United States
 Environmental Protection
 Agency
  Atmospheric Research and
  Exposure Assessment Laboratory
  Research Triangle Park NC 27711
 Research and Development
  EPA/600/S3-90/034
         Sept. 1990
 Project  Summary
 Laboratory and  Field Evaluations
 of  Extrasensiive Sulfur Dioxide
 and Nitrogen  Dioxide  Analyzers
 for Acid  Deposition  Monitoring
EsStsvard E. Bietemani, Jr., Annette H. Green, Robert S. Wright, and
Josepito £. SieKles, III
  Hie objective oi this project was to
evaluate the performance characteristics
of selected extrasensltive NO2 and SO2
analyzers in both laboratory and field
environments. The analyzers that were
evaluated are two Columbia Scientific
Industries (CSI) model  1600 NO2 ana-
lyzers, two Scintrex model LIMA 3 NO2
analyzers, and two Thermo Environmen-
tal (TECO) model 43S SO2 analyzers. The
laboratory test procedures for measuring
their  performance characteristics were
adapted from the equivalency test proce-
dures given in Chapter 40, Part 53 of the
Code of Federal Regulations. The field
evaluation was conducted over a one-
month period  at a site in Research
Triangle Park, North Carolina. The results
of the laboratory and field evaluations are
given in this project summary. Although
EPA  has not established performance ••
specifications for extrasensitive ana-
lyzers, the results of the laboratory and
field evaluations suggest that the analyz-
ers, generally, will perform in an accept-
able fashion for the intended application.
Nevertheless, several  specific areas
needing improvement were uncovered in
the evaluations.
  This Project Summary was developed
by EPA's Atmospheric Research and
Exposure  Assessment Laboratory,
Research Triangle Park, NC, to announce
key findings of the research program that
is fully documented in a separate, report'
of the same title (see Project Report
ordering information at back).

Introduction

  The analyzers that were evaluated are
two Columbia Scientific Industries (CSI),
model 1600 NO2 analyzers, two Scintrex
model LMA 3  NO2 analyzers, and two
Thermo Environmental (TECO) model 43S
SO2 analyzers. The CSI model 1600 NO2
analyzers detect NO2 by catalytic conver-
sion of NO2 to nitric oxide (NO) followed
by the chemiluminescent reaction of NO
with ozone (O3). The Scintrex model LMA
3 NO2 analyzers detect  NO2 by the
chemiluminescent reaction with luminol
in aqueous solution. The TECO model 43S
SO2 analyzers detect SO2 by ultraviolet
fluorescence using a pulsed light source.
The ranges that the instruments were
tested on are as follows:
 CSI model
   1600NO2

Scintrex model
  LMA 3 NO2
 TECO model
  43S SO2
0-0.05 part per million
(ppm) (i.e., 0-50 ppb),

0-50 ppb at recorder
output, (0-20 ppb on
front panel meter), and

0-10 parts per billion
(ppb).

-------
Each instrument also has higher ranges
which were not used  in this evaluation.
The two instruments within each model
are designated as A and B in this Project
Summary.
  Before evaluating the analyzers' perfor-
mance characteristics, it was necessary
to consider  what  characteristics  are
relevant to performance and how those
characteristics can be experimentally
measured. Performance specifications
and test procedures for ambient air quality
monitoring instruments have been estab-
lished in Chapter 40, Part 53 of the Code
of Federal Regulations. These specifica-
tions are referred  to as "equivalency
specifications."  In  general, the proce-
dures used in  the  laboratory and field
evaluations are modified versions of the
procedures given  in the  equivalency
specifications.
  A  fundamental  difference exists
between the purpose  of the equivalency
specifications and  the  purpose of the
present study. The equivalency specifica-
tions were designed to certify that a given
instrument was acceptable for determin-
ing the compliance of a locality with the
National Ambient Air  Quality Standards
for the given  pollutant.  Performance
specifications were established for these
methods and test procedures were given
to determine  whether an instrument
demonstrated acceptable performance.
The purpose of the current study was to
evaluate  instruments  with  substantially
better  sensitivity than  those  used for
compliance monitoring. No similar perfor-
mance specifications  exist for these
extrasensitive  instruments. Thus,  this
study estimated the performance parame-
ters of the  instruments quantitatively
rather  than determining whether these
instruments met predefined  specifica-
tions. For this reason, many of  the tests
described in the Code  of Federal Reg-
ulations were modified for this study.
   In addition to the laboratory  tests for
evaluating performance parameters,  a
field evaluation of  the instruments  was
performed. The field evaluation  was
conducted over a 31 -day period at RTI's
Materials  Exposure  Site in  Research
Triangle Park,  North  Carolina.  The CSI
1600, LM A 3 and TECO 43S extrasensitive
analyzers and other analyzers at the site
were operated  under  typical ambient air
quality  monitoring conditions. This por-
tion  of  the study provided  a  real-world
evaluation of the extrasensitive analyzers'
performance. The primary  goals of the
field evaluation  were to measure the
analyzers' zero and  span drifts, their
precision, and any operational failures.
Procedure

