FFeasibility of high-voltage systems for a very longdrift in liquid argon TPCs
S. Horikawa, A. Badertscher, L. Kaufmann, M. Laﬀranchi,A. Marchionni, M. Messina , G. Natterer and A. Rubbia ETH Zurich, Institute for Particle Physics, CH-8093 Z¨urich, SwitzerlandE-mail:
Designs of high-voltage (HV) systems for creating a drift electric ﬁeld in liquidargon TPCs are reviewed. In ongoing experiments systems capable of ∼
100 kV are realisedfor a drift ﬁeld of 0.5–1 kV/cm over a length of up to 1.5 m. Two of them having diﬀerentapproaches are presented: the ICARUS-T600 detector having a system consisting of an externalpower supply, HV feedthroughs and resistive voltage degraders and the ArDM-1t detector havinga cryogenic Greinacher HV multiplier inside the liquid argon volume. For a giant scale liquidargon TPC, a system providing 2 MV may be required to attain a drift length of ∼
20 m.Feasibility of such a system is evaluated by extrapolating the existing designs.
The intensity of the drift electric ﬁeld is one of the important design parameters for liquid argontime projection chambers (LAr-TPCs). A better collection of ionisation charges is attainedby increasing the ﬁeld intensity because: (1) more electron-ion recombination is prevented and (2) attenuation of the drift electrons due to their attachment to residual electronegativeimpurities such as oxygen decreases. The dependence of the attachment cross section on theelectric ﬁeld is known to be weak in the practical range of the intensity (0.05–1 kV/cm) . Theattenuation then is described well by an exponential decrease with drift time, characterised bythe drift electron lifetime τ which is determined dominantly by the impurity concentration. Themean drift velocity of the electrons increases with increasing electric ﬁeld, leading to the shortercollection time and consequently to the less attenuation. The drift velocity increases by 30% bydoubling the electric ﬁeld intensity from 0.5 to 1 kV/cm, and again 30% from 1 to 2 kV/cm .From a technical point of view, however, diﬃculties increase with increasing ﬁeld intensity whichrequires higher voltages. Therefore, it should be determined by a right compromise between thedetector performance and the practicality of the high voltage. A ﬁeld intensity of 1 kV/cm is areasonable compromise for very long drifts.A drift ﬁeld of 1 kV/cm requires a potential diﬀerence of 100 kV over 1 m. A drift lengthof 20 m which can be the case in giant scale detectors  thus requires 2 MV. There are twodiﬀerent approaches to realise such a high voltage (HV) system for LAr-TPCs. The ﬁrst typeuses an external HV power supply and feed it into the detector volume using feedthroughs as inthe ICARUS-T600 detector . This type has the advantage that the power supply is able to be Now at Albert Einstein Center for Fundamental Physics, Laboratory for High Energy Physics, University ofBern, CH-3012 Bern, Switzerland. a r X i v : . [ phy s i c s . i n s - d e t ] S e p epaired in case of problems without emptying the LAr volume. In addition, the output voltagecan be monitored precisely and continuously during the data taking from the current ﬂowingthrough a resistive voltage degrader. The second type has an internal HV generator directlyinside the LAr volume as the Greinacher HV multiplier of the ArDM-1t detector , which wehave developed at CERN. Its advantages are: (1) all the HV parts are immersed in LAr whichhas a large dielectric strength ( ∼
2. ICARUS-T600 system with an external HV power supply and feedthroughs
The ICARUS-T600 detector  is based essentially on four ionisation chambers occupying twoparallelepiped volumes each having a size of roughly 20 m long, 3.6 m wide and 3.9 m highcontaining approximately 240 tons of LAr. In each volume a central vertical cathode plane isfacing a wire chamber on each side consisting of three planes. The nominal cathode potentialis −
75 kV creating a drift ﬁeld of 0.5 kV/cm over the maximum drift length of 1.5 m, whilethe system has been tested up to 1 kV/cm with −
150 kV without problems. Each of the fourdrift volumes is surrounded by a set of 29 regularly spaced ﬁeld shaping electrodes (Figure 1).Linearly decreasing potentials are distributed to the electrodes with the aid of a resistive voltagedegrader consisting of 30 times 25-MΩ series resistors. The nominal current is 0.1 mA dissipating7.5 W per volume. For redundancy each resistor is made of four 100-MΩ resistors inserted inparallel between adjacent electrodes. Mechanical details of the resistors connecting adjacentﬁeld shaping electrodes are shown in Figure 2. Calculations show that the ﬁeld distortion islimited in the region close to the electrodes in case that two or even three resistors break out offour in the same group. The probability of such a case is anyhow negligibly small.
Field shaping electrodes of theICARUS-T600 detector.
Figure 2. igure 3.
