Printed in: Nucl. Instr. & Meth. A433 (1999) pp 476-481 Single photoelectron detection with a low-pressure gas electron multiplier coupled to a CsI photocathode G. Garty*, A. Breskin, R. Chechik, E. Shefer Weizmann Institute of Science, 76100 Rehovot, Israel Abstract We have operated the Gas Electron Multiplier (GEM) coupled to a CsI photocathode for the detection of single UV photons. Gains of up to 10 000 were observed with a single GEM in pure hydrocarbons in the range 40}400 Torr. Using a double-GEM structure, we have detected single photoelectrons on a strip electrode. We have studied the single-electron detection e$ciency of the GEM, a crucial parameter in its operation in single-photon detection applications, both experimentally and by simulation. 1. Introduction The GEM is a newly developed gaseous electron multiplier, described in detail in Refs. [1}4]. The gas electron multiplication, by a factor of up to 10, takes place within small holes in a dielectric sheet, under a proper electric "eld across the holes. The focusing of the primary electrons and the extraction of the avalanche electrons is controlled by the "elds E and E on both sides of the GEM (see Fig. 1). The study of GEM operation at low gas pressure was proposed for various applications requiring e$cient detection of single electrons, while avoiding secondary processes [4]. A GEM electrode inserted between a radiation converter, such as a photocathode or a gas volume, and a second ampli"cation element considerably reduces second- * Corresponding author. E-mail address: [email protected] (G. Garty) -1- ary e!ects, improving the stability and the detection e$ciency of such detectors. We report here on the study of single-electron detection e$ciency with a single-GEM coupled to a multiwire element, operated at low gas pressure. We also report on the double-GEM structure coupled to readout strips, operated here for the "rst time as a single-electron detector. In our measurements we used two GEM geometries, both manufactured by TECH-ETCH [5] on a 50 lm thick copper-kapton sandwich: E A `transparenta GEM geometry, having a hexagonal lattice of double conical apertures, having an inner diameter of 90 lm and an outer diameter of 110 lm, and a pitch of 140 lm, resulting in a transparency of 56%. E An `opaquea GEM geometry, having a square lattice of apertures, having an inner diameter of 50 lm and an outer diameter of 80 lm, and a pitch of 200 lm, resulting in a transparency of 5%. Fig. 1. The experimental setup with one GEM and a multiwire. (a) Without preampli"cation and (b) with preampli"cation Both GEMs were operated in a #ow mode with high-purity hydrocarbons (40}400 Torr of CH , C H , i-C H and Ar/C H (10/90)). Fig. 1 shows the basic experimental setup. A CsI photocathode, irradiated by an Ar(Hg) UV lamp through a quartz window, induces single photoelectrons. These electrons are focussed into the GEM by an electric "eld E , multiplied in the GEM apertures and transferred via the "eld E to a multiwire (MW) element. For the double-GEM study we used a cascade of two `transparenta GEMs coupled to a printed board strip-readout electrode, located 0.5 mm from the second GEM. The electric "eld con"gurations in the GEM, and the electron transport properties, were simulated using MAXWELL's 3D parameter extractor [6] and GARFIELD [7], respectively. 2. Results and discussion Fig. 2 shows the gain measured on a single `transparenta GEM at various pressures of ethane and methane, under intense UV illumination. It was obtained by measuring the current collected on the GEM's lower face with reversed E (see Fig. 1), and normalizing it to the current i impinging on the GEM. The latter was obtained by measuring the current on the GEM's upper face with both faces at ground potential, and correcting it for the electron collection e$ciency, with the help of our Monte Carlo simulations. In the present condi-2- Fig. 2. Gain curves as function of the potential across the GEM, for di!erent gases and pressures. The graphs result from currents recorded on the GEM faces, using a `transparenta GEM, with electron losses taken into account (see text). tions, this ine$ciency amounts to 20%, most of it due to electrons lost, due to di!usion, to the kapton inside the GEM holes. The "eld E was maintained at the collection plateau in all measurements. Gains of up to 20 000 were measured in 200 Torr ethane before instabilities arose. In pure methane (400 Torr), the maximal stable gain was &1000, due to photon feedback. At low-pressure (40 Torr) gains of a few hundreds to a few thousands were observed, in di!erent gasses. The measured gain curves have no collection plateau, pointing at the poor electron focussing and transmission at low GEM "elds. The simulations show (Table 1) that due to the high electron di!usion at low gas pressures, a signi"cant fraction of the electrons are lost in the holes and on the GEM surfaces, when the "eld across the GEM is small. At GEM gains above 100 the "eld strength is su$cient to e$ciently focus the electrons into the holes. Nevertheless, some electron losses still exist due to the di!usion of electrons onto the GEM surfaces. The GEM e$ciency is crucial for applications where we have to detect only a single or a few electrons. One way to improve the chances of detecting a single primary electron, is to apply a small preampli"cation in the gas volume preceding the GEM. Even a small gain is su$cient to recover possible losses of primary photoelectrons; due to electron di!usion in