Clinica Chimica Acta, 158 (1986) 237-244 237

Elsevier

CCA 03539

Errors in measurements of CO,

with the use of drying agents
Marinos Elia, Terence McDonald and Anthony Crisp
Dunn Clinical Nutrition Centre, 100 Tennis Court Rod Cambridge, CB2 I QL (UK)

(Received January 25th, 1986; revision received April 29th, 1986; accepted after revision May 15th. 1986)

Key worak Drying agents; Indirect calorimetry

To assess the errors in measurements of CO, concentration that arise from the
widespread use of drying agents in clinical and physiological studies, continuous
measurements of CO2 concentration were made with an infra-red analyser before
and after standard gas mixtures were passed through a tube containing one of
several drying agents: calcium chloride; calcium sulphate (Drierite); alumina; silica
gel; and magnesium perchlorate.
The response time of the infra-red analyser was independent of the concentration
of CO, used (O-l%), but it was related to flow rate, dead space and the amount and
type of drying agent used. At a flow rate of 1 litre/min, alumina (145 g anhydrous
weight) trapped virtually all the CO2 (0.8% in air) passing over it; silica gel (145 g
anhydrous weight, mesh 6-20) adsorbed approximately 2 mm01 CO, (38% of total)
over a period of 15 min; calcium sulphate (Drierite) absorbed a variable amount,
depending on the batch; and calcium chloride adsorbed virtually no CO,.
Increasing hydration of the drying agents reduced their capacity to adsorb CO,.
Nitrogen was found to elute CO, from the drying agents at about the same rate as it
had been adsorbed.
It is concluded that awareness of these adsorption/elution phenomena by drying
agents should prevent errors from being made in the calibration of CO, analysers,
in the analysis of CO,, and in the measurement of specific activity or isotopic
enrichment of CO,.

Introduction

A number of CO, breath tests are used clinically for the diagnosis of various
gastroenterological disorders such as malabsorption, the blind loop syndrome and
liver disease [1,2]. Also, measurements of CO* production rate and CO2 specific
activity (or isotopic enrichment) are important in studies of fundamental research

0009-8981/86,/$03.50 0 1986 Elsevier Science Publishers B.V. (Biomedical Division)

238

such as those concerned with protein [3], carbohydrate [4], fat metabolism [5] and
energy expenditure (,Livesey G, Elia M, unpublished data). These measurements are
usually made after the test gases are dried by one of three procedures: cryogenic
precipitation of water vapour [4], dessication by molecular sieving [6,7]; and
dessication ‘by reaction with water [1,8]. The most commonly used processes
probably involve chemical agents which either act as molecular sieves (e.g. silica gel)
[6] or react with water (e.g. calcium chloride and calcium sulphate - Drierite) [l,g].
However, little attention appears to have been given to the possibility that the
drying agents, apart from removing water vapour, may also modify the composition
of the gas mixture. Thus, in several studies, such as those concerned with indirect
calorimetry, few or no details are given about the quantity or type of drying agent
used [9-141. Therefore, the present study was undertaken to assess the potential
errors that may arise from the use of different quantities, batches and types of
chemical drying agents, and to suggest ways in which such errors may be avoided.

Materials and methods

The effect of CO, adsorption by drying agents was assessed by an infra-red CO,
analyser which was incorporated into an indirect calorimeter. Details of the indirect
calorimeter used in this study are given elsewhere [15]. Essentially, the calorimeter is
a microcomputerised open-circuit calorimeter which uses an infra-red CO, analyser
type SSl with a measurement range O-l% (Analytical Development Company Ltd.,
Hoddesdon, Herts., UK) and a paramagnetic oxygen analyser OA 540 (Servomex -
Taylor Instruments Analytics Ltd., Crowborough, Sussex, UK). Flow of gas through
the drying agents and analysers is controlled over a range of 0.3-1.2 litres to an
accuracy of l-6%. This accuracy was checked with a wet-type gas meter (Alexander
Wright & Co. (Westminster) Ltd., UK) which is calibrated to an accuracy of 0.5%.
The standard gases used in the study were a gas mixture of 0.8% CO, in air
(chemically analysed by the Lloyd-Haldane apparatus) and nitrogen (British Oxygen
Company, UK). The concentration of CO2 could be altered by mixing CO, from a
cylinder with a variable amount of room air. The gases could either pass to the
analyser directly (approximate dead space, 30 ml), or indirectly by being allowed to
first pass through a conical mixing flask (approximately 220 ml) and a drying tube
of diameter 2.55 cm and volume of about 200 ml (total dead space, about 450 ml).
Changes in CO, and 0, concentrations were measured by the analysers and
electronically transmitted at programmable time intervals to a visual display unit.
The ingoing gas could be changed automatically from one cylinder to another by
solenoid valves according to a pre-set programme.
The analyser response time was found to be independent of environmental
temperature over the experimental measuring range of 20-27°C. The effect of flow
on response time was assessed by altering the flow rate between 0.3 and 1.2
litres/min. The standard flow rate for all other experiments was 1 litre/min. The
effect of varying the dead space between the source of gas and the analyser was
assessed by substituting the drying tube with a variable length of plastic tube. The
diameter and volume of the tube were measured as described previously [16].
 239

The following drying agents were tested: calcium sulphate, calcium chloride,
silica gel, alumina and magnesium perchlorate. Calcium sulphate, mesh 8 without
indicator was obtained from Fisons (Bishop Meadow Road, Loughborough, Leics.,
LEll ORG, UK), and calcium sulphate, mesh 8 with indicator (Drierite) was
obtained from British Drug Houses (Parham Drive, Boyatt Wood Ind. Est., East-
leigh, Hampshire, SO5 4NU, UK) and Aldrich Chemical Co. Ltd. (The Old
Brickyard - New Road, Gillingham, Dorset, SP8 4.TL, UK) both of which are
supplied by W.A. Hammond Drierite Company (Xenia, OH, USA). The batch-to-
batch variation in the size of Drierite crystals was assessed by comparing the
weights of approximately 50-100 crystals from each source. Calcium chloride, mesh
8 was obtained from Aldrich, and meshes 3-8, 8-16, 12-24 from British Drug
Houses. Silica gel, mesh 6-20 was obtained from British Drug Houses, alumina,
mesh 8-0.25 inch from Sigma Chemical Co. Ltd. (Fancy Road, Poole, Dorset, BH17
7NH, UK) and magnesium perchlorate from Aldrich.
The extent of ‘hydration’ of different drying agents was assessed by heating them
at 250°C until a constant weight was reached.
The extent of sorption of water vapour by different drying agents was assessed in
four ways: (a) observing the colour change of self-indicating dessicants (Drierite,
blue to pink; and silica gel, blue to colourless); (b) mixing those drying agents
without indicators with a few crystals of Drierite which could act as indicators of
the hydration front; (c) observing the change in form, e.g. calcium chloride particles
tend to bind together on hydration; (d) sequential weighing of the drying agents
after hydration with atmospheric water vapour.
Adsorption of CO, by the drying agents was assessed by passing air (1 litre/min)
containing 0.8% CO2 over drying agents that had been previously exposed to
nitrogen to elute-off all adsorbed CO,. Elution of CO, off the drying agents was
assessed by passing nitrogen (1 litre/min) over drying agents that had been
previously equilibrated with 0.8% COz. These procedures were carried out with the
drying agents in the anhydrous form as well as in varying degrees of hydration, The
extent of hydration was controlled by passing atmospheric air at a rate of 1
htre/min over the drying agent, until the desired increase in weight was achieved.
Hydrated Drierite and silica gel were repeatedly heated at 250°C to remove
water from them to allow them to be reused. The effect of this procedure on
subsequent CO2 adsorption was also assessed.

Results

As expected, the response times of the CO, analyser were proportional to dead
space between the source and the analyser, and inversely related to the flow rate.
The initial degree of ‘hydration’ of drying agents (assessed by change in weight
after heating at 250°C) was found to be small and usually less than 1%.
All drying agents were found to be effective in removing water vapour from
atmospheric air, except those that consisted of either large particles of calcium
chloride (BDH mesh 3-8 and Aldrich mesh 8) or small particles of calcium chloride,
such as those of mesh size 12-24 (BDH) which could be blown to the side of the
240

tube by a stream of air. Removal of water vapour by Drier&e and silica gel was also
found to be less effective when these agents were hydrated above 7% and 14%
respectively, and when they are extensively and repeatedly heated to convert them
from a hydrous form to an anhydrous form. The water holding capacity of calcium
chloride was high (up to 50% for BDH mesh 8-16).
Gas adsorption by drying agents was observed with respect to CO2 (see below),
but not with oxygen. The slow response time of the oxygen analyser was not

166-

86-

8 *-
a
a+ 40-

20 -

o-

166-

86-

c 66-
5 -
$ 40-
bp

26-

o-
, I I I I I 8. I I I I I I I I I 21 a,
0 2 4 6 6 10 12 14 16 18 26
Time: minutes

Fig. 1. Upper: The effect of passing nitrogen at a flow rate of 1 litre/min over 145&t g anhydrous
weight of various drying agents, on the response of the CO, analyser. Increased adsorption is indicated
by a slower response. The drying agents were previously exposed and equilibrated with 0.8% CO2 in air.
SiO, = silica gel mesh 6-20, BDH; CaCl, =calcium chloride mesh 8-16, BDH; CaSO., = calcium
sulphate or Drier&. The shaded area represents the variable effect produced by using different batches
of Drierite. Lower: The effect of passing 0.8% CO2 in air at a flow rate of 1 litre/mm over the same
drying agents and alumina (AlaOs) (145 f 2 g). The drying agents were previously exposed to nitrogen to
elute-off any adsorbed COr. The response of the CO2 analyser when no drying agent was used (upper
and lower figures) was virtually the same as when calcium chloride was used (for details see text).
 241

detectably affected by passing gases over drying agents, except as a secondary result
of CO2 adsorption which effectively produces a small increase in oxygen concentra-
tion. The effect of passing a CO,-containing gas mixture over various drying agents
(calcium chloride, alumina, Drierite, magnesium perchlorate) that had been previ-
ously exposed to nitrogen, was virtually identical and opposite to the effect
produced by passing nitrogen over tbe same drying agents that had been previously
exposed to a CO,-containing gas mixture (Fig. 1). This suggests that the phenome-
non is a reversible one which does not involve a permanent chemical change in the
drying agent. The extent of CO, adsorption by different drying agents (145 g
anhydrous weight) when a 0.8% CO, gas mixture is passed over them at a flow rate
of 1 litre/min is illustrated in Fig. 1. Alumina adsorbed virtually all the CO,
passing over it, whilst silica gel adsorbed 38% of the CO, over 15 min. There was
little or no adsorption of CO2 by calcium chloride. The amount adsorbed by
magnesium perchlorate (not shown in Fig. 1) was intermediate between that of
calcium chloride (mesh 8-16) and calcium sulphate (shown in Fig. 1). The variation
in CO, adsorption by different batches (n = 5) of silica gel and calcium chloride was
generally small (the 99% response time of the CO, analyser varied by less than
It 10% in the mean value). However, with Drierite there was more than a 3-fold
batch-to-batch variation in CO2 adsorption, as indicated by the shaded area in Fig.
1. This variation was not due to differences in the size of crystals from different
batches since their mean weight varied by less than 10%. The variation was also not

Timer minute8

Fig. 2. The effect of increasingthe quantityof wg agents in anhydrousform, on CO, adsorptionfrom

a gas mixture of 0.8% CO1 in air flowingover the drying agents at a rate of 1 litre/min. The results are
expressedin terms of the 99% responsetime of the CO1 analyser - the greater the extent of adsorption
the slowerthe response.The drying agents were previouslyexposed to nitrogen to elute-offany adsorbed
CO,. Viiually identical and opposite changes were observedwhen nitrogen was passed over the same
drying agents which had been pre-equilibratedwith 0.8% CO, (data not shown - see text for details).
SiOa= siiica gel mesh 6-10, BDH, CaCl, -calcium chloride mesh 8-16, BDfi Ca!SO.,-calcium
sulphate or Drier&e. The shaded area represents the variable effect produced by different batches of
Drierite.
242

due to contamination of calcium sulphate with variable quantities of basic com-

pounds such as calcium oxide, because the addition to water of crystals from
different sources produced a small and reproducible change in pH (final pH
7.33-7.43).
The effect of increasing the amount of drying agent on the 99% response time of
the CO, analyser is indicated in Fig, 2. The shaded area again represents the
variation produced by different batches of Drier&e. When no drying agent was
included in the drying tube, the 99% response time was about 27 s.
The adsorption of CO, by various drying agents was considerably reduced as the
drying agent became hydrated. This was observed with silica gel and Drierite (Fig.

10 20 30 40 50 60 70 80 90 loll
Time:seconds

w-
$
6
a w-
3
K

ts
0 2 4 6 8 10 12 14 16 18 20
Time : minutes
Fig. 3. The effect of varying the degree of hydration of drying agents on adsorption of CO1 from a gas
mixture of 0.8% COz in air flowing at a rate of 1 litre/min. The results are expressed in terms of the 99%
response time of the CO, analyser - the greater the extent of adsorption the slower the response. The
drying agents were previously exposed to nitrogen to elute-off any adsorbed CO*. The degree of
hydration of the drying agents is indicated in relation to their anhydrous weight (188 g of anhydrous
Drierite (CaSO,) and 121 g of silica gel, mesh 6-20 (BDH).
 243

3) and to a lesser extent with calcium chloride. (Ma~esium perchlorate and

alumina were not tested for this.) The response time of the COr analyser was found
to be slightly faster when 0.8% CO, was passed over hydrated calcium chloride than
when no drying agent was present in the ‘drying’ tube, presumably because of the
reduction in dead space. The pattern of elution of CO, with nitrogen was again
virtually identical and opposite to adsorption of CO,.
Different concentrations of CO2 in air ranging from 0.3 to 0.8% CO, had no
significant effect on the 99% response times of the CO, analyser.

Discussion

This study emphasises the importance of CO, adso~tion/elution phenomena by

drying agents and the potential pitfalls associated with their use in a wide variety of
clinical and physiological studies. Caution is advised in three particular situations.
Firstly, in isotopic experiments involving labelled CO, [l-5], erroneous measure-
ments may arise if the expired air sample is dried with an agent which already
contains significant quantities of adsorbed CO* with a different specific activity or
isotopic en~~~ent as the test gas. For example, 145 g of anhydrous silica gel
adsorbs about 2 mmol of COr when exposed to an 0.8% CO, gas mixture (Fig. l),
and a proportionately greater amount when the concentration of CO, is higher.
Secondly, errors in CO, measurements may arise if CO, analysers are wrongly
calibrated. The time allocated to calibration should not be fixed since the time
required to achieve a steady-state CO2 concentration may vary considerably depend-
ing on such factors as flow rate, dead space, and use of different batches, types and
quantities of drying agents, and on the extent of their hydration.
Thirdly, in studies involving continuous measurements of CO, (e.g. indirect
calorimetry) potential errors due to CO, adsorption/elution phenomena may arise
at the beginning of studies, or when there are changes in CO2 concentrations, as
occur during exercise. In studies carried out in large indirect calorimeters, attempts
to predict actual or ‘up-to-date’ changes in CO, production rates and respiratory
quotients [17,18], will be erroneous and less acute than the real ones, if the gas is
dried with an agent that adsorbs substantial amounts of CO,. Long-term errors may
also potentially arise if test gases are sampled intermittently from different sources
(e.g. animal cages [19]) and passed over the same drying agent.
In all the above situations, errors may be avoided or minim&d by (a) not using
drying agents when possible, e.g. calibration of CO* analysers may be performed by
introducing the standard CO, gas mixture, which is usually devoid of water vapour,
directly into the analyser; (b) using a small quantity of drying agent or one that is
pre-equilibrated with the test gas mixture (e.g. in CO, breath tests); and (c) using an
appropriate drying agent which exhibits few or no CO, adsorption/elution
phenomena. It is suggested that an agent such as calcium chloride mesh 8-16
(BDH) is an appropriate agent for a variety of situations. It shows little or no
adsorption for CO, (cf. alumina, silica gel) and there is little variation in the
behaviour of samples obtained from different batches (cf. Drierite). It is also safe to
use (cf. magnesium perchlorate which may explode when in contact with organic
244

solvents) and has a high water holding capacity (cf. calcium sulphate - Drierite).
However, calcium chloride particles tend to clump together on hydration and this
may offer some increased resistance to the flow of gas and also make regeneration
of the anhydrous form impractical and uneconomic. The obvious practical solution
to this is to discard calcium chloride after use. The agent is cheap and soluble and
therefore easily washed away from its container with water.

Acknowledgement

We wish to thank Mr. P.R. Murgatroyd for helpful discussions.

References

1 Neuman A. Progress report: breath analysis tests in gastroenterology. Gut 1974; 15: 308-323.
2 Hofman AF, Lauterburg BH. Breath tests with isotopes of carbon: progress and potential. J Lab Clin
Med 1977; 90: 405-411.
3 Waterlow JC, Garlick PJ, Millward DJ. Protein turnover in mammalian tissues and the whole body.
Amsterdam, New York, Oxford: North-Holland Publishing Company, 1978.
4 Acheson KJ, Thelin A, Ravussin E, Amaud MJ, Jequier E. Contribution of 500 g naturally labelled
13C dextrin maltose to total carbohydrate utilization and the effect of the antecedent diet in man. Am
J Clin Nutr 1985; 41: 881-890.
5 Nordenstrom J. Utilization of exogenous and endogenous lipids for energy production during total
parenteral nutrition. Acta Clin Stand 1982; Suppl. 510.
6 Golden MHN, Waterlow JC. Total protein synthesis is elderly people: a comparison of results with
[t5N]glycine and [t4C]leucine. Clin Sci 1977; 53: 277-288.
7 Spencer JL, Zikria BA, Kinney JM, Broell JR, Michailoff TM, Lee AB. A system for continuous
measurements of gas exchange. J Appl Physiol 1972; 33: 523-528.
8 Weissman C, Damask MC, Askanazi J, Rosenbaum SH, Kinney JM. Evaluation of a non-invasive
method for the measurement of metabolic rate in humans. Clin Sci 1985; 69: 135-141.
9 Stoker JB, Kappagoda CT, Thenahadu PN, Linden RJ. A simple method for measurement of oxygen
consumption. J Appl Physiol 1973; 35: 748-750.
10 Kinney JM, Morgan AP, Doninques FJ, Gildner KJ. A method for measurement of gas exchange and
expired radioactivity in acutely continuous ill patients. Metabolism 1964; 13: 205-211.
11 Kappagoda CT, Stoker JB, Linden J. A method of continuous measurement of oxygen consumption.
J Appl Physiol 1974; 37: 604-607.
12 Wilmore JH, Costill DC. Semi-automated systems approach to the assessment of oxygen uptake
during exercise. J Appl Physiol 1974; 36: 1618-1620.
13 Wilmore JH, Davis JA, Norton AC. An automated system for assessing metabolic and respiratory
function during exercise. J Appl Physiol 1976; 40: 619-624.
14 Yeh TF, Admani M, Leu S-T, Tan M, Plides RS. Simple daily technique to calculate total and
interior changes of daily 0, consumption and COs production in premature infants. Crit Care Med
1982; 10: 534-538.
15 Crisp JA, Murgatroyd PR. In: van Es AJH, ed. EURO-NUT Report 5. Agricultural University,
Wageningen, The Netherlands. 1985: 44-45.
16 Elia M, Crazier C, Martin S, Neale G. Flow and aspiration of feeds through nasogastric tubes. Clin
Nutr 1984; 2: 159-166.
17 Brown D, Cole TJ, Dauncey MJ, Marrs N, Murgatroyd PR. Analysis of gaseous exchange in open
circuit indirect calorimetry. Med Biol Engin Comput 1984; July: 333-338.
18 Cole TJ, Murgatroyd PR, Brown D, Dauncey MJ. In: Van Es AJH, ed. EURO-NUT Report 5. 1985:
37-40.
19 Tobin G, Hervey GR. In: van Es AJH, ed. EURO-NUT Report 5. 1985: 32-36.