British Journal of Anaesthesia 1998; 80: 213–217
Production of nitrogen dioxide in a delivery system for inhalation of
nitric oxide: a new equation for calculation
L. LINDBERG AND G. RYDGREN
Summary
We have evaluated the kinetics of nitrogen dioxide production in a system for inhalation of nitric
oxide. In addition to a small fraction of contamination of nitrogen dioxide in the nitric oxide
stock gas, a considerable part of the total
concentration of nitrogen dioxide is formed
immediately after mixing of nitric oxide and oxygen. This initial build-up of nitrogen dioxide is
followed by a linear, time-dependent increase in
the concentration of nitrogen dioxide. An equation describing the concentration of nitrogen
dioxide in the delivery system is formulated:
[NO2] kA [NO] kB [NO]2 [O2] kC t [NO]2
[O2], where nitrogen dioxide [NO2] and nitric
oxide [NO] concentrations are in parts per
million (ppm), oxygen concentration [O2] is
expressed as a percentage and contact time (t) is
in seconds. The rate constants are kA5.12 103,
kB 1.41 106 and kC 0.86 106. Calculated
nitrogen dioxide values correlated well with
measured concentrations. This new finding of an
initial build-up of nitrogen dioxide has to be
taken into consideration if the conversion of
nitric oxide to nitrogen dioxide is to be calculated and in the safety guidelines for the use of
nitric oxide. (Br. J. Anaesth. 1988; 80: 213–217)
Keywords: gases non-anaesthetic, nitric oxide; gases nonanaesthetic, nitrogen dioxide; ventilation, mechanical; complications, pulmonary hypertension
Inhaled nitric oxide has emerged as a therapeutic tool
in the management of severe respiratory diseases
associated with hypoxaemia and pulmonary
hypertension.1 The main problem during inhalation
of nitric oxide is its reactivity with oxygen to form the
highly toxic nitrogen dioxide.2 Nitric oxide reacts
with water to form nitric and nitrous acids or it may
be absorbed by antioxidants in the lungs, producing
new free radicals that participate in and delay the
occurrence of lung injuries.3 The oxidation of nitric
oxide has been analysed in sealed chambers and has
been found to be dependent on nitric oxide and oxygen concentrations, and contact time of nitric oxide
with oxygen according to the equation4 5:
−d[NO]/dt = k × [NO]2 × [O2 ]
(1)
The rate constant of this equation has been determined for different humidities and temperatures.6
The equation is used to state safety guidelines for the
use of nitric oxide inhalation7 8 and it has also been
claimed that the equation can be used for calculation
of nitrogen dioxide concentration in mechanical ventilation systems.9
The aim of this study was to analyse the formation
of nitrogen dioxide in a delivery system for inhaled
nitric oxide and to formulate a mathematical
description of nitrogen dioxide production in such a
system.
Materials and methods
A Servo Ventilator 300 (Siemens-Elema AB, Solna,
Sweden) which delivers oxygen and air via two gas
modules with rapid Servo-controlled valves was
used. A third similar gas module delivering nitric
oxide to the proximal part of the inspiratory limb was
attached to the ventilator (fig. 1). Flow in the three
gas modules of the ventilator (air, oxygen and nitric
oxide) is synchronized to permit constant concentrations of nitric oxide, independent of oxygen concentration, tidal volume and airway pressure.10
The gas source was a high-pressure gas cylinder
(160 bar) of nitric oxide (999 20 ppm) mixed in
nitrogen (AGA AB, Lidingö, Sweden). Contamination of nitrogen dioxide in the cylinder was
guaranteed to be less than 20 ppm.
