2/3-5401
IMPROVEMENT OF OPTICAL EFFICIENCY
OF LUMINESCENCE OF A
FLAME PHOTOMETRIC DETECTOR
CONTRACT NO. EHSD 71-50
PREPARED FOR
ENVIRONMENTAL PROTECTION AGENC
RESEARCH TRIANGLE PARK
DURHAM, NORTH CAROLINA 27711
DECEMBER 1972
FINAL REPORT
BAIRD-ATDMIG
GOVERNMENT SYSTEMS DIVISION
125 MIDDLESEX TURNPIKE, BEDFORD, MASSACHUSETTS O173O
-------
IMPROVEMENT OF OPTICAL EFFICIENCY OF LUMINESCENCE
OF A FLAME PHOTOMETRIC DETECTOR
Contract No. EHSD71-50
Prepared for
Environmental Protection Agency
Research Triangle Park
Durham, North Carolina 27711
December 1 972
Final Report
Approved by:
L/L/ .
Arhur W. Hornig
Director of Research
Government Systems Division
Baird-Atomic, Inc.
125 Middlesex Turnpike
Bedford, Massachusetts 01730
-------
TABLE OF CONTENTS
Section Page
1. INTRODUCTION AND SUMMARY 1-1
2. LABORATORY STUDIES 2-1
2.1 Standard Instrument Evaluation 2-1
2. 2 S Emission Studies 2-1
2.3 Background Emission Studies 2-4
2.4 Choice of Monochromator System 2-5
2.5 Multiple Slit Studies 2-6
2. 6 Background Subtract Mechanism 2-7
2.7 Electronic Modifications - 2-7
2. 8 Laboratory Measurements 2-8
2.8.1 Modes of Operation 2-9
2.8.2 Single Slit Mode 2-9
2.8.3 Multi-Slit Mode 2-10
2.8.4 Multi-Slit Differential Mode 2-10
2. 9 Instrument Packaging and Controls 2-11
2.9.1 Electronics 2-12
2.9.2 Burner, Monochromator and Photomultip __. 2-12
2.9.3 Quartz Plate Drive Mechanism 2-13
2.10 Delivery Demonstration at Research Triangle Park 2-13
3. DISCUSSION AND RECOMMENDATIONS 3-1
3.1 Recommendations 3-2
4. OPERATIONAL PROCEDURES 4-1
4. 1 Operating Controls 4-1
4.2 Operation Start-Up 4-1
4. 3 Recorder Input 4-1
4.4 Calibration 4-1
4. 5 Gas Flow System 4-2
4.6 Maintenance 4-2
4. 7 Differential Mode 4-2
ii
-------
LIST OF ILLUSTRATIONS
Figure No.
1 Vertical Spatial Scan
2 Spatial Distribution of S Emission (Clean Burner)
3 Spatial Distribution of S Emission (Dirty Burner)
4 Spectral Distribution of S. Emission with 2 nm Bandwidth
£
5 Spectral Distribution of S Emission/Background
(Burner Center)
6 Spectral Distribution of S Emission at Different
Concentrations
7 Spectral Distribution of Flame Background Emission
8 Spatial Distribution of Flame Background (310 nm line)
9 S Spectra at Exit Plane of Monochromator for
Different Entrance Slits
10 Amplifier/Demodulator Schematic
11 Spectral Distribution of S Emission with 3. 5 nm
Bandwidth
12 Spectral Distribution of S .Emission with 5 nm
Bandwidth
13 Comparison of Multi-Slits and Single Slits
14 SO Monitoring with Multi-Slits in Differential Mode
15 Modified SO Analyzer (Right Side)
16 Modified SO Analyzer (Left Side)
17 Modified SO Analyzer (Top View)
111
-------
1. INTRODUCTION AND SUMMARY
The principal objective of this contract was to increase the optical
efficiency of the Melpar flame photometric detector for the detection of SO .
Such increased efficiency should lead to higher sensitivity, stability and
freedom from interference. Two routes were explored. First, the spatial
distribution of sulfur and background emissions were plotted to establish
regions of greatest signal/background. Second, multi-slits were constructed
to match the structured sulfur emission and allow background subtract. The
spatial studies of sulfur and background emission demonstrated that the opti-
mum signal/background is obtained below the geometric center of the stan-
dard burner. The distribution suggested use of a monochromator with
horizontal entrance slit rather than a filter. Unfortunately the distribution
is not constant with time, being related to the buildup of deposits in the
burner.
A study of the line emission of sulfur demonstrated that multiple
slits could be constructed which would both increase specificity for sulfur
and increase the usable signal. A comparison of the multiple slits with a
single slit of the same spectral bandpass indicated an increase of about
three in sensitivity.
