July 1977
Environmental Protection Technology Series
                                  SULFUR  DIOXIDE
                              Environmental Sciences Research Laboratory
                                  Office of Research and Development
                                  U.S. Environmental Protection Agency
                             Research Triangle Park, North Carolina  27711


Research reports of the Office of Research and Development. U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface m related fields.
The nine series are:

      1.  Environmental Health  Effects Research
      2 . Environmental Protection Technology
      3.  Ecological Research
      4.  Environmental Monitoring
      5.  Socioeconomic Environmental Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7.  Interagency Energy-Environment Research and Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

This report has  been assigned  to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service. Springfield, Virginia 22161.

                    Authur Hornig
                 Baird-Atomic, Inc.
               125 Middlesex Turnpike
            Bedford, Massachusetts  01730
               Contract No. 68-02-0275
                   Project Officer

                  Ralph Baumgardner
     Atmospheric Chemistry and Physics Division
     Environmental Sciences Research Laboratory
    Research Triangle Park, North Carolina  27711


     This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of
trade names or commercial products constitute endorsement or recommendation
for use.


     A prototype flame photometric detector system (FPD) to measure gaseous
sulfur compounds was fabricated utilizing a previously developed correlation
mask optical system and a new flame housing.  Also, a new burner for the FPD
system was optimized to view the excited molecular sulfur emission.  The
sample/hydrogen intake system was also redesigned to operate under a positive
pressure, resulting in improved flame stability.  The prototype detector
system was equipped with a cooled photomultiplier with special optics to
enhance sensitivity.  Initial tests with the completed system indicated the
capability of measuring sulfur dioxide at the part-per-billion level.  Because
of subsequent problems, an absolute calibration of the system was not carried
     This report was submitted in fulfillment of Contract No. 68-02-0275 by
Baird-Atomic, Inc. under the sponsorship of the U.S. Environmental Protection
Agency.  This report covers a period from 1972 to 1975, and work was completed
as of July, 1975.


Number                                                                  Page

  1   Modified burner head 	  5

  2   Schematic of upper burner region 	  8

  3   Dilution control system 	 11

  4   Respcr.se of second burner to 10 ppb SO- and Dilution Air  	 13

  5   Response of the modified Melpar to 10 ppb S0? and
        dilution air	 14
  6   Response of improved burner to 1 ppb S0~ and dilution air  	 16

  7   Optical schematic of new detection system 	 17

  8   Final calibration 	 	 19

  9   Response at Baird Instrument to S0» in ambient laboratory
        air at RTF 	7	 22

 10   Response of Melpar Instrument to S0_ in ambient laboratory
        air at RTF	 22

 Al   Calibration of hydrogen rotameter	 29

 Al   Index of schematics	 33


Abstract 	ill
Figures	 .vi

   1.   Introduction	1
   2.   Laboratory Studies 	2
           Melpar unit repair 	2
           Sample/Hydrogen intake system 	2
           Burner design	3
           Dilution control system	9
           Electronics 	9
           Comparison with Melpar unit 	10
           Measurement at 1 ppb	12
           Final modifications	12
           Final calibration 	18
           Demonstration at Research Triangle Park 	.20

   3.   Conclusions 	23
   A.   Recommendations	24


   A.   Instrument description and operation	 25

                                 SECTION 1
     The prototype flame photometric detectbr (FPD) for measuring sulfur
dioxide, described in this final report, was developed as a logical sequel
to earlier work (under Contract EHSD 71-50) on improving the optical effi-
ciency of luminescence of an existing FPD.  A new flame housing was designed,
fabricated, coupled to the previously developed optics, and tested.  The new
burner was optimized to view the excited sulfur emission (S») with the. pre-
viously developed correlation mask.  The sample/hydrogen intake system was
also redesigned to operate under positive pressure (rather than on suction),
with resulting improved flame stability.  The FPD was also equipped with a
cooled photomultiplier and special optics to improve stability and enhance
     Laboratory tests demonstrated the capability to monitor SO- at part-
per-billion levels; however, the packaged instrument developed problems that
resulted in reduced sensitivity.  The demonstrated sensitivity showed an
improvement over the previous instrument.  Accurate calibration was not
established due to both unresolved instability and limitations of the labor-
atory calibration system available.  (While the tunability of the instrument
suggests monitoring of other molecular species, only S09 was measured in this
     The final instrument permitted two modes of operation.  In the sampling
mode, approximately 700 liters per minute of sample was drawn through the
instrument, of which approximately 0.4 liter per minute was burned for sulfur
detection.  In the calibration mode, the smaller flow could be introduced
directly into the burner base.

                                 SECTION 2
                            LABORATORY STUDIES
     The demonstration of the modified Melpar Unit at the Environmental Pro-
tection Agency, Research Triangle Park, North Carolina (EPA/RTP), at the end
of the previous program was handicapped by several severe leaks that developed
in the burner, possibly during shipping.  The burner was carefully disassembled,
at which tiire it was found that a leak approximately equal to sample flow was
present.  This leak was eliminated by capping the unused flame ionization
detector with a Teflon cap.  Automatic control of block and exhaust line tem-
perature was also restored, thus bringing the instrument back to its original
sensitivity and operation.  This instrument has the monochromator and multi-
slits as developed in the previous contract.