Description of the
Laboratory Evaluation

  The laboratory evaluation  of the NO2
analyzers was done separately from that
of the SO2 analyzers. The laboratory
evaluation consisted of a nine-day test
cycle and separate interference tests. The
test  cycle  was conducted inside an
environmental  chamber  where the
temperature and line voltage were con-
trolled according to the schedule given
in Table 1. Day 0 was for setup and initial
calibration.
  Various performance parameters for the
extrasensitive analyzers were evaluated
on each test  day.  These performance
parameters are defined below:

• Noise  is the  standard deviation  of
  twenty-five consecutive instantaneous
  measurements of zero air taken at two-
  minute intervals;
• Precision is the standard deviation of
  six instantaneous measurements  of
  pollutant  concentrations correspond-
  ing to  0,  20, 50  or 80% full scale (%
  FS) readings. The pollutant concentra-
  tion was alternately spiked positively or
  negatively between each measurement
  and then was allowed to restabilize at
  the original concentration;
• Limit  of detection is three times the
  precision at 0% FS;
• Limit  of quantitation is ten times the
  precision at 0% FS;
• Lower detection  limit is two times the
  noise at 0% FS;
• Zero drift is the shift in an analyzer's
  response to zero air over approximately
  24 hours;
• Span  factor drift  is the shift in an
  analyzer's span  factor  over approxi-
  mately 24 hours.  The daily span factor
  is the slope of the linear regression of
  the mean of the  precision measure-
  ments at 0, 20, 50 and 80% FS;
• Rise time is the  period between the
  initial  analyzer response and the time
  of 95% of the final stable reading after
  a step concentration change from 0 to
  80% FS; and
• Fall time  is the corresponding period
  after a step concentration change from
  80 to 0% FS.

  The test  manifold for the laboratory
evaluation was fabricated from 0.95-cm
ID PFA Teflon™ tubing. Test atmospheres
containing NO2 were prepared by gas-
phase titration and dilution of 5 parts per
Table 1. Laboratory Evaluation Test Schedule
Test
Day
0
1
2
3
4
5
6
7
8
Line Voltage
(Volts AC)
115
125
105
125
105
125
105
125
105
Temperature
(Degrees Celsius)
25
20
20
30
30
20
20
30
30
million (ppm) NO in nitrogen  contained
in a  compressed gas cylinder. Test
atmospheres containing SO2 were pre-
pared  by dilution of  0.8  ppm  SO2  in
nitrogen contained in a co'mpressed gas
cylinder. The diluent gas was ambient air
that had  passed  through a permeation
drier, a heated catalyst bed and chemical
scrubbers. The dilution air  for the NO2
analyzers was' humidified  to  approxi-
mately 20%  relative humidity. The total
sample flowrate was 10 liters per minute.
The diluted pollutants were mixed turbu-
lently  by  a 36-cm length of 0.17-cm  ID
PFA Teflon™ tubing.
  Interference tests were conducted after
the nine-day test cycle had been com-
pleted. In many cases, test atmospheres
containing interferent gases were pro-
duced by dilution of interferent gases
contained in compressed gas cylinders.
O3 was  produced by the ultraviolet
irradiation of air, nitric acid  (HNO3) by a
permeation tube, nitrous acid  (HNO2)  by
a sublimation generator,  peroxyacetyl
nitrate (PAN) by  gaseous dilution of a
PAN/iso-octane solution in Tedlar™ bags,
and water vapor (H2O) by  passing the
dilution air through a glass bulb contain-
ing water.
  In general, the interference tests were
performed in the presence of a pollutant
concentration  equivalent to a 50%  FS
reading. The exceptions were O3, HN03
and HNO2 for the NO2 analyzers. These
interference  tests were performed in the
absence  of  NO2 to eliminate gas-phase
reactions  or because of  experimental
constraints. The CSI NO2 analyzers were
used  to  measure the  PAN and HNO2
concentrations for the interference tests
of the Scintrex NO2 analyzers. Based on
the reported performance of other gas-
phase chemiluminescenceNO2 analyzers
to atmospheres containing PAN or HNO2,
it was assumed that the CSI NO2 analyzers
respond  quantitatively to  these com-
pounds.
  A summary of the results of the labor-
atory  evaluation  (excluding  the interfer-