Left; HV feedthrough. Center; HVcable. Right; the feedthrough with the cableinserted.
Spring contact at the bottomend of the feedthrough and the cup-shapedreceptacle mounted on the cathode.made of a stainless-steel tube, surrounds the PE insulator extending inside the cryostat downto the LAr level. During the ﬁrst tests of the feedthrough occasional micro discharges wereobserved, due to the air presence in the interstices in the cable-to-plug and in the plug-to-socketconnections. The micro discharges immediately disappear by ﬁlling these interstices with drynitrogen gas.A 150-kV/1-mA power supply from Heinzinger electronic GmbH is used for a the operationof T600. It gives a peak-to-peak voltage ripple less than 10 − at maximum load, which isreduced by a factor ﬁve under the operating conditions of T300 (half of T600). The ripplemainly consists of two components, i.e. 50 Hz and 37.45 kHz. Using a proper ﬁlter the highfrequency components can be eliminated and the low frequency component can be reduced bymore than a factor of three.
3. ArDM-1t system with an internal Greinacher HV multiplier
ArDM-1t is a 1-ton LAr-TPC built for a direct detection of WIMPs . Its HV system is basedon an internal Greinacher  (also known as Cockcroft-Walton ) HV multiplier (Figure 5).Figure 6 illustrates the ArDM-1t detector. The cylindrical drift volume is 80 cm in diameterand 124 cm high. A vertical electric ﬁeld is generated with the aid of 30 ring-shaped electrodescalled “ﬁeld shapers” spaced every 40 mm. The cathode grid which is at the highest potential inthe system is located on the bottom. Ionisation electrons drift upwards and are collected by thecharge readout system (near ground potential) on the top. The DC high voltages generated bythe Greinacher circuit are linearly distributed from the cathode to the ﬁrst (top) ﬁeld shaper.The system is capable to generate about −
400 kV resulting in an electric ﬁeld of above 3 kV/cm.For the ﬁrst phase of the experiment it will be operated at −
124 kV to create a drift electricﬁeld of 1 kV/cm.The ArDM Greinacher circuit has 210 stages (Figure 5). It outputs DC voltages on thebottom row of the diagram as denoted with V i . The potential at the i th stage is ideally equal tothe peak-to-peak value of the input AC voltage multiplied by the stage number, i.e. V i = i · V inpp .Each capacitor symbol corresponds to four (two) 82-nF polypropylene capacitors in parallel for Heinzinger electronic GmbH, D-83026 Rosenheim, Germany. Evox Rifa PHE450; KEMET Electronics S.A., CH-1211 Geneva 20, Switzerland. c1 Cs1 Cc2 Cs2 Cc3 Cs3 Cc209 Cs209 Cc210 Cs210 +– V2 V3 V209 V210V1
Diagram of the ArDM Greinacher circuit. Additional capacitors, diodes and the DCvoltage source indicated in red are for adjusting the potential of the lowest stage.
GreinacherChargereadout(Anode)CathodePMTField shaper m Figure 6.
ArDM Greinacher circuit.stages 1 to 170 (171 to 210) and each diode symbol to three avalanche diodes in series. Thecomponents have been chosen for their reliability in LAr. Thanks to the redundancy failuresof a single component do not cause a critical loss of the functionality of the whole circuit. Apicture of the ArDM Greinacher circuit mounted on the detector is shown in Figure 7.Before the ArDM system was realised, substantial R&D eﬀorts were undertaken with severalprototype Greinacher circuits having a smaller number of stages. At the last stage of theR&D a 10-stage prototype was built with exactly the same design and components (i.e. thepolypropylene capacitors and the avalanche diodes) as in ArDM. An extensive study was donewith it in a range of diﬀerent circumstances, i.e. in air and in LAr , where we obtained a goodunderstanding of the fundamental characteristics of the circuit and of the components. In LArthe maximum voltage of 24 kV was reached. BY505, High-voltage soft-recovery rectiﬁer; Philips Semiconductors. No longer in production. reinacher stage number0 20 40 60 80 100 120 140 160 180 200 220 V o l t a g e , V × Figure 8.
Potentials measured at each ﬁeldshaper plotted as a function of Greinacherstage number.
Field shaper number0 5 10 15 20 25 30 V o l t a g e , V × Figure 9.