the gas, at least one of the avalanche-generated secondary electrons will reach Table 1 Simulation of the electron transport of the `transparenta GEM at 40 Torr Ar/C H (10/90): percentage of electrons collected on di!erent electrodes at di!erent "eld con"gurations. The `detection e$ciencya corresponds to the total fraction of electrons producing an avalanche in the GEM and/or MW PP gain 1 6 E V/cm Torr 7.5 7.5 80 80 E V/cm Torr GEM gain Photocathode GEM up Kapton GEM down MW `Detection e7ciencya 33 33 1 500 10% 20% 71% 12% 10% 4% 5% 29% 4% 35% 4% 64% 33 33 1 500 0% 0% 7% 0% 43% 3% 6% 3% 43% 94% 43% 97% a GEM hole. Even though the electrostatic focussing into the GEM is inferior at high electric "eld, E , we expect a rise in the detection e$ciency with increasing preampli"cation up to the point where the preampli"cation is just compensated by the defocusing losses. We measured the relative single-electron detection e$ciency by counting pulses from the GEM#MW structure, under constant UV illumination. Contrary to measurements based on current distribution on the GEM electrodes [4], where the gain and transfer e$ciency are inseparable, measuring single-electron pulses provides a reliable account of the detection e$ciency at di!erent "eld con"gurations of the GEM and its surrounding electrodes. While varying the respective gains on the preampli"cation gap, GEM and MW, the total gain of the detector was maintained constant at 10. This eliminated the dependence on the electronics threshold and reduced secondary e!ects. At this gain we have already observed a saturated single-electron pulse height spectrum (see Fig. 3a). As we have no account of the absolute number of photoelectrons produced, we can only measure the behavior of the relative detection e$ciency of di!erent con"gurations. An absolute measurement can be performed using the electron counting technique [8], namely counting electron swarms of known size, induced by ultrasoft X-rays in a gas volume coupled to a GEM. Fig. 4 shows the results of measurements performed on the two GEM geometries: The `opaquea GEM at 40 Torr isobutane (Fig. 4a), and the `transparenta GEM at 40 Torr Ar/C H (10/90) (Fig. 4b). In the `opaquea GEM, a preampli"cation Fig. 3. Single electron pulse height spectra: (a) saturated spectrum obtained from a GEM # MW structure at 40 Torr of i-C H , using the `opaquea GEM. (b) Exponential spectrum from the double GEM# readout pads in 400 Torr CH , using two `transparenta GEMs. -3- Fig. 4. Relative detection e$ciency of `opaquea (a) and `transparenta (b) GEM: (a) an increase of &40% in counting rate occurs upon preampli"cation by ;6 preceding the GEM; (b) a decrease of a factor of &2 occurs upon preampli"cation ;6; a gain limitation due to ion feedback is seen without preampli"cation at a GEM gain of &200. of 6 raises the relative electron detection e$ciency by &40%. Indeed, our measurements have shown that the relative detection e$ciency does not significantly increase above a preampli"cation gain of 2. In the `transparenta GEM, on the other hand, the situation is reversed, and the detection e$ciency drops drastically, by a factor of 2}3, when adding a preampli"cation. Since no intuitive explanation has been found for this di!erence we have run simulations of the electron transport in the `transparenta GEM, at 40 Torr of Ar/C H (10/90). The radiation conver sion and electron collection gaps were both 1 mm wide. The simulation does not include avalanche formation, but only electron transport and di!usion. We assume that under high GEM gain, any electrons transported into the apertures will pro- -4- duce a detectable avalanche. Under GEM gain 1 only electrons reaching the MW will produce a detectable signal. The results of the simulation are shown in Table 1, where for each electric "eld con"guration we provide the fraction of electrons collected in each region of the detector. The last column provides the `detection e$ciencya which equals to the total fraction of electrons producing an avalanche in the GEM and/or the MW. In the case of no preampli"cation, despite the excellent focussing of electric "eld lines, which was seen from the simulations, the single-electron transport is not good. Many electrons reaching the GEM arrive, through di!usion, onto the top GEM face where they are lost. With increased electric "eld in the GEM, the focussing "eld near the GEM face increases and these electrons are attracted into the apertures. In the case of high GEM gain, we are able to detect electrons reaching the lower face of the GEM, as they, nevertheless, induce an avalanche in the GEM (not included in the simulation), part of which will reach the MW. When we add preampli"cation, the focussing is seen to deteriorate, however, this is more than compensated for by the avalanche statistics. In the simulation of a preampli"ed electron, we tracked six electrons from a distance of 1mm above the GEM and recorded only the one which got farthest down. As only one electron, from the preampli"cation avalanche, reaching the MW is su$cient for detection of the original photoelectron, we are able to overcome the intrinsic ine$ciency of the GEM focussing. The addition of a ;6 preampli"cation, in this case, is seen to lead to an increase of the `detection e$ciencya by &50%, consistent with what we measured for the opaque GEM. This is, however, in disagreement with our measurements performed for the `transparenta GEM, and should be further investigated. A signi"cant gain limiting e!ect was observed in the