     The intake system and burner were mounted in a large hood to allow quick
design charges while working with sulfur dioxide (S0_).  For these studies,
relatively high SO™ concentrations were used to allow visual observation of
emission characteristics with changes in parameters.  SO^ was leaked, drop
by drop, through a silicone trap into the dilution system.  Rate of S0~ flow
was estimated by counting drops.
     The intake consisted of a Teflon-lined aluminum tube 3 inches in diameter
and approximately 14 inches in length.  A six-bladed muffin fan drove sample
air through at approximately 30,000 cm /min.  A 2-inch iris was used at the
exhaust end to adjust back pressure and sample flow to the burner; however,
it was later removed as unnecessary, burner intake being controlled by choice
of inlet aperture.
     The inlet to the burner was located approximately 11-3/4 inches from the
fan, the burner being mounted on the intake tube.  Details of the burner will
be described later, but in essence, a small fraction of the total throughput

of the intake was sent through the burner as sample air, the amount being
determined by a baffle just past the intake and the size and shape of the
input channels.
     In summary, the pressure intake is characterized by a large flow of
sample air, of which a small part is used in the burner.  There is probably
little adsorption of sample on the fan blades or intake walls in this con-
figuration, and there is much greater latitude in choice of actual air volume
into the burner.  The proximity of the burner inlet to the intake fan results
in very rapid response to changes in SCL concentration.

First Burner Study
     The first burner was modeled after the Melpar burner, but modified to
fit onto the pressure intake system.  The base of the burner consisted of a
Teflon cylinder 2 inches in diameter and 2 inches high.  The cylindrical
burner head, containing the hydrogen and air inlets, was also made of Teflon.
This head was threaded to fit into the burner base.  Because of the difficulty
of mating Teflon to Teflon, a flashback resulted on ignition, and the burner
head was melted almost immediately.
     A new burner head was milled from stainless steel, identical to the
original Melpar head, but with a larger air inlet to allow sufficient flow
with the pressure intake.  The shield (and mixing chamber) of the original
burner was omitted to allow experimentation with quartz tubes of various
sizes.  A hollow aluminum ring was press-fitted onto the stainless steel
section to allow formation of a gas sheath around the flame.  Thus the burner
head had three concentric sections:  a large central hole for the air sample,
four symmetric hydrogen inlets very close to the central hole, and a ring of
small holes for a sheath gas such as nitrogen.  A pictorial representation
of the burner is given in Figure 1.
     •First experiments were concerned with establishing a stable flame.  The
air and hydrogen flows were adjusted to approximately 200 cm /min as in the
Melpar unit.  Various short cylinders of quartz were now mounted on the burner
head to replace the metal shield of the original Melpar.  Stability was
monitored by observing the 310-nm (OH) line, using the Farrand monochromator

and a 1P28 detector.  The monochromator was positioned so that the centerline
of the entrance slit lay about 12 nm above the burner head.  Thus the mono-
chromator viewed part of the flame itself (normally shielded in the Melpar)
and some of the region above.  Short bursts of S09 were admitted occasion-
                                                 z    *
ally to allow visualization of the region of highest S™ emission.
     In the first study, nine different quartz shields with outside diameters
of 8, 9, and 10.4 mm, and heights of 6, 9, and 12 mm were placed on the
burner head.  It was observed that maximum flame stability was obtained for
the 8-mm-diameter, 9-mm-height quartz shield.  With the admission of S0»,
the blue emission was found to occur at the upper rim of the quartz shield.
This flame was found to be extraordinarily stable, requiring deliberate
smothering to extinguish.  This is in contrast to the Melpar flame, which
required accurate setting of gases and went out under small perturbations.
     In another series of measurements, quartz tubes of 10.4-mm diameter
and heights of 31 and 45 mm were used to determine the effect of greater
shield lengths on the sulfur emission.  First, the gas flows had to be
adjusted to give optimum results.  Sulfur emission occurred at the top of
the 31-mm tube, but was within the 45-mm tube.  While the longer tubes con-
tained the emission, it was found that the flame had lost its great stability.
     Another series of measurements involved cooling with nitrogen.  First,
nitrogen w.as directed against the side of one of the shorter quartz shields
to determine whether visual emission could be induced at a cold spot on the
quartz.  The results were negative.  Nitrogen was then introduced through
the aluminum annulus, using quartz shields in various positions.  When there
was an inncjr quartz shield (i. e., the nitrogen was cooling the shield from
the outside) no increased emission was observed.  When the inner quartz tube
was removed so that the nitrogen formed a sheath about the hydrogen flame,
increased emission was also not found.  In the latter case, the cooling
seemed to cause the flame to lift due to increased velocity of the gases.
Those negative results were surprising in view of other workers results;
however, they are also tentative since the experiments were conducted with
bursts of SO- and depended on visual observations.

          TOP VIEW
                                  SHEATH GAS
         SECTION A-A

Figure 1. Modified burner head

     The important conclusions we made from measurements on this first
burner were at follows:
     1.   It was possible to obtain an extraordinarily stable flame
          using the pressure intake and a modified Melpar burner.
     2.   The region of maximum sulfur emission depended in detail
          on a large number of parameters that were not adequately
          controlled in the first burner.
     3.   Simple gas cooling methods did not produce an obvious
          visula increase in sulfur emission.

