oN"<
W . * _ «-
United States
Environmental Protection
Agency
Atmospheric Sciences
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S3-86/056 Feb. 1987
Project Summary
Hydroxyl Radical Rate
Constant Intercomparison
Study
E. O. Edney and E. W. Corse
An intercomparison study was con-
ducted to evaluate the OH rate constant
measurement protocol developed by
researchers at the University of Califor-
nia at Riverside. Researchers at the Uni-
versity of North Carolina at Chapel Hill
(UNO, Research Triangle Institute
(RTI), and Northrop Services, Inc.—En-
vironmental Sciences (NSI) used the
protocol to measure the room tempera-
ture OH rate constants for ethane, ben-
zene, chlorobenzene, and sec-butanol.
At least three measurements were
made for each compound, and each ex-
periment was analyzed using the time-
included and time-excluded methods
described in the protocol. The full re-
port of this research project provides
the data collected for each compound.
The room temperature OH rate con-
stants determined by NSI-ES and UNC
were in good agreement and also, with
the exception of the previously unre-
ported rate constant for sec-butanol,
agreed with the literature values.
Agreement was found using both
methods of analysis. The rate constants
for ethane, benzene, and chloroben-
zene obtained by RTI using the time-
excluded method also agreed with the
literature values; however, when the
same data were analyzed using the
time-included method, no such agree-
ment was found. The OH rate constant
for sec-butanol determined by RTI was
approximately one-third the value ob-
tained by NSI and UNC. It was not pos-
sible to identify the source of this dis-
crepancy.
This Project Summary was devel-
oped by EPA's Atmospheric Sciences
Research Laboratory, Research Triangle
Park, NC to announce key findings of
the research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering infor-
mation at back).
Introduction
The reactions of hydroxyl radicals
(OH) with organic compounds play an
important role in atmospheric chem-
istry. Reaction with OH is the dominant
homogeneous gas phase removal
mechanism for many volatile organic
compounds. Rate constants for reaction
with OH must be established to assess
the role of organic compounds in ozone
production and also to determine their
atmospheric persistence, which is of
primary importance when the organic
compounds are toxic chemicals.
Recently, researchers at the
Statewide Air Pollution Research Center
at the University of California at River-
side (UCR) developed for the U.S. Envi-
ronmental Protection Agency (EPA) a
protocol for measuring the OH reaction
rate constants of organic compounds.
To validate the protocol, EPA contracted
with Northrop Services, Inc.—Environ-
mental Sciences (NSI) to conduct a
series of OH rate constant determina-
tions using the UCR protocol. The re-
sults of this study have recently been
published. To further validate the proto-
col, EPA requested NSI to conduct an
intercomparison study. This report
summarizes the results of the study.
Researchers at the University of
North Carolina at Chapel Hill (UNC) and
Research Triangle Institute (RTI), as well
as NSI, participated in the study. Each
group independently measured the
-------
room temperature OH rate constants for
ethane, benzene, chlorobenzene, and
sec-butanol. At least three rate constant
measurements were made for each
compound, and each experiment was
analyzed using the two methods of cal-
culation described in the UCR protocol.
The full report compares the values for
the OH rate constants obtained by each
group. Where possible, the results are
compared with literature values. Details
concerning the exact experimental pro-
cedures employed by each group and
recommendations for improving the
protocol are referenced in the full re-
port.
Experimental Approach
The experimental approach em-
ployed was that described in the UCR
protocol (1 ). Rate constants for OH were
determined by using a relative rate
method, in which the test compound
was irradiated in air in the presence of
methyl nitrite (CH3ONO), nitric oxide
(NO), and a reference compound whose
OH rate constant was well known. The
theoretical chemical reaction sequence
is described in the full report.
If it is assumed that the loss mecha-
nisms for the test compound are reac-
tion with OH and a first-order reaction,
and that the only loss mechanism for
the reference compound is reaction
with OH, the rates of removal for the
test and reference compounds can be
described by the following equations:
[Test] = -kT [Test][OH] - k [Test] (I)
[Ref] = -ka [Ref][OH]
(II)
where [Test] and [Ref] are the concen-
trations of the test and reference com-
pounds, respectively, k \s a first-order
rate constant for the test compound,
and the kR and Rvalues are the OH rate
constants for reaction with the refer-
ence and test compounds, respectively.
