1. Radiation performance of new semiconductor power devices for the LHC experiment upgrades C. Abbate, M. Alderighi, S. Baccaro, G. Busatto, M. Citterio, P. Cova, N. Delmonte, V. De Luca, S. Fiore, S. Gerardin, E. Ghisolfi, F. Giuliani, F. Iannuzzo, A. Lanza, S. Latorre, M. Lazzaroni, G. Meneghesso, A.A. Paccagnella, F. Rampazzo, M. Riva, A. Sanseverino, R. Silvestri, B.G. Spiazzi, F. Velardi, E. Zanoni APOLLO Project INFN Milano, Padova, Pavia, Roma1, ENEA, INAF, FN S.p.A. and Universities of Cassino, Milano, Padova, Parma

2. S. Gerardin on behalf of the APOLLO collaboration 2 Outline  A brief introduction to APOLLO  Radiation sensitivity of SiC devices  Devices and technology  Experimental setup for Single Event Effects (SEE)  Heavy ions irradiation  Radiation sensitivity of GaN devices  Devices and technology  Heavy ions irradiation  Low-energy protons irradiation  Conclusions

3. S. Gerardin on behalf of the APOLLO collaboration 3 APOLLO Alimentatori di POtenza per aLti Livelli di radiaziOne Collaboration among several INFN sections Milano Padova Pavia Roma1 Goal: improvement of the radiation tolerance of power supplies for the LHC phase 2 upgrade High radiation levels during phase 2 Total dose: up to 10kGy(Si) Protons: 2x10 13 protons/cm 2 Neutrons:7.7x10 13 neutrons/cm 2

4. S. Gerardin on behalf of the APOLLO collaboration 4 SiC Power MOSFETs: Devices Devices Planar SiC power MOSFETs Manufactured by CREE and STMicroelectronics Blocking voltage 1200 V Maximum I DS 31,6 A continuous, 60 A pulsed (25°C) SiC Benefits SiC like GaN are wide bandgap materials SiC junction can block high voltage in volumes much smaller than Si counterparts SiC is hard to displacement effects. Minimum energy to displace SiC atoms is larger than in Si and GaAs SEE Sensitive Volumes smaller than Si power MOSFET

5. S. Gerardin on behalf of the APOLLO collaboration 5 SiC Power MOSFETs: Experiments Performed/Ongoing Irradiations High-energy protons (final paper)  rays (final paper) Neutrons (final paper) Heavy ions (this presentation) Irradiations at INFN Laboratori Nazionali di Legnaro

6. S. Gerardin on behalf of the APOLLO collaboration 6 Drain P + N + P _ GateSource N _ Body N + SiC: Single Event Effects An ionizing particle releases charge in the active volume, i.e. the region where high electric field develops in the blocking condition The charge interacts with the device and may cause a strong increase in the local current, destroying its structure (Single Event Burnout, SEB) or may cause a large electric field across the gate oxide capable of destroying it (Single Event Gate Rupture, SEGR) SEB and SEGR can be recognized by increases in the leakage current at the drain and gate, respectively

7. S. Gerardin on behalf of the APOLLO collaboration 7 Cg Cd 50  1 M  Vgs Impacting Ion DUT Vds The current pulses SiC: SEE Setup Bias is applied at the gate and drain Gate and drain leakage currents are monitored during and after the irradiation The drain current pulses associated with the impacts of the particles are logged during the irradiation. The integral of these current pulses supplies the collected charge A post irradiation statistical analysis allows us to find the mean value of the collected charge

8. S. Gerardin on behalf of the APOLLO collaboration 8 The charge collected is larger than the one deposited by the impacting particles (charge amplification) It is lower than the one measured in equivalent Si power devices (MOSFETs and IGBTs), confirming the smaller SEB sensitive volume ST SiC: Heavy-Ion Results 266-MeV Iodine, LET > 60 MeV∙mg -1 ∙cm 2