The differential mode, using the multi-slits, was not so successful
as hoped, for a variety of reasons. First, the quartz plate introduces some
defocussing. Together with the slightly oversized slits employed, this re-
sulted in decreased differential signal.
Defects in the burner and electronics of the standard instrument--
beyond the province of this contract — limited the results to a demonstration
of the successful implementation of the basic principles. Using multi-slits
in the non-differential mode the final s'ensitivity was about an order of -
magnitude greater than the comparison standard instrurrent (or about three
times more sensitivity for SO ).
L*
1-1
-------
2. LABORATORY STUDIES
2. 1 Standard Instrument Evaluation
The basic instrument supplied by EPA was a Melpar Model LL-1100-1B
SO. Analyzer. As received the instrument was very difficult to ignite and
demonstrated very poor sensitivity. The problem was traced to a leaking
sample chamber, resulting in a reduction in the hydrogen/air ratio. While
the problem was temporarily corrected by using a newer, higher quality
O-ring and longer screws, it reoccurred during the entire project.
The close coupling of the pump and the sample/hydrogen sources re-
sulted in very visible "bouncing" of the balls in the gas gauges. A simple
expansion chamber, seven inches long and three inches in diameter, inserted
between the pump and the instrument, increased flow stability and allowed
better setting of flow valves. Subsequent work revealed that even then the
flows could not be accurately reset using the gauges provided.
As the project progressed numerous leaks developed, particularly
around the igniter and flame ionization detector feed-throughs. In each case
the problem was fixed in a temporary manner; however, it is clear that the
basic burner is not reliable.
2. 2 S. Emission Studies
To modify the instrument for spatial studies of the combustion cham-
ber, we were able to reorientate the combustion chamber, heated exhaust
manifold and inlet tubing without shortening any pieces or changing heaters.
With the instrument cover removed, the combustion chamber was positioned
on the left side of the instrument, oriented so it could be viewed by a
monochromator. It was then a simple matter to restore the instrument to
its original condition for comparison.
The burner was mounted on a cross-slide allowing precise positioning
(±0.1mm) in frong of a 1mm aperture; in turn, a quartz lens imaged the
aperture on the entrance slit of a Farrand monochromator. The aperture
2-1
-------
selected a small area of the flame region. Various removable stops near
the lens allowed selection of angular field of view. This arrangement allowed
versatile spatial scanning of the combustion region at intervals of 1mm and
was used for both S. and flame background monitoring.
Using the NBS SO_ permeation tube provided, a concentration of
L*
approximately 1 ppm was used for most studies. First measurements con-
sisted of horizontal and vertical spatial scans across the burner with the
monochromator set at an S emission peak. (At 1 ppm SO. flame background
is no problem. ) Figure 1 shows a typical set of raw data from a raster of
vertical scans over the combustion chamber. The circle in the upper right
hand corner represents the combustion chamber (dimensions X2). The
square protruding from the bottom represents the flame shield; the nib on
the left is the igniter; the line proceeding from the right hand wall represents
the (unused) flame ionization detector. The vertical scans are identified at
the top of the combustion chamber. In the graphs the x-dimension indicates
the vertical distance from bottom to top (XZ).
The data of Figure 1 indicate the variation of S. emission. Of parti-
cular interest are the dips in emission corresponding to the igniter and the
flame ionization detector. Thus, Curve Z (left-most group) shows the effect
of the igniter while the effect of the larger FID element shows up in many
traces. It is evident that the greatest S. emission occurs below the horizon-
L*
tal centerline and very close to the flame shield.
Using data such as in Figure 1 and corresponding horizontal scans,
we had hoped to plot contour graphs of equal emission; however, it was found
that by the time we completed a set of scans we could no longer reproduce
the first scan in amplitude. Thus, the data were unsatisfactory. The non-
repeatability could have been due in part to the short-term stability of our
SO. source, but was mostly due to actual changes within the burner.
The contour graphs of Figure Z were obtained by selecting a particular
emission level and scanning both x and y directions to find all positions for
Z-Z
-------
the given level. In this way the time to complete a given contour was short,
and it was found the contours did form closed loops. Longer term changes
would be reflected in relative error between contour levels.
Contour graphs such as these showed quite unequivocally that there
is a large variation in S. emission over the burner; and if Figure 2 were
representative, it would suggest using a horizontal slit across the two
peaks, or an off-center aperture for use with a filter.
Unfortunately it was found that the pattern of flame emission varies,
within hours, over a wide range, due to turbulence in the flame and build-up
of a deposit on the inner walls of the combustion chamber. After the chamber
has been disassembled and cleaned, the emission pattern again shows a big
change. In Figure 3 we show contours obtained at a slightly later date
which show a large asymmetry in emission. In general, it was found that
the emission pattern depended strongly on the build-up of materials within
the burner.
Other studies included the effect of angular field of view on the emis-
sion pattern and on the changes associated with removal of the FID element
(which is not used). While it is clear that protrusions such as the FID ele-
ment or the igniter do affect the patterns, the variability with buildup of
deposits made it impossible to form a clear picture of the effect of the
protrusions.
In Figure 4 we indicate typical S emission as a function of wave-
length using 2 nm spectral resolution on the monochromator, a 2mm slit
aperture at the face of the burner and approximately 2 ppm SO.. In Figure 5
we illustrate the relative signal and background using 5 nm spectral resolu-
tion and an SO. concentration of approximately 0. 5 ppm. In Figure 6, the
b
spectral resolution is also 5 nm but concentrations vary from approximately
1 to 0. 3 ppm.
The signal appearing at approximately 310 nm in both Figures 5 and
6 is the most prominent flame background, probably due to (OH). Comparing
2-3
-------
the center trace of Figure 6 and Figure 5 (identical spectral resolution and
SO. concentrations), it is evident that the flame peak at 310 nm is much more
L*
pronounced relative to S. in Figure 5 than in Figure 6» Figure 5 was taken
LJ
at the geometrical center of the burner and flame--the region emphasized
by the standard Melpar unit. Figure 6 was taken in a region exhibiting best
S emission. It is quite clear that the signal/background is far from optimum
at the center of the burner.
2. 3 Background Emission Studies
Figures 5 and 6 revealed the principal background signal at 310 nm.
In Figure 7 we show a typical background trace showing the 310 nm signal and
further emissions peaked at about 350 nm and very broadly at about 400 nm.
In this case the spectral bandwidth has been increased to just over 10 nm,
in order to increase sensitivity. From Figures 6 and 7 we find that the ratio
of the highest S peak to the corresponding background is about 800:1 for the
L*
0. 5 ppm SO. concentration. Since the S signal depends on the square of the
SO. concentration, the background is equivalent to approximately 0. 02 ppm
SO . While this is only an estimate, it establishes some idea of background
limitations. This background does not change much when our laboratory air
is filtered through charcoal.
The data of Figure 7 were obtained using a cooled 1P28 photomultiplier
tube. Using an uncooled tube the electronic background interfered with flame
background determination. Thus, for optimum sensitivity the photomultiplier
may have to be cooled, although this requirement may be mitigated by even-
tual optimization of the optical system. In the standard Melpar unit the
cathode of the photomultiplier is actually heated because of close proximity
to the burner chamberl This may well account for the instability and drift
observed during normal operation.
Figure 8 depicts the spatial distribution of the 310 nm background
emission, using a measurement setup1 similar to that used for the S. con-
tours, but with spectral bandwidth just greater than 10 nm. Contrasting this
2-4
-------
set of contours with the S emission contours of Figure 2, it is obvious that
it should be possible to select regions with optimum signal/background ratios.
In Figure 2 the maximum S emission occurs at two points, both about 4mm
L*
below the geometric center of the burner chamber, near the corners of the
flame shield. The flame background peak, on the other hand, is located
above the centerline of the burner.
The original instrument has a 5. 5mm diameter aperture in front of
the burner. This is indicated on both Figure 2 and Figure 8. It is apparent
that it does not include the major sulfur peaks, but is closer to a background
peak. Since the system accepts a divergent beam (about 12°) broader areas
of the flame are actually viewed.
Unfortunately a final signal/background plot (i.e., data of Figure 2
divided by data of Figure 8) cannot be made because the distributions change
within the burner.
The chief background of interest is actually not the 310 nm emission
(which occurs on the short wavelength side of the S. emission) but the 350 nm
peak and the broad peak to longer wavelengths. The 310 nm peak was used
for our measurements because of its greater intensity; however, an indepen-
dent measurement indicated that the 350 nm peak occurs in the same spatial
position as the 310 nm peak.
2. 4 Choice of Monochromator System
As a result of the signal/background studies it was determined that
the chief emission regions for S occurred near the edges of the flame shield,
somewhat below the centerline of the flame chamber; whereas the background
peaked above the centerline of the burner. The optimum viewing region
would then be below the centerline and include both S maximum emission
regions. This could be accomplished using a monochromator with its
entrance slit horizontal and in the plane of the emission maxima. Or it
could be accomplished by viewing only one maximum by means of a lens and
an interference filter. Both systems would allow background subtract;
2-5
-------
however, the filter system would be restricted to one line of S emission
with resulting loss of sensitivity and specificity. The monochromator was
to be preferred for the above reasons plus the versatility of tuning which
might allow application of the same instrument for other materials, e. g, ,
phosphorus.
With the project officer's concurrence it was decided to incorporate
the Farrand monochromator, used in the studies of sections 2.3 and 2.4,
into the final instrument.
2. 5 Multiple Slit Studies
The Farrand single monochromator has a tuning range from 200-700 nm
and a dispersion of 11 nm/mm. With exit slit removed the exit aperture is
approximately 6 rnm, corresponding to a spectral width of approximately
66 nm. From Figure 4 it is evident that the six major S peaks could be
imaged simultaneously if the monochromator is set at 378 nm. The average
distance between the major peaks is 10 nm; however, line shapes are not
identical and it is desirable to tailor a mask to fit the actual emission lines
as observed by a monochromator.
The entrance slit will determine the spectral resolution of the mono-
chromator. In order to determine the optimum setting, pictures were taken
of the S emission lines at the exit plane of the monochromator. Entrance
slits of 2 nm, 3. 5 nm (fabricated specially) and 5 nm were employed for
the data of Figure 9, the monochromator being set at 380 nm for the 2 nm
slit and at 378 nm for the wider slits. SO concentrations were approximately
1 ppm for the 2 nm slit and 0. 75 ppm for the others. From data such as these
it was decided to employ a 3. 5 nm entrance slit to compromise between
higher sensitivity and adequate separation to allow measurements between
the peaks for background subtract. Using separations measured from the
photographs, 3.5 nm slits were etched in brass for use in the exit position
of the monochromator.
2-6
-------
Z. 6 Background Subtract Mechanism
In one mode of operation the S. peaks and valleys are to be imaged
alternately on the multi-slits in order to subtract background and increase
specificity. Several possible methods were investigated including oscilla-
tion of the multi-slits in the monochromator focal plane, rocking the grating
to move the emission image across the multi-slits, and rotation of a quartz
plate in the monochromator to move the image. The last method was
selected for simplicity of mechanical operation and least interference with
monochromator tuning.
A Z mm quartz -plate was mounted in the exit beam of the monochroma-
tor, just before the multi-slit. The monochromator was tuned so that the
six strongest S. lines registered on the multi-slits when the quartz plate
L*
was oriented perpendicular to the beam. As the plate is rotated in the
beam the S lines are displaced until they are not transmitted by the multi-
slits. The quartz plate was driven by a 10 watt 115 volt a. c. synchronous
motor, through a gear train, to oscillate at from 15-18 oscillations per
minute. Oscillation is caused by an eccentric groove in the gear face
which has a pin connected to a guided slide. A pin in the slide connects to
a bell crank on the quartz plate holder. The bell crank has a slot for the
connecting pin to allow for angular adjustment of the quartz plate.
Z. 7 Electronic Modifications
While it was desired to use as much of the original electronics as
possible, it was not feasible to use the original package. The original
amplifier had too long a time constant as compared to the chopping period.
A new electronic package was developed which has a wide-band pre-amplifier
stage, a synchronous demodulator, an output filter with time constants of
0.1, 1. 0 and 10 seconds, and two stages of additional amplification. The
-4 -10
full scale sensitivity is from 10 to 10 amps. Provision was made for
the addition of a log linear amplifier at a later date.
2-7
-------
The first stage of the three-stage amplifier consists of an Analog
Devices 310J operational amplifier with associated feedback resistor to
make a transresistance circuit. The output of this stage passes through a
mechanical chopper synchronized with the optical chopper so that the de-
sired S signal is detected when the quartz plate is perpendicular to the
L*
optical beam. The chopped output is applied to the positive and re gative
inputs of an AD43J Analog Device through a time constant network. The
AD43J is used as a differential detector with unity gain. The differential
output is coupled to a Fairchild 741C amplifier which allows adjusting gain
by a factor of six for calibration purposes. The output of the 741C is con-
nected to a voltage divider, giving both one volt and one millivolt outputs.
These outputs may be used to drive appropriate high impedance recorders
or a voltmeter. Schematics are given in Fig. 10,
During the course of our measurements we often used laboratory
high voltage and low voltage supplies in preference to the supply incorporated
in the Melpar unit. The built-in high voltage supply had a 120 cycle ripple
which contributed to background. However, for final test we returned to
the built-in supplies, since they sufficed to demonstrate the operation of
the modified instrument.
2,8 Laboratory Measurements
During laboratory measurements prior to final assembly, after
final assembly and at a demonstration at Research Triangle Park, many
difficulties occurred with the basic instrument. These served not only to
delay performance checks, but to make detailed signal/noise and sensitivity
measurements difficult. The principal problems included recurring leaks
about gaskets, igniter feedthrough and FID feedthrough, shorting out of
block and exhaust heater control (necessitating a manual system), and
somewhat unstable power supplies. As a result the laboratory measurements
were often made with substitute amplifiers, etc. Nevertheless, they demon-
strated the successful operational mode of the instrument.