Potentials measured at eachﬁeld shaper as a function of the ﬁeld shapernumber.The full 210-stage system was ﬁrst tested in air up to −
10 kV cathode voltage. Figure 8shows the potentials measured at each ﬁeld shaper plotted as a function of the correspondingGreinacher stage number. Under real conditions the output voltage does not depend linearlyon the stage number. This non-linearity is explained by the “shunt” capacitance between thetop and bottom rows of the circuit as in Figure 5 (see also Section 4), which is essentially dueto the diode capacitance and the stray capacitance determined by the geometry of the circuit.Fitting the measured curve with a model  as shown in blue in Figure 8 the value of theshunt capacitance was evaluated to be 2.35 pF. Note that the inﬂuence of the resistive load ofthe circuit (mainly the reverse resistance of the diode chain) to this eﬀect is negligibly smallin this system compared to that of the shunt capacitance. The non-linearity however can becompensated simply by choosing an appropriate Greinacher stage connecting to each ﬁeld shaperwhen the number of Greinacher stages N Gr is signiﬁcantly larger than that of ﬁeld shapers N fs .This is the case in ArDM ( N Gr = 7 N fs ) and the potentials are distributed linearly to the ﬁeldshapers as shown in Figure 9, where the measured potentials are plotted now as a function ofthe ﬁeld shaper number (No. 31 corresponding to the cathode).The ArDM Greinacher circuit is designed so that the resistive load is minimised in order toallow operation at low frequency (50 Hz). Hence, once it is charged up, it keeps the high voltageeven if the input AC voltage source is switched oﬀ. Its natural discharging is dominated by theleakage current through the diode chain and was measured to be consistent with an exponentialdecay with a lifetime of 19 hours in air at room temperature. It can be even slower in LAr becausethe leakage current is much smaller at 87 K. The total energy stored in the system is 22 J at 124kV output voltage and can be ∼ ∼
400 kV. For safety and also for protecting the apparatusincluding the circuit itself in case of any unexpected events, it is therefore very important tohave a system capable of actively and rapidly discharging the Greinacher circuit. Figure 10illustrates the schematic diagram of the discharging system in ArDM. The whole circuit can bedischarged by touching the cathode with an oval-shaped contact grounded through the resistorchain having a total resistance of 1.8 GΩ, which consists of nine 200-MΩ HV resistors. Thedischarge current and consequently the cathode potential before discharging can be measuredhrough the voltage across the 220-kΩ readout resistor connected in series to the chain. Figure11 illustrates the mechanical realisation of the system. The discharging contact and the chaincan rotate around the polyethylene axis rod steered by a rotary motion vacuum feedthrough.This technique was tested using a 10-stage prototype Greinacher circuit in liquid nitrogen. A100-MΩ resistor of the same type as in ArDM was used as the discharging resistor with a 200-kΩreadout resistor. Figure 12 shows the voltage across the readout resistor at discharging from − τ = RC agreed with naive calculations from the resistance and the capacitance ofthe system, verifying the small temperature coeﬃcients of the components. A simulation usingNgspice shown in red in the same plot agreed with the measurement as well. Measuring device V Discharge contact D e t e c t o r v e ss e l × M Ω k Ω Ω M Ω Gas Discharge Tube(Surge arrester)FeedthroughCathode HV Wall ground (Oscilloscope)
Schematic diagram of the ArDMGreinacher discharging system.
The ArDM Greinacher discharg-ing system mechanics.
4. Extrapolation to a megavolt system
Keeping 1 kV/cm a very long drift of electrons over 20 m requires 2 MV. In order to realise such aHV system with an external power supply, a feedthrough which withstands megavolts is needed.Such a feedthrough can in principle be built by scaling the existing ICARUS design. One mayalso exploit expertise from industries, where devices for above 1 MV are in development e.g. forHVDC (high voltage direct current) power transmission systems . Nevertheless, realising aHV generator for 2 MV is very challenging. Use of LAr can be very proﬁtable in this contextsince it has a dielectric strength of larger than 1 MV/cm and is known as one of the best insulatormaterials. A Greinacher HV multiplier in pure LAr can therefore be an interesting option forsuch a HV generator in particular for a giant scale LAr-TPC. HV generators are often the mostcrucial source of high frequency noise which readout electronics suﬀer from. As mentioned abovethe ArDM-type system having an internal Greinacher circuit does not require any resistive loadand thus can be operated at very low frequencies. This feature can be a remarkable advantageparticularly for megavolt systems since a low pass ﬁlter for such high voltages can potentiallybe a critical issue. Ngspice circuit simulator. http://ngspice.sourceforge.net/ ime, s0 2 4 6 8 V o l t a g e , V Cs Cs C p C p Cs Cs C p C p v(n-1) v(n) v(n+1)i(n-1) i(n) i(n+1)i(n-1) i(n) i(n+1) Figure 12.
Discharging curve of the 10-stageGreinacher circuit measured in liquid nitrogenover the readout resistor (see text).
