1. High Voltage MUX for ATLAS Tracker Upgrade EG Villani STFC RAL on behalf of the ATLAS HVMUX group IPRD16 Siena, Oct. 2016

2. Introduction: ATLAS Upgrade HVMUX
Outlook
Introduction: ATLAS Upgrade HVMUX
HVMUX project design and radiation test results
Summary & conclusions 1

3. ATLAS Phase II Tracker Upgrade
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
Challenges facing HL-LHC silicon detector upgrades
Higher Occupancies ( 200 interactions / bunch crossing)
Finer Segmentation
Higher Particle Fluences ( 1014 outmost layers to  1016 innermost layers
Increased Radiation Tolerance ( 10 increase in dose w.r.t. ATLAS )
Larger Area (~200 m2)
Cheaper Sensors
More Channels
Efficient power/bias distribution / low material budget
From 1E33 cm-2 s-1
…to 5E34 cm-2 s-1 2

4. HV distribution in ATLAS Upgrade
Current SCT uses independent powering for the 4088 detector modules.
Each sensor has its own independent HV bias line.
One HV bias line for each sensor is somewhat an ‘ideal’ solution:
High Redundancy
Individual enabling or disabling of sensors and current monitoring
But the increased number of sensors in the Upgraded Tracker (>10 k modules in barrel upgrade vs. 2 k in present barrel) implies a trade off among material budget, complexity of power distribution and number of HV bias lines.
Amend the statement about increased number of sensors and existing HV conductors 3

5. High Voltage Multiplexing
Use single (or more) HV bus to bias all 13 sensors in a ½ stave and use one HV switch for each sensor to disable malfunctioning sensors: High Voltage Multiplexing ‘HVMUX’
The HV switch is DCS controlled
Additional circuitry to measure bias currents on each sensor, will likely be implemented in AMAC ASIC
HV (-500V)
DCS controlled HV switch 4

6. HVMUX strips implementation
DC2DC
HVMUX switch area
20 mm
9 mm
RAL ABC130 Barrel Short Strip Module. The HVMUX will sit on the lower side of the power board.
The final version will incorporate power and HVMUX elements in the AMAC (Autonomous Monitor And Control) ASIC. 5

7. HVMUX devices requirements
High Voltage switches strip detector requirements:
Must be rated to 500V plus a safety margin
Must be radiation hard, nominal maximum expected 1x1015 neq/cm2 ,  30Mrad (Si) for strip end cap. Multiply by 2 to include safety margin
On-state impedance Ron << 1kΩ // Ion  10mA (for irradiated strip sensors)
Off-state impedance Roff >> 1GΩ // Ilkg << Isens
Must be unaffected by magnetic field
Must maintain satisfactory performance at -30 C
Must be small (mass/area constraint) and cheap (around 1E4 needed)
Several different type of devices investigated, including Si, SiC, GaN.
Currently GaN seems very promising. 6

8. HVMUX schematic design
Oscillator (will be included in AMAC)
AC coupled Charge pump step-up converter
Load Res 150k
LV meas
HV meas
GaN FET
RC filter
Protection Zener
HV (-550V)
PGA26E16BV
Schottky array
HVMUX board ver. 2.0 with Panasonic PGA26E19 (600V MAX) in die form,
can accommodate Si PJFET from CNM (later slides) 7

9. HVMUX test R2=R3=2.2k; R4=100k; NO C8 ON transient OFF transient
Time plots of output voltage (i.e. measured at HVout node).
HVin set to – 550V
Turning ON Settling time  45 ms MAX, including turning ON of OSC
Turning OFF Settling time  20 ms MAX including turning OFF of OSC
With Vcc = 2.0V, VR4  2.03V
Maximum Leakage current through HV switch in OFF state:  V 8

10. HV GaN FET devices
GaN shows direct, higher BG and DE w.r.t. to many semiconductors, rendering it inherently more radiation hard.
Its high critical field makes it ideal for HV applications.
An HEMT GaN relies on a 2DEG formed by strained GaN under an AlGaN layer. The 2DEG has very high conductivity, leading to high current density achievable.
Interrupting the 2DEG using a pGaN makes an equivalent of an enhancement MOS transistor 9

11. HVMUX devices n real time radiation tests
Cables  10 m
Keithley 2602
Keithley HV 2410
Keithley 7002 with:
10ch HV switch card ch LV switch card 7011
55 mm
USB2 GPIB
HVMUX
DUTs BOARD
77 mm
USB
30 mm
reactor
GaN FET EPC2027 and PGA26E19BV irradiated and monitored in real time during irradiation with neutrons at TRIGA reactor (Ljubliana) 10

12. HVMUX devices n real time radiation tests
USB
The HVMUX boards is inserted into the TIC channel:
The source meters/PC placed on top of the reactor
Additional devices irradiated ‘passively’ and measured after deactivation
Diodes, EPC2027, GS66502B, PGA26E19BV 11

