A study of HFO-1234ze (1,3,3,3-Tetrafluoropropene) as an eco-friendly replacement in RPC detectors
L. Benussi, S. Bianco, M. Ferrini, L. Passamonti, D. Pierluigi, D. Piccolo, A. Russo, G. Saviano
aa r X i v : . [ phy s i c s . i n s - d e t ] M a y INFN-14-14/LNF
11 novembre 2014
A study of HFO-1234ze(1,3,3,3-Tetrafluoropropene) as an eco-friendlyreplacement in RPC detectors
L. Benussi , S. Bianco , M. Ferrini , L. Passamonti , D. Pierluigi , D. Piccolo , A. Russo , G. Saviano Abstract
The operations of Resistive Plate Chambers in LHC experiments require F-based gases for optimal perfor-mance. Recent regulations demand the use of environmentally unfriendly F-based gases to be limited orbanned. This report shows results of studies on performance of RPCs operated with a potential eco-friendlygas candidate 1,3,3,3-Tetrafluoropropene, commercially known as HFO-1234ze. Laboratori Nazionali di Frascati dell’INFN, Italy; Laboratori Nazionali di Frascati dell’INFN and Facolta’ di Ingegneria Roma1, Italy
Introduction
Resistive Plate Chambers (RPCs) [ 1] are widley used in High Energy Phyiscs applications and in partic-ular in the LHC experiments. For applications where high background rates are expected they have to beoperated in avalanche mode in order to keep the total produced charge low with benefits in terms of agingand rate capability. This is usually obtained with suitable gas mixtures that prevents the transition fromavalanche to stremer keeping the detection efficiency above 90 %. The use of F-based gases, usually usedin refrigerants, have shown so far to give the best performance. The RPC system of the CMS [ 2] experimentis operated with a gas mixture of C H F SF CO (GWP = 1), and the effect on the ozone layer is measured in Ozone Depletion Potential (ODP),normalized to the effect of CCl F (ODP = 1). The European Community has prohibited the productionand use of gas mixtures with Global Warming Power > C H F and SF gases used in the RPCs for example present a GWP=1430 and 23900 respec-tively and need to be replaced with components with lower GWP.An overview of potential gas candidates can be found in [ 5]. In this note we will concentrate on theuse of HFO1234ze component in the RPC gas mixture. First results of RPCs operated with this componentand in streamer mode have been already presented recently [ 4]. In this note independent tests are performedand additional gas combinations are investigated. An array composed of 12 single-gap square (50 x 50) cm RPC detectors is in operation at Frascati nationallaboratories of INFN and arranged as a cosmic ray hodoscope. The system is a copy of the Gas GainMonitoring (GGM) system of the Compact Muon Solenoid (CMS) RPC muon detector at the Large HadronCollider (LHC) of CERN, Geneva, Switzerland described elsewhere [ 6] [ 7]. The system is located insidea T, RH controlled box and flushed with different gas mixtures humidified at about 40% (Fig.1).Each chamber of the GGM system consists of a single gap with double sided pad read- out: twocopper pads are glued on the two opposite external side of the gap. Two foam planes are used to reducethe capacity coupling between the pad and the copper shields used as screen for the system. The signal isread-out by a transformer based circuit (fig. 2) that allows to algebraically subtract the two signals, whichhave opposite polarities, and to obtain an output signal with subtraction of the coherent noise, with animprovement of about a factor 4 of the signal to noise ratio. The trigger is provided by scintillator layerslocated on the top and bottom of the RPC stack with an average event rate of about 1 Hz, corresponding toabout 20 minutes for 1000 events. 2igure 1: Schematic layout of the GGM system of the CMS RPC detector reproduced in the FrascatiLaboratories.
Measurements of the RPC performance as a function of different gas mixtures are obtained with a simplereadout of the signals with a digital oscilloscope. In the tests performed so far we acquired the signals fromonly two RPCs. Both detectors are operated with the same gas mixture but one of the two is maintained ata fixed working voltage and used as additional offline trigger to the events.The signal pick-up extracted from the pad of each detector is sent to an oscilloscope lecroy with 5GSamples and 1 GHz band width externally triggered by the scintillator system. Measurements are taken atdifferent Effective High Voltages normalized at the pressure P = 990 mbar and temperature T = 290 ◦ K according to the standard formula 1. HV eff = HV × P /P × T /T (1)For each HV setting about 1000 waveforms are saved on the disk for offline analysis.In order to reduce the environmental noise picked up from the system a 20 Mhz filter is applied oneach RPC channel signal. The use of the filter in addition to the fact that we read a pad electrode, modifiesthe signal shape increasing the raising and falling time of the electric pulse. The overall induced charge isanyway not affected by this shape modifications and the timing of the signals is weakly affected by using alow electronic threshold to define the time of the signal.In figure 3 a typical set of waveforms is shown. A total time window of about 1 µ s is acquired andfor analysis purpose it is divided in four regions. 3igure 2: Scheme of double pad readout of RPCs used for the testsFigure 3: Typical RPC signals stored from oscilloscope. One RPC signal is registred in two diferent chan-nels with different y scale (red and black tracks). The green track corresponds to the trigger RPC signal.4igure 4: Left: Distribution of the minimum of the voltage in the “control region”. A threshold of -0.4mV is enough to discriminate fake signals due to noise induced baseline variations. Right: Distributionof the induced charge integrated in a 150 ns time window of the “control region”. A threshold of 0.3 pCin addition to the minimum voltage threshold reduce the fake signals and simulate an electronic thresholdsimilar to that used in the CMS experiment.The Region of 150 ns around the trigger is the ”signal window” and is used to measure the signalcharacteristics. The 10 ns region before the ”signal window” (”baseline window”) is used for baseline eval-uation. For each of the three channels the voltage baseline is evaluated averaging the voltage measurementsin this region in which signals are absent and then subtracted to the measurement in the ”signal window”.The 150 ns before the ”baseline window” defined as ”control region” is used to check the width of thecharge distribution and of the minimum voltage in absence of signals in order to define a correct thresholdto identify RPC signals. Finally the 150 ns above the ”signal region” (“delayed region”) is used to identifydelayed streamers.The data are saved on the disk in form of ascii files storing the time-voltage waveform history.Analysis is performed offline with ROOT software.For each event after having corrected the measurements in the ”signal window” for the baselineextracted from the ”baseline window” the integral of the signal in the 150 ns ”signal window” is convertedin pC in order to plot the induced charge. In figures 4 (left) and 4 (right) it is shown the minimum voltage andthe integral of the charge, after the baseline correction, in the ”control region” while in figure 5 (left) and 5(right) the induced charge in the ”signal region” respectively for gas mxiture with high streamer probabilityand for the CMS standard gas mixture. The two different horizontal scales are used to better identify theavalanche and streamer peaks. As can be seen from the plot of fig.4 in the control region a threshold of -0.4mV on the minimum voltage plot is enough to exclude fake signals due to pedestal variations and to takethe RPC signals. In addition we also require that the integrated charge is above 0.3 pC that is very close tothe present threshold of the electronics used in the CMS experiment [ 8]. To evaluate the streamer probability we define a streamer as a signal with integrated charge above20 pC in the 300 ns time window defined by the ”signal” + ”delayed” regions. Finally the time resolution The threshold applied in CMS ranges between 150 and 175 fC, but in our layout the induced charge isabout twice the charge induced on CMS strips becauses of the double pad readout.5 ntries 674Mean 2.311RMS 2.118 charge (pC)0 5 10 15 200246810121416182022