Equations I and II can be combined
and integrated. The result is
, [Test], k, -i [Ref],
llnTfelrf,=rRlln-[RW,+k> W)
where [Test], and [Ref],o are the con-
centrations of the test compound and
the reference compound at time t0, re-
spectively, and [Test], and [Ref], are the
respective concentrations at time t. If 7/f
In ([Test], /[7esf],) is plotted against 7/f
ln([Ref]tjon.Ref],), a straight line should
be obtained with slope equal to kTlkfi
and intercept equal to k. The value for
the test compound OH rate constant is
immediately obtained from the slope
because the reference compound OH
rate constant is known.
Equation III can be reduced further if it
is assumed that the only removal mech-
anism for the test compound is reaction
with OH, i.e., k = 0. Under these circum-
stances Equation III can be simplified,
and the following result is obtained:
[Tesf]^ k
ln~[TeWt=TRln
(IV)
Equation IV differs from Equation III in
that the former is not explicitly time de-
pendent. If ln([Test],o/[Test]t) is plotted
against ln([Ref],o/[Ref],), a straight line
should be obtained with slope equal to
kT/kR, the same result found using
Equation III, but here, a value of zero is
found for the intercept. If the compound
under investigation satisfies the as-
sumptions stipulated for the application
of Equation IV, the rate constants ob-
tained using Equations III or IV should
be equal. Each rate constant measure-
ment made during this study was ana-
lyzed using both Equations III and IV,
and analyses based on Equations III and
IV are denoted as the Tl method (time-
included method) and the TE method
(time-excluded method), respectively.
Both Equations III and IV are in the
form of the straight line equation,
y = mx + b. Least squares linear regres-
sion analyses were conducted to obtain
values for the slope (m), intercept (b),
and the correlation coefficient squared
(r2) for each experiment. In addition,
standard deviations for the slope (am)
and intercept (ab) were calculated.
Experimental Method
All irradiations were conducted in
pillow-shaped 2-mil FEP Teflon bags
that were constructed by heat-sealing
three sides of two 4-ft x 8-ft sheets of
FEP Teflon. Swagelok fittings were
mounted in the wall of each bag and
served as reaction bag ports for filling
and evacuating the bag, as well as for
sampling its contents.
The NSI reaction chamber consisted
of a wooden cylindrical frame split
lengthwise with two light banks that
contained five lamps each mounted on
the inner side of the frame. Each light
bank contained a mixture of sun lamps
and black lamps. A 180-cfm blower was
mounted at the top of the frame and
was used to remove heat from the
chamber. The reaction chambers used
by RTI and UNC were designed and con-
structed by NSI. The chamber was a 2-t
ft x 2-ft x 4-ft aluminum box. LightJ
banks were mounted on two of the
inner sides of the chamber and con-
tained a mixture of sun lamps and black
lamps. Each bank contained positions
for six lamps. A 1/80-hp ventilation
blower was mounted on the top of the
chamber. Thermometers were mounted
inside both types of reaction chambers.
The detection systems and the meth-
ods for sampling were selected by the
individual research groups. UNC used a
combination of three automated Carle
Model 211 packed column FID gas chro-
matographs to monitor the concentra-
tions of test and reference compounds.
Automatic gas sample loop injections
were used to sample the bag contents.
RTI used Perkin Elmer 3920B and Sigma
300 FID gas chromatographs to monitor
the contents of the reaction bag. The
gas chromatographs were connected in
series with the reaction bag, and sam-
pling was accomplished by pumping a
gas sample through a connected Teflon
tube and manually injecting the sample
into the gas chromatographs. NSI used
Perkin Elmer Model 900 and GOW-MAC
Model 750 gas chromatographs. Gas
sampling valves were used to sample
the contents of the bag during the
ethane, benzene, and chlorobenzene
experiments. sec-Butanol was sampled
by bubbling a 5-L volume through an
impinger containing 2 mL of CH3OH. A
10-|j.L aliquot of the solution was in-
jected onto the gas chromatograph. Ad-
ditional experimental details, which in-
clude the GC columns and conditions
employed by each group, are refer-
enced in the full report.
NOX concentrations were determined
by both NSI and UNC using a Bendix
Model 8101-B NO/NO2/NOX analyzer. In
addition, UNC used a Bendix Model
8002 analyzer for monitoring 03. RTI did
not monitor NOX and O3 during the irra-
diations and relied on a high NO con-
centration (—15 ppm) to prevent 03 for-
mation. 03 concentrations were not
monitored by NSI because the NO con-
centrations measured during the irradi-
ations prevented its formation.