GEM operated at low gas pressure [4], due to ions accelerated within the GEM apertures and sputtering its upper face, under high electric "eld. The resulting secondary electrons are then focussed into the GEM apertures, inducing secondary avalanches, at a typical delay on the order of microseconds after the primary avalanche. The Fig. 5. Charge pulses from a GEM#MW structure at a total gain of 10 in 40 Torr Ar/C H (10/90). (a) No preampli"cation: multiple ion feedback is evident. (b) A preampli"cation gain of 6: no ion feedback observed. Fig. 6. Pulses from the double GEM structure coupled to printed board readout strips, operated in 400 Torr CH . (a) A preampli"cation gain of 100 followed by a gain of 100 on each GEM, with a low (collection) "eld above the readout strips. (b) A preampli"cation gain of 10 followed by a gain of 500 on the "rst GEM, a gain of 20 on the second GEM and a gain of 2 above the readout strips, yielding faster and larger pulses. Pulses recorded using a fast (t (0.5 ns) current ampli"er, Z "50 ), gain"200. preampli"cation preceding the GEM, by modifying the "eld line maps, was seen to greatly decrease this e!ect by diverting the ions to the photocathode. Fig. 5 depicts charge pulses obtained from the GEM#MWPC structure at 40 Torr of Ar/C H (10/90) and a total gain of 10 (GEM gain 200). In (a) multiple ion feedback events are evident, reducing the stability of the detector. When we add a preampli"cation "eld corresponding to a gain of 6, (b) the ion feedback is totally eliminated. In order to take advantage of the GEM properties, and since a single GEM does not provide su$cient gain for single-electron detection, we attempted to detect single photoelectrons using a double-GEM followed by a readout strip electrode, as demonstrated previously by Sauli [3]. We have operated this structure without or with small additional multiplication between the last GEM and the strips. In Fig. 6 we see single-electron pulses obtained from a double-GEM coupled to the strip readout electrode, at 400 Torr of methane. Fig. 6a shows pulses recorded when the strips are operated as a collection electrode; the pulses are rather slow and of small amplitude. When adding even a small ampli"cation across the GEM-strip gap (gain of 2, shown in (b)), the signals are faster and of higher amplitude. A typical pulse height spectrum of this structure is seen in Fig. 3b, with a gain of 100 on both GEMs and a gain of 2}3 on the readout trips. 3. Conclusions We have studied single- and double-GEM detector structures, coupled to a CsI photocathode, at low gas pressures, up to 400 Torr of hydrocarbons. Stable gains of up to 10 were reached with single photoelectrons, in a single GEM. We have observed saturated pulse height spectra with the GEM followed by a multiwire chamber. We have shown that a double-GEM followed by a readout strip electrode provides su$cient gain for the detection of single electrons. -5- The accent in this work was put on understanding the single-electron transport from the photocathode into the GEM and their detection e$ciency. This is of crucial importance for applications in imaging or counting of single photons or electrons. We have measured the relative singleelectron detection e$ciency for two types of GEMs, as a function of its gain and of the optional additional preampli"cation. From our simulations we expect that the addition of a small preampli"cation gain will dramatically improve the single-electron detection e$ciency, as the di!usion of the preampli"ed avalanche should compensate for the electron losses. In reality this was con"rmed only in the `opaquea GEM where electron focussing is expected to be, apriori, poor. With the `transparenta GEM, where we expect the initial e$ciency to be higher, the single-photon detection e$ciency was seen to drop drastically with the addition of preampli"cation. This drastic drop is not yet understood and is still under investigation. It was observed that preampli"cation in the gap preceding the GEM, reduces gain limitations due to ion feedback e!ects, but on the account of a larger probability of damaging the photocathode. In the future, we plan to study the absolute single-electron detection e$ciency, using the electron counting method. In addition, we are studying other detector/photocathode arrangements, as for -6- example depositing the photocathode directly on the top face of the GEM. Acknowledgements We wish to thank F. Sauli for helpful discussions and R. Veenhof and S. Iranzo for their extended help with the simulations. The work was partially supported by the Israel Science Foundation. The GEM electrodes were generously supplied by TECH ETCH Inc., for which we are very grateful. A. Breskin is the W.P. Reuther professor of research in the peaceful use of atomic energy. References [1] F. Sauli, Nucl. Instr. and Meth. A 386 (1997) 531. [2] F. Sauli, Nucl. Instr. and Meth. A 433 (1999) 464 and references therein. [3] A. Bressan, R. De Oliveira, A. Gandi, J.C. Labbe, L. Ropelewski, F. Sauli, D. Mormann, T. Muller, H.J. Simonis, Nucl. Instr. and Meth. A 425 (1999) 254. [4] R. Chechik, A. Breskin, G. Garty, J. Mattout, F. Sauli, E. Shefer, Nucl. Instr. and Meth. A 419 (1998) 423 and references therein. [5] TECH-ETCH Inc. http://www.tech-etch.com/. [6] Ansoft Corporation, http://www.ansoft.com. [7] R. Veenhof, CERN, Gar"eld, http://consult.cern.ch/ writeup/gar"eld/. [8] A. Pansky, A. Breskin, R. Chechik, G. Garty, E. Klein, Nucl. Instr. and Meth. A 392 (1997) 465}470.
© Copyright 2026 Paperzz