Entries 674Mean 2.311RMS 2.118 charge efficient channel 3 (pedestal subtracted)
Figure 5: Left: Typical distribution of the induced charge integrated in a 150 ns time window of the “signalregion” for one of the gas mixtures tested and working in streamer mode. The peak of the streamers isclearly seen for induced charges above 20 pC. Right: induced charge distribution for the CMS standard gasmixture showing the avalanche peakis obtained measuring the time over a threshold of -0.4 mV of signals coming from two different RPCs andplotting the difference. The values coming from the distribution should then be divided for a factor of theorder of √ in order to extract the time resolution.Gas mixtures are prepared via gas flowmeters and then analyzed in output to the RPC system with agas chromatograph to evaluate the tested working mixture. In all the tests we started from the standard gasmixture used in the RPC CMS system ( C H F % , isobuthane 4.5 % , SF % ) and then we changethe amount of C H F , SF and add fractions of HFO1234ze and Argon keeping almost unchanged theisobuthane content. However small variations between the expected and the measured fraction of eachcomponent can be found. For each test we quote the mixture measured with gas chromatograph. SF scan In order to validate the method a first set of measurement have been performed starting from the CMSstandard gas mixture and varying the amount of SF . Figures 6, 7 and 8 show respectively the efficiencyand streamer probability, the induced charge and the time over threshold difference between the RPC undertest and a reference RPC at fixed voltage as a function of High voltage for the different gas mixtures tested.As expected reducing the amount of SF in the mixture the working point is shifted to lower values but thestreamer probability increase. To evaluate the impact of the new HFO1234ze component in the RPC gas mixture in a first test we replacedthe SF with the HFO. Efficiency/streamer probability, induced charge and time resolution of two RPCsare compared respectively in fig 9 10 11 for standard CMS gas mixture and for the case in which SF hasbeen replaced by the HFO. The addition of a small quantity of HFO does not have the same impact as the SF for what concerns streamer suppression. 6 V (Volts)8800 9000 9200 9400 9600 9800 10000 10200 10400 E ff i c i en cy ( % ) RPC 3: SF6 0.28%RPC 3: SF6 0.15%RPC 3: NO SF6 s t r ea m e r p r ob a b ili t y ( % ) Figure 6: Efficiency and streamer probability vs effective HV for a mixture composed by 4.5 % ofisobuthane, about 95.2 % of R134a and varying the amount of SF between 0 and 0.28 %. The amount ofR134a is adjust accordingly. Streamer probability is indicate with dashed line and the scale is on the righty axis HV (Volts)8800 9000 9200 9400 9600 9800 10000 10200 10400 A v e r age c ha r ge ( p C ) RPC 3: SF6 0.28%RPC 3: SF6 0.15%RPC 3: NO SF6
Figure 7: Integrated induced charge vs effective HV for a misture composed by 4.5 % of isobuthane, about95.2 % of R134a and varying the amount of SF between 0 and 0.28 %. The amount of R134a is adjustaccordingly 7 V (Volts)9000 9200 9400 9600 9800 10000 10200 10400 c on v o l u t ed t i m e r e s o l u t i on ( n s ) SF6 0.28%SF6 0.15%
Figure 8: Difference between the time over threshold of two RPCs vs effective HV for a misture composedby 4.5 % of isobuthane, about 95.2 % of R134a and varying the amount of SF between 0.15 and 0.28 %.The amount of R134a is adjust accordingly. One of the two RPCs is fixed at 9.2 kV while the HV of theRPC under test is changed. To get the real time resolution the values should be divided for something ofthe order of √ . HV (Volts)8800 9000 9200 9400 9600 9800 10000 10200 10400 E ff i c i en cy ( % ) RPC 3: SF6 0.28%RPC 3: HFO 0.28% s t r ea m e r p r ob a b ili t y ( % ) Figure 9: Efficiency and streamer probability vs effective HV for a mixture composed by 4.5 % ofisobuthane, 95.2 % of R134a and SF V (Volts)8800 9000 9200 9400 9600 9800 10000 10200 10400 A v e r age c ha r ge ( p C ) RPC 3: SF6 0.28%RPC 3: HFO 0.28%
