness)
No standard required
Noise (three tests) No standard required
Absorbance zero stability (drift) No standard required
Stray light KCl solution (1988/2005 Ph. Eur. test) Stray light standards (B2500099)
Stray light standards (B2500099)
Stray light NaI (ASTM®) Stray light NaNO2 (ASTM )
Stray light standards (B2500099)
Resolution (toluene in methanol – Ph. Eur. 2005)
Toluene in methanol solution
Photometric accuracy – K2Cr207 at 430 nm
(Ph. Eur. 2005) Solution of K2Cr207
For the high performance instruments, there is a collection
of Service Methods. These are similar in nature to the IPV
methods but there are more of them (reflecting the additional complexity of these instruments) and they are regular
analytical methods arranged into a collection rather than
being part of a separate software module. There are also
additional service methods for some major accessories such
as the integrating spheres and the Universal Reflectance
Accessory (URA).
Figure 12. Service methods for the high performance instruments.
Calibration utility for the high performance
LAMBDA instruments
The high performance LAMBDA instruments also have an
onboard facility to adjust the instrument’s slit and wavelength settings. This calibration utility can be accessed by
right-clicking on the instrument icon (in the instruments
node of UV WinLab). Unlike other tests described here, it
will measure and correct for any wavelength shift detected
– it can be regarded as a kind of “autotune” routine. It also
calibrates the continuously variable slit mechanism as well as
checking the 0%T offset. The wavelength calibration is based
on the 656.1 nm deuterium emission line (the 486.0 nm line
can be also used optionally). The NIR monochromator can
also be checked using the second order 1312.6 nm line.
Please note that the photomultiplier is used for all tests
(even the NIR wavelength accuracy – due to the fact that it
is a second order effect and so the detector is still actually
detecting at 656.1 nm). Where possible, it is best to calibrate
the instrument using the standard detector rather than with
an accessory. This is because very narrow slits are used for
the wavelength calibration and so sensitivity is important. For
dedicated “S” and “R” versions of the LAMBDA 650 and
750, the calibration routine in v6.0.4 (released in October
2011) has been enhanced in order to allow successful
calibration when the sphere or URA are installed in place
of the standard detector.
Figure 14. Common non-disposable cuvette material transmittance profiles
(graph courtesy of Hellma GmbH).
Figure 13. Calibration utility screenshot.
Validating solvents
Setting tolerances
When devising a validation test, it is necessary to set realistic
tolerances for the test, otherwise it will fail unnecessarily.
The tolerance should comprise of three parts which should
be added together. These are:
• The uncertainty of measurement given on the standard
certificate (or published emission wavelength if using a
source emission line).
• The manufacturer’s specification for the instrument that
relates to that particular standard (e.g. a wavelength
accuracy test will be influenced by the instrument’s
wavelength accuracy specification as well as the tolerance
of the standard).
• Any influence to the measurement coming from the
operator or other parts of the system.
Validating assays
So far, the instrument validation itself has been discussed.
An instrument may be working perfectly but may still not
be fit for purpose of a particular assay. This may be because
the instrument has unsuitable specifications for that particular
determination. As a trivial example, a perfectly functional
instrument covering the range 190 to 900 nm would be
unsuitable for measuring absorbance at 1000 nm.
Cuvette validation
Testing of the type of cuvette chosen for a particular assay
should also be included in any validation of the assay. This
should be in terms of design (macro, semi-micro, micro,
ultra-micro, whether the walls are blackened, etc.), construction
material (quartz, fused silica, NIR silica, glass, etc.) and the
manufacturing tolerance of the nominal pathlength (usually
to German norm DIN 58963-1/2).
Water is a commonly used solvent in UV/Visible absorption
spectroscopy. It is a near perfect material offering transparency over of the UV/Visible range (but having stronger
absorption bands in the NIR region) and ideal for the wide
range of water-soluble analytes.
Some materials (e.g. organic species) will not dissolve in
water and others (mainly biological compounds) may require
other substances to be present in order to stabilize the
material (e.g. buffer solutions).
It is necessary, therefore, to check that the analysis is being
conducted at a wavelength where the solvent is sufficiently
transparent for the assay to be stable and reproducible. This
can be done by assessing the cut-off point of the solvent.