As accurate measurement was of the utmost
importance and we wanted to rule out bias from a
single technique of measuring nitric oxide and nitrogen dioxide, we chose to analyse these gases both
side-stream and main-stream at a fixed distance from
the mixing point. Nitric oxide and nitrogen dioxide
were analysed side-stream with a chemiluminescence
analyser (ML 9841A Monitor Laboratories, LS Lear
Siegler; Measurement Controls Corporation, Englewood, CO, USA) and main-stream with electrochemical fuel cells (CiTicels, City Technology Ltd,
City Technology Centre, Portsmouth, UK). Calibration of the chemiluminescence analyser and the electrochemical fuel cells was performed with dry
calibration gases containing nitric oxide (82.2 ppm
2%) and nitrogen dioxide (6.7 ppm 2%) in
nitrogen (AGA AB, Lidingö, Sweden). The chemiluminescence analyser is zeroed automatically. The
electrochemical fuel cells were zeroed to dry air
hourly. The system was flushed with fresh nitric
oxide–nitrogen four times before the measurements
started to eliminate interference from formed and
trapped nitrogen dioxide in the high-pressure tubing.
LARS LINDBERG, MD, PHD, Department of Anaesthesiology and
Intensive Care, University Hospital of Lund, S-221 85 Lund,
Sweden. GÖRAN RYDGREN, PHD, Servotek AB, Arlöv, Sweden.
Accepted for publication: September 10, 1997.
Correspondence to L. L.
214
Figure 1 Schematic diagram of the ventilator with paediatric
silicon tubes inserted for administering and measuring
concentrations of nitric oxide and nitrogen dioxide.
Chemiluminescence analysers use a technique in
which nitric oxide reacts with ozone to form excited
nitrogen dioxide, which produces a luminescence
that is measured spectroscopically. Nitric oxide is
measured directly and nitrogen oxides (NOX) are
measured indirectly after passage through a conversion chamber where nitrogen oxides are converted to
nitric oxide before measurement. The concentration
of nitric oxide produced after the gas has passed this
conversion chamber minus the nitric oxide concentration that was directly measured yields the nitrogen
dioxide concentration in the sampled gas. Oxygen,
carbon dioxide and water vapour can absorb the
luminescence, limit the luminescent energy produced and thereby cause underestimation of the
actual concentration of nitric oxide and nitrogen
dioxide. This process has been referred to as
“quenching” and may be more obvious in the measurement of nitrogen dioxide than of nitric oxide.11
The chemiluminescence analyser was calibrated with
nitric oxide–nitrogen dioxide in nitrogen. Measurements were made in gas mixtures of oxygen which
introduces quenching and small errors in the
measured concentrations of nitric oxide and nitrogen
oxides. In order to equate the two measuring
techniques, nitric oxide and nitrogen dioxide were
measured with the chemiluminescence analyser and
electrochemical fuel cells at a standardized point and
the measured concentrations were plotted against
each other. A linear compensation was made for the
difference between the measured concentrations.
Paediatric silicon tubes (diameter 0.009 m) of
length 0.9 or 1.47 m were connected to the
ventilator. Minute volume ventilation was set at 1, 4,
and 7 litre min1 and nitric oxide concentration to 10,
20, 40, 60, 80 and 100 ppm at four different FIO2 values (0.21, 0.4, 0.7 and 0.9). To avoid contamination
with room air, we added a 0.6-m tube after the monitors, and the system was open to atmospheric air to
avoid pressure dependency (fig. 1).
Contact time of nitric oxide with oxygen was
calculated by dividing the volume of the tube system
with mean gas flow. Nitrogen dioxide concentrations
were plotted against contact time of nitric oxide and
oxygen in the inspiratory tube at nitric oxide 10, 50,
and 100 ppm and an FIO2 of 0.9.
The oxidation of nitric oxide to nitrogen dioxide
and the ultimate nitrogen dioxide concentration in
the delivery system was analysed and described by
the following formula:
British Journal of Anaesthesia
[NO 2 ] = k A × [NO] + kB × [NO]2 × [O 2 ] + kC × t ×
[NO]2 × [O 2 ]
(2)
where nitrogen dioxide [NO2] and nitric oxide [NO]
concentrations are in parts per million (ppm), oxygen
concentration [O2] is expressed as a percentages and
contact time (t) is in seconds. kA, kB and kC are rate
constants. The rate constants were determined from
210 different combinations of nitric oxide and
oxygen concentrations and contact times by curve
fitting analysis. A linear regression analysis was
performed between calculated concentrations of
nitrogen dioxide and measured values.