Figure 10: Integrated induced charge vs effective HV for a misture composed by 4.5 % of isobuthane, 95.2% of R134a and SF HV (Volts)8800 9000 9200 9400 9600 9800 10000 10200 10400 c on v o l u t ed t i m e r e s o l u t i on ( n s ) RPC 3: SF6 0.28%RPC 3: HFO 0.28%
Figure 11: Difference between the time over threshold of two RPCs vs effective HV for a misture composedby 4.5 % of isobuthane, 95.2 % of R134a and SF √ .9 V (Volts)8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 E ff i c i en cy ( % ) Standard gas mixtureHFO 23% R134 70%, iso 4.2%, SF6 0.34%HFO 23% R134 70%, iso 4.2%HFO 45% R134 55.2%, iso 4.8% s t r ea m e r p r ob a b ili t y ( % ) Figure 12: Efficiency and streamer probability vs effective HV for different mistures composed byisobuthane, R134a, HFO1234ze with and without SF . Streamer probability is indicate with dashed lineand the scale is on the right y axisA second set of measurement have been performed with gas mixtures in which isobuthane has beenmaintained fixed around 4.5 % and varying the amount of HFO-1234ze vs C H F and the SF content.Figures 12, 13 and 14, show the impact of the addition of 23 % and 45 % of HFO-1234ze replacing C H F , with and without SF addition. An amount of about 23 % is enough to move the working point of about1500 Volts while the fraction of streamers is maintained low if SF is added to the mixture.In any case the HFO1234ze component shows improved quenching characteristics with respect toR134a and replacing the full amount of C H F with HFO-1234ze would require to increase the workingvoltage too much stressing the HV system so that our tests stopped at a fraction of 45 % HFO-1234ze.
We explored the effect of adding Argon to the HFO-1234ze based gas mixture in order to reduce the workingpoint and eliminate the C H F component in the mixture. Previews results /citeref:liberti shown that amixture of Argon with small percentage of HFO-1234ze give encouraging results in streamer mode. Westarted from the already tested mixtures and then increased the fraction of HFO-1234ze to analyze theperformance.Figures 15, 16, 17, 18, 19 and 20 show as usual efficiency/streamer probability, induced charge andtime resolution for different mixture compositions. Addition of HFO-1234ze move as usual the workingpoint to higher values but is not sufficient to reduce the fraction of streamers keeping the efficiency above90 % at least with the tested electronic threshold. Reducing the electronic threshold could recover detectionefficiency in the range where streamer probability is still low, but such tests were not possible with the10 V (Volts)8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 T o t a l c ha r ge ( p C ) Standard gas mixtureHFO 23% R134 70%, iso 4.2%, SF6 0.34%HFO 23% R134 70%, iso 4.2%HFO 45% R134 55.2%, iso 4.8%
Figure 13: Integrated induced charge vs effective HV for different mistures composed by isobuthane,R134a, HFO1234ze with and without SF HV (Volts)8500 9000 9500 10000 10500 11000 11500 12000 12500 13000 t i m e d i ff e r en c e ( n s ) Standard gas mixtureHFOze 23% R134 70%, iso 4.2%, SF6 0.34%HFOze 23% R134 70%, iso 4.2%HFOze 45% R134 55.2%, iso 4.8%
Figure 14: Difference between the time over threshold of two RPCs vs effective HV for different misturescomposed by isobuthane, R134a, HFO1234ze with and without SF . To get the real time resolution thevalues should be divided for something of the order of √ .11 V (Volts)6000 7000 8000 9000 10000 11000 E ff i c i en cy ( % ) HFO 34% R134 22%, Ar 41%, iso 3%HFO 55% Ar 41% iso 4%HFO 55% Ar 40% iso 4.5% SF6 0.5%HFO 20% Ar 80% s t r ea m e r p r ob a b ili t y ( % ) Figure 15: Efficiency and streamer probability vs effective HV for different mixtures adding Argon to theHFO1234ze based mixtures. Streamer probability is indicate with dashed line and the scale is on the righty axispresent setup due to the environmental noise limiting the threshold to aplly.
