CLIN. CHEM.

32/1, 162-165 (1986)

Monitoring the Stability of Wavelength Calibration of Spectrophotometers

William J. Korzun”2 and W. Greg Miller1

The difterence in absorbance (A) between equimolar acid but also in the magnitude of the absorbance difference (M)
and alkaline solutions of methyl red, at a wavelength near the at a wavelength near the isosbestic point, as seen in Figure
isosbestic point of the indicator, is reproducible. Furthermore, 1. In a non-scanning instrument, it is more convenient to
this A issensitiveto changes in the wavelength calibration measure .A than the isosbestic point itselL
of the instrument used to make the measurement. The iA of Here we report how the A near the isosbestic point
methyl red can be used to monitor wavelength accuracy in between an acid and alkaline solution of methyl red, 2-{[4-
both manual and automated spectrophotometnc instruments. (dimethylamino)phenyl}azo}benzoic acid, can be used to
Although this measurement does not establish wavelength detect changes in the wavelength-calibration status of spec-
calibration, it is useful for monitoring the wavelength accura- trophotometers. We used this technique to monitor the
wavelength accuracy of the automated spectrophotometric
cy of previously calibrated, automated spectrophotometers
system in a “Cobas Bio” centrifugal analyzer.
that do not easily lend themselves to calibration checks by
conventional techniques. Materials and Methods
AdditIonal Keyphrases: methyl red variation, source of Instruments
centrifugal analyzer . spectrophotometry
We calibrated Models 24 and 25 spectrophotometers
(Beckman Instruments Inc., Fullerton, CA 92634) with the
Wavelength accuracy is one of many factors that can
656.3-nm and 468.1-nm emission lines of the deuterium
significantly affect the overall accuracy of spectrophotomet-
lamps that are housed in the instruments. For all absor-
nc methods, as exemplified by the wavelength sensitivity of bance measurements with these instruments, we used
the determination of serum alkaline phosphatase activity
square, 1.0-cm (light path) quartz cuvettes matched to
(1). Commonly used techniques for calibrating the wave- within 0.002 A. We calibrated a Stasar Ill spectrophotome-
length settings of spectrophotometers-use ofthe character- ter (Gilford Instruments, Oberlin, OH 44074) with the 536-
istic absorption maxima of holmium oxide or didymium nm and 361-nm absorbance maxima of the holmium oxide
ifiters and the characteristic emission lines of mercury or
ifiter supplied with the instrument. For absorbance mea-
deuterium lamps (2-7)--are suitable for quickly and easily surements we used a 1.0-cm flow-through cuvette supplied
checking the wavelength calibration of most spectrophoto-
with the instrument. We calibrated a Coleman Jr. HA
meters. However, the spectrophotometric modules of many spectrophotometer (Perkin-Elmer Corp., Norwalk, CT
automated chemical analyzers are not amenable to calibra- 06856), using the 585-nm absorbance maximum of the
tion checks by such techniques; or they may require the
didymiuni ifiter supplied with the instrument. Absorbance
attention of factory-trained personnel.
measurements were made with round glass cuvettes, 12 mm
To circumvent the use of lamps and filters, some have
in diameter and matched to within 0.002 A. A Cobas Bio
proposed using the isosbestic points ofpH indicators and (or) centrifugal analyzer system (Roche Analytical Instruments
hemoglobin derivatives as reference points for the calibra- Inc.,Nutley, NJ 07110) was calibrated by the manufactur-
tion of the wavelength settings (8-10). However, for this
er’s service engineer with the 537-nm and 361-nm absor-
approach to be useful, the isosbestic points of the species
being utilized must be determined with a calibrated refer-
ence instrument. Furthermore, the wavelength at which an
isosbestic point is observed in a test instrument will depend
upon the bandpaaa and the degree of asymmetry in both the
band spectrum of the chromogen and the slitdistribution of
the instrument (3, 4). The errors in wavelength calibration
introducted by these factors may be substantial in wide-
bandpass spectrophotometers.
Nonetheless, isosbestic points can be very useful in rou-
tine quality control of the wavelength accuracy of spectro-
photometers between recalibrations. Once an instrument
has been calibrated by a standard technique, and isosbestic
points for a chromogen have been determined with that
instrument, then the wavelengths at which the isosbestic
points are observed should remain constant unless the
wavelength calibration changes. A change in the wave-
length calibration will lead to a change not only in the
observed wavelength of the isosbestic point of a chromogen 350 400 Sbo 600 700
nm
‘Section of ClinicalChemistry, Department of Pathology, Medi- Fig. 1. Absorbance spectra of methyl red at(1) pH 2.5and ( pH 11.5,
cal College of Virginia, Box 597, Richmond, VA 23298. illustrating the observed isosbestic point (I) and #{163}4
2vious address: Department of Medical Technology, Universi-
Spectra were recorded with a Beckman Model 25 spectmphotometer: wave-
ty of South Alabama, Mobile, AL 36688. length drive at 100 nm/mm, chait drive at 5 cm/mm, and A 0-2.0. Methyl red
Received July 18, 1985; accepted September 12, 1985. concentration in the cuvette, 9.8 mg/L

162 CLINICAL CHEMISTRY, Vol.32, No. 1, 1986

bance maTima of a holmium oxide filter. Absorbance was
cuvette as described earlier. Thus, the dilution factors for
measured in the disposable plastic cuvette rotors. each reagent were the same as those generated by the
program in the Cobas Bio.
Reagents For each ofthe experiments described, the concentrations
To prepare 10 and 60 mgfL stock solutions of methyl red, of HC1 or NaOH in the cuvettes were great enough so that
we dissolved 10 and 60 mg of its sodium salt (Sigma small uncertainties in pH would have an unmeasurably
Chemical Co., St. Louis, MO 63178) in de-ionized water in 1- small effect upon the absorbance spectra of the methyl red
L volumetric flasks. These solutions were stored in stop- solutions.
pared glass bottles at room temperature protected from
light. NaOH and HC1 solutions, both in 0.1 mol/L and 30 Experimental Design
mmol/L concentrations, were prepared from reagent-grade In the first experiment, we measured the #{163}4
of methyl
chemicals in de-ionized water in volumetric flasks, and red at 465 tun with the Model 25 and Stasar Ill, and at 475
stored at room temperature in polypropylene (NaOH) or ma with the Jr. HA on 30 different days over a five-month
glass (HC1) bottles. period, to show that this value is sensitive to the wave-
Measurement of iA length-calibration status of a spectrophotometer. We used
different wavelengths because the isosbestic point of methyl
For the Model 25, Stasar ifi, and Jr. HA instruments, we red was observed at 470 nm with the Jr. HA, as compared
measured tA as follows: pipet 3.0 mL of the 10 mg/L methyl with 461.5 nm with the Model 25 and 461 nm with the
red solution into each of two cuvettes (or test tubes, for the Stasar ifi. The Jr. HA has a 20-nm bandpass. On each of the
Stasar ifi), followed by 50 zL of the 0.1 mol/L HC1 for one of first 10 days of this experiment, the wavelength calibration
the cuvettes and 50 ML of the 0.1 mol/L NaOH for the other. of each instrument was verified as above. On days 11-20,
Cover the cuvettes with Parafllm, and mix by inversion. the wavelength settings of the Stasar ifi and Jr. HA
Measure the absorbance of each cuvette at the wavelengths instruments were internally altered to yield a +2-nm error.
indicated below (see Experimental Design) vs de-ionized To serve as a control, the calibration of the Model 25
water as a reference blank. Calculate by subtracting
#{163}4 the spectrophotometer was not altered on these days. On days
absorbance of the alkaline methyl red from the absorbance 21-30, the Stasar ifi and Jr. HA were recalibrated to their
of the acid methyl red. “true” settings.
To measure with the Cobas
#{163}4 Bio, we used the following In the second experiment we wanted to verify that results
program: obtained with the Cobas Bio would be similar to those
Parameter Description Enby obtained with the instruments in the first experiment. The
a reactiondirection 11(+) isosbestic point for methyl red observed with the Cobas Bio
1 units 10 (M) was between 461 and 462 ma, which is within 1 nxn of that
2 calculation factor 1000 observed with the Model 25 and Stasar ifi. We measured the
3-5 standard concn. 0 ofmethyl
#{163}4 red with the Cobas Bio at 470 ma. We chose to
6 limit 0 use the longer wavelength, as well as the reagent modifies-
7 temp., #{176}C 25 tions discussed above, to miximize without exceeding
#{163}4
8 type of analysis 6 the solubifity of methyl red in dilute HC1 solutions, while
9 wavelength, nm 470 keeping the absorbance measurements between 0.1 and 1.0
10 sample vol., ML 50 A. Maximizing the #{163}4 of methyl red allowed maximum
11 diluent vol., pL 30 sensitivity to wavelength errors, within the range of mini-
12 reagent vol., pL 5 mum photometric imprecision. The wavelength accuracy
13 incubation time, a 10 was verified as above.
14 start reagent vol., ML 50 Using the Cobas Bio, we measured the #{163}4
of methyl red
15 time of first reading, s 0.5 on 15 different days during six weeks. On the same 15 days,
16 time interval, s 10 we also measured the #{163}4 at 468 nm, simulating the effect of
17 no. of readings 1 a +2-nm wavelength error on the absorbance measure-
18 blanking mode 0 ments.
19 printout mode 1 Finally, we investigated the long-term performance of the
To execute the analysis, we load the 30 mmol/L solutions of #{163}4
measurement in the Cobas Bio, establishing a central
HC1 and NaOH into sample cups on the sample wheel and 95% reference interval for the #{163}4of methyl red from 33
the 60 xng/L methyl red into the “second reagent” well of the replicate measurements made over the course of two weeks.
reagent boat, leaving the “first reagent” well empty. After Whenever we made more than one determination on the
the instrument pipets 50 ML of HC1 and NaOH solutions same day, we implemented a test program involving a
plus 30 pL of de-ionized water into their respective cuvette different wavelength in between those determinations. This
positions in the rotor, the absorbances of the diluted HC1 introduced the wavelength repeatability of the Cobas Bio, a
and NaOH are measured and stored for use as sample major source of day-to-day imprecision in measuring #{163}4,
blanks. Then 50 ML of methyl red and 20 ML of de-ionized into each determination. Over the next eight months, we
water are pipetted into both cuvette positions, the final periodically measured #{163}4
at 470 ma, using both the Cobas
absorbance of each is measured, the absorbance of the Bio and the Model 24 spectrophotometer on the same days.
respective sample bank is subtracted, and the net absor- Before each measurement, we verified the wavelength cali-
bance of each is printed out. #{163}4
is calculated manually as bration of the Model 24 by use of deuterium lamp emission;
described above. Both pipetting steps are carried out by the these #{163}4
values served as a reference against which we
sample pipettor, which is maintained to have imprecision could evaluate changes in the #{163}4
observed with the Cobas
(CV) of less than 1%. Bio.
To measure #{163}4 with the Model 24 Spectrophotometer, we
pipetted 1.0 mL each of the 60 mg/L methyl red solution, do- Results and Discussion
ionized water, and either MCI or NaOH (30 mmol/L) into The results of the first experiment (Table 1) illustrate the
each of two cuvettes, then measured the absorbance of each sensitivity of the #{163}4 of methyl red to small errors in

CLINICAL CHEMISTRY, Vol. 32, No. 1, 1986 163

 wavelength. the Stasar ifi, the #{163}4
With of methyl red at 465 fled with the holmium oxide ifiter three times during this
ma shifted by more than 20-fold the mean day-to-day eight-month interval: on May 7, July 5, and on October 10.
variation when only a 2-ma error in wavelength
in #{163}4 The #{163}4
of the same solution, measured on the same day in
calibration was introduced. With the Jr. IIA, at 475 urn, the the Model 24 spectrophotometer, exhibited parallel behav-
shifted by more than 12-fold its mean day-to-day
#{163}4 varia- ior. The difference in the absolute values between the two
tion in #{163}4with the same magnitude of wavelength error. In instruments is primarily ascribable to the difference in light
both instruments, recalibrating the wavelength settings path (1.0 cm for the Model 24, 0.6 cm for the Cobas Bio). The
returned the #{163}4 values very close to their original ranges. trend in #{163}4 values from May 21 through June 11 was
The discrepancies between the original range and those attributed to a deterioration in the stock solution of methyl
obtained after recalibration indicate the tolerable impreci- red. Measurements of the #{163}4 of a stock solution freshly
sion with which the wavelength settings in those instru- prepared on June 22 were all within the 95% reference
ments can be calibrated. The reproducibility of the #{163}4 interval for the old stock solution.
determined with the Model 25, for which we did not alter the Although we used the Model 24 as reference instrument
wavelength, verifies that the shifts in #{163}4 on the other two for this investigation, a shift in observed #{163}4
values owing to
instruments were not a reagent-related phenomenon. Table reagent deterioration can easily be distinguished from a
1 also shows that the Cobas Bio measurements of the #{163}4 of shift in wavelength calibration, without using a reference
methyl red were sensitive to the wavelength being used to instrument. Figure 1 shows that if the wavelength used to
make the measurement. A +2-nm error in wavelength measure changes, the absorbance values of the acid and
#{163}4
decreases the #{163}4 values by approximately 8.5 times the alkaline solutions of methyl red will change in opposite
mean day-to-day variation of the measurement. directions. If, however, the methyl red solution deteriorates,
Figure 2 illustrates the precision with which the measure- the absorbance values of the acid and alkaline solutions will
ment of #{163}4 can monitor the wavelength calibration of the both decrease. This was observed during the trend of May
Cobas Bio. The mean and standard deviation of 33 measure- 21-June 11. Thus, measuring the absorbance values of both
ments of #{163}4, made before April 15, was 0.343 ± 0.010. For the acid and alkaline methyl red solutions, rather than
eight months, #{163}4 remained within the 95% reference inter- using one as a reference blank for the other, provides a
val except for the two measurements made on June 8 and valuable troubleshooting technique for differentiating in-
11. The wavelength calibration of the Cobas Bio was veri- strumental from reagent problems.
Figure 2 shows that the #{163}4 values measured with both
instruments decreased by approximately 6% from June 22
Table 1. Effect of a 2-nm Error in Wavelength on to December 19. This trend may reflect a gradual deteriora-
A for Methyl Red tion of the methyl red solution. It should be noted that with
only a 2-ma change in wavelength, we observed an average
Instrument and Range Mean day-to-day
decrease in #{163}4 of 32% (data not shown) with the Cobas Bio.
wavelength (nm) used measured Thus, any gradual deterioration of the methyl red solution is
Model 25 small relative to the sensitivity of the reagent to wavelength
465 0.069-0.073 ±0.002 errors in a spectrophotometer. Under the storage conditions
465 0.069-0.075 indicated above, three separate stock solutions of methyl red
465a 0.069-0.073 have lasted an average of 10 months each.
Stasar Ill The identification of additional pH indicators with isos-
465 0.105-0.110 ±0.002
bestic points in different regions of the ultraviolet-visible
+2 nm error 0.062-0.067
Recalibrated 0.109-0.114
spectrum would provide a means to monitor wavelength
Jr. hA calibration with #{163}4 measurements at more than one wave-
475 0.098-0.108 ±0.003 length. The advantage of measuring the #{163}4,
rather than the
+2 nm error 0.062-0.070 isosbestic point itself, is that only two absorbance measure-
Recalibrated 0.096-0.104 ments, both made at a single wavelength, are required. In a
Cobas Bio non-scanning instrument, several absorbance measure-
470 0.332-0.365 ±0.013
+2 nm error’ 0.206-0.255 ments are required to locate an isosbestic point precisely.
aMode; 25 was not altered, to serve as a control. bsimulated error We conclude that measuring #{163}4
is a simple method for
introduced by changing the wavelength setting. n 10 each, except for Cobas
=
monitoring the wavelength calibration of both manual and
Bio, n = 15 each. automated spectrophotometers. It is particularly useful in
equipment where the spectrophotometric module is not
readily accessible to checking with filters or discharge
Se lamps.

.50(
Portions of this investigation were supported by the Department
of Medical Technology, University of South Alabama; the Universi-
.46(
AA ty of South Alabama Research Committee; and the Department of
4lOnm Pathology, Medical College of Virginia.
.35C

.34C References
1. Lott JA, Turner K, Scott J. Factors affecting measurement of
.30C V total alkaline phosphatase activity in human serum, especially
4/15 5/15 Sf15 7/15 6)15 9/15 1015 Il/iS 12115 wavelength accuracy. Clin Chem 24, 938-941 (1978).
DATE 2. Rand RN. Practical spectrophotometric standards. Clin Chem
15, 839-863 (1969).
for methyl red as measured
Fig. 2. Precision of #{163}4 with the Cobas Bio
and the Model 24 spectrophotometer 3. Aiman DH, Billmeyer FW Jr. A review of wavelength calibra-
The dashed hnes indicate the 95% reference interval for the Cobas Bio,
tion methods for visible-range photoelectric spectrophotometers. J
determined before the first date shown. The a,mw indicates the date after which a Chem Educ 52, A281-A290 (1975).
fresh stock solution of methyl red was introduced 4. Alman DH, Billmeyer FW Jr. A review of wavelength calibra-

164 CLINICAL CHEMISTRY, Vol. 32, No. 1, 1986

tion methods for visible-range photoelectric spectrophotometers 8. Parthasarathy NV, Sanghi I. A simple technique for the calibra-
(concluded). J Chem Educ 52, A315-A321 (1975). tion of the wavelength scale of spectrophotometers. Nature (Lon-
& West MA, Kemp DR Practical standards for UV absorption and don) 182, 44 (1958).
fluorescence spectrophotometry. Am Lab 9, 37-49 (1977). 9. Fog J. Calibration of the wavelength scale on spectrophoto-
6. Lucas DH, Blank RE. Spectrophotometric standards in the meters by the “indicator method.” Scand J Clin and Lab Invest 14,
clinical laboratory. Am Lab 9,77-89(1977). 320 (1962).
7. Frings CS, Broussard LA. Calibration and monitoring of spectro- 10. Hoxter G. Suggested isosbestic wavelength calibration in clini-
meters and spectrophotometers. Clin Chem 25, 1013-1017 (1979). cal analyses. Clin Chem 25, 143-146 (1979).

CLIN. CHEM. 32/1, 165-169 (1986)

The EPOS Automated Selective Chemistry Analyzer Evaluated

Godfrey C. Moses, GeraldIne 0. Lightle, James F. Tuckerman, and A. Ralph Henderson1

We evaluated the analytical performance of the EPOS (Eppen- curcy vapor lamp and double interference-filter combina-
dorf Patient Oriented System) Automated Selective Chemis- tion, can be used for analyses at five different wavelengths.
try Analyzer, using the following tests for serum analytes: Because reactants are mixed by air pulsed by radiowaves,
alanine and aspartate aminotransferases, lactate dehydro- no moving parts touch the reaction mixture during the
genase, creatine kinase, gamma-glutamyltransferase, alka- mixing process. The U-shaped quartz cuvet facilitates sam-
line phosphatase, and glucose. Results from the os corre- ple or reagent delivery and mixing.
lated well with those from comparison instruments (r Other attractive features of the Eros are negligible sam-

0.990). Precision and linearity limits were excellent for all ple-to-sample carryover; automatic dilution of samples hav-
ing high enzyme activity or high analyte concentration;
tests; linearity of the optical and pipetting systems was
automatic cuvette washing, rinsing, and drying; and the
satisfactory. Reagent carryover was negligible. Sample-to-
detection of nonlinear reactions, substrate depletion, and
sample carryover was less than 1% for all tests, but only
endogenous interferences. These features are extremely
lactate dehydrogenase was less than the manufacturer’s
important when one is measuring enzyme activities in
specified 0.5%. Volumes aspirated and dispensed by the serum. In addition, the analyzer requires little daily mainte-
sample and reagent II pipetting systems differed significantly nance, needs no external drain or plumbing, and can be set
from preset values, especially at lower settings; the reagent I to various assay temperatures (20 #{176}C
to 40#{176}C,
in increments
system was satisfactory at all volumes tested. Minimal daily of 1 #{176}C).
maintenance and an external data-reduction system make Our objectives were to assess the suitability of the instru-
the EPOS a practical alternative to other bench-top chemistry ment for performing routine clinical tests (primarily enzyme
analyzers. determinations in our laboratory, but many other analytea
can be handled by the zros) and to determine its analytical
Addftlonal Keyphraee: discrete analysis performance in user-defined applications such as analyses
for glucose.
The Eppendorf Patient Oriented System (Eros; EM Diag-
nostic Systems, Inc., Gibbstown, NJ 08027) is a bench-top, Materials and Methods
automated discrete analyzer with a 100-test capacity. Tests
Mode of Operation of EROS
are performed in one of two operatiohal modes: time-
sharing, used for most routine analyses, and serial, used Moving counterclockwise, cuvets are passed into and out
primarily for reactions with fast initial and decay rates. The of the light path (rotor position 20). Reagent one, the
zroe is filly computer-controlled (Intel 8085 microprocessor) sample, and reagent two are automatically added to the
and has a maximum throughput of 300 tests per hour. The cuvets at rotor positions 16, 15, and 12 through 1, respec-
analyzer performs end-point, kinetic, enzymoimmuno-, and tively, the exact position used for adding reagent two
turbidimetric assays, with or without corrections for sample depending on the requirements of the assay reaction. Cuvet
or reagent blanks. contents are mixed by pulses of air pressure generated by a
The ES has several special technical features. The use of tiny speaker vibrating at radiowave frequency; the ifiled
three separate pipetting probes for independent processing cuvet is then advanced to rotor position 20, where the
of sample and two reagents minimizes errors from pipetting reaction progress is monitored photometrically. Cuvets are
cross-contamination. The spectrophotometric unit, a mer- automatically washed, rinsed, and air-dried at rotor posi-
tions 19, 18, and 17. To perform a given analysis in the arcs,
the user selects a pre-coded method from among the 30 that
are available for the basic analyzer or from the 100 that are
Department of Clinical Biochemistry, University Hospital (Uni-
versity of Western Ontario), P.O. Box 5339, Postal Stn. A., London, available when the computer terminal is used. After sam-
Ontario, Canada N6A 5A5. ples, quality control sera and (or) standards, and reagents
‘Address correspondence to this author. are placed in appropriate positions in the analyzer, the user
Received May 20, 1985; accepted September 16, 1985. pushes the “start” button.

CLINICAL CHEMISTRY, Vol. 32, No. 1, 1986 165