Second Burner Study
     The-first burner design was based on the Melpar unit with certain
changes to accomodate a pressure system.  The resulting burner demonstrated
a greater stability than was possible with the Melpar vacuum unit.  The
second burner design is based in a general way on a burner design by Gilbert.
This design is characterized by a water-cooled chimney to enhance sulfur
emission.   Sample air is introduced around a central hydrogen jet, in con-
trast  to the first burner.  A schematic of the burner is given in Figure 2.
     First measurements were made at relatively high SO™ concentrations,
usually exceeding 150 ppb, in order to allow visual estimation of emission
geometry and intensity.  Lower concentrations were detected with a mono-
chromatorphotomultiplier combination.  This latter system was set up so
that the emitting region of the burner (see Figure 2) could be scanned
spectrally or spatially.  For spectral scans, the monochromator slit was
oriented perpendicular to the axis of the burner to give best spatial
resolution.  For higher sensitivity measurement, the monochromator was
rotated 90° to include as large a portion of the burner chimney as possible.
Various spectral bandwidths from 2 to 20 nm could be chosen.  The multiple
slit developed in the first project was also available.
     The first water-cooled jacket was approximately 15 cm long, with the
burner chimney extending several centimeters out the end.  The later, more
compact version, approximately 10 cm long, resulted in no loss of emission
 Gilbert, P. T.  Nonmetals.  In:  Analytical Flame Spectroscopy, R. Mavrodineanu,
 ed., New Fork, Springer-Verlag, 1970, p. 246.

efficiency and was therefore adopted for most measurements.
     Cooling water temperature was varied from 20 to 70°C, with a broad
optimum in the 40 to 50°C region.  With our particular cooling bath, a
minimum flow of 150 ml/min was necessary to prevent heating of the bath.
A flow of 350 ml/min has been adopted as a nominal flow.
     Total gas flow (air plus hydrogen) and the air/hydrogen ratio were
varied independently.  At high total gas flows, the emission extended higher
up the chimney but with no evident increase in brightness.  The actual flame
region was longer, and emission from the (OH) line at 310 nm was scattered
into the viewing region.  At low total flows, the flame became unstable.
Average requirements for the burner of Figure 2 were approximately 300 to
400 ml H~ per minute and 250 to 275 ml sample per minute.
     The burner was constructed to allow relatively easy replacement of the
burner chimney.  This allowed some variation in chimney geometry.  The con-
struction, evident in Figure 2, is an essential part of the design, affecting
the cooling of the reactant flame gases.  Using tubes of 7-mm I.D. and 8.5-mm
O.D., various constrictions were fabricated.  Chimneys with no constriction
resulted in very weak emission.  A constriction with an inside diameter of
several millimeters seemed optimum.  The position of the constriction relative
to the end of the chimney also had an effect, a spacing of 15 mm being near
optimum.  A chimney with a 10-mm I.D. was constructed, but resulted in
slightly decreased emission.
     Spatial scans of the chimney above the constriction revealed that the
sulfur emission remained essentially constant over the region between the
metal end pieces of the water jacket.  Spectral scans indicated that there
was no discernible flame background in the sulfur region from 360 to 420 nm.
     A simple opaque cylinder was fitted around the water-cooled region with
a slit opening along one side.  An aluminum reflector was placed opposite
the slit to enhance emission through the slit.  The monochromator was placed
with its entrance slit about 24 mm from the axis of the chimney and parallel
to the axis of the chimney.  The monochromator was mounted on a double cross-
feed, allowing vertical and horizontal positioning for optimization.  The
f/3.5 aperture of the monochromator was overfilled by the chimney emission,
making distance from the axis uncritical.

                      :.:::..- *&
                             INNER QUARTZ TUBE
                             7-mm I.D., 8.5-mm 0.0.
                                    "0" RING SEAL
                                     "0" RING SEAL
                                  OUTER QUARTZ TUBE
                                 EMISSION AREA
                                 FOR DETECTOR
     Figure 2. Cross section of upper burner region.

     Accurate calibration in the part-per-billion region is difficult.  Our
approach has been to use as simple a system as possible, producing stable
samples in the concentration range of interest without demanding too great
a degree of accuracy.  Using a standard permeation tube with a measured
output of 2.91 yg/min at 22°C, known SO- concentrations were produced using
the dilution system shown in Figure 3.  A flow of 10 liters/min over the
permeation tube produced a concentration of approximately 110 ppb.  Diluting
1 liter/min of this concentration with 10 liters of zero air produced a
concentration of approximately 10 ppb.  Alternately, diluting 0.1 liter/min
of the original concentration with 10.9 liters/min of dry air produced a
concentration of approximately 1 ppb.
     Problems with this system included the following:
     1.   The lengthy tubing resulted in some SO- buildup in time,
          especially after use of high concentrations.
     2.   While the tubing was Teflon, several connectors were of
          glass or polypropylene, causing further buildup.  More
          seriously, two inexpensive acrylic gauges were used in the
          sample line.  This plastic is known to collect SO-.
     3.   Because the controlled temperature bath was in use to supply
          water for chimney cooling, the permeation tube in its Teflon-
          lined outer tube was left at ambient temperatures in the hood.
          Turbulence in the hood could have been responsible for some of
          the long-time fluctuations observed.