The rate constant in equation (1) was also
determined from nitric oxide and oxygen concentrations and contact times by curve fitting analysis. A
linear regression analysis was performed between
calculated and measured concentrations of nitrogen
dioxide. The concentration of oxygen was converted
to ppm and time to minutes to agree with previously
published rate constants.9
STATISTICAL ANALYSIS
Curve fitting analysis and linear regression were performed in Statistica 5.0 for Windows (StatSoft, Inc.
Tulsa, OK, USA). Difference between methods was
assessed by calculating bias (mean difference between two values) and precision (SD of the mean differences) based on the method proposed by Bland
and Altman.12
Results
The concentration of nitric oxide measured by the
chemiluminescence analyser was underestimated by
3.5% and the concentration of nitrogen oxides
(NOX) by 3.4% compared with the electrochemical
fuel cells. The measured concentrations of nitric
oxide and nitrogen oxides were linearly corrected to
compensate for these underestimations. After this
correction for oxygen-dependent quenching in the
chemiluminescence analyser, concentrations of
nitric oxide and nitrogen dioxide measured by the
Figure 2 Difference in nitrogen dioxide (NO2) concentrations
measured by the chemiluminescence analyser after correction for
quenching and the electrochemical fuel cell, plotted against the
mean of the two methods. The limits of agreement ( 0.18 ppm
( 2SD)) are small enough to show that the measurements have
good accuracy.
Nitrogen dioxide production
Figure 3 Nitrogen dioxide (NO2) concentrations measured by
the electrochemical fuel cell at different contact times between
nitric oxide and oxygen, at nitric oxide concentrations of 10, 50
and 100 ppm and an FI O2 of 0.9. Note the intercept at time zero.
215
The concentration of nitrogen dioxide increased
linearly during the contact time of 0.5–3.5 s. At an
FIO2 of 0.9, nitrogen dioxide was produced at an
approximate rate of 0.045, 0.21 and 0.8 ppm s1 with
nitric oxide 10, 50 and 100 ppm, respectively (fig. 3).
Calculated initial concentrations of nitrogen dioxide
found by extrapolation of measured concentrations
of nitrogen dioxide in figure 3 to contact time zero
were 0.031, 0.51 and 1.75 ppm with nitric oxide 10,
50 and 100 ppm, respectively.
The rate constants in equation (2) at 21⬚C dry
gases were: kA5.12103, kB1.41106 and kC
0.86 106.
Calculated nitrogen dioxide concentrations using
equation (2) correlated well with measured nitrogen
dioxide concentrations in the system (r2 0.98) and
showed good agreement when the difference between calculated and measured values was plotted
against measured concentration of nitrogen dioxide.
The limits of agreement were 0.29 and 0.22 ppm
(fig. 4).
The rate constant in equation (1) at 21⬚C dry gases
was 1.8910 6 ppm 1 % 1 s 1 (11.310 9 ppm 2 min 1).
Calculated nitrogen dioxide concentrations using
equation (1) correlated less than equation (2) with
measured concentrations (r2 0.82). A plot of the
difference between calculated and measured values
against measured concentrations of nitrogen dioxide
illustrates the lack of agreement. The limits of agreement were 1.14 and 0.50 ppm (fig. 5).
Nitrogen dioxide concentrations at the Y-piece,
with a 0.9-m long paediatric tube and a flow of 2 litre
min1 at different mixtures of nitric oxide and oxygen,
were calculated according to equation (2) describing
nitrogen dioxide production in the delivery system
(table 1). Nitrogen dioxide concentrations of 1 and 2
ppm are indicated in order to highlight the
concentrations of nitric oxide and oxygen at which
these levels were reached.
Figure 4 Difference between values of nitrogen dioxide (NO2)
calculated by equation (2) [NO2] kA [NO] kB [NO]2 [O2] kC t [NO]2 [O2] and measured, plotted against
measured concentrations of NO2. The limits of agreement (0.29
and 0.22 ppm) are small enough to conclude that the
agreement between calculated and measured values is good.
