Plane-Parallel Spark Counters

In recent years, spark counters with plane-parallel electrodes have come into use 2—9.

Interest in them is due mainly to the fact that, because of the small delay of the discharge relative to the moment at which the ionizing particle passes, and because of the rapid development of the discharge, they are suitable for measuring very small time intervals (\(\sim 10^{-9}\) sec). In ordinary cylindrical Geiger–Müller counters, the discharge delay is determined by the drift time of the primary electrons to the counter wire, where the electric-field gradient is large and where the impact ionization leading to the discharge begins. The mean delay time in such counters is \(\sim 10^{-7}\)—\(10^{-8}\) sec.

In spark counters, the electric-field gradient at any point of the working volume is sufficient for the onset of impact ionization and the formation of an avalanche. Impact ionization and photoionization form a streamer—a conducting channel along which spark breakdown occurs. The formation time of a streamer is very short; with a distance between the electrodes of \(2.5\) mm, it is \(\sim 10^{-9}\) sec\(^1\).

The amplitude of the pulses of a spark counter reaches several hundred volts; therefore they are easily recorded without additional amplification. It should be noted that the spark discharge developing along the streamer channel is visually observable and makes it possible to determine the point at which the particle passed through the counter.

To quench the discharge in the counter, quenching electronic circuits are usually used 3, 4, which remove the voltage from the counter for a certain time \(T_0\) *).

*) A quenching resistance may be used for this purpose 5.

Usually \(T_0 \simeq 0.01—0.05\) sec. Such a comparatively long dead time may, in some cases, limit their use for recording high intensities. In recent works\(^9\) it has been possible to reduce this time to \(5\cdot 10^{-3}—10^{-4}\) sec.

Fig. 1. Plateau for a counter filled with argon and oxygen. Time \(T_0 = 0.05\) sec.

In \(^8\) a counter is described with two flat copper electrodes of rectangular shape with an area of \(35\text{ cm}^2\); the distance between the electrodes is \(2.5\) mm. The counter was filled with a mixture of oxygen (pressure 6 mm Hg) and argon to a total pressure of \(1/2\) atm. Its counting characteristic is presented in Fig. 1. Let us note that the slope of the plateau depends on \(T_0\); when the dead time \(T_0\) is decreased, the slope increases.

An important characteristic of the spark counter is the magnitude of the delay time of the pulse relative to the moment of passage of the charged particle. It is difficult to measure this time in a single counter; therefore the relative delay of pulses arising when one cosmic particle passes through two counters placed directly one above the other was measured. Measurement of the delay times was carried out with the aid of the so-called Nedermeier chronograph, substantially improved by the author\(^2\).

The work presents histograms of the distribution of delays for several fixed values of the overvoltage on the counters (\(\Delta V = 250\) V, 500 V, 750 V and 900 V). It turned out that the widths of these distributions depend substantially on the magnitude of the overvoltage. If each distribution is characterized by such a value \(\Delta T\) that half of the observed delays do not exceed \(\Delta T\), then at \(\Delta V = 250\) V, \(\Delta T \simeq 17\cdot 10^{-9}\) sec, while at \(\Delta V = 900\) V it is considerably smaller and equal to \(\Delta T \simeq 5\cdot 10^{-9}\) sec.

A more careful investigation of pulse delays in spark counters was carried out in \(4—6\).

Fig. 2. Distribution of delays in a spark counter at overvoltages equal to 300 V and 1000 V.

In \(^4\) counters were used with circular electrodes made of thin copper foil, positioned at a distance of 4 mm; the relative delay of pulses in two counters was investigated during the passage of \(\beta\)-rays from \(P^{32}\). The efficiency of the counter was 98%.

The typical distributions obtained for the delays for overvoltages on the counters equal to 300 V and 1000 V are shown in Fig. 2. Here the abscissa gives the magnitude of the delay, and the ordinate gives the number of recorded events. The half-widths of these curves are equal to \(6\cdot 10^{-9}\) sec and \(18\cdot 10^{-9}\) sec for overvoltages of 1000 V and 300 V, respectively, which agrees with the result\({}^{3}\).

The same authors in\({}^{5}\) give oscillograms of the pulses of a spark counter; an estimate of the duration of the pulse front led to a value \(\sim 3\cdot 10^{-9}\) sec.

In\({}^{6}\) delays were investigated in counters with electrodes of rectangular shape. In a glass cylinder 55 mm in diameter, two flat copper electrodes of size \(11.5 \times 4.5\) cm were placed. The distance between

Fig. 3. Histogram of delays in a spark counter.

the electrodes was 2.5 mm. The counter was filled with argon and saturated alcohol vapor to a total pressure of 60 cm Hg. The plateau value of the counter is about 1000 V, the plateau slope is 4% per 100 V, the efficiency is \(\sim 98\%\); the service life of the counter with continuous registration of cosmic radiation at sea level is more than 3 months, and after washing and refilling the counter is again suitable for operation.

The author, as in\({}^{3}\), measured the relative delay of pulses in two counters placed one above the other at a distance of 6 cm, but instead of a chronograph used a special fast oscillograph, on which pulses from both counters were recorded.