     Final electronics for the prototype were to be constructed using those
parts from the Melpar unit that were useful.  For laboratory tests, a
Keighley Model 416 High Speed Picoammeter was used for all measurements.
Occasionally, the last stage of the Melpar modified electronics was used in
conjunction with an x-y recorder to give 5-second time constants.  (The
Keithley is limited in this respect.)  With enhanced flame stability, the
photomultiplier noise becomes of some importance.  After a demonstration
showed the dramatic decrease in dark current possible, an SSR Low Noise

Tube Housing, Model 1150A, was purchased, along with an EMI 9781A.  The dark
current wa:; approximately 6 x 10    amperes, after aging with high voltage in
the darkt
     One goal of this project was to determine the improvement of signal/
noise over the previous design.  At the beginning of the present effort, the
modified Melpar was repaired as discussed in section 2.1.  Based on our
measurements at RTP, we believe that the present modified Melpar, equipped
with 2-nm entrance slits and 3.5-nm multiple exit slits, was slightly more
sensitive than a standard Melpar unit.  We therefore compared the new burner
with the modified Melpar.  The same slits and electronics were used in each
case, except for the photomultiplier.  The Melpar unit was operated in parallel
with the n>aw burner, sampling excess gas from the last mixing stage.
     Figure 4 and 5 show laboratory data on a series of measurements taken
alternately on our second burner and the modified Melpar.  The time scale,
from right to left, was one-half minute per major division.  The detector
was an EMI 9781A in the SSR 1150A low noise housing.  High voltage was 750
volts, supplied by a Baird 312A.  The output of the Keithley 416 was coupled
into the Last two stages of the Melpar revised electronics to allow a
5-second time constant on the recorder.
     In Figure 4, we show the response of the current burner at a nominal
concentration of 10 ppb, using the 0.3 x 10   ampere scale on the electrometer.
The average signal was quite constant over the 6-1/2-minute period, the peak
noise amounting to approximately 10 percent of the average.  The signal from
the dilution air (hopefully SO.-free) is indicated in the second section of
this trace.  Its amplitude was less than 10 percent of that of the 10 ppb
sulfur, except for a rise in the last 2 minutes.
     In the modified Melpar unit on the same current scale, the sulfur
signal was indistinguishable from background, despite the fact that the
phototube in the Melpar (EMI 9526B) has a gain about ten times greater than
the EMI 9781A used with the new burner.  Increasing the gain to the
0.1 x 10   ampere scale (a current gain of thirty), we see in Figure 5 the
results for the same 10 ppb SO- measured by the Melpar.  The signal from
dilution air (not shown) was approximately zero on the chart.  The signal


                         FLOW METER
                  0.5 liters/min

                  I VALVE
                                 Figure 3. Dilution control system.

from the S09 was characterized by a general upward drift over the 24 minutes
of observation.  Superposed were broad oscillations with a period of approx-
imately 5 minutes.  The amplitude of the oscillations was near 50 percent of
the average signal.
     It should be noted that apparent rapid response between zero gas and
sample was :aot real; the recorder was shut off during switching and some time
given to establish equilibrium.  In practice, we noted a relatively rapid
buildup whe.i increasing concentration, but a much slower decrease on reducing
concentration.  We feel that this was due to accumulation in rather long
mixing and sample lines.  That the lines did clear out is indicated in the
low responsa of the dilution air used after sufficient time.
     In contrasting the sulfur signals from Figures 4 and 5, it is clear that
the sensitivity of the new burner was much greater than the older Melpar,
yielding a relatively steady output, whereas the Melpar output was hardly
usable at this gain level.

     An important goal of this project was to demonstrate usable sensitivity
at levels under 1 ppb.  In order to test this, we endeavored to produce
approximately full-scale chart deflection at this concentration, with usable
signal-to-noise ratio.  Data for such a measurement are contained in Figure 6.
Figure 6b is a continuation of Figure 6a.  The setup was similar to the pre-
vious measurement, except that the current scale was 1 x 10   amperes.  Note
the dramatic improvement over the 10-ppb measurement of Figure 4.
     Following the measurements at 1 ppb discussed in the previous section,
numerous refinements were made, including modification of the burner base
to allow better mixing in the "calibrate" mode.  The instrument was then
assembled in the housing with covers for final tests.  The exhaust was
redirected into the sample flow to minimize water buildup.  A final cali-
bration was next attempted.
     The calibration was unseccessful due to unexpected drift in the baseline
signal.  This was attributed to temperature fluctuations in the housing,
giving rise to drift in photomultiplier dark-current.  While a fan helped

                                      1 min
   Figure 4. Response of second burner to 10 ppb S02 and
   dilution air (0.3 x 1(>9 ampere scale).

I    I    I    I     I    I    I    I    I    I    I     I    I    I	I
     Figure 5.  Response of the modified Melpar to 10 ppb SO2 and dilution air (0.1 x 10-10 ampere scale).

to reduce the drift, it was still too large to permit reproducible measure-
ments in the 1-ppb region.  Rather than ship the instrument in this condition,
additional modifications were proposed, which we felt would correct the
problems.  The most important suggestion was to stabilize the temperature
of the photomultiplier through use of a thermostatted housing.
     The original photomultiplier (EMI 9781A) was mounted inside the instru-
ment with not enough clearance to permit installation of an insulated housing.
Therefore it was recommended that the photomultiplier be mounted directly
above the multiple exit slits of the polychromator, necessitating projection
through the top cover.  Since this new geometry demanded new optical coupling,
it was recommended that further improvement be achieved by use of a new
selected low-noise photomultiplier in a thermostatted housing.  This would
not only reduce temperature drifts, but would also further increase signal/
noise, thus enhancing useful sensitivity.  The recommended replacement photo-
multiplier was an EMI 9789QA, mounted in an RFI/2010 cooler.  This photo-
multiplier is unusual in that it has a very small  (10 mm) photocathode and
correspondingly low dark current.  It was proposed that the "frost-free"
window in the housing incorporate quartz lenses to image the multiple exit
slits on the photocathode.  An optical schematic of the new detection system
is given in Figure 7.
     Recommendations incorporating these suggestions, together with request
for further funding to allow installation and test, were made in July 1973.
There followed lengthy consultations and negotiations, during which time the
instrument was placed in storage.  Finally, negotiations that stipulated
that the government would furnish the suggested cooled detector system were
completed.  The latter was received, with special frost-free window, in
June 1974.
     It had been hoped that installation of the new detector system would be
straightforward and that immediate improvement in instrument sensitivity
would be realized.  Following final calibration, the instrument would be
tested and demonstrated at RTF.  Unfortunately, this was not to be the case.
For reasons not understood by us, a decrease in sensitivity was observed.
Since time and funding did not permit a thorough recheck of all portions of
the instrument to locate the problem(s), the final instrument was shipped
and demonstrated at RTF in July 1975.