Performance of 2 mm wide single gap Resistive Plate Chambers, operated with several HFO1234ze basedgas mixtures, have been presented here in terms of efficiency, streamer probability, integrated charge andtime resolution. The HFO1234ze shows large quenching characteristics that could be interesting for RPCoperations. At the same time it cannot be used to replace the R134a in large fractions as the working pointis shifted to very high voltages. Several mixtures with different fractions of HFO1234ze and Argon havebeen also tested giving incouraging results but still the full efficiency is note reached in absence of largestreamer fractions, at least with the electronic thresholds used in CMS. More tests will be necessary to fullycharacterize this new gas component for his use inside RPCs.
References [1] R. Santonico and R. Cardarelli, “Development of Resistive Plate Counters”, Nucl. Instr. Meth., vol.187, pp. 377-380 (1981)[2] CMS Collaboration, The CMS experiment at the CERN LHC JINST 3 (2008) S08004.[3] CMS Collaboration, “The performance of the CMS muon detector in protonproton collisions at √ s =7 TeV at the LHC”, JINST 8 (2013) P11002[4] Liberti et al. submitted on proceedings of rpc2014 workshop12 V (Volts)6000 7000 8000 9000 10000 11000 E ff i c i en cy ( % ) HFO 30% Ar 65%, iso 4% SF6 1%HFO 34% Ar 62%, iso 3.3% SF6 0,7%HFO 45% Ar 50%, iso 4% SF6 1% s t r ea m e r p r ob a b ili t y ( % ) Figure 16: Efficiency and streamer probability vs effective HV for different mixtures adding Argon to theHFO1234ze based mixtures. Streamer probability is indicate with dashed line and the scale is on the righty axis[5] L. Benussi et al. , “Properties of potential eco-friendly gas replacements for particle detectors”, LNFnote INFN-14-13/LNF. [arXiv:1505.00701[physics.ins-det]].[6] L. Benussi, S. Bianco, S. Colafranceschi, D. Colonna, L. Daniello, F. L. Fabbri, M. Giardoni andB. Ortenzi et al. , “The CMS RPC gas gain monitoring system: An Overview and preliminary results”,Nucl. Instrum. Meth. A (2009) 805 [arXiv:0812.1108 [physics.ins-det]].[7] S. Colafranceschi, et al. , “Performance of the Gas Gain Monitoring system of the CMS RPC muondetector and effective working point fine tuning”, JINST (2012) P12004 [PoS RPC (2012) 046][arXiv:1209.3893 [hep-ex]].[8] M. Abbrescia et al. , “New developments on front-end electronics for the CMS Resistive Plate Cham-bers”, Nuclear Instruments and Methods in Physics Research A 456 (2000) 143-14913 V (Volts)6000 7000 8000 9000 10000 11000 T o t a l c ha r ge ( p C ) HFO 34% R134 22%, Ar 41%, iso 3%HFO 55% Ar 41% iso 4%HFO 55% Ar 40% iso 4.5% SF6 0.5%HFO 20% Ar 80%
Figure 17: Intergrated induced charge vs effective HV for different mixtures adding Argon to theHFO1234ze based mixtures. Streamer probability is indicate with dashed line and the scale is on the righty axis
HV (Volts)6000 7000 8000 9000 10000 11000 T o t a l c ha r ge ( p C ) HFO 30% Ar 65%, iso 4% SF6 1%HFO 34% Ar 62% iso 3.3% SF6 0.7%HFO 45% Ar 50% iso 4% SF6 1%
Figure 18: Intergrated induced charge vs effective HV for different mixtures adding Argon to theHFO1234ze based mixtures. Streamer probability is indicate with dashed line and the scale is on the righty axis 14
V (Volts)6000 7000 8000 9000 10000 11000 t i m e d i ff e r en c e ( n s ) HFOze 34% R134 22%, Ar 41%, iso 3%HFOze 55% Ar 41% iso 4%HFOze 55% Ar 40% iso 4.5% SF6 0.5%HFOze 20% Ar 80%
Figure 19: Difference between the time over threshold of two RPCs vs effective HV for different mixturesadding Argon to the HFO1234ze based mixtures. Streamer probability is indicate with dashed line and thescale is on the right y axis