The most commonly used definition is the point at which
the solvent has a transmittance of 10% (1 Absorbance)
in a 1 cm cuvette. This can be ascertained quite easily by
scanning the solvent or buffer on its own in a 1 cm quartz
cuvette (using air as the autozero) and observing the point
at which the spectrum reaches 1 absorbance.
Beyond the cut-off point, the solvent becomes completely
opaque and the instrument will try as best it can to
continue measuring (particularly if solvent is in both sample
and reference beams) but what is observed is not useful
analytically and routine methods should be constructed in
such a way that they don’t collect data where the solvent
is beyond its cut-off point as it will only cause confusion
in trying to interpret meaningless results. Measuring the
sample in energy (single beam) mode will give a clearer
picture of what the instrument’s detector is observing.
Solvent manufacturers offer different purity grades of
solvents and the UV cut-off point will normally be stated
on the bottle. Some approximate cut-off points of common
solvents are shown in the table below.
Cut-off (nm)
(>1 AU in 1 cm cuvette)
Ethyl Acetate
Using statistics to evaluate the data
Statistical analysis techniques provide a useful “toolkit” in
the assessment of validation data – both in terms of the
instrument calibration and also with respect to a particular
It is also necessary to determine the robustness of a
calibration or assay in terms of its accuracy and precision.
The diagram below gives a good visual representation of
these two terms.
Mean Value
The arithmetic mean value of a set of data points.
Standard Deviation
sx =
(xi – x)
Relative Standard Deviation (%RSD)
RSD(%) =
sx • 100
Residual Error (of linearity data or a calibration curve)
sy =
(y*i – yi)2
sx =
(xi – x)
n Number of data points.
xi The analytical value of the data point (e.g. concentration).
yi The ordinate value of the data point (e.g. absorbance).
yi* Corresponding point on the calibration curve.
df Number of degrees of freedom (linear curve without
intercept = 1, linear curve plus intercept term = 2,
quadratic curve without intercept = 2, quadratic curve
with intercept = 3).
Error assessment
Random errors
Figure 15. Pictorial representation of accuracy and precision.
Assessing precision is useful when dealing with autosamplers –
either “sippers” or carousel type devices. These devices rely
on pumping the sample through a flowcell and are, therefore, prone to carryover. Although it is possible to devise operating parameters that will avoid this (e.g. minimum pumping
times), they will be affected by things such as sample viscosity,
temperature, etc.
Statistical assessment of the measurement result
The following statistical formulas are useful in assessing
precision and accuracy:
Random errors are caused by unavoidable deviations of
the analytical results obtained with n determinations of
the same sample from the related value (mean value of
unknown samples, known value with reference samples).
A measure for the random error is the repeatability, estimated
from measurement values of completely independently
repeated measurements. Each repeated measurement must
be carried out as if it were a separate unknown sample.
Systematic errors, bias
Systematic errors (bias) are the deviations of the mean value
obtained from n repeated measurements (expectation) from
the related value (true or conventional true value). If there
are systematic errors, this difference is bigger than the range
of confidence of the mean value. The respective range of
confidence is calculated from the standard deviation according
to the given statistical accuracy. The systematic deviations
are also characterized by the term trueness.
Shewhart Charts
Shewhart Charts (or Means Control Charts) are a useful tool
to trend data – either within a particular analytical batch
(if there are a statistically significant number of samples) or
over a time period. They are useful for trending both analytical
data and the results from calibrations so that any change
in instrument performance can be monitored and the result
of any change in the measurement system (such as a lamp
change) can be assessed. UV WinLab uses a database as the
primary repository for data and so it is possible to query this
database using a combination of search criteria such as user,
date range, instrument, keywords in either the method or
task (data) name in order to select data that fits the search
In the example shown, the deviation from the deuterium
emission line at 656.1 nm is trended to show the mean
deviation is approximately 0.016 nm (specification is 0.08 nm
for this particular instrument – a LAMBDA 1050 UV/Vis/NIR
spectrophotometer). Statistical limits (based on 2 and 3
standard deviations – 2σ and 3σ) are also shown on the
Figure 16. Shewhart chart showing deviation from deuterium emission line at
656.1 nm.
