High Voltage Multiplexing for ATLAS Tracker Upgrade EG Villani on behalf of the ATLAS HV group STFC Rutherford Appleton Laboratory

Outlook Introduction: ATLAS Upgrade HV mux needs HV project description: devices and control circuitry Summary & conclusions 1

ATLAS Phase II Tracker Upgrade 2

3 Challenges facing HL-LHC silicon detector upgrades Higher Occupancies (  150 interactions / bunch crossing) ⤷ Finer Segmentation Higher Particle Fluences (  outmost layers to  innermost layers ⤷ Increased Radiation Tolerance (  10 increase in dose w.r.t. ATLAS ) Larger Area (~200 m 2 ) ⤷ Cheaper Sensors More Channels ⤷ Efficient power/bias distribution / low material budget Phase 2 (HL-LHC) Replacement of the present Transition Radiation Tracker (TRT) and Silicon Tracker (SCT) with an all-silicon strip tracker Conceptual Tracker Layout Short Strip (2.4 cm)  -strips (stereo layers): Long Strip (4.8 cm)  -strips (stereo layers): r = 38, 50, 62 cm r = 74, 100 cm From 1E33 cm -2 s -1 …to 5E34 cm -2 s -1

The Stave concept and HV distribution in ATLAS Upgrade 4 ~ 1.2 meters Bus cable Hybrids Coolant tube structure Carbon honeycomb or foam Carbon fibre facing Stave Cross-section A Stavelet is a shortened stave prototype with up to four modules on each side, used for preliminary tests, including power distribution Designed to reduce radiation length  Minimize material by shortening cooling path  48 Modules glued directly to a stave core with embedded pipes Designed for mass production  Simplified build procedure  Minimize specialist components  Minimize cost

HV distribution in ATLAS Upgrade The ‘ideal’ solution would be one HV bias line for each sensor: High Redundancy; Individual enabling or disabling of sensors and current monitoring; But the increased number of sensors in the Upgraded Tracker implies a trade off among material budget, complexity of power distribution and number of HV bias lines. Use single (or more) HV line to power all 12 sensors in a ½ stave and use one HV switch under DCS control for each sensor to disable malfunctioning detectors. 5

HV distribution in ATLAS Upgrade DC Reference Approach Each Sensor sees a different bias voltage (over-deplete top sensors in serial power chain by ~30V) Current measuring circuit may be placed on each hybrid Some mixing of HV and LV (DC) currents AC Reference Approach All detectors see same bias voltage Current measuring circuit most naturally located on End-of-Stave Card No mixing of HV and LV (DC) currents 6 HV SW

HV distribution in ATLAS Upgrade 7

HV distribution in ATLAS Upgrade: devices High Voltage switches requirements: Must be rated to 600V (or more for pixels) plus a safety margin Must be radiation hard (rules out most of Si based devices and optocouplers) Off-state impedance R off >> 1GΩ On-state impedance R on 1mA Must be non-magnetic (rules out electromechanical switches) Must maintain satisfactory performance at -30  C Must be small and cheap 8

HV distribution in ATLAS Upgrade: devices High Voltage switches considered: Si based devices: Bipolar transistors: main effect of radiation damage is lowering of gain and relatively high base current required MOS transistors: high voltage power MOSFET usually have thick gate oxide that makes them not rad – hard ( possible exception – see later slides) Si JFET: potentially rad – hard but hard - to - find Non – Si based devices: SiC based devices: on the market there are already examples of High Power rated devices SiC JFET (Semisouth) BJT (Fairchild/Transic), SiC SJT (Genesic) GaN devices: (Transphorm, Panasonic, Infineon ) as for SiC devices there are switches operating at kV’s 9 Materials Property Si SiC-4H GaN Band Gap (eV) low leakage; higher displacement threshold (more rad-hard) Critical Field 1E6 V/cm higher BV voltage/thinner; Electron Mobility (cm2/V-sec) Electron Saturation Velocity (106 cm/sec) higher current density; Thermal Conductivity (Watts/cm2 K) easier cooling;

HV distribution in ATLAS Upgrade: SiC device tests Initial 4-switch box with SemiSouth SiC JFET SJEP170 with slow controlled circuitry was tested to switch a stavelet; no additional noise seen 10

HV distribution in ATLAS Upgrade: SiC device tests Semisouth SJEP170 JFET characterization tests Pre and post irradiation tests results (>30Mrad gamma) V ds (V) I ds (pA) I ds (A) V gs (V) 11 Post irradiation Pre irradiation Post irradiation

