2422 F. T. Gucker, Jr., C. T. O’Konski, . B. Pickard and J. N. Pitts, Jr. Vol.

69

stants are presented in Table II. A graph of log —

propionic acid in methyl and ethyl alcohol-water

K versus the temperature shows that the values of solutions by Patterson and Felsing4 and in di-
the ionization constants reach a maximum in each oxane-water solutions by Harned and Dedell8
of the solvent mixtures, as well as in water, at were plotted on the same graph and were found to
about 20°; this is shown in Fig. 2. fall very nearly on the same line. Thus, even
Values of the standard free energy, heat content, though the Born equation is not rigorously obeyed,
heat capacity and entropy of the ionization process it appears that the dielectric constant of the
may be calculated by means of the equations medium is more important than the nature of the
A' + D'T + C'T2 added organic solvent in determining the extent
AFf =

of ionization of this weak acid.

AH? = A’ —
C'T2
AC?--2 C'T Summary
AS?
= Z>' —
2CT 1. The ionization constants of propionic acid in
where A', C' and D' are 2.3026 R times A, C, and 5, 10 and 20 weight per cent, isopropyl alcohol
solutions have been determined from 0 to 40° by
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

D, respectively.
A graph of —log K versus the reciprocal of the the use of cells without liquid junction.
dielectric constant was made to test the validity 2. Equations are presented which express the
of the Born7 equation. The curve obtained ionization constants as functions of the tempera-
Downloaded via UNIV OF EDINBURGH on May 9, 2023 at 11:25:37 (UTC).

differed only slightly from a straight line over the ture. From these, the values of the standard
range of dielectric constants involved. The thermodynamic quantities may be calculated.
values obtained for the ionization constant of (8) Harned and Dedell, This Journal, 63, 3308 (1941).
(7) Born, Z. Physik, 1, 45 (1920). Austin, Texas Received May 27, 1947

[Contribution from the Department of Chemistry, Northwestern University]

A Photoelectronic Counter for Colloidal Particles1
By Frank T. Gucker, Jr.,2 Chester T. O’Konski,2 Hugh B. Pickard3 and James N. Pitts, Jr.4
During the Second World War, gas mask filters A dioctyl phthalate (DOP) smoke of about 0.8 µ
were developed to such high state of efficiency
a diameter was chosen for the first experiments.
that they passed only a small fraction of the most Preliminary tests with this smoke in our photo-
penetrating aerosols (about 0.3 micron (µ) diame- electric smoke penetrometer showed that the de-
ter). Recently we have described8 an electronic tectable limit of sensitivity corresponded to the
photometer, designed as a smoke penetrometer light from only ten smoke particles, hence the
for the comparisons of concentrations of very di- counting of individual particles of this size did not
lute aerosols, with a limit of sensitivity of 10_9g./ seem impossible. The change from right-angle
liter. If somewhat coarser test smokes are used, light scattering to the more intense forward-angle
the penetration might be of the order of a few scattering, and the development of a stable high-
particles per minute, which would be beyond the gain pulse amplifier seemed a logical line of attack.
limit of accuracy of this or any apparatus measur- By June 1945 we had developed a photoelectronic
ing steady illumination, since the scattered light apparatus, with an electrical background of 1
would come in pulses. In the summer of 1944 we count per minute or less, which would count indi-
undertook the development of an instrument vidual DOP particles down to 0.6 µ diameter.
which would count individual particles and give We are describing the apparatus here since, in ad-
rapid quantitative tests of the best filters. dition to serving as a supersensitive smoke-filter
(1) This paper is based chiefly upon work done for the Office of penetrometer, it can be adapted to many other
Scientific Research and Development under Contract OEMsr-282 uses in colloidal chemistry, bacteriology, biology
with Northwestern University. The later development of the in- and industry.
strument was carried out for the Army Service Forces under Con-
tracts WA-18-064 CWS-137 and -160 with Northwestern Uni- Apparatus
versity. This paper was presented before the Division of Physical
and Inorganic Chemistry at the 111th meeting of the American The apparatus consists of three parts: (1) a
Chemical Society, in Atlantic City, N. J., April, 1947. cell in which a stream of the very dilute smoke,
(2) Present address: Department of Chemistry, Indiana Uni-
versity, Bloomington, Indiana. passing through an intense light beam, scatters
(3) Present address: Department of Chemistry, University of flashes of light forward upon a photosensitive
Maryland, College Park, Md. cell, (2) a stable pulse amplifier with a maximum
(4) Present address: Department of Chemistry, University of
California at Los Angeles.
voltage gain of about 300,000, and (3) a thyratron
(">) F. T. Gucker, Jr., Hugh B. Pickard, and Chester T. O'Konski,
trigger circuit which actuates a mechanical coun-
This Journal, 69, 429 (1947), and F. T. Gucker, Jr., Electronics, 20, ter. The smoke cell and amplifier are mounted on
107 Í1U47J. a chassis base, 7.6 X 25.4 X 35.6 cm. (3 X 10 X
Oct., 1947 Photoelectronic Counter for Colloidal Particles 2423

14 in.). A cabinet, 30.5 X 50.8 X 30.5 cm. (12 X which fits closely over A' and is slotted at the
20 X 12 in.), houses the power supply and the left end for the stud and knurled nut H which
time-recorder circuit, which may be used to re- fastens it in place. The bulb D is focused by ro-
cord automatically the time for 100 counts. The tation and horizontal adjustment in E, vertical
two units are connected by a 4-conductor shielded motion of F, and longitudinal motion of G.
cable with shielded locking-type microphone con- The optical system is similar to that devised in
nectors on the rear of the chassis bases. 1941 by LaMer and Sinclair5 for their study of
The Smoke Cell Unit.—This unit, made of smokes, carried out under the National Defense
brass with most joints silver-soldered, is shown Research Committee at Columbia University. A
in Fig. 1. conical shadow is cut from the converging light
Vibrations in the optical system may cause beam by a disk of black paper, I, 3.9 cm. in diame-
variations in the background illumination which ter cemented to the outer face of the glass plate C.
give spurious counts. To reduce these back- The light beam, outlined by the dashed lines, gives
ground counts, the optical system was made rigid dark-field illumination of a space about 4 mm. in
and the whole unit was mounted on a heavy base diameter at J.
plate Q. This was bolted to the chassis, but in- The flow system in the cell is specially designed
sulated from mechanical vibrations by a piece of so that each particle must pass through the light
6-mm. felt under each end, and felt washers and beam with no chance for recirculation. The
rubber sleeves around each of the 4 bolts. smoke stream, entering through the smaller of
The smoke cell A is closed at the left end by a two concentric tubes, Ti, machined to a knife edge
carefully machined cap B. The right end is closed at the top, is surrounded by a streaming sheath of
with a glass plate C cemented in place with glyp- air flowing from T2 at the same linear rate of 200
tal (glycerol-phthalic anhydride polymer). The cm. per second. Visual tests with concentrated
projecting sleeve A' is threaded to hold the mount smoke showed that the stream from Ti to the exit
for the 2 aspheric condensing lenses, Li, L2, each tube Ts is well defined and uniformly smaller in
with a diameter of 6.4 cm. and a focal length of cross-section than the intensely illuminated focal
about 5.7 cm. The 6-v., 50-c. p. automobile region J. Each smoke particle is illuminated for
headlight bulb D is mounted rigidly in the split about 0.003 sec., and throws a flash of light into
block, E, which connects its outer terminal to the conical shadow to the left of J. Most of this
ground. A flexible lead soldered to the central light, outlined by the dotted lines, is collected by
lamp terminal is connected to one pole of a large the plano-convex lens L3, of about 25 mm. focal
6-volt storage battery, the other pole of which is length, sealed into B with glyptal cement. This
grounded. Two screws passing through vertical scattered light, falling on the photosensitive cell
slots in the plate F fasten it to the cylinder G K, produces an electrical impulse for each par-
2424 F. T. Gucker, Jr., C. T. O’Konski, . B. Pickard and J. N. Pitts, Jr. Vol. 69

tide. The diaphragm M, placed as near the that the pulses from 0.8 µ DOP particles were only
smoke tubes as possible, shields the lens L3 from about 50% above the amplitude of these back-
most of the light which might be scattered from ground pulses, and many particles were missed.
the surface of the plate C or lens L2. The disk of The setting of the grid bias was extremely critical,
black paper 1' cuts down the stray light reflected since a small decrease gave many background
from the surface of C, and the tube O and dia- counts, and a small increase cut off more particle
phragm N also help to limit the stray light reach- counts. The success of the arrangement was
ing the cell K. The inside surfaces of the lens doubtful.
mount, smoke cell, and tube O, and the baffles M Fortunately, at this time the photoconductive
and N are painted with dull black optical lacquer. "thalofide” cell, with a special thallous sulfide
The vents T4 and T5, connected together with surface, developed by Dr. R. J. Cashman6 and
rubber tubing, and to the filtered air line through associates in the Department of Physics at North-
the bottom of Te, can be used to flush the whole western University, was made available to us by
cell before use. Division 16 of the NDRC. Preliminary tests in-
To insure rigidity of the optical system, the dicated a more favorable signal-to-noise ratio for
right end of the smoke cell is fastened to plate Pi the thalofide cell, which was then incorporated
with three cap screws, one shown in Fig. 1. Pi is into the instrument. The signal-to-noise ratio was
bolted to the bed plate Q, the top of which was increased to 2 or 3 and the unit became more
ground flat. The other end of the cell is fastened stable.
with the hexagonal nut R and washer S to plate The first thalofide cells used in the instrument
P2, forming one end of a rectangular box desic- were especially chosen with a low dark resistance
cated by a small can of silica gel, held in place by to match the 2.5-megohm input impedance of the
a spring clamp U. This box contains the photo- amplifier. Later tests of the input circuit showed
sensitive tube, input resistors and condensers, and that the resistance of the cell most frequently used
type-1603 tube which serves as the first amplifier. was only 1.5 megohms in the dark, and dropped to
This tube is covered with a standard shield, 0.6 megohm under the background illumination
painted black on the inside to prevent the light in the smoke cell. Later we tried two thalofide
of the heater from reaching the cell K. The cells with dark resistances of 7 and 40 megohms
socket for this tube is mounted on a brass plug V, and corresponding resistances of 5 and 18 meg-
holding a bakelite disk W fitted with 4 hollow ohms, respectively, under background illumina-
brass prongs X through which the leads pass to tion. The signal-to-noise ratio did not change ap-
the voltage supply and the next amplifier tube. preciably, but the values of both signal and
A drop of solder seals the end of each prong, gas- noise increased substantially so that only about
kets S, Y, Z are coated with “Seal-Tite” cement, one-half the amplification was required for com-
and the other seams are closed with glyptal ce- parable operation. Therefore, the higher-resist-
ment. A soft rubber bulb from an Orsat gas-an- ance units are preferable although the circuit im-
alysis apparatus, connected to the box and partly pedances are not balanced.
inflated when the lid is screwed down on the gas- In first comparing the photomultiplier and
ket Z, reduces a tendency to "breathing” of humid thalofide cells, we found that the 931-tube photo-
air when the temperature of the box changes. current due to background light was ten times
The desiccated box assures performance of the that due to a DOP particle of 0.8 µ diameter while
amplifier in a very humid atmosphere. The first the ratio was about 800 to 1 in the thalofide cell,
model of our instrument functions satisfactorily probably due to stray light of long wave length to
without desiccation in our air-conditioned lab- which this cell is more sensitive. However, the
oratory. The cell K is surrounded by a light-tight background noise of the thalofide cell was inde-
cylindrical case, with a tube projecting from the pendent of background illumination up to these
side to match O, to which it is connected by a levels, while that of the photomultiplier was a
movable section of tubing telescoping over the function of background illumination.
outside. More recently we have tested both type-931
The Photosensitive Cell.—In seeking a cell and type-931-A photomultipliers operated at
with the necessary high sensitivity and high considerably reduced gain, with 30 to 50 volts per
signal-to-noise ratio we tried several RCA type- stage supplied by a very well regulated variable
931 electron-multiplier phototubes in the first high-voltage supply. The photomultiplier was
model of this instrument. The 931 tube, oper- followed by the amplifier described below, set at a
ated at high gain with 112 volts per stage, was gain of about 200,000. We reduced the back-
followed by an audio-frequency amplifier having a ground light by introducing the disk of optical
maximum gain of 10,000, and a thyratron with black paper I' (Fig. 1) and lining the tube O with
adjustable grid bias. In order to obtain counts optical paper. With this arrangement the signal-
with the smoke particles, the negative grid bias (6) R. J. Cashman (a) OSRD Report No. 5997 “Development of
had to be reduced to a point where the background Stable Thallous Sulfide Photoconductive Cells for Detection of Near
Infra-red Radiation.” (b) Abstract in J. Opt. Soc. Am., 36, 356A
was several counts per minute. With the best 931 (1946). (c) More detailed article in Proc. of the National Elec-
tube available, oscilloscope observations showed tronics Conference, 2, 171 (1946).
Oct., 1947 Photoelectronic Counter for Colloidal Particles 2425

Fig. 2.—Photoelectronic counter circuit.

to-noise ratio was increased to about 2, so that the tive bias on the type-885 thyratron tube. Since
photomultiplier system proved comparable with one side of the power line is grounded, and the
the thalofide, as shown in the section describing amplifier chassis also is connected to a water pipe
comparative tests. or other convenient ground, a pilot lamp is con-
Power Supply.—The development of a stable nected as shown in Fig. 2, to give a red warning
power supply for everything except the photo- light when the cord is plugged into the 115-volt
tube involved considerable experimentation. receptacle with the wrong polarity.
The final circuit, occupying the lower lefthand The Photocircuit and Amplifier.—This circuit,
section marked off by long-dashed lines in Fig. 2, shown above and to the right of the long-dashed
was mounted on a chassis in the bottom of the lines in Fig. 2, is mounted on the chassis with the
cabinet housing the time-recorder circuit. The smoke-cell unit. To reduce the noise, Shallcross
full-wave type-83 mercury-vapor rectifier supplies Akra-ohm 5-megohm wirewound resistors are
two filter systems. The first, consisting of a 2- used in the phototube and 1603-grid circuits. A
section choke-input filter followed by a VR-105 and 1603 tube is used in the first stage because of its
VR-150 in series, supplies the main power output of rigid construction and low microphonics. The
255 volts for the thyratron plate, time-recorder cir- plate voltage is supplied by the VR-105 tube which
cuit, and the second VR stage supplying the 6SJ7 also furnishes about 20 volts for the screen grid,
amplifiers. The second independent filter, con- through the voltage divider shown.
sisting of a 2-section choke-input filter followed The second and third amplifier stages use 6SJ7
by a VR-150 tube, supplies the 1603 tube through tubes, the screen grids and plates of which are sup-
a second VR-105 stage. This arrangement ade- plied with 210 and 105 v., respectively, from the
quately decouples the 1603 tube from the output second VR stage connected to the main power
of the second 6SJ7 and reduces the ripple from the supply. The gain of these tubes is controlled by
rectifier. varying the negative feedback with the 2000-ohm
A 6X5 tube used as a half-wave rectifier is con- potentiometers in the cathode circuits. Cathode
nected directly to the 115-volt a. c. power sup- resistors provide the biasing voltages for all
ply. The rectified voltage is applied to a con- three stages. The 1-megohm input resistors of the
denser-input filter, followed by a VR-150 tube 6SJ7 tubes are shunted with 50-MMFD condens-
connected to a circuit which regulates the nega- ers to decrease the gain at high frequencies and
2426 F. T. Gucker, Jr., C. T. O’Konski, . B. Pickard and J. N. Pitts, Jr. Vol. 69

thus increase the stability of the amplifier. The controls is measured by means of an oscilloscope
amplification at 10,000 cycles per second is about connected between T3 and ground. Knowing the
half that at 10 to 1000. The values of the resistors input resistance R of the oscilloscope and the out-
and shunting condensers in the cathode circuits, put impedance (105 ohms) of the circuit, the volt-
and of the grid resistors and coupling condensers, age amplification factor A is given by the equation
were chosen to give large time constants, thus ex- A =2 X 104(1 + ( 5/!?)) £/e
tending the amplifier range down to 10 c. p. s. with A setting of the gain controls giving an over-all
little attenuation. This allows reproduction of
the low-frequency components of the non-sinu- amplification of 200,000 is satisfactory for opera-
tion of the apparatus.
soidal pulse from the photocell and eliminates the
Thyratron “Trigger” Circuit.—A type-885
pulse multiplication we observed with a more con- thyratron tube is used with a self-quenching
ventional amplifier.
In assembling the amplifier, precautions were plate circuit7 consisting of an 8 MH radio-fre-
taken to reduce pickup and positive feedback quency choke and a 0.25 MFD condenser. The
which could cause ripple, instability or oscilla- grid bias is obtained .from the VR-105 tube,
tions. The phototube and 1603-type amplifier through a circuit arrangement to compensate for
variations in individual tubes of both types. The
are effectively shielded by the desiccated brass
ratio r of the 5K and 500-ohm resistors is measured
box, as shown in Fig. 2. The grid leads of the potentiometrically within 1%. A potentiometer is
other tubes are short and surrounded by woven then connected from T4 to ground, set at a value of
shields, securely grounded to the chassis. The (100/r) v., and balanced by adjusting the 10-K re-
bottom of the: amplifier chassis is covered with a sistor. The voltage drop across the whole 50-K
piece of sheet iron. discriminator potentiometer is then 100 v., and its
Amplifier Test Circuit.—The 120-cycle ripple dial reads directly in volts. The discriminator
across the first condenser of the main power knob is removed and the shaft rotated until the
supply is picked up by means of the 0.05 MFD tube fires. The knob is then replaced, pointing to
condenser and fed through a voltage divider to
0, so that any subsequent setting gives directly,
the binding post in Fig. 2. The ripple voltage within 1 v., the voltage of the minimum pulse re-
e is measured with a calibrated oscilloscope con-
nected between T2 and ground. The test voltage quired to fire the tube. The rest of the circuit as-
sociated with the count-reset switch will be dis-
is then fed into the grid of the 1603 tube, and the cussed under the next section.
output voltage E for various settings of the gain The Time-Recorder Circuit.—This circuit,
shown in Fig. 3, is arranged to measure simul-
taneously the duration of an experiment and
the number of counts. The type S-10 timer,
built by the Standard Electric Time Co. of
Springfield, Massachusetts, operates from a syn-
chronous motor, through a clutch accurate to
0.01 sec. The type-73511 impulse counter, made
by the Central Scientific Co., indicates impulses
up to 100, and then closes an auxiliary switch S
which may be used to actuate a 5-place magnetic
counter operated from the 115-v. a. c. line. The
impulse counters eventually become worn and
inaccurate, hence two are used to give a continu-
ous comparison of their behavior. Since most of
our tests involved a relatively small number of
counts, we developed the time-recorder circuit to
measure automatically the time for 100 counts.
This arrangement gives in each experiment the
same probable error of 0.67/1001·7’ or 6.7%.
The automatic-manual switch is a 4-gang 2-posi-
tion non-shorting-type rotary switch, shown in Fig.
.3 as the 4-pole double-throw switch Si. Momen-
tarily depressing the start button operates the leaf
switches Lx and L2, the function of which is ex-
plained below, and then turns on S2, starting the
timer and connecting the thyratron output to the
counter. S2 consists of 2 type-BZ-RLX micro-
switches, made by the Micro Switch Corp., Free-
port, Illinois, mounted on a 10,000-ohm No.
A23307 niicroswiteh relay made by C. P. Clare &
(7) J. R. Dunning, Rev. Sci. Inst., 6, 387 (1934).
Oct., 1947 Photoblectronic Counter for Colloidal Particles 2427

Co., Chicago, Illinois. In the starting position, veloped by LaMer and Sinclair.8 The uniformity
the switch S in the counter is closed, and the 0.1- and size of the smoke was judged by illuminating
MFD condenser is charged to +175 v. through the it with a beam of parallel light and observing the
10-megohm resistor. When 100 counts are re- spectra scattered at different angles to the incident
corded, S is closed, discharging the condenser beam. The observations were made through a
through the 0.2-megohm resistor. This momen- Polaroid disk arranged to pass light with the elec-
tarily swings the starter anode positive and fires the tric vector perpendicular to the plane of observa-
OA4-G tube. The anode current operates the re- tion, defined by the incident beam and the line of
lay which opens the microswitches, stopping the sight. Clear spectra are only obtained with ho-
timer and disconnecting the counter. The timer mogeneous smokes, and the number of orders ob-
records the time for 100 counts. As the start but- served is practically equal to the radius of the
ton is pressed for the next experiment, Li first is smoke particles, in tenth-microns. Our smokes
opened to extinguish the OA4-G tube, then L2 is were 3 or 4 order, i. e., 0.3 or 0.4 µ in radius. This
closed to reset the timer, and finally S2 is turned method of determining particle size, based on an
on' opening the timer reset circuit and starting the interpretation of Gustav Mie’s theory of light
timer and counter. The 0.1-megohm resistor scattering,9 was developed by LaMer and Sinclair10
across the leads to the time recorder (Fig. 2) pre- in their work for the NDRC.
vents a spurious count when S2 first is closed and When we found that our apparatus counted
the 0.25-MFD condenser becomes charged. DOP smokes as small as 0.3 µ radius, we then car-
The 4-circuit 3-position shorting-type selector ried out some experiments with a smoke of spores
switch S3 is connected with the off position in the of Bacillus globigii (BG), set up with an apparatus
center. The center contact of section C is furnished by the group with which we cooperated
grounded to discharge the 0.1-MFD condenser at Camp Detrick. The aqueous suspension of the
and prevent firing the OA4-G tube when the spores at 1 to 200 X 10s per ml. was sprayed
switch is turned to position 1 or 2. The 10-K re- through a nebulizer11 into a mixing chamber where
sistor limits the discharge current and protects it was mixed with a stream of dry air as shown in
the contacts of S. Fig. 4 to give an aerosol of dry spores. Tests at
When Si is turned for manual operation the Camp Detrick showed that practically all the
hundreds counter circuit is connected to section D of resulting particles contained only single spores.
S3, and the common recorder-switch lead is The spores are ellipsoidal, with a major axis aver-
grounded. The auxiliary hundreds counter is aging about 1.2 µ and a minor axis of about 0.8 µ
actuated each time S is closed. The experiment is as shown by electron photomicrographs. The
started as before by means of the start button and aerosol could be made up to 107 spores £er liter, or
run for any desired length of time. By switching about 5µg./liter, but the usual test smoke con-
to automatic, the timer may be stopped at any tained about 106 spores (0.5 µg) per liter. Al-
even hundred counts. though care was taken to purify the original aque-
When the counting rate is too slow to justify ous suspension, it was found to contain appreci-
waiting for 100 counts, the time for a given num- able foreign particles from the nutrient medium.
ber is noted, and the counter is brought to the 100 The smoke generator and flow line were ar-
mark by throwing the 2-pole 3-posjtion lever-type ranged to furnish smoke to the counter at a pres-
switch of Fig. 2 to the reset position. This con- sure slightly above atmospheric. If the smoke is
nects the thyratron grid to a small relaxation os- sucked through the cell, special care must be
cillator consisting of the 0.25-watt neon glow tube, taken to seal all joints and all connections per-
G-megohm resistor, 0.02-MFD condenser and 1- fectly, since a small percentage of room air will
megohm resistor, shown in Fig. 2. This circuit give a very high background count. Unfiltered
sends about 15 pulses per second to the grid of the laboratory air passed through the cell may con-
thyratron tube, rapidly bringing the counter to tain far more than 1000 particles per liter. This
the 100 mark. The proper thyratron grid bias number cannot be readily counted with the in-
is obtained by adjusting the grounded 50-K re- strument.
sistor to a position which gives satisfactory reset
operation. If the resistance is too high, the thyra- Experimental Methods
tron will not fire; if too low, the grid will not re- Adjustment of the Smoke Cell.—The 50-
gain control after each pulse. c. p. lamp is clamped with the V-shaped filament
The counter may be stopped at any time by in the plane of the optical axis. The cell is re-
turning the lever switch from count to off. The moved from its base, swept out with air, and
connection between the off contact and the VR- adjusted in the dark room.
150 tube maintains a charge on the 0.01-MFD Air at 3 liters per minute (1. p. m.) is passed
condenser and prevents a spurious impulse when (8) Unpublished report, 1941.
the switch is turned to the count position. (9) G. Míe, Ann. Physik, 25, 377 (1908).
Smoke (10) Unpublished report, 1941. V. K. LaMer and Irving Johnson,
Generators.—Preliminary experi- This Journal, 67, 2055 (1945), "Observations on the Angular
ments were made with dioctyl phthalate (DOP) Scattering of Light by Sulfur Sols.”
smokes made with a generator of the type de- (11) Made by the Vapo-Nefrin Co., Upper Darby, Pa.
2428 F. T. Gucker, Jr., C. T. O’Konski, . B. Pickard and J. N. Pitts, Jr. Vol. 69

through T2 (Fig. 1), while 4 1. p. m. is drawn out from the calibration curve of the oscilloscope. We
through T3, Thus 11. p. m. is drawn through Ti, found a normal background level of less than 1
which may be connected through a piece of rubber count in ten seconds above 10 volts, at an ampli-
tubing to a flask filled with cigaret smoke. An fication of 200,000. These pulses are separated
image of the illuminated smoke stream at J is from the particle counts by proper setting of the
thrown upon a thin sheet of paper fastened over discriminator.
the end of the diaphragm support 0. The bulb is The appearance of the high-frequency electrical
adjusted to form a uniform image of the smoke background pulses is very different from the par-
stream, white in the middle and somewhat red ticle pulse of about 0.003 sec. duration, and this
around the outside, in the center of the shadow of helps distinguish electrical background from dust
the diaphragm N. After adjusting the lamp, the particles. When a thalofide cell is used, turn-
tube O is removed and the cell viewed through the ing off the 50-c. p. light only changes the oscillo-
lens L3. Some background illumination will be scope pattern appreciably if the pulses are from
seen, but there should be no bright regions. A dust counts or a defective cell.
bright ring around the edge of the black disk I in-
dicates that its size should be increased. This Tests of the Counter with Thalofide Cell
disk should be just large enough to cut off all Two questions arose in connection with the
illumination of the cell K by the direct beam.
Arrangement of Apparatus.—The electrical functioning of the instrument: (1) Will it give a
true indication of the particulate concentration of
circuits are connected as shown in Figs. 2 and the very dilute aerosols of the type encountered in
3, with all switches turned off. First the filter testing? (2) What is the practical range of
transformer, heaters switch is closed, turning on the instrument in its present form, i. e., without a
the transformer and heaters of the rectifiers and
the thyratron. After about fifteen seconds, when scaling circuit or rate meter?
The first question is hard to answer unequivo-
the filament of the 83-tube has warmed up, the rec-
t ifier-power switch can be closed. Then the battery cally, in the absence of a proved standard of com-
is connected to the 50-c. p. light, and the heaters parison. Obviously particles below a certain size
will be missed because of the rapid decrease of
and amplifier supply switches are turned on. The
10-K resistor across the off position of this last light scattering with size. However, if every par-
ticle in a very dilute aerosol is safely above the
switch prevents the overloading of the first bank critical size, and if the whole cross section of the
of VR tubes whenever the amplifier unit is dis- smoke stream at J is illuminated with equal in-
connected.
In making measurements with the counter, the tensity, no particle should escape the counter. To
determine whether or not any of the spores in the
smoke is introduced at 11. p. m., measured by dif- BG test smoke failed to register in the counter,
ference to avoid using a flow meter in the smoke
line. The ra.te of flow of the air sheath is set at comparisons were made at Camp Detrick between
the counts with our apparatus and the number of
3 1. p. m., and the smoke adjusted to bring the ef-
viable spores collected on a cotton impinger, de-
fluent flow to 4 1. p. m. Both are measured on termined by the number of colonies grown on agar
calibrated flow meters.
Before use, the cell is flushed out with carefully plates incubated twenty-four hours. In these
filtered air at 3 1. p. m. until the dust count has early experiments, we obtained counts exceeding
the number of viable spores, thus suggesting the
fallen to 1 per minute or less. The complete re-
moval of dust in the air sheath is necessary when presence of additional particles presumably in-
cluding some non-viable spores, and also extrane-
measuring background counts ih the instrument. ous matter. This hypothesis was later confirmed
We found that 2 sheets of the best grade service-
canister filter paper, or 2 complete pleated filters by studies of electron microphotographs. Later
fastened into a canister were satisfactory, after experiments showed less counts than viable
air had been passed through them at 3 l. p. m. for spores, and indicated that a certain fraction of the
several hours, or preferably overnight. particles do not register on the counter.
Since the counter and the biological tests meas-
Oscilloscope Tests.·—If the background ex- ure different quantities, filter tests with either sys-
ceeds 1 count per minute, an oscilloscope is con- tem must be made self-consistent. Counts from
nected across the output of the amplifier, with the raw smoke were above the range of our instru-
the thyratron circuit turned off. Careful shield- ment, hence we reduced them to a known fraction
ing and removal of a. c. appliances will reduce by passing the smoke through a filter of measured
the 60-cycle ripple to a few volts with the random transmission, or a smoke-dilution apparatus.
background fluctuations appearing as a heavy Two methods were used to determine the upper
“hash” superposed upon it. The background limit of the present counter, both depending on
noise level can be evaluated by covering the screen comparisons with the photoelectric penetrom-
with a horizontal strip of paper, and raising this eter6 previously developed.
until the number of pips per minute appearing The first method employed a constant filter
above it is reduced satisfactorily. The voltage pad of low penetration. The influent smoke was
corresponding to the height of the strip is read reduced by an initial filter to a point where it
Oct., 1947 Photoelectronic Counter for Colloidal Particles 2429

gave a reading on the penetrometer which could liter, and a background count less than 1 per min-
be measured with an accuracy of 5 or 10%. The ute, the range of the instrument is about 3 to 1200
filter pad was chosen to give an effluent smoke of micro per cent. By decreasing the influent con-
10-20 counts per minute in the counter. In a centration with filters of known penetrations,
series of tests, the influent concentration was in- measurements may be extended to any desired
creased in steps, by changing the initial filter, and filter, but the counter is less convenient than the
measured by the penetrometer current, C in µµ- penetrometer for measurements in the range where
amp. (10-12 amp.), while the effluent particulate the latter can be used.
concentration c was measured with the counter.
The ratio (C/c) is plotted against counting rate c,
and extrapolated to 0, where (C/c)0 represents the
limiting ratio at a very low counting rate. As
the counting rate increases (C/c) increases and the
value of (C/c)/(C/c)t> is the count-rate factor
necessary for the counter at the corresponding
rate. With DOP smokes this method checked
satisfactorily with the second method described
below, but with BG smoke it was unreliable, due
probably to clogging of the filter at the higher
influent concentrations. Later, however, we
developed a smoke-dilution apparatus which
obviated this difficulty.
The second method, which gave consistent re- Fig. 5.—Count-rate curve for BG smoke, showing re-
sults with both smokes, employs a variable filter sults obtained by Sgt. A. Cohen and Cpl. W. Voelker on
pad, made of individual sheets the penetrations of three consecutive days: O, run 1; run 2; run 3.
©,

which are measured at higher concentrations with

the smoke penetrometer, and a dilute influent Comparative Tesjs with Thalofide. and Photo-
smoke, of constant concentration, checked with multiplier Tubes
the penetrometer. The experimental arrange- In the summer of 1946, a counter using a thalo-
ment is shown schematically in Fig. 4. Counts fide cell was built in our Laboratory by Mr. Leo
are made on the effluent smoke from the pads.
E. Farr for the use of Dr. Ronald M. Ferry of the
Counts per minute (equal to counts per liter) c2 Harvard Medical School, At that time we were
divided by the fractional penetration P measured
with the penetrometer, should give a constant experimenting with 931-A tubes in our apparatus,
and we made a careful series of tests to compare
value of the influent counts, cx c2/P. Actually =
the behavior of the two instruments and to ob-
Ci calculated in. this way decreases with counting
tain the count-rate curves, using the smoke pene-
rate c2, due to missing of counts at high rates. trometer readings calibrated in terms of spore
Extrapolation of calculated values of cx to low concentrations. The BG generator was operated
counting rates gives a value which can be used with a constant pressure of 6 lb. per sq. in. on the
to determine the count-rate factor as before.
atomizer, and a constant flow of 25 1. p. m. of mix-
BLOW OFF
ing air, giving a total flow of 301. p. m. A suspen-
sion of BG obtained from Camp Detrick had
FIXED VARIABLE
EXHAUST
been found there to contain 4 X 109 viable spores
SPRAY FROM flter FILTER
ASPIRATOR ^
Mixing per ml. Portions of this were diluted with water
iHAMBER 35 L PER M. 32
L PER M
to give relative concentrations /ie, y8, y4, */2
1

DRY
and 3/4 of the original suspension. It was thus
possible to produce raw smokes in which the rela-
AIR

TO COUNTER
tive concentrations of viable spores, and presum-
TO
I L PER M.
ably of other particulate matter, varied over a
FLOWMETER
3 L PER M.
iPENETROMETER considerable range, while the relative humidity,
which might affect the scattering power of the
Fig. 4.—Flow line for testing counter.
particles, remained constant.
Figure 5 shows the results obtained by Sgt. A. A sample of the raw smoke was passed through
Cohen and Cpl. W. Voelker at Camp Detrick. our penetrometer, in which the photocurrent was
The values of cx calculated from counting rates up measured 10 times or more, at about one-minute
to 500 per minute were averaged, since they intervals over the time required to atomize a
showed no systematic trend. This average, di- measured quantity of suspension. Another por-
vided by Ci at each rate, gives the count-rate fac- tion of the raw smoke passed through a quantita-
tor shown in Fig. 5. Counts up to 1000 per min- tive 3-stage diluting system designed to avoid the
ute can be made with no factor, and the practical selective removal of different-sized particles in a
range of the counter is about 3 to 1200 counts per filter. Tests of the stages individually showed that
minute. With an inlet concentration of 10s per the diluting system reduced a smoke, to 0.030%
2430 F. T. Gucker, Jr., C. T. O’Konski, . B. Pickard and J. N. Pitts, Jr. Vol. 69

of its original concentration. Samples of the The results enabled us to determine the ratio
diluted smoke were passed through the two of counts to spores from the equation
counters, in which the counting rates were meas- counts/spores =
c/CrD
ured about 10 times for each suspension. Both
counters were operated with gains adjusted to
D is the dilution, expressed as a fraction. Substi-
about 3 counts per minute background, and with tuting the values of D 0.00030, r
=
33600 =

discriminators set about 20 v. above the firing counts/spores =

0.099c/C
point. The average value of C/c for the thalofide
From the data we calculated the ratio r be- counter was 0.31, while for the photomultiplier
tween concentration s (spores per liter of aerosol) counter it was 0.21, giving a ratio of counts to
and penetrometer current C in µµ amp. from the viable spores of 0.32 in the first case, and 0.47 in
equation the second. Later readjustment of the lens in the
nS/FCt
r =
s/C —
thalofide counter to focus more of the red on the
where is the number of ml. of suspension, con-
n
cell increased the counting rate by a factor of
taining 5 spores/ml., atomized in t minutes, F is nearly two, giving a value of about 0.6 count per
the total smoke flow through the generator, in viable spore.
These results indicate that the present appara-
1. p. m., and C is the average current. We deter- tus does not respond to every particle in the
mined Ct by graphical integration under the curve smoke. This fact was obscured by a numerical
of C versus i. error in the calculations outlined above, which
The results given in Table I show only random was not discovered until the article was in proof.
deviations from the average value of r, over a 16- At the present time (September, 1947) a small
fold range of smoke concentration, indicating that optical probe has been developed, and we have
the penetrometer readings and dilution factor are made a preliminary survey of the 4-mm. field
self-consistent over this range. illuminating the smoke stream at J. This field
is much less uniform than it had appeared to our
Table I
earlier visual observations. Probably a fraction
Penetrometer Calibration in Spores per Liter per of the particles are not illuminated strongly
Micromicroampere
enough to register on the cell. We find that a
Deviation
r
concentrated arc lamp12 and a pair of achromatic
“

10-· n Cl nS/FCT from r, %

lenses give a much more uniform illumination.
.5

0.25 2.0 444 37600 + 12% It is hoped that this new optical system will
.25 2.0 575 28500 -15 eliminate the counting error, and also will reduce
.5 2.0 862 38700 + 15 the stray light.
1.0 2.0 2190 30400 -
9.5
2.0 2.0 4190 31800 -
5.4 Uses of the Particle Counter
3.0 2.0 5760 34700 + 3.3 Beside its use as a penetrometer to test the best
4.0 2.0 8080 33000 -
1.7 filters which can be produced, the counter has a
4.0 1.0 3960 33800 + 0.6 number of other applications. One of these in-
.
spores X 1 X 1
µµ amp. 7 =
33600 d= 7.8 struments is now in use by a group at Harvard
University, where its utility is being appraised in
The data necessary to calculate the count-rate a study of bacterial aerosols. It may have wide
factor are given in the first 5 columns of Table II. applications in bacteriology. In testing contami-
Table II
Calculation of Count-rate Factors for BG Smoke in Counters Using Thalofide and Photomultiplier Cells
,---Av. counts per liter-» ,--Av. C/c-* /--Count-rate factor*»-
Av. C 931-A Thalofide 931-A Thalofide 931-A Thalofide
(µµ .) C-l» C-2 C-l C-2 C-l C-2 C-l C-2 C-l C-2 C-l C-2
23.5 110 106 69 78 0.21 0.22 0.36 0.31 1.00 1.09 1.09 1.07
42.8 204 202 .21 147.21 .30 158 .28 1.00 0.95 0.91 0.97
94.8 449 340 432 .21 293.22 .33 .28 1.00 1.00 1.00
.97
194 718 576 773 .27 537.27 .36 .33 1.28 1.23 1.14
1.09
308 987 8321039 .31 711.30 .43 .37 1.47 1.36 1.28
1.30
377 1079 9891180 35 852.32 .44 .38 . 1.67 1.46 1.31
1.33
“
C-l and C-2 refer to counters 1 and 2. 6 Based on limiting values of 0.21, 0.22, 0.33, and 0.29, respectively, for C/c
obtained by averaging the results at the 3 lowest concentrations.

The first three values of C/c were found to be nation of a system where sterile conditions are
constant for both counters, hence the instruments desired, the dust in a trace of room air will show
are self-consistent up to about 350 counts per up instantly in the counter, whereas the accom-
minute. The average of these 3 values in each panying bacterial contamination will not appear
column is used in calculating the count-rate fac- until hours later in the ordinary plate counts.
tors shown ir. the last 4 columns. (12) Made by the Western Union Telegraph Company.
Oct., 1947 Photoelectronic Counter for Colloidal Particles 2431

The counter could also be adapted to liquid sys- visual observation or a thyratron circuit to operate
tems and used in making counts of suspended cells a mechanical counter. An electrical discriminator
of various kinds, which could be distinguished by limits the counting to particles above a chosen
different settings of the discriminator if they dif- size, so that the apparatus serves to determine
fered enough in their light-scattering properties. relative size distribution. Guyton’s counter was
The 931-A tubes, with which we are now experi- not sensitive to particles less than about 2.5 µ in
menting, have a much better high-frequency re- diameter. He stated that this limitation was
sponse than the thalofide cell. If we can refine the due to the amplifying circuit, which he con-
optical system and reduce stray light greatly, the sidered capable of improvement so that it
background noise of the 931-A tube can be cut be- would respond to particles of 1 or even 0.3 µ
low that of the thalofide cell and the counting diameter. The simplicity of the pulse pickup in
range extended to smaller particles. We are also Guyton’s counter should encourage further study
working to determine the size of smoke particles to determine the theory of its operation and its
individually and rapidly by means of these tech- practical limitations. Only after further study
niques. If uniform illumination is achieved, the of the photoelectronic counter also will it be
size of the light pulse will characterize a spherical possible to compare the advantages and disad-
smoke particle. Otherwise, simultaneous deter- vantages of the two systems.
mination of the pulses sent at two different angles Acknowledgments.—It is a pleasure to ac-
to the light beam, or of the components of the knowledge our indebtedness to Mr. Norman
pulse polarized perpendicular and parallel to the Nachtrieb for the construction of the original
plane of observation, will allow a calculation of smoke cell unit, and to Mr. John E. Kamper for
the particle size from the Mie9 theory of scatter- the construction of the later model, to Mr.
ing. A rapid method of determining particle-size Wendell Weber for assistance in some of the tests
distribution may answer a number of puzzling of the instruments, and to Mr. Elmer Engquist
questions about the difference in the filtration of and Miss Betty Hengeveld for help in the prepara-
smokes which appear to have the same average tion of the drawings.
particle size.
Summary
Comparison with Guyton’s Particle Counter We have described the construction and opera-
Apparently the only other automatic particle tion of a photoelectronic counter, utilizing the
counter comparable with the present apparatus flashes of light scattered by colloidal particles
was developed by Arthur C. Guyton13 at Camp traversing a small space under brilliant dark-field
Detrick between June and October, 1945. The illumination. The stray background count may
principle of operation is entirely different from be made less than 1 per minute. The device was
ours. In his apparatus, the test smoke is forced tested with uniform dioctyl phthalate aerosols
through a fine jet at high velocity, to impinge of diameter down to 0.6 µ (10-13 g. per particle)
upon a metal collector placed very close to the jet. and with aerosols of ellipsoidal spores of Bacillus
A solid dielectric smoke particle transmits to the globigii averaging about 0.8 X 1.2 µ (5 X 10-13 g.
collector an electrical pulse of about 50 micro- per particle). The practical range of the counter,
seconds duration, the voltage of which is propor- without a scaling circuit, is from 3 to 1200 counts
tional to the square of the diameter of the particle per minute. At present, a certain fraction of the
Just how the pulse is produced is not clearly particles is missed, apparently because the smoke
understood, but apparently it is due to static stream is not uniformly illuminated.
electricity generated by friction against the col- Tests of respirator filters can be made with a
lector. .Solid conducting particles and aqueous sensitivity of 10~8% using concentrated aerosols
drops produce weak pulses. However, charging of dioctyl phthalate which may contain up to 1010
the collector increases these pulses in direct pro- particles per liter.
portion to the potential applied. The original The principle of the apparatus should have wide
pulses are put through a 4-stage amplifier with a application in colloidal chemistry, bacteriology,
gain of 100,000, followed by an oscilloscope for biology and industry.
(13) A. C. Guyton, J. In,!. Hog. Toxicol., 28, 133 (10-40). Evanston, Illinois Received March 20, 1947