Figure 3 presents a histogram of the relative delays in the counters at an overvoltage \(\Delta V = 500\) V and a dead time \(T_0 = 0.01\) sec. Along the abscissa is plotted the magnitude of the relative delay, along the ordinate—the number of delay events (in an interval of \(5\cdot 10^{-9}\) sec). A total of 2000 events were recorded. In the obtained distribution the author introduced a correction taking into account coincidences in the counters caused by showers, penetrating associated particles (see\({}^{7}\)), etc. This measurement background is represented in the histogram by the events below the line \(AB\). The author takes the half-width of the smooth envelope curve of the obtained distribution (above the line \(AB\)) as the resolving time of the counter; for the counters studied it is approximately \(\sim 5\cdot 10^{-9}\) sec. Thus, the data of various works on the delay of pulses in spark counters are in full agreement with one another.

It is noted that the delay can be reduced by decreasing the interelectrode distance and increasing the overvoltage on the counters. However, the implementation of such counter designs apparently encounters certain difficulties\({}^{6}\).

In two communications published in September 1953, a brief theoretical consideration is given of the operation of a spark counter.1 A counter design is described with electrodes measuring \(10 \times 2.8\ \text{cm}^2\). The counters were filled with a mixture of argon (\(34\ \text{cm}\) Hg) and alcohol (\(4\ \text{cm}\) Hg). They had a lifetime considerably longer than those described earlier in works 2–7. For the present one can indicate only the lower limit of their lifetime; counters manufactured a year ago operated continuously for 6 months and continue to operate effectively.

The following characteristics of these counters are given:

1. The plateau is \(800\text{--}900\ \text{V}\) at \(T_0 = 5 \cdot 10^{-3}\ \text{s}\) and decreases to \(600\text{--}700\ \text{V}\) at \(T_0 = 10^{-4}\ \text{s}\).
2. Resolving time \(\sim 10^{-9}\ \text{s}\).
3. The pulse magnitude is several hundred volts.
4. The counters are suitable for operation at temperatures from \(+10^\circ\text{C}\) to \(+40^\circ\text{C}\).

Fig. 4.

A photograph of the spark discharge in the counter is also given; from the photograph it is possible, with good accuracy (within \(1\ \text{mm}^3\)), to determine the place where the particle passed through the counter.

The authors carried out an ingenious method of photographing the discharge, making one of the electrodes semitransparent and photographing through it the working volume of the counter. In this case a single photograph is sufficient to determine the point where the particle trajectory intersects the electrode surface.

Spark counters can be used in experiments with cosmic rays and in nuclear physics. They make it possible to record quite accurately the trajectory of a particle and the moment of its passage through the instrument. At the same time they make it possible to measure quite accurately small time intervals \(\sim 10^{-9}\ \text{s}\). As an example, data are given on the measurement of the average velocity of particles of cosmic radiation at sea level.1 The mean time of relative delay of pulses in two counters arranged one above the other at a distance of \(6\ \text{cm}\) was \((+0.42 \pm 0.22)\cdot 10^{-9}\ \text{s}\), and at a distance of \(42.2\ \text{cm}\), \((+1.72 \pm 0.24)\cdot 10^{-9}\ \text{s}\). Assuming that in the second case the delay time increased by \((1.30 \pm 0.33)\cdot 10^{-9}\ \text{s}\) because the recorded particles traversed an additional path between the counters equal to \(36.2\ \text{cm}\), one can estimate the mean value of the particle velocity. It turns out to be \(\beta = 0.93 \pm 0.25\). As an example of use-

...of spark counters one may also cite work^7, whose author used spark counters to study associated particles in cosmic radiation at sea level. Two counters were placed in a horizontal plane, and the number of coincidences was measured as a function of the distance between them*). Simultaneously with the measurement of the number of double coincidences, the relative delay of the pulses in the two counters was measured in each case of coincidence. The histogram in Fig. 4,b shows the distribution of relative delays for distances between the counters equal to 15 cm and 21 cm (in air and under lead).

For comparison, Fig. 4,a gives a histogram of delays for the case in which the counters were placed one above the other and were traversed by a single cosmic particle.

It is seen that distribution (b) is considerably broader than (a). The mean value of the delay time when the counters were arranged in a horizontal plane, where pairs or groups of particles were recorded, was \(\sim 2\cdot 10^{-8}\) sec.

In the author’s opinion, this indicates the existence in cosmic radiation at sea level of associated particles traveling with an average relative delay of \(\sim 2\cdot 10^{-8}\) sec.

However, the physical interpretation of this experimental result still remains ambiguous.

M. D.

REFERENCES CITED

1. N. A. Kaptsov, Electronics, Gostekhizdat, Moscow, 1953, p. 350.
2. J. W. Keuffel, R. S. I 20, No. 3, 197 (1949).
3. J. W. Keuffel, R. S. I 20, No. 3, 202 (1949).
4. Leon Madausky and R. W. Pidd, R. S. I. 21, No. 5, 407 (1950).
5. R. W. Pidd and Leon Madausky, Phys. Rev. 75, No. 8, 1175 (1949).
6. E. Robinson, Proc. Phys. Soc. 66, No. 397, A, 73 (1953).
7. E. Robinson, Proc. Phys. Soc. 66, No. 397, A, 79 (1953).
8. F. Bella and C. Franzinetti, Nuovo Cimento 10, No. 9, 1335 (1953).
9. F. Bella, C. Franzinetti and D. W. Lee, Nuovo Cimento 10, No. 9, 1338 (1953).

1. Reference number as printed on the page. ↩↩