12 HV distribution in ATLAS Upgrade: SiC device tests Los Alamos Sep 2012 SemiSouth SJEP170 Run: Irradiation up to 5E14 1MeV n-eq Beam profile

13 HV distribution in ATLAS Upgrade: SiC device tests Fast pulse test: a (negligible) increase in on-state resistance is observed following irradiation ID [A]

14 HV distribution in ATLAS Upgrade: SiC device tests Off-state leakage current preliminary test results very good Unfortunately, SemiSouth went out of business in 2012 IS [pA] IG [pA]

#4 Silicon High Voltage N JFETs 2N6449 devices mounted on a PCB, #3 bare dies JFETs of the same type and #3 PCB samples were irradiated to 1/ p/cm 2 with 26 MeV protons (doses of around 1MGy in 10 minutes) at Birmingham University, UK The JFETs mounted on PCB were previously characterized at RAL: for V ds = 250 V and V gs = -9 V maximum I ds < 200 pA for the 4 devices tested Onset of Breakdown at V dg = 300 V, as per DS The JFETs show an average DC resistance of [V gs = 0 V, T = 22  C] The bare dies JFETs were irradiated to study their gamma spectrum emission 800 μm 15 Si JFET HV distribution in ATLAS Upgrade: Si device tests

Si JFET # 4 I - V curves: UNIRRADIATED I DS, I GS vs. V GS (left), IRRADIATED I DS, I GS vs. V GS (right) Maximum I ds compliance: 2 mA Maximum I gs compliance: 0.1 mA UNIRRADIATED I ds = 50 [V GS =-9 V V ds =250 V] IRRADIATED I ds = 37 [V GS =-11 V ds =250 V] Leakage current increases by around 3 o.f.m. (but so would leakage current of sensors); Main issue is the increased R dson (  kohm to  100’s khom) preventing their use as  mA switches; Interfet (manufacturer) potentially interested in fabricating P type JFET (which may be more rad – hard) I DS (A) V GS (V) I GS (A) I DS (A) V GS (V) I GS (A) 16 HV distribution in ATLAS Upgrade: Si device tests

HV MUX control scheme Negative HV multiplier filter HV JFET DEPL V source Regardless of the devices used as HV switches, a control circuitry, referenced to a high potential, to enable them is needed An investigated option consists of an AC coupled control switch based upon a voltage multiplier (it works with depletion and enhancement mode devices depending on the polarity of the diodes ) -HV To Detector 17

Negative HV multiplier filter HV JFET V source Average current consumption: 50 μA The voltage needed is generated using a 2.5 V (or lower amplitude) square wave AC. Only the coupling capacitors from V src need to be rated for HV. The (in the example shown) negative voltage is generated ONLY when the JFET needs shutting off: no power is needed during normal operation (i.e. when the JFET switch is ON). To Detector 18 HV MUX control scheme

* The voltage across R2 is measured vs. amplitude and frequency of V in ( square wave, 50% duty cycle) and for V high = [0, -300] V * Applying -300 V a slight decrease in abs(V out ) is noticed (some leakage current over the board surface is the likely cause) ‘DABO’ connection Voltage MPY ‘MOBO’ f in = 50 kHz V bias =0V f in = 100 kHz V bias =0V f in = 50 kHz V bias =-300V V MPYV out V in Multimeter : Fluke 287 Signal generator: Tektronix AFG3252 HV PSU: EA-BS315-04B (#2 in series to get 300V) (HV PSU) (Sign. Gen) (Meter) 19 HV MUX control scheme test

Conclusions High Voltage distribution via HV switches and DCS control is being investigated A number of devices, based upon Si and wider bandgap materials, are being investigated (Panasonic, Transphorm, Genesic, Cree, Interfet) but no final solution yet The control circuitry to enable and disable the HV switches also being investigated. 20

I Backup slides A cascode scheme allows using lower voltage devices to achieve high voltage switching More components, more area required Increased rds on

Some works on the SPICE model for the FSICBH057A120 SiC BJT, provided by Dave, will run some simulations with the control circuitry. * The I b seems to be a little high for the control circuit as is, need some modifications to it (DGT, 2 nd stage, cascode) Drain current with 500 Ohm Base Series Resistor (Rc = 231k) Base current with 500 Ohm Base Series Resistor (Rc = 231k) II Backup slides