       VJ	H(
                                                                      1 min
I    I     I    I    I     I
             Figure 6.  Response of improved burner to 1 ppb S02 and dilution air (1 x 1f>9 ampere scale).

                        4-in. FOCAL
                            MULTIPLE SLITS
          Figure 7. Optical schematic of new detection
          system (approximately to scale).

     A great deal of effort was expended, in the brief time available, to
locate the source of the problem.  It was found that it was best to reduce
the high voltage  on the new phototnultiplier to 1100 volts, since spurious
noise spikes appeared at higher voltages.  It was also observed that no
measureablia improvement was noted when the photomultiplier was cooled to
-20°C.  Therefore a setting close to 0°C was adopted.  The temperature of
the cooling water for the quartz chimney was varied from ambient to 70°C,
with the h:lgher temperatures being only slightly better.  The burner was
disassembled and cleaned repeatedly, with no improvement.  The polychromator
was checked, and all optics found clean.  In short, the sensitivity reported
in the previous section was never regained.  Nevertheless, even the reduced
performance: was believed to be a significant improvement over the earlier

     On 27 June 1975, Dr. R. K. Stevens of EPA/RTP visited the Baird-Atomic
laboratory to observe the instrument and arrange for delivery to RTF.  It
was agreed that the instrument would be calibrated for a most sensitive scale
of 10 ppb or less.  Final calibration was carried out by Dr. L. Giering.
(L. Campbell, who had done much of the earlier work, had left for other
     Final calibration was hampered by an imprecise source of calibrated
SO., an unknown (and possibly varying) background of SO,, in the mixing air,
and some erratic behavior of the instrument.  Nevertheless, over a period of
several days just prior to shipping the instrument for demonstration to RTF,
data were taken at nominal concentrations ranging from approximately 6 to
18 ppb.  (Data at higher concentrations were also taken but were discarded
due to an obvious breakdown in the SO™ calibration system.)  While these data
are felt to be of limited accuracy and cover only a restricted concentration
range, they provide information on the final sensitivity of the instrument
and the nature of the dependence of SO. concentration.  The data are plotted
in Figure 8.
     Theory predicts a response that depends on the square of the concentration
of total S02«  If the mixing air for preparing known SO- concentrations with
a permeation tube also contains a measureable background of SO- (or other











iu   10
P   9

3   8

*   7
O         _
      1                2         3      4     5   6   7  8  9 10               20        30

                                     NQMiHAL S02 CONCENTRATION, ppb
                                 Figure 8. Final calibration.

sulfur-containing molecule), then the apparent response to calibrated S0~
departs from quadratic at concentrations approaching that in the mixing
air; at concentrations much less than in the mixing air, the dependence
should be linear.
     The limited data of Figure 8 do not allow too much analysis; however,
several conclusions may be drawn.  The apparent slope lies between 1.5 and
1.8.  The two data points for the lowest"concentrations indicate an increasing
slope at lower concentrations.  This is probably not true, but reflects the
problems with the calibration technique.  The response for the data point was
38 percent of full scale on the 10   amp range.  Assuming a quadratic depen-
dence, this; would predict a full scale reading of just over 3 ppb or the 10
amp scale.

     The instrument was installed and demonstrated at EPA/RTP on 24 and 25
July 1975 by Drs. A. Hornig and L. Giering.  While the instrument survived
shipment very well, other problems arose at the laboratory.  The calibrated
SO- system available was not able to deliver the required 0.5 liter/min at a
pressure sufficient to allow proper use of the calibration mode.  There was,
of course, no calibrated SO- source capable of delivering approximately
700 liters/min for the ambient sampling mode.  Therefore, the only measurements
possible that satisfied the design inputs involved measurement of ambient
laboratory air.
     After uncrating and assembling the instrument, first measurements were
made in comparison with a Melpar unit on ambient laboratory air.  Figure 9
contains a tracing made from data taken during this test.  Note that the
instrument  'zero" is obtained by detuning the monochromator to approximately
200 nm, well out of the sulfur emission region.
     The Melpar unit made use of prototype "zero air" generator, under test
at RTF.  This allowed selection of known concentrations from "zero air" to
many parts per billion SO-.  While response to known sulfur concentrations
seemed reasonable, when ambient laboratory air was monitored the reading was
less than that for "zero air", as indicated in Figure 10.  This anomalous
result could not be explained; however, the Melpar unit could not detect the

sulfur background present in the laboratory.  It should also be noted that
there is a periodic fluctuation in the signal.
     The second day was spent trying to calibrate our instrument using the
calibrated supply.  The "zero air" generator did not operate properly when
attached to the Baird prototype, overheating m ny times and melting connectors.
In comparison with the Melpar unit, the Baird instrument displayed poorer
signal/noise, along with unexplained periodic changes in sensitivity.  It is
unclear what part of this was due to makeshift sampling arrangements.  Some
of the sensitivity variation seemed related to accumulation of water in the

                                          	LABORATORY AIR
                          ZERO (200 nm)	\
Figure 9.  Response of Baird instrument to S02 in ambient laboratory air at RTP.
                                   	ZERO AIR
            Figure 10. Response of Melpar instrument to S02 in
            ambient laboratory air at RTP.

                                  SECTION 3
     This program has produced a novel instrument that differs in several
important ways from its predecessors.  First, the instrument employs a poly-
chromator with correlation mask to detect only major S_ emission lines,
resulting in high specificity for sulfur compounds.  An oscillating mechanism
allows further enhanced specificity in the difference mode by comparing peaks
and valleys of the emission spectrum.  While the instrument was demonstrated
only for sulfur, its tunability permits extension to measurement of phos-
phorous, arsenic, and selenium compounds, and perhaps others.  The re-
designed burner operates under pressure, sampling only a small fraction of
the sample volume.  This results in an exceedingly stable flame, rapid
response to changes- in concentration, and small sample buildup.  The cooled
flame, viewed at some distance from the actual hydrogen flame, results in
high sensitivity for sulfur and low flame background.
     Several unsolved problems remain.  The S0_ calibration system was in-
adequate to produce reproducible standard concentrations in the 1-ppb range.
Proper calibration at this level will require much greater sophistication.
There remains the unsolved problem of deterioration of sensitivity after the
year's storage and reactivation of the instrument.  Because the calibration
mode was added after design of the primary sampling mode, the present instru-
ment does not allow easy interchange between the two and requires awkward use
of gauges.  The difference mode has not been optimized and has electronic and
mechanical problems.  The sample flow rate of 700 liters/min is overly large
and could be substantially reduced, using a smaller manifold and fan.

                                  SECTION 4
     In light of the unexplained deterioration in performance of the instru-
ment as delivered, it is first recommended that the entire instrument be
rebuilt, correcting the problems listed in the previous section, and making
it easier to operate in both calibrate and sampling modes.  Prior to final
redesign, several methods of enhancing sensitivity should be explored, in-
cluding auxilliary optical excitation and use of selective surface sensi-
tizers to increase sulfur emission.   Finally, the use of the instrument for
other molecular species should be explored.

                                 APPENDIX A
     The instrument consists of a main housing with several accessories.  The
main unit houses the sample intake, the burner, detector system and necessary-
electronics to analyze for sulfur dioxide.  A fan in the main housing pulls
a large volume of sample (_700 liters/min) through a 1-inch duct.  A small
fraction of this (0.3 to 0.4 liters/min) is introduced into an enclosed
burner together with hydrogen from an auxiliary tank.  Hydrogen flow is set
by a valve and gauge in the front panel to approximately 200 ml/min.  In the
resulting hydrogen-rich flame, sulfur dioxide is reduced to form excited
molecular sulfur, S_.  The deep blue chemiluminescence is observed above the
visible portion of the flame in a water-cooled quartz column.  A poly-
chromator with multiple exit slits matching S~ band structure sums this
emission, which is amplified by a standard photometer.  The excess hydrogen
and moisture is redirected into the main sample exhaust.  The burner also
has a calibration mode allowing direct introduction of known S0« concen-
     There are two basic modes of operation.  Highest sensitivity is obtained
in the normal mode when the polychromator is tuned to the multiple line
emission of S~.  In the difference mode an internal motor tunes the poly-
             L                                         *
chromator alternately to the peaks and valleys of the S_ emission spectrum,
the output being the difference between peak and valley readings.  This mode
achieves higher selectivity at the cost of sensitivity.  The overall signal
level decreases by about a factor of two.
     The photometer consists of a Melpar amplifier and high voltage supply
modified to permit difference measurements..  A cooled photomultiplier, added
in the last phase of this project, protrudes above the main housing.  Outputs
of 1 millivolt or 1 volt full scale are available for display on a strip chart
recorder (not supplied).


     External cooling water is required for both the thermoelectric cooler and
the quartz column.  For best results, a cooling bath should be used to main-
tain the quartz column at 50°C.  The controller for the cooled photomulti-
plier tube is in a separate housing.  Variation in photomultiplier temper-
ature from -20° to 10°C showed little influence in signal/noise ratio; the
more improtant function of the cooler is to maintain the .photomultiplier at a
constant temperature.
Ambient Airflow System
     An ambient air sample is supplied by a Cyclonair Blower, CHE 1, Rotron,
Inc., Woodstock, New York.  The air intake is approximately 700 liters/min.
A small fraction (0.3 to 0.4 liters/min) of this large air intake, which is
directed at the base of the burner, is sampled through slotted holes in the
base of the burner.  The flow to the burner is regulated in the sample
mixing chairber by opening or closing these holes with the flow valve.  A
more complete description of the burner can be found in A.3.
Calibrated Airflow System
     A calibrated S0_ sample can be introduced into the burner by attaching
the Teflon line in the rear of the instrument to a system that supplies the
desired concentration of SO- at a flow of 0.350 to 0.375 liters/min.
Sample Flow Control
     A lever on the burner assembly allows switching from the calibrated to
the ambient air sample mode.  As the lever is moved from position C (cali-
bration) to position A (ambient), the flow valve moves so as to close the
calibration air inlet and open the ambient air inlet.  Small variations
about position A serve to regulate the airflow in the ambient mode.
Hydrogen Gi.s Flow
     The hydrogen for the burner is supplied through a rotameter in the
instrument to the base of the burner where it is mixed with air.  The flow
of hydrogen should be 0.425 to 0.475 liters/min.  The calibration curve for
the rotameter is given in Figure A.I.