Transmission line equivalent to aGreinacher circuit.The feasibility of extrapolating the ArDM Greinacher system to a megavolt generator hasbeen studied based on the test results as described above. Theoretical studies based on atransmission-line model  suggest that the maximum output voltage that can be attained bya Greinacher circuit occurs for an inﬁnite number of stages and is given by V max = Eγ , γ ≈ (cid:114) C p C s , (1)where E (= V inpp /
2) is the input voltage, C s the capacitor capacitance and C p ( (cid:28) C s ) the shuntcapacitance (Figure 13). To attain larger V max one has to go for larger E , or/and smaller γ byincreasing C s or decreasing C p . The total output voltage V of an N -stage circuit can then beobtained by V = V max tanh(2 γN ). To attain a desired output voltage V for given E and C p it isfavourable to choose N and C s so that the total capacitor capacitance N · C s can be minimised.This corresponds to 2 γN ≈ .
42 or V ≈ . · V max .The results of the extrapolations are summarised in Table 1 with the actual ArDM parametersas reference. For the extrapolations to 10-m drift use of the same polypropylene capacitors asin ArDM is assumed and the input voltage thus is limited to their maximum operation voltage2 E = 2 . C p = 2 .
35 pF is assumed. Thecapacitor capacitance C s is then increased until V max = V / .
89 is reached for the desired outputvoltage V . A 1-MV system for 10-m drift looks very feasible but the total capacitance becomesabout 20 times larger than ArDM. The extrapolation to a 2-MV system for 20-m drift was doneassuming E can be increased by a factor of √ C p can be reducedby a factor of 2, leading to the twice larger V max with the same C s . The former can be realisedwith a diﬀerent capacitor type or by developing similar polypropylene capacitors having a higheroperating voltage. The latter can be achieved by optimising the geometry of the system, which ispossible since the space density of the circuit components becomes signiﬁcantly lower for a giant-scale detector. These calculations demonstrate that it is feasible with engineering optimisationto build an ArDM-type Greinacher circuit which can generate 2 MV. able 1. Extrapolation of the ArDM Greinacher circuit to a long drift up to 20 m.Unit ArDM Extrapolationﬁrst phase nominalDrift length m 1.24 1.24 10 20Electric ﬁeld V/cm 1k 3.05k 1k 1kTotal output voltage V 124k 378k 1M 2MInput voltage V pp − in = 2 E V 820 2.5k 2.5k 3.5kShunt capacitance, C p F 2.35p 2.35p 2.35p 1.18pCapacitor, C s F 328n/164n 328n/164n 1.90 µ µ Number of stages, N – 210 210 638 903 N per 10 cm – 16.9 16.9 6.38 4.51Total capacitance F 125 µ µ µ µ µ µ Total charge C 73.6m 312m 6.06 12.1Total energy J 21.7 948 7.58k 21.5k
5. Conclusions and outlook
ICARUS-T600 has a well established HV system with an external power supply, feedthroughsand resistive voltage degraders for a nominal operation voltage of 75 kV with a possibility for150 kV. The ArDM internal Greinacher HV multiplier was developed for ∼
400 kV operatingvoltage. Its fundamental characteristics has been well understood in the ﬁrst tests in air andthe system will be tested at 124 kV in LAr in 2010. A very long drift in giant scale LAr-TPCsrequires megavolt systems. A feedthrough withstanding megavolts can in principle be realisedby scaling up the existing ICARUS design. Calculations showed that a Greinacher HV multiplierof the ArDM type can be extrapolated to 1–2 MV with some engineering optimisations.
This work was supported by ETH Z¨urich and the Swiss National Science Foundation (SNF).We are grateful to F. Sergiampietri of INFN for the thorough and detailed information on theICARUS-T600 HV system and also for the stimulating discussions about various possibilities ofthe future HV systems for a long drift in LAr-TPCs.
References  S. Amoruso et al. [ICARUS Collaboration], Nucl. Instrum. Meth. A (2004) 275. S. Amoruso et al. , Nucl. Instrum. Meth. A (2004) 68. W. Walkowiak, Nucl. Instrum. Meth. A (2000) 288. A. Rubbia, J. Phys. Conf. Ser. (2009) 012020 [arXiv:0908.1286 [hep-ph]]. S. Amerio et al. [ICARUS Collaboration], Nucl. Instrum. Meth. A (2004) 329. A. Rubbia, J. Phys. Conf. Ser. (2006) 129 [arXiv:hep-ph/0510320]. H. Greinacher, Z. Physik (1921) 195. J. D. Cockcroft and E. T. S. Walton, Proc. Roy. Soc. (London) (1932) 619. L. Kaufmann, “Detector Performance and Background Studies for the ArDM Experiment,” Diss. ETH No.17806 (2008). M. M. Weiner, Rev. Sci. Instrum. (1969) 330; E. Everhart and P. Lorrain, Rev. Sci. Instrum. (1953)221. For example:(1953)221. For example: