Coincidence Experiments for Noise Reduction in Scintillation Counting by Hartmut Kallman and Carl A. Accardo

Coincidence Experiments for Noise Reduction in Scintillation Counting by Hartmut Kallmann and Carl A. Accardo

An overview of this paper from a 1998 letter to Dr. Richard Bingham by Dr. Carl Accardo
"The first was the arrival of Dr. Hartmut Kallmann, in 1948 to the Signal Corps Engineering Laboratories. He was a distinguished German scientist who had rejuvenated the idea that high energy radiation (betas, gammas et al) would upon being absorbed in certain materials (called scintillators) produce a detectable quantity of light emission by means of a high-gain photomultiplier. Kallmann was an extremely energetic researcher and while I was associated with him at Evans, we performed basic research on a coincidence technique to extend the lower limit of the detection of the very weakest scintillations masked by the noise of the photomultiplier. Kallmann was the honored guest at the First Scintillation Convention Symposium held at Oak Ridge, Tennessee in 1949 or 1950 (I cannot remember the year). He later accepted an appointment as a full professor of Physics at New York University where he directed a leading laboratory in the fields of organic luminescence, solid state physics, and he trained many graduate students for careers in these disciplines. Dr. George Brucker (still at Ft. Monmouth) and myself obtained graduate degrees at New York University under his mentorship. There is a more complete biography included in the Kallmann obituary published in Physics Today in October, 1978. "
Thanks to Dr. Carl A. Accardo for providing a copy of this paper.


Camp Evans Technical Publication THE REVIEW OF SCIENTIFIC INSTRUMENTS, Vol. 21, No. 1, January, 1950 Page 48 - 51

Coincidence Experiments for Noise Reduction in Scintillation Counting

HARTMUT KALLMANN* AND CARL A. ACCARDO
Signal Corps Engineering Laboratories, Fort Monmouth, New Jersey
(Received August 1, 1949)

A method is described which permits discrimination between noise pulses and pulses of small light flashes.
The arrangement consists of three multipliers facing the three faces of one and the same crystal. One
multiplier acts on the "Y" axis, another on the X axis and the third one on the Z (intensity) axis of an
oscilloscope. Light flashes induce pulses in all multipliers simultaneously, whereas noise-pulses occur at
random and produce triple-coincidence pulses only very rarely . Triple coincidences are easily recorded with
the scope. Pictures are presented illustrating the operation of the arrangement. The original noise being of the
order of three thousand per second was dropped to one noise pulse in three seconds .


THE main difficulty in scintillation work is that when light flashes of small intensity are recorded, the counting rate is limited by the number of noise pulses of comparable size occurring in the multiplier tube itself. These pulses are caused by single electrons released from the photo-cathode by thermal collisions; their multiplication in the tube creates a pulse. It has been found by Morton and Mitchell 1 that these single electron pulses are not of equal size but exhibit a whole distribution of pulse sizes. Thus, it may occur that pulses produced by a single electron are even ten times larger than the average pulse size produced by single electrons. On the other hand, a pulse induced by the simultaneous emission of ten electrons from the photocathode by a light flash undergoes fluctuations, and it may be that its actual pulse height is considerably less than its medium pulse height; therefore, even pulses originating in the simultaneous emission of several electrons cannot be distinguished with certainty from noise pulses. I. DISCRIMINATION BETWEEN NOISE PULSES AND LIGHT FLASHES In this paper, two methods are described which make it possible to distinguish between noise pulses and pulses produced by light flashes from a crystal . For this purpose a crystal was used which emitted its light to three multipliers simultaneously as illustrated in Fig. 1. The	faces of the crystal are chosen so that all light proceeding from them can geometrically enter the multiplier window. The multipliers of Type 931A are operated as shown in Fig. 1 . One multiplier, called the "Y" multiplier, is connected to the Y axis of the scope and causes deflections of the cathode-ray beam in this direction . The second, the X multiplier, is connected to the X axis of the scope and causes deflections in this direction . This deflection may be produced by the pulse itself which is directly fed into the X plates through an amplifier ; then the X deflections are proportional to the pulse intensity (arrangement B) or the pulse may be used to trigger a sweep in this direction giving deflections independent of the pulse intensity (arrangement A). The third, the Z and,r multiplier, is connected to the Z axis amplifier of the scope and regulates the cathode-ray beam intensity according to the size of its pulses. The cathode-ray beam operates simultaneously in all three directions only if all three multipliers produce a pulse at the same time and only then a deflection of the beam in the Y direction is to be seen when using method A. Noise pulses occurring in the multipliers are entirely distributed at random and are independent of each other. If the duration of a pulse is given by T, and if the number of noise pulses per second is described by n, the probability of noise pulses occurring in two multipliers simultaneously is given by nr and that of noise pulses coinciding in all three multipliers is given by .

* Now at Washington Square College of Arts and Science, New York University, New York.
1 G. A. Morton and J . A. Mitchell, Nucleonics 4, 16 (1949) .

**49 NOISE REDUCTION IN SCINTILLATION COUNTING**
From this it follows that the number of accidental triple coincidences or the number of events caused by noise pulses in which all three multipliers operate simultaneously is given by , with nT sufficiently small, as is usually the case. The number of triple coincidences caused by noise pulses is therefore much smaller than the noise number in one single tube. If, however, some exterior radiation induces a light flash in the crystal this will induce pulses in all multipliers at the same time, therefore the number of light flashes counted is not reduced by this device . By this method light flashes of small intensity proceeding from the crystal can easily be counted even if noise pulses are much more frequent than the light flashes. With a pulse duration of 10-6 second and a noise number of 1000 per second, the number of counted noise pulses is reduced by a factor of 10-6. This means that only after 1000 seconds one accidental triple coincidence occurs in the device . Exterior radiation with such a small counting rate could be detected by this device . II. EXPERIMENTAL RESULTS Experiments were carried through according to arrangements A and B and with a small stilbene crystal operating on all three multipliers. The crystal was not very transparent but exhibited considerable light scattering, therefore, the light flashes excited within the crystal did not emit equal light intensities to all multipliers, nevertheless, the pulses of gamma-photons were always indicated by the operation of all three multipliers. Gamma-intensities which were difficult to detect with one multiplier tube since they were covered by noise pulses were easily detected with this triple coincidence arrangement . The counting of pulses was performed photographically . A. Recording of Y-Pulses in Arrangement A With arrangement A, the fluorescent screen of the scope was shielded and only a small strip of the screen parallel to the Y axis was visible, on which the pulses produced by light flashes occur. The film was made to run continuously parallel to the X axis. The oscilloscope used had a frequency limit of about 0.5 microsecond and	a sweep duration of 5 microseconds. According to these limits the duration of the pulses was chosen of the order of half a microsecond. If, for instance, a light flash starts a sweep, a deflection of the cathode-ray beam is seen half a microsecond after the beginning of the sweep since the Y pulses were delayed by half a microsecond to make them appear fully on the scope. With a high noise rate it is possible that an accidental noise coincidence in the Y and Z multiplier will occur during the 5 microsecond duration of a sweep started by a light flash or a noise pulse in the X multiplier. Such an accidental coincidence pulse in the Y and Z multiplier is registered on the scope by a visible deflection of the cathode-ray beam, however, this deflection does not generally occur at the same position on the scope as the gamma-pulses but appears on another position since they are randomly distributed over this time interval . These pulses were canceled by the described shielding of the scope. To begin with the noise pulses of the tube were counted by connecting one multiplier to all three axis and were found to be of the order of 3000 per second. Then this arrangement was adjusted in such a way that all pulses excited by light photons were also recorded. Figure 2 describes a "still picture" of these noise pulses with an exposure time of 1/100 of a sec. Numerous small pulses are hidden by the strong intensity at small deflections. Figure 3 demonstrates a similar experiment with an additional exterior gamma-radiation of about 10,000 absorbed quanta per minute. On the other hand Figs. 4 and 5 show still pictures with the triple coincidence arrangement A with an exposure time of 1/10 of a second. Figure 4, without exterior radiation, is blank since the noise counting rate with the triple coincidence arrangement is very low. Figure 5 shows pulses induced by exterior gamma-radiation . Visual counting and counting with a running film