     The burner assembly (see section A.9, drawing number 5409-1000) attaches
to the blower and sits in the main stream of the ambient air intake.  The
bottom base of the burner is a cone-shaped piece with four slots (90° apart)
to allow a small fraction of the large volume of intake air to enter the
burner.  The cone shaped burner base also has four holes (90° apart), set in
a circle concentric with a circle through the centers of the four slots,
approximately 1/2 inch from the edge of the burner base.  These four holes
allow only calibration samples to enter the burner and are not connected to
the ambient air source.  The hydrogen enters the burner through a stainless
steel nozzle that fits into the center of the piece cut with four holes and
four slots.  Hydrogen is not premixed with the sulfur dioxide sample.  A
Teflon seal fits on top of the burner base with a series of holes and slots
cut out to match the burner base top.  A stainless steel plate with four
holes acts as a flow valve as it either partially or fully opens or closes
the four slots or four holes.  A mixing chamber and the base section of the
burner sit on top of the cone-shaped burner base.  The hydrogen nozzle is
held in place by a spring washer and retainer nut.  In assembling the burner,
it is important to make sure the tip of the nozzle is aligned with the line
scribed on the inside of the inside of the burner.  An o-ring and two Teflon
plugs seal and secure the burner to the quartz chimney.  A wire passing
through one of the Teflon plugs is used for flame ignition.
     The cooled quartz, column is designed with a constriction to cool and
enhance the emission of the reactant flame gases.  The constriction is
located approximately 15 mm from the end of the chimney.  The design of the
upper burner region including the quartz column is shown in Figure 2.
     A rectangular block equipped with a water inlet and outlet serves as a
cooling jacket for the quartz column.  A viewing window cut out of the cooling
jacket is mounted directly to the entrance slit of the monochromator.  In
order to remove the water fromed in the reaction with sulfur dioxide and
hydrogen, a tube is attached to the exhaust end of the quartz chimney.  The
tube is directed into the 1-inch duct near its exit, creating a slight vacuum
to air in the removal of water.

     A modified Farrand Monochromator with a single 5-mm entrance slit and a
five-slit exit mask is used.  A Gencon cooled photomultiplier tube housing
equipped with an EMI 9789 QA tube is used for detection.  The "frost-free"
window in the housing incorporates quartz lenses that focus a reduced image
of the exit, slits into the small (10-mm) photocathode.
     Final electronics for the prototype were constructed from various Melpar
parts.  The amplifier is mounted in a shielded sub-assembly with three access
holes to adjust amplifier offset and gain.  The gain settings on the front
panel apply to the internal calibration of the amplifier and to all amplifier
settings, except for the Ion-linear scale, which has been disconnected.  The
original tvo-position time constant toggle switch has been replaced with a
four-position switch.  The first three positions allow time constant selection
of 0.1, l.C, and 10.0 seconds.  The fourth position is open for an additional
time constant selection (utilizing an external capacitor).
     In the differential mode, the S~ peaks and valleys are imaged alterna-
tively on the multislit mask.  In this mode, background is automatically sub-
tracted and specificity is improved.  To accomplish this, a 4-mm-thick quartz
plate is mounted in the exit beam of the monochromator, just before the multi-
slit.  The nronochromator is tuned so that the five strongest S? lines are
registered on the multislit mask when the quartz plate is oriented perpen-
dicular to the beam.  As the plate is rotated in the beam, the sulfur lines
are displaced until they are not transmitted by the multislit mask.
     .A brief description of the front panel indicators and controls is given
Panel .labels                                      Function
     HYDROGEN                         Adjusts the flow of hydrogen gas to the
     POWER                            Delivers 110 V a.c. to the instrument via
                                      a three position switch
                                          Center  -  Apparatus off
                                          Down    -  Electronic component
                                          Up      -  Operation

200           300

   H2 FLOW, cm3
                 Figure A-1.  Calibration of hydrogen rotameter.

Panel labelss








Ignites the hydrogen-air mixture by a
high voltage discharge.
Provides up to 5 x 10   amps buckout

Allows selection of 0.5 sec, 1.0 sec,
and 10 sec time constants.  Two spare
positions are available for additional
external capacitors.

Provides linear^current,amplification
ranges for 10    to 10   amps full

Provides internal signals to allow
periodic calibration of the photo-

Actuates oscillatory quartz plate
mechanism to allow difference measure-
ments .

Provides 1 mV and 1 V full scale sig-
nals for an external recorder.

(No longer used.)

(No longer used.)
The modified Farrand monochromator
has been adjusted so that the five
principal S- emission lines fall on
the multiple exit slit.  Readjustment
is possible by maximizing signal while
retuning the monochromator.  Using a
combination of the large coarse tuning
dial and the center fine tuning screw.
This adjustment must be made with the
cover off.

In order to disengage the differential
mode completely, the chopper drive
must be in the maximum upward position.
This adjustment must be made with the
top cover off.

Panel labels                                          Function

     SAMPLE CONTROL                   A lever on the burner assembly allows
                                      switching from ambient air (position
                                      A) to calibration air (position C).


     Prior to start of ari monitoring, the following should be done:

1.   Connect hydrogen supply line to inlet on left side of panel.

2.   Connect water lines from a constant temperature bath to the burner
     chimney.  Set water temperature at approximately 50°C.