The CH3ONO was prepared by drop-
wise addition of 50% sulfuric acid
(H2S04) to a stirred, saturated solution
of sodium nitrite (NaN02) in methanol
(CH3OH). A nitrogen (N2) stream was
used to transfer the CH3ONO from the
reaction system to a liquid N2 cold trap,
where it was collected. The transfer line
consisted of two bubblers in series in
-------
front of the cold trap to remove H2SO4
nd H20. The first bubbler contained a
saturated solution of sodium hydroxide
(NaOH), and the second contained Dri-
erite desiccant.
A typical OH rate constant determina-
tion was begun by filling the bag with
clean air, mixing its contents, and evac-
uating it a number of times. The bag
was then filled with clean air to its final
volume (-100 L). During the filling proc-
ess, NO, CH3ONO, and the test and ref-
erence hydrocarbons were introduced
into the bag. All chemicals used during
the study had purities greater than 99%
and were not purified further. A series
of GC samples were taken before the
irradiation began in order to determine
the stability of the test and reference
compounds in the dark. In general, the
sampling continued until successive
peak heights of the compounds differed
by less than 2%. During the irradiation,
GC samples were taken as frequently as
possible. Typical irradiation times
ranged from 30 min to 60 min. The tem-
perature was recorded each time a GC
sample was taken.
Results and Discussion
OH rate constant data for ethane, ben-
zene, chlorobenzene, and sec-butanol
are provided in the appendix of the full
report. The appendix contains for each
experiment the average temperature,
the reference compound employed, the
initial concentrations of the test and ref-
erence compounds, the slopes and in-
tercepts and their standard deviations,
and the square of the correlation coeffi-
cient obtained using both the Tl and TE
methods. The results for each com-
pound and the reference rate constants
used to convert the slopes into OH rate
constants are also summarized in the
full report. The reference OH rate con-
stants employed, with the exception of
the benzene rate constant employed by
UNC, are those recommended in the lit-
erature. The OH rate constant for ben-
zene used by UNC is the value the inves-
tigators determined during the study.
The full report also provides, in tabu-
lar form, final summaries for the rate
constants determined by using the two
methods of calculation. In these tables,
the temperatures are the averages of
the individual temperatures shown for
each experiment in the appendix. N is
the number of experiments performed
by each group for each compound. The
values for the rate constants and the
square of the correlation coefficients
are averages obtained from the individ-
ual experiments. The standard devia-
tions in the rate constants are based on
the deviations of the rate constants
from the above-mentioned average rate
constants.
Ethane
The Tl method OH rate constants
were as follows.
(Northrop) 0.264 ± 0.019 x 10~12 cm3/
molecule-s, r2 = 0.9501
(UNC) 0.273 ± 0.059 x 10~12 cm3/
molecule-s, r2 = 0.9783
(RTI) 0.188 ± 0.495 x 10"12 cm3/
molecule-s, r2 = 0.4851
The r2 values obtained by Northrop and
UNC indicate that the data can be ade-
quately described by a straight line;
however, the r2 value found by RTI is
low. The poor fit obtained by RTI is also
reflected in the large deviation (263%) in
the rate constant obtained. Even though
the value of the rate constant is at the
lower limit for which the protocol can be
employed, good results are expected
because ethane concentrations can be
measured accurately and it is unlikely
that ethane undergoes reactions other
than that with OH. However, RTI re-
ported substantial problems with in-
creases in the ethane concentrations at
the start of the irradiation. RTI attributed
the increases to mixing problems.
The values obtained for the OH rate
constant using the TE methods were as
follows.
(Northrop) 0.273 ± 0.010 x 10~12 cm3/
molecules-s, r2 = 0.9904
(UNC) 0.302 ± 0.050 x 10~12 cm3/
molecule-s, r2 = 0.9522
(RTI) 0.255 ± 0.026 x TO'12 cm3/
molecule-s, r2 = 0.8305
The results obtained by Northrop and
UNC are in reasonable agreement with
those found using the Tl method. The
RTI value for r2 is still relatively low, but
far better than that obtained using the Tl
method.