Discussion
Figure 5 Difference between values of nitrogen dioxide (NO2)
calculated by equation (1) d[NO]/dt k [NO]2 [O2] and
measured, plotted against measured concentrations of NO2. The
limits of agreement (1.14 and 0.50 ppm) and the plotted values
illustrate the lack of agreement between calculated and measured
values.
electrochemical fuel cell and the chemiluminescence
analyser agreed well. Difference in nitrogen dioxide
concentration measured by the chemiluminescence
analyser after correction, and the electrochemical
fuel cell, was within 0.18 ppm (2 SD) (fig. 2).
We have found that the previously published
equation (1)5 describing the oxidation of nitric oxide
in a closed system with constant concentrations of
nitric oxide and oxygen is unsuitable for describing
the conversion of nitric oxide to nitrogen dioxide in
our delivery system for inhalation of nitric oxide. Our
measurements indicate that the oxidation of nitric
oxide takes place in two steps, a rapid build-up reaction followed by a linear, time-dependent reaction. A
new equation describing the oxidation of nitric oxide
to nitrogen dioxide was formulated as a result of the
following.
A small contamination of nitrogen dioxide from
the nitric oxide stock gas tank contributed to the
nitrogen dioxide concentration. As contamination of
nitrogen dioxide is dependent on the stock gas
concentration of nitric oxide, it was assumed to be
proportional to the stock gas concentration of nitric
oxide:
(2a)
k A × [NO]
An initial build-up of nitrogen dioxide became
apparent when the linear, time-dependent increase in
nitrogen dioxide concentration was extrapolated to
the Y axis (fig. 3). The influence of nitric oxide and
oxygen concentrations on this initial build-up of
216
British Journal of Anaesthesia
Table 1 Calculation of nitrogen dioxide concentrations in the delivery system at different nitric oxide and oxygen concentrations, at an
inspiratory flow of 2 litre min1 (neonatal) and with a paediatric tube (diameter 0.009 m) with a length of 0.9 m. Nitrogen dioxide
concentrations of 1 and 2 ppm are indicated in bold type and bold and underlined type, respectively
Oxygen concentration (%)
NO (ppm)
21
35
50
60
70
80
90
10
20
30
40
50
60
70
80
90
100
0.06
0.12
0.20
0.30
0.40
0.52
0.66
0.80
0.96
1.13
0.06
0.14
0.24
0.37
0.51
0.68
0.87
1.07
1.31
1.56
0.06
0.16
0.28
0.44
0.63
0.84
1.09
1.37
1.68
2.02
0.07
0.17
0.31
0.49
0.70
0.95
1.24
1.56
1.92
2.32
0.07
0.18
0.34
0.54
0.78
1.06
1.39
1.76
2.17
2.63
0.07
0.20
0.37
0.59
0.86
1.17
1.54
1.96
2.42
2.93
0.08
0.21
0.39
0.64
0.93
1.28
1.69
2.15
2.67
3.24
nitrogen dioxide, which appears at a contact time of
0–0.5 s, was analysed. Different exponential coefficients for the concentrations of nitric oxide and oxygen were tested in a non-linear multiple regression
analysis. The build-up of nitrogen dioxide was found
to be directly proportional to the concentration of
oxygen and to the square of the nitric oxide concentration, and under this condition was independent of
time:
kB × [NO]2 × [O2 ]
(2b)
The linear and time-dependent increase in the
concentration of nitrogen dioxide is in agreement
with the previously published equation describing
the oxidation of nitric oxide with oxygen:
kC × t × [NO]2 × [O2 ]
(2c)
Theoretically the equation is:
kC × (1 − e −t / τ ) × [NO]2 × [O2 ]
(2d)
where τ time constant and the factor (1-et/τ) shows
that as time advances, the rate of change decreases
towards zero and the final nitrogen dioxide concentration approaches a final limiting value. The graph of
equation (2d) is therefore non-linear. In this case the
non-linear behaviour corresponds to the consumption and limited amount of nitric oxide in the gas mix.