-------
ence tests) is given in Table 2. The results
of the  interference  tests are given  in
Table 3.

Description of the
Field Evaluation

  Following the laboratory evaluation, the
extrasensitive analyzers underwent a 31 -
day field evaluation at a  monitoring site
on the RTI campus. The  purpose of the
field evaluation was to test the analyzers'
performance under realistic  field condi-
tions.  The field  evaluation tested for
operational failures, zero and span factor
drifts and precision.
  The monitoring site is a grassy, open,
2.4-hectare area located near a four-lane
highway. The extrasensitive analyzers
were housed in a temperature-controlled
trailer. Continuous monitors for several air
pollutants and meteorological parameters
are also operated  in this trailer in con-
junction with other RTI projects.
  The extrasensitive  analyzers were
calibrated on a daily basis for the first 17
days of the field evaluation and on a bi-
daily basis for the remainder of the field
evaluation. The calibration apparatus was
the same as was used during the labor-
atory evaluation with the exceptions that
the NO2 and SO2 analyzers were cali-
brated simultaneously and that the diluent
air was not humidified. The SO2 analyzers'
data were corrected for the NO remaining
from the gas phase titration. In general,
the extrasensitive analyzers'  zero  and
span  pots were not adjusted during the
field evaluation.
  The extrasensitive analyzers sampled
ambient air during those portions of the
field evaluation when they were not being
calibrated. The purpose of these ambient
air quality measurements was to produce
matched sets of NO2  and  SO2 data to
compare the analyzers and from  which
analyzer precision  estimates  could  be
calculated. The NO2 analyzers produced
483 sets of hourly mean  concentrations,
and the SO2 analyzers produced 463 sets
of hourly mean concentrations. The SO2
analyzers' data were corrected for the NO
interference  using the relative interfer-
ence  responses to NO and the NO
measurements from  the  CSI NO2 ana-
lyzers.


Maintenance and Failures

  During the laboratory  evaluation, the
CSI 1600 and  TECO 43S  analyzers
required  only scheduled maintenance
and experienced no operational failures.
One LMA 3 analyzer (i.e., LMA B)  expe-
rienced problems with fluid backup in the
air intake trap. The problem was traced
to worn pump tubing and a misaligned
tensioning bracket on the pump. Realign-
ing the bracket and replacing the tubing
at two-week  intervals solved that prob-
lem. Erratic responses  were noted after
several  days  operation  of the LMA  3
 analyzers. This problem was eliminated
 by backflushing the LMA 3 analyzers on
 a daily basis.
  During the field evaluation, the TECO
43S and CS11600 analyzers required only
scheduled maintenance. One CSI 1600
analyzer  (i.e.,  CSI A)  experienced  a
temporary failure in its NOX channel due
to an air conditioner  failure at the field
site. All  three analyzers experienced
noticeable but  temporary changes in
calibration during  the  air conditioner
failure. One LMA 3 analyzer (i.e.,  LMA A)
had a solution backup into its ozone trap.
It is possible that an earlier replacement
in pump tubing could have prevented this
problem. The other LMA 3 analyzer (i.e.,
LMA B) exhibited a significant span factor
drift that appeared to be associated with
excessive wear of the pump tubing. This
problem is discussed further below.