That the r2 obtained using the TE
method is substantially greater than
that found with the Tl method may be,
at first, somewhat surprising since the
Tl calculation method appears more
complete because it takes into account
first-order processes for ethane. How-
ever, this distinction may be mislead-
ing. The slopes obtained using each
method are based on a least squares
regression analysis of the data. An ex-
amination of this analytical method
shows that the magnitude of the calcu-
lated slopes and intercepts and the val-
ues for r2 are strongly influenced by the
large data points in the set.
The relative importance of a given
data point in the two calculation meth-
ods differs. The independent variable
used in the TE method, /n([7esf]fo/
[Test],), monotonically increases dur-
ing the irradiation because the test com-
pound concentration decreases
throughout the experiment. However,
the situation is reversed if one uses the
Tl method, in which the independent
variable is 1/t ln([Test]t l[Test]t). In the
absence of reactions of the test com-
pound other than that with OH, it can be
shown that 1/t ln([Test]to/[Test]t) is
equal to the average OH concentration
times the OH rate constant during the
time interval from 0 to t. The dominant
source of OH in this system is the pho-
tolysis of CH3ONO, and because the
CH3ONO concentration decreases dur-
ing the irradiation and the sinks for OH
(in particular, the concentration of NO2)
increase, it is reasonable to expect that
the average OH concentration should
decrease as a function of time. If this is
the case, the slope obtained using the Tl
method depends most strongly on the
short-time irradiation points, as op-
posed to the TE method, where the
long-time irradiation points influence
the calculation of the slope, intercept,
and r2, and hence the rate constant.
The discrepancy in the RTI results is
consistent with this argument. The
good agreement found between the
two methods used by Northrop and
UNC suggests that during these experi-
ments the dominant removal mecha-
nism for ethane and the reference com-
pound was reaction with OH. However,
in the past, Northrop has encountered
problems similar to those experienced
by RTI.
Reports in the literature recommend a
value of 0.275 x 10~12 cm3/molecule-s
for the room temperature (25°C) OH rate
constant, with an uncertainty of ±20%.
With the exception of the RTI value ob-
tained using the Tl method, the experi-
mentally determined values found in
this study are in agreement with the lit-
erature values, as well as with each
other.
Benzene
The OH rate constants and the associ-
ated values determined for r2 using the
Tl method are as follows:
(Northrop) 0.876 ± 0.276 x 1Q-12 cm3/
molecule-s, r2 = 0.9731
(UNC) 1.041 ±0.071 x 10~12 cm3/
molecule-s, r2 = 0.9657
(RTI) 0.544 ± 0.043 x 10~12 cm3/
molecule-s, r2 = 0.6646
-------
The values obtained using the TE
method were as follows.
(Northrop) 0.949 ± 0.183 x 10~12 cm3/
molecule-s, r2 = 0.9921
(UNC) 1.070 ± 0.075 x TO'12 cm3/
molecule-s, r2 = 0.988
(RTI) 1.054 ± 0.047 x 10'12 cm3/
molecule-s, r2 = 0.9642
All of the values overlap except for the
RTI value found using the Tl method;
however, the standard deviation of this
determination is misleading. The RTI re-
sults displayed in the full report show
that although the Tl method deviations
for experiments RTI-1 and RTI-2 are rel-
atively small (11% and 7%, respec-
tively), the value obtained in RTI-3 is
large (86%). The small (8%) deviation in
the average for the three experiments
may be fortuitous. The improvement in
the precision of the results using the TE
method over that of the Tl method may
be associated with factors cited in the
discussion of the RTI ethane results.
The recommended value for the room
temperature (25°C) OH rate constant for
benzene is 1.28 ± 10~12cm3/molecule-s,
with an estimated uncertainty of ±30%.
The results obtained here are slightly
below this value, but with the exception
of the RTI Tl method result, there is
overlap when the error bars are in-
cluded. A comparison between the RTI
results and those obtained by Northrop
and UNC is further complicated be-
cause of the elevated temperature
(33.8°C) at which RTI investigators con-
ducted their experiments.
Chlorobenzene
The following OH rate constants were
found for chlorobenzene using the Tl
method.
(Northrop) 0.789 ± 0.190 x 10~12 cm3/
molecule-s, r2 = 0.9169
(UNC) 0.784 ± 0.129 x 10'12 cm3/
molecule-s, r2 = 0.9692
(RTI) 0.707 ± 0.306 x 10~12 cm3/
molcule-s, r2 = 0.6116
The RTI experiments were again con-
ducted at an elevated temperature
(35.8°C). The corresponding values
found by the TE method were as fol-
lows.
(Northrop) 0.593 ± 0.116 x 10~12 cm3/
molecule-s, r2 = 0.9777
(UNC) 0.756 ± 0.101 x 10~12 cm3/
molecule-s, r2 = 0.9662
(RTI) 0.696 ± 0.129 x 1Q-12 cm3/
molecule-s, r2 = 0.9754
All of the results overlap; however,
there are large uncertainties in the RTI
(43%) and NSI (27%) Tl method values.
As was postulated for the RTI Tl method
determinations for ethane and benzene,
the uncertainty may be due to a hetero-
geneous reaction that occurs early in
the irradiation and that cannot be de-
scribed by the parameterization of theTI
method.
The values of the rate constant are in
reasonable agreement with two re-
ported measurements that are refer-
enced in the full report. One group ob-
tained a value of 0.88 ±0.11 x 10~12
cm3/molecule-s at 26 ± 2°C using the
UCR protocol, and another reported a
value of 0.67 ± 0.05 x 10~12 cm3/
molecule-s at 23°C using the flash pho-
tolysis resonance fluorescence method.
sec-Butanol
The Tl method OH rate constants de-
termined were as follows.
(Northrop) 10.30 ± 2.48 x 10~12 cm3/
molecule-s, r2 = 0.9692
(UNC) 11.55± 1.77 x 10~12 cm3/
molecule-s, r2 = 0.9854
(RTI) 4.01 ± 1.38 x 10-12 cm3/
molecule-s, r2 = 0.7581
The corresponding TE method rate con-
stants were as follows.
(Northrop) 9.40 ± 0.87 x 10~12 cm3/
molecule-s, r2 = 0.9894
(UNC) 7.37 ± 1.65 x 1Q-12 cm3/
molecule-s, r2 = 0.9917
(RTI) 2.71 ± 0.20 x 10~12 cm3/
molecule-s, r2 = 0.9681
The UNC and Northrop values over-
lap, but they are approximately a factor
of three larger than the values found by
RTI. Although this rate constant has not
been previously measured, it can be es-
timated using the predictive methods
developed recently by Atkinson. A value
of 8 x 10~12 cm3/molecule-s is found
using the technique. This result is in rea-
sonable agreement with the Northrop
and UNC measurements. There are no
obvious reasons for the low RTI values.
The major difference between the ex-
perimental approaches used by NSI and
UNC and that used by RTI is that the
Northrop and UNC experiments were
conducted at 24°C, whereas the RTI ex-
periments were conducted at an aver-
age temperature of 34°C. Because it is
expected that the OH rate constant in-
creases as a function of increasing tem-
perature, the difference in temperature
makes the discrepancy more difficult to
explain.
Conclusions and
Recommendations
The room temperature OH rate con-
stants determined with the UCR proto-
col for ethane, benzene, chlorobenzene,
and sec-butanol by NSI and UNC werJ
in good agreement and also, with the
exception of the previously unreported
rate constant for sec-butanol, agreed
with the values recommended in the lit-
erature. The agreement was found
using both the TE and Tl methods. The
rate constants for ethane, benzene, and
chlorobenzene obtained by RTI using
the TE method also agreed with the lit-
erature values; however, when the
same data were analyzed using the Tl
method, no such agreement was found.
The OH rate constant for sec-butanol
determined by RTI was approximately
one-third the value obtained by
Northrop and UNC. It was not possible
to identify the source of this dis-
crepancy.
The results of this intercomparison
study suggest that additional studies
should be conducted to expand the data
base and thus assess accurately the
validity of the protocol. Special empha-
sis should be placed on determining
which of the two calculation methods
should be employed.
-------
Edward 0. Edney and E. W. Corse are with Northrop Services. Inc., Research
Triangle Park, NC 27709.
J. J. Bufalini is the EPA Project Officer, see below.
The complete report, entitled "Hydroxyl Radical Rate Constant Intercomparsion
Study," (Order No. PB 87-111 142/AS; Cost: $11.95. subject to change)
will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 221611
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United Stales
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-86/056
0000329 PS
U S ENVIR PROTECTION AGENCY
H8irDi.kS«atlT.E«T
CHICAGO It- 60604
------- |