Standards Available from PerkinElmer
The following standards materials are available from PerkinElmer:
Part Number
Photometric and Wavelength Accuracy Reference Material Set
Set of secondary reference materials traceable to NIST® to verify the ordinate and
abscissa (wavelength) accuracy of UV/Vis and UV/Vis/NIR spectrometers. The set
comprises four glass filters mounted in precision aluminum spring mounts. They are
designed for use with the standard 10 mm cuvette holders of most UV/Vis and UV/Vis/
NIR spectrometers. The set is provided in a robust aluminum container. There are three
neutral density glass filters for checking the ordinate accuracy and a holmium oxide glass
filter for checking the wavelength accuracy of a spectrometer. Each filter has a unique
identification number. The certified ordinate and abscissa values and tolerances of each
filter are quoted in the accompanying Certificate of Calibration. Measured wavelengths
for the three neutral density filters are 440.0, 546.1, 635.0, 1700, and 2300 nm. The glass
filters are chosen to provide nominal absorbance values of 0.3 A (50%), 0.5 A (30%) and
1 A (10%).
German PTB
Holmium Oxide Glass Filter
U.S. National Institute of
Holmium oxide glass filter for checking the wavelength accuracy of a UV/Vis or UV/
Standards and Technology (NIST®)
Vis/NIR spectrometer. The filter is mounted in precision aluminum spring mount and is provided in a robust aluminum container, with a Certificate of Calibration.
Stray Light Solutions Reference Material Set
Set comprising four solutions in sealed cuvettes – one reference solution and three test solutions. The cuvettes are designed for use with the standard 10 mm cuvette holders of most UV/Vis and UV/Vis/NIR spectrometers. The test solutions comprising sodium
iodide, sodium nitrite and potassium chloride, which act as high-pass filters with very
precise cut-off wavelengths. Potassium Chloride (KCL) for measuring stray at 200 nm,
Sodium Iodide (NaI) for measuring stray light at 220 nm, and Sodium Nitrite (NaNO2)
for measuring stray light at 340 and 370 nm. The set is provided in a robust aluminum
container, with a Certificate of Calibration.
Instrument calibrated using
NIST® 930D standards
Standards Available from PerkinElmer, continued
Part Number
Pharmacopoeia Reference Material Set
Set comprising six solutions in sealed cuvettes – two reference solutions and four test solutions to verify calibration in accordance with pharmacopoeia monographs:
• Ordinate (photometric) accuracy; potassium dichromate. Measured wavelengths are
235, 257, 313 and 350 nm
• Abscissa (wavelength) accuracy; holmium perchlorate
• Level of stray radiation; potassium chloride for measuring stray radiation at 200 nm
• Resolution; toluene in hexane
Instrument calibrated using
NIST® 930D standards
The cuvettes are designed for use with the standard 10 mm cuvette holders of most
UV/Vis and UV/Vis/NIR spectrometers. The set is provided in two robust aluminum
containers with Certificates of Calibration.
PELA9057 (1.25" dia.)
Calibrated Spectralon® Diffuse Reflectance Standard
250 - 2500 nm, reported every 50 nm. CD-ROM provided with data every 1 nm.
U.S. National Institute of
Standards and Technology (NIST®)
PELA9058 (2" dia.)
(1.25" dia.)
(2" dia.)
(1.25" dia.)
(2" dia.)
Spectralon® UV/Vis/NIR Diffuse Reflectance Reference Materials
Spectralon® diffuse reflectance reference material sets consist of a diffuse white reference
material and a selection of three diffuse gray reference materials. Each reference material
in the set is supplied with complete diffuse reflectance data from 250 nm to 2500 nm, and
is mounted in an anodized aluminum frame. The set is packed in an airtight storage case.
U.S. National Institute of
Standards and Technology (NIST®)
• Durable, chemically inert
• Reflectance values – 99%, 75%, 50% and 2%
• Washable
Spectralon® UV/Vis/NIR Diffuse Reflectance Reference Materials
U.S. National Institute of
Spectralon® diffuse reflectance reference material sets consist of a diffuse white reference Standards and Technology (NIST®)
material and a selection of seven diffuse gray reference materials.