When all nitric oxide is consumed no more nitrogen
dioxide can be formed. For short contact times and
small values of nitrogen dioxide, such as appear in a
delivery system, the graph appears linear.
The initial production of nitrogen dioxide has not
been described in earlier kinetic studies on the reaction between nitric oxide and oxygen.13 However, it
has been noticed that there is an almost instantaneous increase in nitrogen dioxide concentration
during air pollution with nitric oxide.14 One possible
reason why this early formation of nitrogen dioxide
has not been observed is that kinetic analysis has
been carried out in a closed system with a constant
concentration of nitric oxide and oxygen in order to
avoid the instability that appears during the mixing
procedure of the gases. Another reason is that very
fast nitrogen dioxide analysers are not available. We
were able to estimate the immediate build-up phase
by extrapolation.
It seems plausible that the initial production of
nitrogen dioxide occurs at the mixing point as a
result of a high mixing stock gas concentration of
nitric oxide in a fairly high and constant concentration of oxygen. However, it has been noticed that
nitrogen dioxide immediately re-forms after a soda
lime absorber15 and that some surfaces have catalytic
activity and can increase the rate of nitric oxide
oxidation.14 We used silicon tubes that are claimed
not to affect nitric oxide or nitrogen dioxide
concentrations.15 This implies that the mixing
concentrations of nitric oxide and oxygen may affect
nitrogen dioxide production, but that a rapidly
achieved equilibrium in the reaction between nitric
oxide and oxygen may explain the initial increase in
concentration of nitrogen dioxide.
The agreement between measured and calculated
concentrations of nitrogen dioxide also indicates the
accuracy of equation (2) and our findings. Equation
(2) makes it possible to calculate accurately the concentration of nitrogen dioxide at different set
concentrations of nitric oxide and oxygen at different
contact times. Documentation of the expected
concentration of nitrogen dioxide is of value when set
nitric oxide and oxygen concentrations are decided.
In table 1 we have calculated nitrogen dioxide
concentrations at different set concentrations of
nitric oxide and oxygen with settings appropriate to
paediatric ventilation, where a high concentration of
nitrogen dioxide can be expected. The delivered
nitrogen dioxide concentration in the system was
always less than 1 ppm if nitric oxide was less than 50
ppm, independent of oxygen concentration, but
additional formation takes place in the airway of the
patient also, and therefore the alveolar concentration
must be expected to be higher.
Equation (1) has been used to calculate and
recommend safety guidelines for the use of nitric
oxide inhalation. We determined the rate constant of
equation (1) and found expectedly that the equation
did not accurately describe the concentration of
nitrogen dioxide in the delivery system. The rate
constant in equation (1) in the current study is higher
than has been published previously,5–8 but similar to
the rate constant which Nishimura and co-workers
found when nitric oxide was blended with air before
introduction into a ventilator.9 The differences
between the rate constants could be because we used
a dynamic system where the final concentration of
nitrogen dioxide was used for calculation.
Because of the unexpected and previously undescribed finding of a very early high concentration of
nitrogen dioxide it was important that accurate
measurements were performed. In order to rule out
bias from one single technique of measurement we
chose to use a chemiluminescence analyser and electrochemical fuel cells. However, both techniques
have problems and to equate them we had to
compare and analyse measured concentrations.
Nitrogen dioxide production
Electrochemical fuel cells and chemiluminescence
analysers are at present the methods of choice to
measure nitric oxide and nitrogen dioxide. The
chemiluminescence analyser is usually preferred as it
offers a more accurate reading of nitric oxide,
especially during prolonged measurements. The
chemiluminescence technique was designed for environmental monitoring and was not intended to be
used in high oxygen environments, but it is also used
by gas manufacturers to measure nitric oxide
concentration. The effect of quenching was the main
reason for the underestimation of measured concentrations of nitric oxide and nitrogen dioxide by the
chemiluminescence analyser, but after correction
there was good agreement between the chemiluminescence analyser and the electrochemical fuel cells.