3.   Connect high voltage input and signal output cables from the photo-
     multiplier to the electronics module.

4.   Connect cooling water fro thermoelectric cooler to photomultiplier

5.   Connect the instrument to a 110-volt power source.

6.   Set the range selector switch to the_appropriate linear (amplification
     scale.  For ambient air, this is 10   amps.  The calibrate switch
     should be in the normal position and the time constant/switch set
     at 1 sec.
7.   For calibration air, the sample line must be connected to the back
     panel and the lever on top of the burner must be in position C.


1.   Turn instrument power switch to "down" position to allow the electronic
     components of the system to stabilize.
2.   Dark current check:  Turn range switch to 10   amps and time constant
     switch to 1 sec.  Adjust operational amplifier_offset to read zero on
     recorder or voltmeter,  Turn gain switch to 11   amps and time con-
     stant switch to 10 sec and adjust dark current to zero.  Recheck "0"
     setting in 10   position.

Water Cooling—

Turn on water to cooling jacket for thermoelectric cooler then turn on thermo-
electric cooler.  (Initially, the current reading should be 7 to 8 amps; at
a reading of 1 amp, the tube has reached thermal stability.)  Turn on chimney
cooling water.

1.   Turn nain power switch to "up" position.  (Blower can extinguish flame if
     started after ignition.)
2.   Ambient mode:  With hydrogen off, adjust airflow to 0.375 to 0.400 liter/
     min by moving the lever on top of the burner.  Turn on hydrogen supply
     with needle valve closed.  Slowly open needle valve and press ignition
     switch.  Ignition is confirmed by a persistent meter reading above the
     dark current.  Adjust hydrogen flow to 0.425 liter/min.
3.   Calibration mode:  Switch burner valve to calibrate position (position C).
     Disconnect glass tube from end of burner exhaust to main exhaust manifold.
     Check for zero flow through burner by inserting flow gauge (Lab Crest
     flow meter #448-118).  Open calibration air supply and adjust air flow
     to 0.375 to 0.45 liter/min.  Ignition procedure is the same as the ambient
     mode.  Adjust hydrogen flow to 0.425 liter/min.  Ignition is confirmed by
     a pernistent meter reading above dark current.

1.   Turn off hydrogen supply.
2.   Turn off calibration air, or blower if monitoring ambient air.
3.   Turn off SO- monitor power switch.

4.   Turn off all cooling water.

1.   Airflow in excess of 0.500 liter/min produces high temperature capable
     of fracturing the quartz burner.
2.   Maxiimm measurable flow of hydrogen is 0.500 liter/min (6.5 on the flow
3.   Large injection of SO- will result in build up of SO  in the calibration

     Table A-l indexes the fifteen mechanical and one electrical schematics.
The scheire.tics are tabulated by drawing number and title.


Drawing number                                         Title

5409-0100                               Right side—Low-level SO- monitor

5409-010:.                               Front panel—Low-level SO- monitor
5409-0101!                               Prototype—Low-level SO- monitor
5409-1000                               Burner Assembly
5409-1001                               Burner Cooler Assembly


                       TABLE A-l.  INDEX OF SCHEMATICS
Drawing number                                         Title

5409-1002                               Burner Cooler
5409-1003                               Cap, Burner Cooler
5409-1004                               Mixing Chamber
5409-1005                               Flow Valve
5409-1006                               Retainer Nut
5409-1007                               H2 Nozzel
5409-1008                               Burner Base
5409-1009                               Burner Support Bracket
5409-1010                               Flow Valve Adjustments—Parts and
5409-1011                               Sample Manifold
5409-1012 (a) .                          SC>2 Monitor Wiring Diagram

                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
 1. RI-1PORT NO.
                                                            3. RECIPIENT'S ACCESSION-NO.

                                                            5. REPORT DATE
                                 6. PERFORMING ORGANIZATION CODE
                                                            8. PERFORMING ORGANIZATION REPORT NO,
  Authur Hornig
  Baird-Atomic, Inc.
  125  Middlesex Turnpike
  Bedford,  Massachusetts
                                 10. PROGRAM ELEMENT NO.
                                 1AA010 26AAP (FY-74)

 Environmental Sciences Research Laboratory -RTF, NC
 Office of Research and Development
 U.S.  Environmental Protection Agency
 Research Triangle Park,  NC27711   	
                                 13. TYPE OF REPORT AND PERIOD COVERED
                                 14. SPONSORING AGENCY CODE
             A prototype flame  photometric detector  system (FPD) to measure
        gaseous sulfur compounds was fabricated utilizing a previously developed
        correlaticn mask optical system and a new  flame housing.  Also,  a new
        burner fcr the FPD  system was optimized to view the excited molecular
        sulfur emission.  The sample/hydrogen intake system was also  redesigned
        to operate under a  positive pressure, resulting in improved flame
        stability.  The prototype detector system  was equipped with a cooled
        photomulf.iplier with special optics to enhance sensitivity.   Initial
        tests wit:h the completed system indicated  the capability of measuring
        sulfur dioxide at the part-per-billion level.  Because of subsequent
        problems., an absolute calibration of the system was not carried  out.
                                KEY WORDS AND DOCUMENT ANALYSIS
                    b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
 * Air pollution
 * Sulfur dioxide
 * Instruments
 * Flame photometry
 * Optical  equipment
                                                     2 OF

                    19. SECURITY CLASS (ThisReport)
              21. NO. OF PAGES
                                              20. SECURITY CLASS (Thispage)
                                                                         22. PRICE
EPA Form 2220-1 (9-73)