**H. KALLMAN AND C. A. ACCARDO 50**
showed one noise pulse every 4 or 5 seconds . With a pulse duration of a microsecond and with 3000 noise pulses per second one triple coincidence would have been expected every 25 seconds according to the above described equations . The difference between the observed number of accidental triple coincidence and the calculated one may be due to a difference in the number of noise pulses occurring in the different multipliers . Besides this, the width of the visible strip on the scope was a little wider than the width of the pulses, as can be seen in Figs. 2, 3, and 5. As a result such accidental coincidences could also be seen or recorded which occurred during a time duration somewhat longer than that of the actual pulse. A part of the coincidences observed without exterior gamma-radiation are undoubtedly real light flashes from the crystal caused by contamination of the counting chamber or by cosmicradiation particles. If the triggering of the beam intensity was switched off and if the cathode-ray beam intensity turned on so high that the beam could be permanently seen, 50 pulses per second were approxi- proximately counted. These were accidentally occurring noise pulse coincidences in the X and Y circuits. B. Recording with Arrangement B The investigations according to method B are described in Figs. 6 through 10. Figures 6 and 7 check the experimental arrangement. They give a picture of the screen when only one multiplier was connected to all three axis. In Fig . 7 not only the pulses produced by the exterior radiation but also the noise pulses cause deflection of the cathode-ray beam. Since the amplifiers in the Y and X axis were adjusted to equal amplification the deflection of the pulse gives straight lines in the 45° direction, in these pictures each pulse is recorded by curves with two traces, a bright trace and one not quite so bright . These two curves occur from the rise and the decay of the pulse (the latter part the bright trace.) That they do not coincide is due to a small difference in the amplifiers used. In this picture, Fig . 6, again only noise pulses of larger intensity can be seen clearly, due to strong intensities of the cathode-beam traces at small deflections . Figure 7 describes a picture in which pulses of exterior radiation as well as noise pulses are seen. It is hard to distinguish between noise or light pulses as observed in Figs. 6 and 7. Figure 8 describes a photographic picture with the same arrangement but with a pulse running over the Y axis delayed by half a microsecond. For this reason the pulse does not start in the 45° direction but starts from	the X axis and then exhibits a sharp deflection in the Y direction corresponding to the sharp rise of the pulse, and then the pulse decays not in a 45° direction but in the direction the tangent of which is given by where r is the decay time of the pulses and A describes the time delay impressed on the Y axis. These curves show how easily coincidence or small time differences between pulses can be determined by this method. Figures 9 and 10 describe the same operation with the triple coincidence arrangement B ; in this case in all three directions different multipliers are operating. It is seen from the photographic pictures that the rise curve and the decay curve do not even nearly coincide in this case ; this is due to the fact that the preamplifiers used for this work were not identical but exhibited a little difference in rise times. It is further noted that the decay curve is not represented by a straight line but by an inflected curve and the tangent of the angle of inclination varies along the curve. But it is found that the curves start from the same point ; this means that there is no time delay between the X and Y pulses. The inflection of the curve is completely due to small difference in time constants of the multiplier and preamplifier circuits used. If this difference in time constants is given by A the tangent varies as These pictures, Fig. 9 and Fig . 10, further indicate that the intensity ratio between the Y and the X pulses is not the same for different pulses but varies from pulse to pulse se pro m e intensity of the light flashes proceeding from the crystal to the different multipliers is not the same.This intensity ratio can be determined by connecting the point of largest deflection with the origin. If the same pictures were taken without exterior radiation, they were again completely blank, all noise pulses were eliminated, since noise pulses coinciding in the Y and Z circuit or the X and Z circuit respectively

**51 NOISE REDUCTION IN SCINTILLATION COUNTING**
are generally deflected only in the Y or X direction but not under a definite angle between the Y and X axis. The method developed makes it possible to compare the light emission of a crystal through the different faces of this crystal. III. CONCLUSIONS These methods are not only operable with one single crystal but also with different crystals excited by the same radiation . In this case each multiplier is excited by a different crystal. The noise pulses are eliminated in the same ratio as previously described since only pulses are recorded in which all multipliers operate simultaneously . In this case not only three but a great number of crystals may be used and connected to the Y axis over delay lines with different time delay. Using arrangement A all these multipliers excite deflection in the Y direction which are seen on the screen with different fixed distances on the driven sweep. In this way one photographic picture of the light emission of different crystals energized, for instance, by one event of cosmic radiation can be taken. A similar arrangement can be used for many crystals in	case of arrangement B ; the pulses of different crystals are then fed to the plus and minus Y and plus and minus X axis with equal time delays . In this way the light flashes proceeding from nine crystals can be recorded simultaneously on one picture . The limit of this method is given by the intensity of the light which hits the photo-cathode of the different multipliers. If the light intensity hitting the photocathodes from one light flash is so small that each light flash does not release one electron, then the number of counted light flashes will decrease. Only when the light flashes release at an average at least one electron is there no decrease in the number of counted light flashes. If the probability that one light flash releases one electron is given by P aid if this number is smaller than one, then the number of counted light flashes is diminished by a factor of P2. Nevertheless, this method of triple coincidence will be advantageous as long as the factor P which diminishes the number of counted light flashes is larger than the factor nr which diminishes the number of noise pulses .

Page updated February 17, 2004 page created February 17, 2004

Back to the InfoAge Homepage

Back to the Press Index

Camp Evans

Technical Publication