Each reference material in the set is supplied with complete diffuse reflectance data from
250 nm to 2500 nm, and is mounted in an anodized aluminum frame. The set is packed
in an airtight storage case.
• Durable, chemically inert
• Reflectance values – 99%, 80%, 60%, 40%, 20%, 10%, 5% and 2%
• Washable
Set of Four Color Standards (Red, Green, Blue, and Yellow)
(1.25" dia.)
PELA9019 (2" dia.)
U.S. National Institute of
Standards and Technology (NIST®)
Set of Four Color and Four Grey 2" Standards (Red, Green, Blue, Yellow, Plus
Four Greys) (99%, 50%, 20%, 2%) U.S. National Institute of
Standards and Technology (NIST®)
Set of Eight Color 2" Standards (Red, Green, Blue, Yellow, Orange, Cyan, Violet
and Purple)
U.S. National Institute of
Standards and Technology (NIST®)
UV/Vis/NIR Second Surface Reference Mirror
Metal-dielectric multi-layer second surface reference mirror design optimized for stability and maximum reflectance in the UV/Vis/NIR range from 200 nm – 2,600 nm
for checking the accuracy of absolute and to correct the spectra of non-absolute
specular reflectance accessories.
• Easy to clean, protective 2 mm silica front plate
• Smooth reflectance curve without water related absorption bands in the NIR range
• Size 50 mm x 100 mm
• Calibration at 8 degrees incidence in the wavelength range 200 nm - 2,500 nm,
calibration uncertainty typical < 0.25%
Manufactured by OMT Solutions,
Eindhoven, Netherlands
This is not traceable to a national
standard as it offers greater
accuracy than any traceable
material commonly available
standard material.
Delivered in a lightweight protector case, accompanied by the calibration report and a
CD-ROM with the calibration data in different Excel® formats.
Standards Available from PerkinElmer, continued
Part Number
Disposable Reference Mirror for Reflectance Spectroscopy
This mirror is necessary to check the accuracy of absolute and to correct the spectra of non-absolute specular reflectance accessories. The mirror is a secondary reference material and is calibrated according to GLP regulations, wavelength range 250 to 2,500 nm.
U.S. National Institute of
Standards and Technology (NIST®)
Recalibration of standards
Most standards can be recalibrated. In some cases (e.g.
aluminum mirror), the standard itself is an inexpensive
material and also gets scratched over time and so it is not
practical to have it recalibrated. It is, therefore, referred to as
Some liquid standards can present issues with couriers/ airlines
when being sent back for recalibration due to the materials
(albeit only a trace) that they contain.
Further reading
Miller, J.N., “Basic Statistical Methods for Analytical Chemistry,
Part 2. Calibration and Regression Methods, A Review” Analyst,
116 (1995).
Demey, R.C., Sinclair R., Visible and Ultraviolet Spectroscopy,
John Wiley and Sons, New York, 1987.
Nijnatten, P.A., “Calibration of Neutral Density Glass Filters to
Produce Transmittance Standards” TNO Institute of Applied
Physics, Eindhoven, The Netherlands (now contactable via
MacDonald, R.P., Uses for a homium oxide filter in spectrophotometry. ClinChem 10: 1117-1120, 1964.
Slavin, W., Stray light in ultraviolet, visible and near-infrared
spectrophotometry. Anal Chem 35: 561-566, 1963.
Standard definitions of terms and symbols relating to molecular
spectroscopy (E131-71), in Manual on Recommended Practices
in Spectroscopy. Philadelphia, PA: American Society of Testing
Materials Committee E13, 1969, pp 10-34.
Standard method of estimating stray radiant energy (E38772), in Analytical Methods of Spectroscopy, Philadelphia, PA:
American Society of Testing Materials Committee E-13, Annual
Book of ASTM® Standards, 1974, pp 291-302.
PerkinElmer, Inc.
940 Winter Street
Waltham, MA 02451 USA
P: (800) 762-4000 or
(+1) 203-925-4602
For a complete listing of our global offices, visit
Copyright ©2012, PerkinElmer, Inc. All rights reserved. PerkinElmer® is a registered trademark of PerkinElmer, Inc. All other trademarks are the property of their respective owners.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF