Infrared CO2 analyzer error:
an effect of background gas (N2 and O2 )
R. ARIELI, O. ERTRACHT, AND Y. DASKALOVIC
Israel Naval Medical Institute, Israel Defense Forces Medical Corps, Haifa 31080, Israel
carbon dioxide calibration; hyperoxia; alveolar-arterial
difference
(CO2 ) is a very important gas in human
physiology and is extensively monitored in medical
treatment, physiological studies, and various environmental situations. Inspired gas enriched with O2 is
used in many fields of human activity. For example,
combinations of various levels of O2 and CO2 are
common in diving and in medical treatment. At our
institute, where we practice hyperbaric O2 therapy and
study the physiology of closed-circuit diving, combinations of various levels of O2 and CO2 are common.
Infrared CO2 analyzers are the most widely used. In
different situations, we came up with conflicting results
by using both our mass spectrometer and infrared CO2
analyzer after calibration with certified commercial
calibrating mixtures. Discrepancies in the readings of
the CO2 concentration in a gas mixture and unacceptable respiratory ratios are two examples. We therefore
set out to study this conflict by using the Wösthoff
precision pumps (H. Wösthoff, Bochum, Germany) to
produce various precise concentrations of CO2 with
either N2 or O2 as the background gas. These precision
pumps are known for their accuracy, and an analysis of
a gas mixture prepared by the precision pumps that we
performed by using the Micro-Scholander gas analyzer
was proven accurate to within 0.01–0.02%. Infrared
CO2 analyzers’ instruction manuals do not indicate any
effect of N2 or O2 on CO2 sensing.
CARBON DIOXIDE
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
http://www.jap.org
METHODS
Three infrared CO2 analyzers and a mass spectrometer
were used.
Ametek CD-3A (Ametek, Thermox Instruments Division,
Pittsburgh, PA). The instruction manual for the Ametek
CD-3A (1) suggests zero calibration with either air or N2 and
span calibration with a CO2-containing mixture (CO2 .4%
and up to 15%). There is no reference to interference by O2
or N2.
Servomex 1440 (Crowborough, East Sussex, UK). The instruction manual for the Servomex PA 404 CO2 analyzer
suggests ‘‘narrow-band interference filters and a solid-state
detector provide a measurement which is inherently less
affected by cross-sensitivity’’ (10). The manual does suggest
some background effect of gases on the O2 sensor, but there is
no mention of the CO2 sensor. It is suggested that ‘‘these
effects can be compensated by either using the background
gas as a zero or by offsetting the N2 zero point by the amount
of error induced by the background gas’’ (10). There is no
mention of cross sensitivity of N2 or O2 on CO2 sensing.
Dräger Multiwarn P CO2 (Lübeck, Germany). The instruction manual for the Dräger Multiwarn P CO2 (6) suggests zero
calibration with pure N2 and span calibration with a CO2containing mixture. They suggest calibrating with a ‘‘concentration corresponding to the typical values which are to be
measured’’ (6). Nothing is mentioned regarding cross sensitivity of N2 or O2.
The mass spectrometer used was the CaSE 9000T/BG
(Biggin Hill, UK).
We used Wösthoff precision pumps to produce various
concentrations of CO2 with either pure O2 or pure N2. The CO2
analyzer was calibrated with a 5% mixture coming from the
pumps. The outflow (550 ml/min) from the precision pumps
passed through a rubber glove (to smooth out the pressure
fluctuations), from which it went on to the analyzer sensor via
a hole cut in one of the fingers. The values read for various
concentrations were recorded. To rule out asymmetry between the two pumps, one was used for CO2 and the other for
the background gas, and after a few readings they were
switched. The results of changing the pumps when CO2 was
read by using the Ametek CD-3A proved symmetry. In the
test, we calibrated the CO2 analyzer with 0 and 5% CO2
produced by the precision pumps by using pure N2 as the
background gas. We then produced, in ascending and descending order, various concentrations of CO2 with the precision
pumps, reading the CO2 concentration on the analyzer. We
then changed the background gas to O2 and read the CO2
concentrations for various combinations of the precision
pumps’ input. The background gas was then changed back to
N2, and we read a few more concentrations to make sure that
no drift had occurred in the interim. After the completion of
this series, we changed the background gas. We calibrated the
analyzer with O2 as the background gas and proceeded with
the previous protocol, interchanging O2 and N2.
RESULTS
The CO2 readings obtained from the CaSE mass
spectrometer (calibrated with a certified commercial
8750-7587/99 $5.00 Copyright r 1999 the American Physiological Society
647
Downloaded from http://jap.physiology.org/ by 10.220.33.1 on June 14, 2017
Arieli, R., O. Ertracht, and Y. Daskalovic. Infrared CO2
analyzer error: an effect of background gas (N2 and O2 ). J.
Appl. Physiol. 86(2): 647–650, 1999.—Three infrared CO2
analyzers were tested for the effect of background gases: the
Ametek CD-3A (Ametek, Thermox Instruments Division,
Pittsburgh, PA), the Dräger Multiwarn P CO2 (Dräger, Lübeck, Germany), and the Servomex 1440 (Servomex, Crowborough, East Sussex, UK). Various CO2 concentrations were
prepared with Wösthoff precision pumps (H. Wösthoff, Bochum, Germany). Calibration with a different background
gas (O2 or N2 ) caused a similar but systematic error in the
CO2 readings of all three analyzers. When the CO2 analyzers
were calibrated with N2 as the background gas, the CO2
reading in an O2-enriched atmosphere was 8% lower than the
true value. Conversely, calibration with O2 as the background
gas resulted in a 10% overestimation of CO2 levels when N2
was the background gas. This error may be important in a few
fields of respiratory physiology.
648
INFRARED CO2 ANALYZERS
Fig. 1. CaSE mass spectrometer readings of CO2 concentrations.
Difference between CO2 concentration readings and precision pumps’
input is shown as a function of CO2 input from H. Wösthoff precision
pumps. Both pure O2 and pure N2 were used as background gas for
pure CO2 (s and j, respectively).
of CO2 concentrations for the other background gas (the
gas that had not been used for calibration) was similar
to that seen in the Ametek CD-3A readings: underestimation when the background gas was O2 (Fig. 2E) and
overestimation when the background gas was N2 (Fig.
2F). The deviations were corrected by subtracting the
deviation for the calibration background gas from the
deviation of the noncalibration background gas (q in
Fig. 2, E and F). The results gave a linear relationship
for the corrected deviation as slope 5 20.085, r2 5 0.97,
and slope 5 0.09, r2 5 0.78 in Fig. 2, E and F,
respectively. The nonlinear concentration readings of
the Dräger analyzer may be inferred from their instruction manual, which mentions a possible error of 65%
for the values 0 to the calibration value and 610% from
the calibration value to 1.5 times this value (6). In all
the analyzers, the zero setting was not sufficient on
recalibration with another background gas; sensitivity
had to be calibrated as well.
DISCUSSION
A CO2 infrared analyzer responds differently, depending on whether the background is N2 or O2. Because
most common commercial calibrating tanks contain a
high percentage of N2, CO2 readings in an O2-enriched
atmosphere are underestimated. For the three analyzers we tested (Ametek, Servomex, Dräger), this underestimation was 20.075, 20.085, and 20.080, respectively, for each 1% CO2. Therefore, calibrating an
infrared CO2 analyzer with N2 as the background gas
would result in an 8% underestimation of CO2 levels in
O2 as background. For example, when the analyzer
reading is 4.6% CO2, the true concentration is 5%.
Conversely, calibrating the three analyzers with O2 as
the background gas gave an overestimation of 0.11,
0.09, and 0.10, respectively for each 1% CO2 in N2. Thus
calibrating the infrared CO2 analyzer with O2 as the
background gas will result in a 10% overestimation of
CO2 levels in a N2-rich atmosphere.
There are a few fields in respiratory physiology
where an error in the CO2 reading could seriously affect
the outcome. In hyperbaric physiology, a small change
in CO2 concentration can make a large difference,
because PCO2 is the product of the fraction of CO2 and
pressure. CO2 is an important factor in O2 toxicity (3),
which will be enhanced by an inspired PCO2 as small as
1 kPa (2). CO2 is an important respiratory drive in
hyperbaric conditions, and professional divers tend to
hypoventilate and retain CO2 in their tissues (7).
Diving with breathing mixtures containing various
concentrations of O2 (Nitrox or closed-circuit diving)
would affect the CO2 monitored with infrared analyzers. The alveolar-arterial CO2 difference is negligibly
small in the healthy lung and is therefore used to
calculate alveolar PCO2 from arterial PCO2 (5). An
8–10% error would represent a serious flaw in the
computation of alveolar-arterial gas exchange. Various
tests use the switch from an air- to an O2-filled spirometer (4), and the CO2 reading would be affected if an
infrared analyzer were used. In clinical medicine, if the
Downloaded from http://jap.physiology.org/ by 10.220.33.1 on June 14, 2017
mixture, Scott Specialty Gases, Plumsteadville, PA),
with either N2 or O2 as the background gas, are shown
in Fig. 1. In Figs. 1 and 2, we present the difference
between the analyzer readings of CO2 and the precision
pumps’ input, on the y-axis, as a function of the
percentage of CO2 in the precision pumps’ input. It can
be seen that the CO2 readings deviate only slightly
from the input of the precision pumps and are similar
for both background gases. When the Ametek CD-3A
was calibrated by using N2 as the background gas,
other CO2 readings showed small deviations from the
expected concentrations (Fig. 2A). However, CO2 readings deviated to a greater degree as a function of input
CO2 concentration when the background gas was O2.
The slope of the decrease was 20.075 (%error/%CO2,
r2 5 0.99). There was no difference between new CO2
readings and those from the initial set, when the
background gas was changed back to N2. Calibration of
the Ametek CD-3A with O2 as the background gas gave
the opposite results (Fig. 2B). The different CO2 readings with O2 as the background gas yielded small
deviations from the expected values, but replacing the
O2 with N2 produced positive deviations that are linearly related to the input CO2 concentration (slope 5
0.11, r2 5 0.98). Results similar to those obtained with
the Ametek CD-3A were derived by using the Servomex
1440 (Fig. 2, C and D). The slope for the deviation of
CO2 concentrations in O2 from the input when the
calibration background gas was N2 was 20.080 (r2 5
0.99), and, in N2 when the calibration background gas
was O2, was 0.10 (r2 5 0.99). Similar results were
obtained with the Dräger Multiwarn P analyzer (Fig. 2,
E and F). When the background gas was the same as
had been used for calibration, the CO2 readings yielded
a reclining s-shaped response that crossed the zero at
the calibration concentration. However, the deviation
INFRARED CO2 ANALYZERS
649
analyzer is calibrated with N2 as the background gas,
alveolar PCO2 may be underestimated in patients
breathing an enriched O2 mixture. In artificial respiration, for example, when the ventilator is set to give 5%
CO2 in the alveolar gas, the true value may be close to
4.6%.
Lauber et al. (8) tested the accuracy of various
infrared CO2 analyzers as affected by an array of
background gases and conditions. They concluded that
‘‘all tested analyzers were found to be safe for clinical
use,’’ allowing for a 12% error. However, the error
established by the present study, which is probably
related to the collision-broadening effect of O2 (9),
should be considered in scientific research. Presently,
only a gas mixture with the same O2/N2 ratio as that
used in the planned experiment can be used as a
calibrating mixture in infrared CO2 analyzers.
The authors thank R. Lincoln for skillful editing.
The opinions and assertions contained herein are the private ones
of the authors and are not to be construed as official or as reflecting
the views of the Israel Naval Medical Institute.
Address for reprint requests: R. Arieli, Israel Naval Medical
Institute, PO Box 8040, Haifa 31080, Israel.
Received 13 March 1998; accepted in final form 4 September 1998.
REFERENCES
1. Ametek/Thermox. Ametek CD-3A. Pittsburg, PA: Ametek/
Thermox, 1986.
2. Arieli, R. Hazards of CNS oxygen toxicity and the effects of
metabolic rate and CO2 production (Abstract). Annu. Meet. Israel
Physiol. Soc. Maale Hahamisha, Israel, November 1997.
Downloaded from http://jap.physiology.org/ by 10.220.33.1 on June 14, 2017
Fig. 2. Ametek CD-3A (A and B), Servomex 1440 (C and
D), and Dräger Multiwarn P (E and F) readings of CO2
concentrations. Difference between CO2 concentration
readings and precision pumps’ input is shown as a
function of CO2 input from H. Wösthoff precision pumps.
Right: results from calibration with O2 as background
gas; left: results from calibration with N2 as background
gas. j, CO2 readings when N2 was background gas; s,
readings when background gas was O2; q, differences
between errors for N2 and O2 with Dräger Multiwarn P.
650
INFRARED CO2 ANALYZERS
3. Clark, J. M. Oxygen toxicity. In: The Physiology and Medicine of
Diving (4th ed.), edited by P. B. Bennett and D. H. Elliott.
London: Saunders, 1993, p. 121–169.
4. Cunningham, D. J. C., P. A. Robbins, and C. B. Wolff.
Integration of respiratory responses to changes in alveolar
partial pressures of CO2 and O2 and in arterial pH. In: Handbook
of Physiology. The Respiratory System. Control of Breathing.
Bethesda, MD: Am. Physiol. Soc., 1986, sect. 3, vol. II, pt. 2,
chapt. 15, p. 475–528.
5. Dejours, P. Principles of Comparative Respiratory Physiology.
Amsterdam: North-Holland, 1975, p. 102.
6. Dräger. Dräger multiwarn p. Lübeck, Germany: Dräger, 1991.
7. Lanphier, E. H., and E. M. Camporesi. Respiration and
exertion. In: The Physiology and Medicine of Diving (4th ed.),
edited by P. B. Bennett and D. H. Elliott. London: Saunders,
1993, p. 77–120.
8. Lauber, R., B. Seeberger, and A. M. Zbinden. Carbon dioxide
analysers: accuracy, alarm limits and effects of interfering gases.
Can. J. Anaesth. 42: 643–656, 1995.
9. Raemer, D. B., and I. Calalang. Accuracy of end-tidal carbon
dioxide tension analyzers. J. Clin. Monit. 7: 195–208, 1991.
10. Servomex. Servomex 1440. Crowborough, UK: Servomex,
1986.
Downloaded from http://jap.physiology.org/ by 10.220.33.1 on June 14, 2017