Intra-Model Agreement

  From the field evaluation data, the intra-
model  agreement is  estimated  by the
concentration difference  between the
hourly average ambient air concentra-
tions indicated by the two analyzers within
a given model. The 30-day, overall mean
concentration difference for the  LMA 3
analyzers is 0.733 ppb. On a daily basis,
the mean concentration difference varies
from —1.577 to 1.793 ppb. The pattern of
the day-to-day  variation  in the mean
concentration  differences  parallels the
Table 2. Summary of Results for the Laboratory Evaluation of Extrasensitive SOZ and NO2 Analyzers (Excluding Interference Tests)
Performance
Parameter
Mean Noise (am)
Mean Noise (pm)
Mean Precision (0% FS)
Mean Precision (20% FS)
Mean Precision (50% FS)
Mean Precision (80% FS)
Limit of Detection"
Limit of Quantitatione
Lower Detection Limit
Mean Daily Zero Drift (am)
Mean Daily Zero Drift (pm)
Mean Daily Span Drift (20% FS)
Mean Daily Span Drift (50% FS)
Mean Daily Span Drift (80% FS)
Mean Daily Span Factor Drift
Mean Rise Time
Mean Fall Time
Nitrogen Dioxide Analyzers
Units
ppb
PPb
ppb
PPb
ppb
ppb
ppb
ppb
PPb
PPb
ppb
%
%
%
%
min
mm
LMA A
0.003
0.002
0.003
0.040
0.065
0.105
0.009
0.030
0.005
-0.00
-0.00
-1.3
0.4
0.7
-1.0
1.8
0.2
LMAB
0.004
0.003
0.003
0.050
0.078
0.165
0.009
0.030
0.006
0.00
-0.00
-3.3
-1.2
-0.5
-0.0
1.9
0.3
CSI A
0.280
0.303
0.221
0.300
0.300
0.297
0.663
2.210
0.560
0.01
-0.00
0.3
0.6
0.5
-0.6
4.9
4.2
CSIB
0.335
0.358
0.363
0.408
0.364
0.426
1.089
3.630
0.670
0.04
0.02
0.1
0.6
0.3
-0.4
4.3
3.8
Sulfur Dioxide Analyzers
TECO A TECO B
0.032
0.033
0.029
0.038
0.046
0.052
0.088
0.292
0.065
0.010"
0.011"
0.1"
0.4"
0.5"
-0.4
3.5
3.4
0.016
0.015
0.016
0.017
0.027
0.035
0.049
0.163
0.031
0.003"
0.003"
0.7"
0.7"
0.8"
-0.9
3.4
3.5
a These values are calculated from the mean precision at 0% FS. Slightly higher values for the LMA 3 and TECO analyzers are obtained when
 the concentration-dependent precision estimates are used.
"Higher values for the zero and span drifts are obtained when Days 3 and 7 are excluded from the calculations. These revised zero and span
 drifts are equivalent to approximately —1.1%FS per day.

-------
Tab/e 3. Interference Test Results for Extrasensitive Analyzers
                                   Relative Interferent Response
                             (A [NOJ or JSP J Response/A. [Interferent])
Interferent
Species
02
C02
H2O
SO2
HCI
NH3
03«
PAN
HN03"
HNO-f

02
CO2
NO
H2S
H20
Xy/ene
/.M/M
3X10(-8)
-4X10(-6)
-1X10(-7)
1 X 10(-3)
2X 10(-3)
ND
ND
7X 10(-1)
ND
, 4X10(-1)
TECOA
ND
ND
3X1'0(~2)
1X10(-3)
ND
1 X 10(-2)
LMAB
3X10 (-8)
-5X 10 (-6)
-2X 10 (-7)
ND
6X10 (-4)
ND
ND
6X 10(-1)
ND
6X10(-1)
TECOA
ND
ND
2X10 (-2)
1 X 10(-3)
-7X 10(~9)
8X10(-3)
CSIA
ND
ND
-6X 10(-8)
1 X 10 (-3)
ND
6X10 (-2)
ND
NA
1 X 10(0)
NA







CSIB
ND
ND
-7X 10(-8)
; ND
ND
2X 10(-1)
ND
NA
1 X 10(0)
NA
I
l'

'
I


I
ND = Not detected. The change in the analyzer response is less than the 95% confidence interval
 for the precision at 50% full scale.
NA = Not applicable. The CS11600 analyzers were used to measure the interferent concentration.
• The WO2 concentration — 0 ppb for this interference test. For the remaining interference tests,
 the pollutant (i.e., NO2 or SOJ concentrations corresponded to analyzer readings of 50% full
 scale.
pattern of the day-to-day variation in LMA
B's span factor (see  below). The  two
phenomena appear to be related in some
manner.
  The overall mean concentration differ-
ence for the CS11600 analyzers is -0.259
ppb. On a daily basis, the mean concen-
tration difference varies  from —1.134 to
0.212 ppb.  The daily  mean values  are
generally consistent for the first 15 days
of the field evaluation until the  air
conditioner failure and show greater
variability afterwards.
  The overall mean concentration differ-
ence for the TECO 43S analyzers is 0.036
ppb. On a daily basis, the mean concen-
tration difference varies  from —0.008 to
0.112 ppb. Relatively little day-to-day
variation is seen  in  these  daily mean
values.