13. HVMUX devices n real time radiation tests
G4/FLUKA simulations of n spectrum in Upgraded Strip barrels 1 & 2 12

14. HVMUX devices n real time radiation tests
expected strips n fluence
MAX n fluence
time
1E15
2E15
3E15
4E15
5E15
6E15
7E15
8E15
9E15
1E16
50’’x21
x25
x25
x25
x25
x25
x25
x25
x25
x25
x25
170’’
24’
48’
72’
96’
120’
144’
168’
192’
216’
240’
Irradiation test: 25 cycles of ‘rds on’ test followed by 1 cycle of ‘HV test’, for a total time, for each cycle, of about 1440’’. The timing sequence is controlled by a Labview program on PC
During the ‘rds on’ phase a current of 10 mA is injected into the devices, turned on with a Vgs = 2V. During the ‘HV test’ the devices are turned off, a HV is applied and Vgs is swept [0,2]V
Total fluence of 1E16 n / (4*hr), with a fluence of 1E15 every 1440 seconds, or 24 mins.
The minimum ‘fluence resolution' during the ‘rds on’ test approx. 3.5E13.
The total fluence of fast neutrons around 3.27E15. 13

15. HVMUX devices n real time radiation tests
HVMUX board
Insertion in TIC channel
Reactor on!
Cerenkov light
The response of the devices during n irradiation is monitored in real time 14

16. HVMUX devices n radiation tests results
IM16-17
VR4
EPC2027 – rds ON
PGA26E19BV – rds ON
All the 4 EPC2027 (300V MAX) and the 4 PGA26E19BV (600V MAX) devices survived the irradiation test
Plots show average and deviation of measured Vds and Igs vs. n fluence
Negligible changes in rdson
ON state below 5 μA, slight decrease in the PGA26E19BV 15

17. HVMUX devices n radiation tests results
IGS(A)
IDS(A)
VGS(V)
Φ=0
IM16-17
Vds=550V
Vds=300V
Φ=1E15
Φ=5E15
Φ=1.1E16
EPC2027 – HV Ids vs. Vgs
PGA26E19BV – HV Ids vs. Vgs
VR4
During the HV test, the Vgs of the EPC2027 swept between [0, 2.5]V/50mV
Plots show average and deviation of Ids and Igs vs. Vgs vs. fluence
Some changes in Vth of PGA26 (increases by approx. 250mV)
Bigger changes in Vth of EPC2027(decreases by approx. 750mV)
Minor changes in leakage Ids of PGA26, slightly bigger in the EPC2027
Minor Igs leakage of PGA26, more pronounced in the EPC2027
N.B. The pre-irradiation plots (yellow in the picture) were obtained using a different HV board holding the devices (i.e. not the same used for the online test)-> actual leakage current of irradiated devices is less than what shown on the plot 16

18. HVMUX devices n radiation tests results
Plots of #4 BAT54SDW-TP Schottky devices irradiated (6E15 n) and non-irradiated.
Average Plots of #2 5V1 Zener devices irradiated (6E15 n) and non-irradiated. 17

19. HV MUX next steps Received 130 bare dies from Panasonic of PGA26E19BV
Next TID radiation test on batch of devices (GaN, diodes, etc.) will include proton, gamma and pions (completed some days ago). Further test with n to validate their radiation hardness to bulk damage
Proton irradiation at Birmingham and Gamma irradiation at BNL to evaluated TID effects on GaN devices
SEE effects and long term reliability for GaN devices will be investigated
At the same time a Si based solution is being investigated in collaboration with CNM (Barcelona) for a P-type JFET rated for High Voltage. Prototype devices already fabricated and radiation tests in progress 18

20. Summary & Conclusions HVMUX project progressing well
After several tests and investigation of different devices, very promising results from GaN devices: prototypes work reliably up to 550V even after 1E16 n irradiation. Additional GaN devices, rated for lower voltage, successfully tested with protons
Extensive irradiation campaign ongoing to validate radiation hardness, SEE sensitivity and long term reliability (and long term availability of devices)
A Si PJFET is also being investigated
Prototype HVMUX switch built and successfully tested. Detailed tests of HVMUX switch on a strip module planned for Atlas Upgrade Week THANK YOU 19

21. Backup slides EPC2012 test results with protons 26MeV
Vgs(V)
Average Ig and 1 (device ON).
Average Ids and Igs Vds=200V (device OFF).
EPC2012 test results with protons 26MeV
GaN devices rated for up to 200V ( up to 600V would be needed for HV MUX but stacked configuration possible – see later slides) I

22. Backup slides 270V Zeners EPC2027s
We could use HV devices rated for lower voltage than needed and stack them on each other to achieve higher voltage switching
the biasing circuit needs careful designing, to avoid overvoltages and / or excessive leakage
Modeled circuit with parasitic resistor values taken from measurements of our own EPC devices.
270V Zeners
EPC2027s II

23. Backup slides Rload = 100 K III SW OFF SW ON – 130V – 230V – 330V
HV ON transient
Rload = 100 K
Time plot of output voltage (i.e. measured across Rload) for different Vin
VinMAX set to – 530V, Icmpl = 6 mA
Turning ON Settling time  100 ms
Turning OFF Settling time  200 ms
With Vreg = 2.5V, VR4  3.3V, IU1  1.4 mA
Maximum Leakage current :  V
III

24. Backup slides Vgs(V) Igs(A) Ids(A) IV
Average Plots of #3 devices irradiated offline and #3 devices non-irradiated PGA26E19BV. Neutron fluence 6E15.
At Vgs = 2V the Igs is < 1 uA, average Vds = Ids = 10mA IV

25. HV devices investigated
HV Si, SiC and GaN based devices are being investigated
FAILED
IRR- to be tested
PASS – need conf.
T.B.T.
PASS – N.A. V