Safety guidelines for the use of nitric oxide have
referred to kinetic gas reaction analysis in closed systems of constant concentrations of nitric oxide and
oxygen, and have therefore exclusively noted the
time-dependent oxidation of nitric oxide. An initial
build-up of nitrogen dioxide, which seems to appear
immediately after mixing of nitric oxide and oxygen,
has to be taken into consideration and will result in
higher concentrations of nitrogen dioxide than have
been assumed previously.
Acknowledgment
We thank Sven-Gunnar Olsson MDhc, who provided the
necessary equipment and Lars Nordström MD, PhD, for critical
reading and advice.
References
1. Body SC, Hartigan PM, Shernan SK, Formanek V, Hurford
WE. Nitric oxide: Delivery, measurement, and clinical
application. Journal of Cardiothoracic and Vascular Anesthesia
1995; 9: 748–763.
2. Gaston B, Drazen JM, Loscalzo J, Stamler JS. The biology of
nitrogen oxides in the airways. American Journal of Respiratory
217
and Critical Care Medicine 1994; 149: 538–551.
3. Rasmussen TR, Kjaergaard SK, Tarp U, Pedersen OF.
Delayed effects of NO2 exposure on alveolar permeability and
glutathione peroxidase in healthy humans. American Review of
Respiratory Diseases 1992; 146: 654–659.
4. Treacy JC, Daniels F. Kinetic study of the oxidation of nitric
oxide with oxygen in the pressure range 1 to 20 Mm. Journal
of the American Chemical Society 1955; 77: 2033–2036.
5. Glasson WA, Tuesday CS. The atmospheric thermal oxidation of nitric oxide. Journal of the American Chemical Society
1963; 85: 2901–2904.
6. Miyamoto K, Aida A, Nishimura M, Nakano T, Kawakami Y,
Ohmori Y, Ando S, Ichida T. Effects of humidity and
temperature on nitrogen dioxide formation from nitric oxide.
Lancet 1994; 343: 1099–1100.
7. Bouchet M, Renaudin M, Raveau C, Mercier J, Dehan M,
Zupan V. Safety requirement for use of inhaled nitric oxide in
neonates. Lancet 1993; 341: 968–969.
8. Foubert L, Fleming B, Latimer R, Jonas M, Oduro A, Borland
C, Higenbottam T. Safety guidelines for use of nitric oxide.
Lancet 1992; 339: 1615–1616.
9. Nishimura M, Hess D, Kacmarek RM, Ritz R, Hurford WE.
Nitrogen dioxide production during mechanical ventilation
with nitric oxide in adults. Effects of ventilator internal
volume, air versus nitrogen dilution, minute ventilation, and
inspired oxygen fraction. Anesthesiology 1995; 82: 1246–1254.
10. Lindberg L, Rydgren G, Larsson A, Olsson S, Nordström L.
A delivery system for inhalation of nitric oxide evaluated with
chemiluminescence, electrochemical fuel cells, and capnography. Critical Care Medicine 1997; 25: 190–196.
11. Etches PC, Harris ML, McKinley R, Finer NN. Clinical
monitoring of inhaled nitric oxide: comparison of chemiluminescent and electrochemical sensors. Biomedical Instrumentation and Technology 1995; 29: 134–140.
12. Bland JM, Altman DG. Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet 1986; 86: 307–310.
13. Greig JD, Hall PG. Infra-red spectrophotometric study of the
oxidation of nitric oxide. Transactions of the Faraday Society
1966; 63: 652–659.
14. Lindqvist O, Ljungström E, Svensson R. Low temperature
thermal oxidation of nitric oxide in polluted air. Atmospheric
Environment 1982; 16: 1957–1972.
15. Westfelt UN, Lundin S, Stenqvist O. Safety aspects of delivery
and monitoring of nitric oxide during mechanical ventilation.
Acta Anaesthesiologica Scandinavica 1996; 40: 302–310.