Noise

  The LMA 3 NO2 analyzers had consid-
erably less noise than the CSI 1600 NO2
analyzers. The noise  level for the TECO
43S SO2 analyzers was between  that of
the two NO2 analyzer models. During the
field  evaluation,  ambient air  measure-
ments using the LMA 3  analyzers exhi-
bited more short-term variability than the
corresponding measurements for the CSI
1600 analyzers. However, the LMA  3
analyzers' responses were less  noisy
during field calibration than those of the
CSI  1600  analyzers.  It appears that the
LMA 3 analyzers  were  responding  to
short-term fluctuations in the ambient air
NO2 concentration which were  being
integrated over longer time periods by the
CS11600 analyzers.

Precision                 I

  In the laboratory evaluation, the LMA
3  NO2 analyzers  had lower  precision
values (i.e., better  precision), expressed
as standard deviation than the  CSI1600
NO2 analyzers. The precision values for
the LMA 3 NO2 analyzers increased with
increasing concentration. The  precision
values for the CS11600 NO2 analyzers did
not exhibit a concentration dependence.
The precision values for the TECO 43S
SO2 analyzers were lower tha.n those of
the NO2 analyzers and  showed a con-
centration dependence.
  From the  field  evaluation data, esti-
mates of the intra-model  precision and
the  effective  operating precision  are
 calculated  from ambient  air measure-
 ments. The first estimate is obtained by
 dividing the  standard  deviation of the
 hourly mean concentration  differences
 between the two analyzers within a given
 model by  the  square  root  of 2.  The
 precision estimate is somewhat in  error
 for the LMA 3 and TECO 43S analyzers
 because  of its assumption that the
 precision is independent of concentra-
 tion. The laboratory evaluation demon-
1 strated that the precision of these  ana-
 lyzers is concentration  dependent. The
 second estimate  is obtained from the
 hourly mean concentrations using the
 principal component analysis program
 that was  developed by  Holland  and
 McElroy. The second estimate includes
 both the intra-model precision and the
 uncertainty associated with  assumed
 random interference effects.
   The overall standard  deviation of the
 concentration differences jbetween the
 two LMA 3 analyzers is 0.930 ppb, which
 results in an intra-model precision esti-
 mate  of 0.658 ppb. On a daily basis, the
 standard deviation varies  from 0.091 to
 1.901  ppb. This precision estimate may
 be influenced by the day-to-day variation
 in LMA B's span factor (see below).
   The overall standard  deviation of the
 concentration differences between the
 two CSI 1600  analyzers  is 0.554  ppb,
 which results in an intra-model precision
 estimate of 0.392 ppb. On a daily basis,
 the standard deviation varies from  0.149
 to 1.039 ppb.
   The overall standard  deviation of the
 concentration differences ibetween the
 two TECO 43S analyzers is 0.042 ppb
 which results in an intra-model precision
 estimate of 0.030 ppb. On a daily basis,
 the standard deviation varies from  0.012
 to 0.061 ppb.
   It should  be noted that; these  intra-
 model precision  estimates cannot be
 compared validly to those obtained during
 the laboratory evaluation. The  field
 precision estimate is for hourly average
 measurements of varying concentrations
 and the laboratory precision estimate is
 for instantaneous measurements  of a
 constant concentration. Nevertheless, the
 precision estimate for the  LMA 3  ana-
 lyzers obtained from the field evaluation
 data is approximately an order of  mag-
 nitude greater than those obtained from
 the laboratory evaluation  data.  The
 difference between the laboratory and the
 field precision estimates that is seen for
 the LMA 3 analyzers is  not seen for the
 CSI 1600 and TECO 43S analyzers. The
 laboratory and field precision estimates

-------
 are in good agreement for the latter ana-
 lyzers.                         ,
   The effective operating precisions of
 the extrasensitive analyzers during the
 field evaluation were calculated using the
 principal  component analysis  program
 that Holland and McElroy developed. The
 hourly mean concentrations were used in
 the calculations. This statistical  method
 has a number of assumptions which must
 be satisfied  if its  results are  to  be
 statistically valid. Among these assump-
 tions are:

 • At least three analyzers' data must be
   compared;
 • The range of measured concentrations
   is large relative to the precision;
 • The precision must be concentration-
   independent;
 • The interference effect must be random
   and independent of the pollutant con-
   centration; and
 • The analyzers to be compared must be
   sufficiently different in their sensitivity
   to interfering substances.