2-
-------
2. 8. 1 Modes of Operation
The improved instrument will have several possible modes of opera-
tion. First the instrume nt may be used in a mode similar to the original
instrument, using fixed single slits. These slits may be set to match a parti-
cular line, or may be wide enough to encompass some or all of the lines.
For extreme sensitivity (and least specificity) the bandpass may be greater
than 60 nm and include all six major lines.
The instrument may also be used with the multi-slits in the output to
give better specificity for S emission. The entrance slit may be matched to
the multi-slit (3. 5 nm) for best specificity, or increased to 5 nm for in-
creased sensitivity. This mode is still non-differential. The quartz plate
is adjusted perpendicular to the beam and the oscillator is not used.
Finally, using the multi-slits and a suitable entrance slit, the
oscillator may be used, and the output is the smoothed difference between
measurements at S-, peaks and valleys. This last should offer the highest
specificity against interferents, with some loss of sensitivity.
The above discussion describes the application of the instrument to
sulfur detection. By tuning the monochromator to other wavelengths, and
with suitable selection of slits, the instrument is applicable to other materials,
e.g. phosphorus.
Z.8.2 Single Slit Mode
It is usually best to match input and output slits of the monochromator.
In Fig. 4 we displayed the spectral scan of a 2 ppm concentration of SO
viewed with 2 nm slits. It will be observed that the lines are well resolved,
with minima not far from the base. Fig. 11 contains similar data for 3. 5 nm
slits and a concentration of 0. 5 ppm. The sensitivity is clearly higher
while the resolution remains fairly good. The minima approach the base
less closely than in Fig. 4. Finally, Fig. 12 shows similar data for 5 nm
slits. The overall signal is greater, but there is now a large unresolved
envelope.
2-9
-------
For greatest sensitivity (at the loss of specificity) even larger
slits could be used. Thus using 20 nm slits produces about four times
greater signal than 10 nm slits.
2.8.3 Multi-Slit Mode
Using the line separation data of Fig. 9 and 3. 5 nm spectral width,
multi-slits were constructed for the six largest central lines of S emission.
L*
(Grooves were first milled into an aluminum block and photographed.
Slits were then etched in a nickel-plated copper of 0. 003 inch thickness. )
The results were checked by comparing the throughput using the
3.5 nm entrance slit and multi-slits (with the monochromator set to pass
the six lines), and the throughput using 3. 5 nm single entrance and exit
slits. These results are indicated in Fig. 13. Here the lower trace shows
single-slit output while the upper curve shows a scan through the maximum
for the multi-slit mode. Gain is the same for both traces. The increase
in sensitivity for monitoring six lines over the single largest line (measured
by peak height) is about three, which is somewhat lower than if all lines
were imaged perfectly. The increase is also about three if a peak to
valley measurement is used. We n ote that the minima using multi-slits
are not quite so deep as hoped for, indicating the multi-slits are not
perfectly designed. As will be discussed later, there is also a loss of
image definition caused by the quartz plate.
We have also obtained data for various larger entrance slits used in
conjunction with the multi-slits. The output increases more or less line-
arly with input slit aperture; however, specificity is lost rapidly (as •
measured by minima when tuning with multi-slits in place).
2.8.4 Multi-Slit Differential Mode
In this mode the signal is sampled alternately when the S peaks are
LJ
imaged onto the multi-slits, and when they are shifted approximately 5 nm.
The sampling period is approximately 30° for both peaks and valleys, phasing
being accomplished by mechanical positioning of a reed switch. From the
2-10
-------
data accumulated it appeared best to use a 5 nm fixed entrance slit in com-
bination with the 3. 5 nm multi-slits. (Spectral widths can be translated into
actual dimensions using the dispersion of 11 nm/mm. )
Using filtered sample air, the output is set to zero by adjusting the
reed switch. S emission is then read as before. Filtering is accomplished
after difference measurements. Signal to noise is decreased not only because
of the use of narrow slits but because of the short duty cycle. (Information
is gathered for approximately one-sixth of a cycle. )
Fig. 14 is a tracing of cycling through a 10 ppb SO sample and fil-
- 8
tered laboratory air. The sensitivity of the instrument is set at 10 amperes
full scale and a 10 second time constant is used. The low-frequency noise
is attributable largely to the variation in phasing in the device, although
power supply instability is also evident.
2. 9 Instrument Packaging and Controls
With some rearrangement of components it was possible to utilize
the original instrument case. Side access holes had to be added to incor-
porate the quartz plate drive mechanism and allow slits to be changed. The
original front panel was used, with relabeling of switches to conform to new
functions. The final instrument is pictured in Figures 15-17.
In Fig. 15 we see the instrument with right panel removed, showing
the new electronics sub-chassis as well as the original front panel. Figure
16 shows a view of the left panel with quartz-plate oscillating mechanism and
slit access holes. A metal dust cover for the drive assembly has been re-
moved. Finally, Fig. 17 is a top view showing the layout of principal com-
ponents. The necessary rearrangements of components will be discussed
below.
2-11
-------
2.9.1 Electronics
The new electronic package has the same overall dimensions as the
original. The amplifier is mounted in a shielded subassembly, with three
access holes to adjust amplifier offset and gain. The subassembly is mounted
directly to the rear of the front panel, utilizing the original cut-outs. The
original scales apply to the internal calibration and to all amplifier settings
except the log linear (which is omitted). The original two-position time
constant toggle switch has been replaced with a four-position switch. The
first three positions albw time constant selection of 0.1 seconds, 1. 0 seconds
and 10 seconds. The fourth position allows conversion of the instrument from
the differential mode to a straight-through amplifier. See Fig. 15 for details.
2. 9. 2 Burner, Monochromator and Photomultiplier Detector
The bulk of the monochromator necessitated repositioning of several
components. The monochromator is mounted so that the entrance slit is
horizontal and looking at the lower center region of the burner. The burner
is mounted directly onto the monochromator with an asbestos gasket for
thermal isolation. The exit slit of the monochromator faces upwards,
necessitating the acdition of a front-surface aluminized mirror to couple
the output to the original photomultiplier tube. The latter is now mounted
horizontally to fit into the original enclosure. This part of the layout can
be seen in Fig. 17. Both entrance and exit slits can be replaced through
openings in the left side of the cabinet (Fig. 16). Standard slits of 2, 5, 10
and 20 nm together with a special 3. 5 nm slit can be used in the entrance
position. The exit position can utilize either the special multi-slit or any
of the above.
The monochromator dial indicator has been modified with a fine
tuning adjust located in the center of the calibrated dial to allow changes as
small as 0.1 nm. With the fine adjust in its maximum counterclockwise
position the center wavelength is as indicated on the large dial. The fine
tuning adjust knob can be seen in Fig. 17: it is available only with the top
to left side cover removed.
2-12
-------
The burner position was changed from rear center to front left cen-
ter to allow mounting on the monochromator. Exhaust lines had to be
changed to accommodate the new burner location; however, the location
of the sample control valve did not change, and the same rear panel location
was used for the connection of the exhaust to the pump.
The temperature control was relocated to allow space for the
monochromator and to make adjustment of the temperature convenient.
The sample air and hydrogen inputs were moved from the rear panel to
the front left side, reducing the length of the connecting hose to the flow indicators
to approximately 8 inches. This reduction reduces recovery time with
large sample concentration changes and minimizes contamination of lines.
2. 9. 3 Quartz Plate Drive Mechanism
It was necessary to mount the motor and gears for the oscillating
quartz plate mechanism outside of the original case. This can be seen in
Figs. 16 and 17. The unit is normally covered with a metal dust shield.
The oscillating mechanism can be deactivated by means of a toggle switch
on the rear panel, allowing use of the instrument in the original non-differen-
tial mode.
2.10 Delivery Demonstration at Research Triangle Park
The demonstration at RTP was not so successful as we had hoped due
to the recurrence of perennial problems with the basic instrument. A short
time prior to the demonstration another leak developed in the igniter feed-
through. Previous trouble with the automatic block/exhaust heater forced
use of a manual control. Upon arrival at RTP we found alignment of the
quartz plate, etc. , had been affected by shipment. Further, unknown to us
at the time, a further leak had developed approximately equal in flow to
sample intake.
The basic unit provided by RTP for comparison was an older model
which proved far more stable than the unit originally supplied to us. Initial
2-13
-------
measurements revealed our unit was far down in sensitivity and signal to
noise. After intensive troubleshooting the instrument was demonstrated in
its several modes as discussed earlier. The results of the test may be sum-
marized as follows:
1. The incorporation of a monochromator in the FPD was successful
in concept and execution. The versatility in choice of bandpass, single or
multi-slits, and tunability, was demonstrated.
2. The limitations of the performance of the revised instrument can
be traced to recurrent leak problems in the sampling system and a poor
burner geometry for coupling to the monochromator.
3. After troubleshooting on the spot, the final sensitivity and
signal/noise ratio of the revised instrument were slightly superior to those
of the comparison instrument. Multi-slits in the non-differential mode were
used for this comparison.
4. The differential mode had considerably less sensitivity than the
non-differential.
5. Selectivity comparisons were inconclusive due to long sampling
lines and uncontrolled sampling me thods.
Subsequent to the demonstration the large unknown leak was found
and repaired, increasing sensitivity and signal to noise by about an order
of magnitude.
2-14
-------
3. DISCUSSION AND RECOMMENDATIONS
The principal objective of this contract was to increase the optical
efficiency of the Melpar flame photometric detector for the detection of SO .
C*
Such increased efficiency should lead to higher sensitivity, stability
and freedom from interference. Two routes were explored. First, the
spatial distribution of sulfur and background emission were plotted to deter-
mine regions of greatest signal/background. Second, multi-slits were
constructed to match the structured sulfur emission and allow background
subtract.
An assumption made at the beginning of this work was that the basic
instrument, including electronics and burner, was well designed and that
we would not attempt to improve on it. Our experience has proven this
false, since we have had to patch up innumerable leaks and make do with
an inadequate burner.
The spatial studies of sulfur and background emission demonstrated
that the optimum signal/background is obtained below the geometric center
of the standard burner. The distribution suggested use of a monochromator
with horizontal slit rather than a filter. Unfortunately the distribution
is not constant with time, being related to the buildup of deposits in the
burner.
While the signal/background ratio is increased with the monochromator,
the total light input is decreased considerably, limiting any dramatic in-
crease in sensitivity. The obvious next step is to redesign the burner to
maximize coupling of signal to a monochromator slit.
A study of the line emission of sulfur demonstrated that multiple
slits could be constructed which would both increase specificity for sulfur
and increase the usable signal. A comparison of the multiple slits with a
single slit of same spectral width indicated an increase of about three in
sensitivity. The multi-slits fabricated at Baird-Atomic are not optimum;
however, they demonstrated the correctness of the principle.
3-1
-------
The differential mode, using the multi-slits, was not so successful
as hoped, for a variety of reasons. First, the method chosen to shift the
spectrum across the multi-slits involved some difficulty. The quartz plate
introduces some defocussing. Further, phasing, or balancing of peak
and valley signals, was not well enough controlled using reed switches,
resulting in some instability. Finally, the greater bandwidth required
resulted in increased noise. A further consideration involves choice of
slit widths. For the non-differential application of multi-slits a 5 nm en-
trance slit and 3.5 nm multi-slits were chosen as a good compromise be-
tween high specificity and high throughput. In the differential mode it
would have been better to use narrower slits since peak-valley differences
are quite sensitive to spectral width and precision of registration.
A plague of burner leaks and electronic problems in the basic instru-
ment, particularly in the latter part of the study, made it impossible to
determine the increase in sensitivity afforded by the innovations introduced.
Using a 5 nm entrance slit and a 3. 5 nm multi-slits in the non-differential
mode, the demonstration showed a better signal/noise than the comparison
standard instrument. The test data were not precise, but suggest an im-
provement of about a factor of three. Since this measurement was made
with an unknown leak present, we feel the improvement was probably well
over an order of magnitude.
In conclusion, the work of this contract has successfully demonstrated
the use of a monochromator and multi-slits to increase both the optical
efficiency and specificity for sulfur detection in a standard flame photometric
detector.
3. 1 Re commendations
Based on the results of this program the following recommendations
are made to realize fully the potential of techniques developed:
1. Redesign the burner to optimize coupling between sulfur emission
and a monochromator slit.
3-2
-------
2. Redesign the burner to improve flame stability and eliminate
leaks.
3. Evaluate improvement in signal/noise and establish minimum
detectable level of SO .
4. Investigate applicability of instrument to other materials, e. g. ,
phosphorus, arsenic and selenium.
3-3
-------
4.
OPERATIONAL PROCEDURES
4. 2
4.4
a. H Flow Control
Because of the similarity to the basic Melpar unit, most ope rational
procedures are similar to those for the model LL-1100-1B SO Analyzer.
L*
Reference should be made to the Instruction Manual, especially Section III.
4.1 Operating Controls
Same
b. Sample Air Flow Control Same
c. Power Switch
d. Ignition Switch
e. Temperature Control
f. Range Selector
g. Calibrate Switch
h. Time Constant
i. Buckout
j. Thermocouples
Operation Start-Up
a. Set-up Provisions
b. Operation
4. 3 Recorder Input
Calibration
Same
Same
Same, but relocated
Same, but no LOG position
Same
Changed to rotary switch with time
constants of 0. 5 sec, 1 sec and 10 sec.
Two spare positions available for
other time constants.
Same
Same (but temperature control is
manual using a Variac)
Same, except set RANGE SELECTOR
switch to appropriate linear scale.
All items the same except linear
scale only.
Same, except omit reference to LOG
position.
Final calibration incomplete due to
various possible modes of operation
and difficulties with basic instrument.
4-1
-------
4. 5 Gas Flow System Same (Section IV of Melpar manual)
4. 6 Maintenance Same (Section V of Melpar manual)
4. 7 Differential Mode
The differential mode is in operation when the chopper toggle switch
on the rear of the instrument is in the on position.
To shift from differential mode to straight operational mode, turn
chopper switch off. Remove dust cover on chopper assembly and rotate
chopper drive so that the upper reed switch is in a closed position. The
detector output is then connected directly to the operation amplifier for a
direct output to recorder. The chopper switch in off position closes opera-
tional amplifier positive input to ground.
4-2
-------
FIGURE 1
VERTICAL SPATIAL SCAN
-------
BAIRO ATOMIC
SPATIAL DISTRIBUTION OF S2 EMISSION
(Clean Burner)
Slits: 10-20/1 mm2 aperture
Concentration: ~1.0ppm
Scale: 1 cm = 1 mm
IGNITER
8
-------
BAIRD ATOMIC
IGNITER
SPATIAL DISTRIBUTION OF $2 EMISSION
(Dirty Burner)
Slits: 10-20/1 mm2 aperture
Concentration: ~ 1.0 ppm
Scale: 1 cm = 1 mm
BURNER
SHIELD
-------
BAIRD ATOMIC
SPECTRAL DISTRIBUTION OF S- EMISSION
Slits: 2-2/2 mm col. slit
Concentration: ~2.0ppm
200
300
400 500
WAVELENGTH (nanometers)
600
-------
FIGURE 5
SPECTRAL DISTRIBUTION OF S,
EMISSION/BACKGROUND l
(Burner Center)
Slits: 5-5/2 mm col. slit
Concentration: ~ 0.5 ppm
TOO
WAVELENGTH (nanometers)
-------
FIGURE 6
SPECTRAL DISTRIBUTION OF S2 EMISSION
AT DIFFERENT CONCENTRATIONS
Slits: 5-5/2 mm col. slit
Concentrations: A ~ 1.0 ppm
B ~ 0.5 ppm
C ~ 0.33 ppm
200
300
400 500
WAVELENGTH (nanometers)
600
700
-------
FIGURE 7
SPECTRAL DISTRIBUTION OF FLAME
BACKGROUND EMISSION
Slits: 10-20
700
WAVELENGTH ( nanometers )
-------
BAIRD ATOMIC
SPATIAL DISTRIBUTION OF FLAME
BACKGROUND (310 nm line)
Slits: 10-20/1 mm2 aperture
Sample: - ambient air
Scale: 1 cm = 1 mm
BURNER SHIELD
-------
2 nm entrance slit,
1.0 ppm SO,-, 380 nm center wavelength
3.5 nm entrance slit,
0.75 ppm SO9, 378 nm center wavelength
5 nm entrance slit.
0.75 ppm SCX, 378 nm center wavelength photos enlarged 3X
Figure 9. So spectra at exit plane of monochromator for different entrance slits
-------
RANGE
IOOOM
f 15
1
8UCKOUT IOOM
* _L*
r 1- IMF
PM i
-15
1
4.99K
IOK
IK
IM
IOOK
^. 1 M ,
IOM
^lOOM^
^ iqooM^
— ^t NORM
F 1
-
'
•—
v^ •
*~ 1
IOM
s
s
s
s
—
•—
•—
r5
i
M
OOK
***, 9
OK
K
-w* — i
""^x
BAL^,
\
IOOK
I
fS»-
r*—
)J
OUT
X
1
+15 -15
r -s
IO-«
io-7
io-«
lO-io
x io-'°
IX 10-'
x io-7
i •». io-«
_5
x IQ-*
IK
9.09K
IK
n,
T
.
.0033MF
49.9 K
IMF IOMF
49.9K
f 1 J
-H IMF IOMF
L 1 |
IOOMF| {
1-, '
LI I ,
LJ 1
5IIK
5IIK ~
0033H,-5"
MF 1 TR
HORT
IK
I SEC
MO SEC
CALIBRATE
FIGURE 10. AMPLIFIER/DEMODULATOR SCHEMATIC
-------
FIGURE 11
SPECTRAL DISTRIBUTION OF $ EMISSION
Slits: 3.5 - 3.5/2mm col. slit
Concentration : 0.5 ppm
-------
FIGURE 12
SPECTRAL DISTRIBUTION OF S EMISSION
1 Slits: 5-5/ 2mm col. slit
Concentration : 0.5 ppm
-------
FIGURE 13
COMPARISON OF MULTI-SLITS AND SINGLE SLITS
Slits: A 3.5 - mulH
B 3.5-3.5
-------
FIGURE 14
SO MONITORING WITH MULTI-SLITS IN DIFFERENTIAL MODE
BAIRD-ATDMIC
Slits: 3.5/multi-slits
Concentration: lOppb
f.
-------
Figure 15. Modified SO- Analyzer (Right Side)
-------
Figure 16. Modified SCL Analyzer (Left Side)
-------
Figure 17. Modified SCL Analyzer (Top View)
------- |