 These assumptions are violated to some
 degree by the extrasensitive analyzer data
 from  the  field evaluation.  Accordingly,
 precision estimates  from principal com-
 ponent analysis of these data are some-
 what questionable.
   The effective operating precisions for
 the two LMA 3 analyzers are 0.504 and
 0.628 ppb, which compare well  with the
 corresponding  intra-model precision of
 0.658 ppb. These precision estimates may
 be partially compromised  by the span
 factor drift that was  observed in LMA B.
 The frequent calibration of these analyz-
 ers tended to reduce the effect of this drift
 on the NO2 concentration measurements
 and the precision estimates obtained from
 these analyzers.
  The  effective operating precisions for
the two CS11600 analyzers are 0.284 and
0.185 ppb, which are appreciably lower
than the  corresponding  intra-model
precision of 0.392 ppb. These values are
considerably lower than the value of 2.7
ppb obtained by Holland and McElroy for
the same analyzer model. Their data set
was collected on the analyzer's 0 to 500
ppb range, while this data set was
collected on its 0 to 50 ppb range. The
range that is used may have an effect on
the analyzer's precision.
  The principal component analysis of the
SO2 data included measurements from
the RTI trailer's TECO  model 43A SO2
analyzer. This analyzer  was calibrated
using  a different calibration apparatus
 from that used  for the  extrasensitive
 analyzers. It was operated on its 0 to 200
 ppb FS range. This inclusion was neces-
 sary to satisfy the requirement of principal
 component analysis that  at least three
 analyzers' data must be compared.
   The effective operating  precisions for
 the TECO 43S analyzers are 0.0003 and
 0.0003 ppb, which are approximately two
 orders of magnitude less  than the cor-
 responding intra-model precision of 0.030
 ppb.  It is hypothesized that the small
 precision estimate from principal compo-
 nent analysis is a result of violating the
 assumption  that  the analyzers to  be
 compared must be different. The effective
 operating precision for RTI's TECO 43A
 analyzer is 0.111 ppb.
 Detection Limits

   The detection limits for the LMA 3 NO2
 analyzers were lower than those of the
 CSI 1600 NO2 analyzers. There was only
 a very small difference  between  the
 detection limit estimates for the LMA 3
 NO2 and TECO 43S analyzers with and
 without compensation for their increase
 in  precision  with  concentration. The
 detection limits for the TECO 43S SO2
 analyzers were between those for the two
 NO2 analyzer models.
Zero and Span Drifts

  Analysis of the zero and span factors
that were obtained during the laboratory
evaluation for the CS11600 NO2 analyzers
did not show any obvious dependence
on  room temperature or line voltage. In
addition, no time-dependent change (i.e.,
secular trend) of either zero values  or
span factors was seen for the CSI 1600
analyzer.
  The LMA  3 NO2  analyzer showed a
slight dependence  of  zero  values  on
temperature (approximately  —0.003  to
-0.006 ppb/degree  C). This is in agree-
ment with the value of —0.009 ppb/degree
C obtained by Bubacz and Pleil. The LMA
3 analyzers did not show the temperature-
dependent span  drift  seen in earlier
models by Bubacz and Pleil.  No secular
trend  or voltage dependence in zero or
span  drift  was  seen for the LMA 3
analyzer.
  The TECO 43S analyzers showed a
secular trend in their zero values  of
approximately -0.1  ppb/day during  the
laboratory evaluation. This made the span
 values appear to also drift because the
 span drift calculations do not provide for
 a zero-drift correction. In  addition, the
 TECO 43S analyzers' zero values showed
 a slight  dependence  on  temperature
 (approximately 0.024 ppb/degree C). No
 change in span factors with time, temper-
 ature, or line voltage was  seen for the
 TECO 43S analyzer.
   During the field evaluation, no secular
 trends were observed in the zero values
 for the CSI 1600 and LMA 3 analyzers.
 The  TECO  43S analyzers  showed  a
 secular trend  in their zero values of
 approximately -0.01  ppb/day. This trend
 is approximately an  order of magnitude
 less  than  that observed  during  the
 laboratory evaluation  in  these same
 analyzers. No secular trends were
 observed in the span factors for the CSI
 1600 analyzers and one of the two LMA
 3 analyzers. The other  LMA 3  analyzer
 (i.e., LMA B) exhibited a span factor drift
 of approximately 2.2%/day. It is hypothe-
 sized that this drift was  caused  by
 excessive wear in the analyzer's pump
 tubing because large span factor shifts
 were observed when the pump tubing
 was replaced in this analyzer. The  two
 TECO  43S analyzers  exhibited span
 factor drifts of  approximately -0.13 and
 -0.28%/day.
 Rise and Fall Times

  All the extrasensitive analyzers' rise and
 fall times were within 5 minutes. The LMA
 3 NO2 analyzers' values were noticeably
 faster than those of the other extrasen-
 sitive analyzers. Additionally, the fall times
 for the LMA 3 NO2 analyzers were much
 faster than the corresponding rise times.
  The LMA 3 NO2 and TECO  43S SO2
 analyzers took long periods of  time (i.e.,
 several hours) to rise from approximately
 95 to 100% of their final, stable readings •
 during each test day's initial change from
 an overnight zero air exposure to an 80%
 full scale concentration. This  pheno-
 menon was not observed for the CS11600
 NO2 analyzers. In the case of  the LMA
 3 NO2 analyzers, it is hypothesized that
their ozone  traps required conditioning.
 In the case  of the TECO 43S SO2 ana-
 lyzers, it is hypothesized  that this phe-
 nomenon  is  associated with the daily
initial conditioning of the calibration
apparatus and the sample lines after the
overnight zero air exposure. This pheno-
menon was not observed after subse-
quent step concentration changes during
each test day.

-------
Interferences

  Each of the extrasensitive  analyzer
models responded to one or more inter-
ferents.  The  LMA  3  NO2 analyzers
responded to  oxygen,  carbon  dioxide,
water, PAN, and nitrous acid. The relative
response  to PAN was lower  than  the
quantitative response reported by Wen-
dell et at. The nitrous acid interference
has not been reported previously in the
literature. Interference  from  O3  on
luminol-based chemiluminescent analyz-
ers, which was reported in the literature,
was not seen in this study with the LMA
3 analyzers. It appears that the 03 traps
which  were supplied by Scintrex  are
effective in eliminating the O3 interference
in the LMA 3 NO2 analyzers. The CS11600
N02 analyzers showed  responses  to
water,  ammonia,  PAN,  nitric  acid, and
nitrous acid. The ammonia response was
not found in a previous post-designation
equivalency testing of the CSI 1600 NO2
analyzer. It is hypothesized that insuffi-
cient test manifold stabilization time was
allowed in the earlier testing. The TECO
43S SO2  analyzer responded to nitric
oxide  and  xylene.  The nitric  oxide
response was greater than what has been
reported by TECO for the model 43A SO2
instrument

Ambient Air Quality
Measurements

  The  field  evaluation was conducted
from April 26 through  May  27, 1988.
During this period, the CS11600 and LMA
3 analyzers produced 483 sets of matched
hourly mean NO2 concentrations. The
TECO  43S  and  TECO 43A  analyzers
produced 463 sets  of matched  hourly
mean  SO2 concentrations during  the
same period. Overall mean concentra-
tions for the field evaluation are given in
Table 4.

Tables 4. Overall Mean Concentrations During
the Field Evaluation
Parameter
NOs,
SO2
Analyzer
CSI A
CSIB '
LMA A
LMAB
TECO A
TECOB
RTI-SO2
Overall Mean
Concentration
13.022 ppb
13.314 ppb
11. 309 ppb
10.512 ppb
1.691 ppb
1.653 ppb
1.398 ppb
The SO2 data are in good agreement with
SO2 data collected with a filter pack at
a nearby EPA National  Dry Deposition
Network site in Research Triaingle Park.
The EPA site measured an overall mean
SO2  concentration of 1.61  ppb for the
period from April 26 through May 24,1988.


Calibration Apparatus

  It was necessary to design and fabri-
cate a  calibration  apparatus  for the
laboratory and field evaluations that is
capable of generating ppb-level SO2 and
NO2 concentrations. In general, the effort
was  successful. The  laboratory evalua-
tion  demonstrated that the apparatus
could generate N02 at low concentrations
via  gas-phase titration.  It  alsio  demon-
strated that  the turbulent mixing loop
permits the rapid concentration changes
needed for response time measurements.
The  daily span  factors from  the field
evaluation demonstrated that the calibra-
tion apparatus has good repeatability on
a day-to-day basis. However, it was not
possible to evaluate the  accuiracy of the
concentrations that the calibration appa-
ratus generated.
  The only negative feature of the cal-
ibration apparatus that was discovered is
the long stabilization time (approximately
2 hours) for the TECO 43S SO2 analyzers.
After an overnight zero air exposure and
the initial change to an 80% full scale
concentration,  the TECO 43S SO2 ana-
lyzer took a long time to go from 95% to
100% of the final, stable  residing. This
phenomenon  was not observed  for
subsequent step concentration changes
during each test day.  It  is  hypothesized
that the long stabilization time is due to
the need to condition the PlrA Teflon™
walls of the calibration apparatus after the
overnight zero air exposure.
  A similar phenomenon was observed
for the LMA 3 NO2 analyzers, but it is not
attributed  to  the calibration apparatus
because it was not seen for the CS11600
NO2 analyzers. It is hypothesized that the
LMA 3  NO2  analyzers'  ozone traps
required conditioning.

Conclusions

  EPA has not established performance
specifications for extrasensitive analyzers
to be used in acid deposition monitoring.
As a result, conclusions concerning the
specific applicability and acceptability of
these analyzers cannot be drawn. Never-
theless,  the results of the laboratory and
field evaluations suggest that Ithe analyz-
ers will perform in a generally acceptable
fashion for  the intended  application.
Several specific areas of the analyzers'
performance that need improvement were
uncovered in the evaluations.

Recommendations

  Performance specifications should be
established for extrasensitive analyzers to
be used in acid deposition monitoring.
Users of monitoring data obtained  with
these extrasensitive analyzers should be
made aware of the performance charac-
teristics of the analyzers (especially with
respect to interference effects).
  To prevent changes in O2 concentration
from affecting analyzer response, calibra-
tion  mixtures for all  three  types of
instruments should be made using dilu-
tion  with air rather than  nitrogen or ,
another dilutant. In addition, the LMA 3
NO2  analyzer is sufficiently sensitive to
O2 concentration that the percentage of,
O2 in synthetic zero air must be closely
specified. Similarly, the CO2 and water
concentrations of calibration mixtures for
the LMA 3 NO2 analyzer should be similar
to those in ambient air to prevent changes
in analyzer  response due  to  differing
concentrations of  CO2 and water.  The
water concentration of calibration mix-
tures for the CSI 1600  NO2  analyzer
should be similar to that in ambient air
to prevent interference effects. Sites using
the  TECO  43S  SO2 analyzer should
monitor NO concentrations to be able to
detect and  possibly correct for NO
interference.
  Sites  using the LMA 3 NO2 analyzer
should backflush the analyzer daily and
change the pump tubing every two weeks
to prevent operational failures. All three
types of instruments should be operated
in buildings with heating and air condi-
tioning, as evidenced by the temporary
change in zero and span values which
occurred when the air conditioner at the
Materials Exposure Site malfunctioned.
  Further investigations  of some of the
instrumental characteristics revealed in
this  laboratory evaluation should be
pursued. In  particular, more  extensive
interference tests  may  be needed for
those compounds which demonstrated
an interference in this study. Additional
work is needed to identify the cause of
the span drift observed in one LMA 3 NO2
analyzer during the field evaluation.

-------

-------
  £ £ Rickman, Jr., A. H. Green, R. S. Wright, and J. E. Sickles, II are with Research
       Triangle Institute, Research Triangle Park, NC 27711.
  Darryl J. von Lehmden is the EPA Project Officer (see below).
  The complete report, entitled "Laboratory and Field Evaluations of Extrasensitive
       Sulfur Dioxide and Nitrogen Dioxide Analyzers for Acid Deposition Monitoring,"
       (Order No. PB 90-201 062/AS; Cost: $23.00, subject to change) will be available
       only from:
           National Technical Information Service
           5285 Port Royal Road
           Springfield, VA 22161
           Telephone: 703-487-4650                                 \
  The EPA Project Officer can be contacted at
           Atmospheric Research and Exposure Assessment Laboratory
           U.S. Environmental Protection Agency
           Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-90/034

-------