9. S. Gerardin on behalf of the APOLLO collaboration 9 Operating area is limited by SEGR. SEB is less of a concern The lower sensitivity to SEB, at the LET used for the irradiation, is paid off with a larger sensitivity to SEGR The gate oxide, apart from the interface, is similar to that used in standard Si MOSFETs ST SiC: Safe Operating Area 266-MeV Iodine, LET > 60 MeV∙mg -1 ∙cm 2

10. S. Gerardin on behalf of the APOLLO collaboration 10 SRIM simulation of Bromine @ 20 - 550MeV The thickness of the surface structure (metal + polysilicon + SiO 2 ) is about 5  m The thickness of the epitaxial layer is about 10  m CREE SiC: Energy Deposition

11. S. Gerardin on behalf of the APOLLO collaboration 11 Measured mean value of the collected charge vs. drain voltage used to bias the device during the irradiation Each point corresponds to the mean value of the collected charge over a population of 1500 events Charge amplification is very strong for 79 Br at 240MeV for which damages can be observed in the vertical device structure CREE SiC: Collected Charge

12. S. Gerardin on behalf of the APOLLO collaboration 12 I DSS and I GSS after each 20-MeV Br irradiation step I DSS and I GSS after each 60-MeV Br irradiation step SiC: SEGR and SEB after 20MeV and 60MeV Br No damage was found during irradiation with 79 Br @ 20MeV SEGR (I GSS >1  A) occurred during irradiation at V DS =750V with 79 Br @ 60MeV

13. S. Gerardin on behalf of the APOLLO collaboration 13 I DSS and I GSS after each 240 MeV Br irradiation step I DSS and I GSS after each 550 MeV Br irradiation step SiC: SEGR and SEB after 240MeV and 550MeV Br A SEGR (I GSS >1  A) is detected in the irradiation with 79 Br @ 240MeV and 550 MeV at V DS =150V and V DS =100V The results confirm the weakness of the gate oxide of SiC power MOSFETs

14. S. Gerardin on behalf of the APOLLO collaboration 14 GaN HEMTs: Devices GaN High Electron Mobility Transistors Enhancement-mode GaN transistors Manufactured by Efficient Power Conversion (EPC) Blocking voltage 40/200 V Maximum I DS is 12/33 A continuous, 60/150 A pulsed From EPC website

15. S. Gerardin on behalf of the APOLLO collaboration 15 GaN HEMTs: Potential for Radiation Hardness Material GaN is quite hard to displacement effects. Minimum energy to displace GaN atoms is larger than in Si and GaAs, close to SiC Heterostructure Channel is formed through band engineering In principle, no dielectric layers are used underneath the gate, so tolerance to total dose is expected to be excellent From EPC website

16. S. Gerardin on behalf of the APOLLO collaboration 16 GaN HEMTs: Potential for Radiation Hardness (2) Questions marks No information are provided as to how enhancement- mode has been achieved Some additional layers have been probably introduced to engineer the positive threshold voltage What’s the impact on radiation hardness? From EPC website

17. S. Gerardin on behalf of the APOLLO collaboration 17 GaN HEMTs: Experiments Performed/Ongoing Irradiations  rays (final paper) Heavy ions (this presentation) High-energy protons (this presentation) Low-energy 1.8-MeV protons (this presentation)

18. S. Gerardin on behalf of the APOLLO collaboration 18 GaN HEMTs: Experiments (3) Heavy-ion irradiation at Laboratori Nazionali del Sud Maximum LET of about 50 MeV∙mg -1 ∙cm 2 High-energy proton irradiation at Laboratori Nazionali del Sud 60 MeV protons Fluence up to 2∙10 13 p/cm 2 Low-energy proton irradiation at Laboratori Nazionali di Legnaro 1.8 MeV protons Fluence up to 4∙10 14 p/cm 2 Energy is too low, to give rise to secondary particles  no indirect SEE Ionization and displacement damage

19. S. Gerardin on behalf of the APOLLO collaboration 19 100V devices Used ions: 266-MeV Iodine and 240-MeV Bromine No SEEs Large charge amplification  evidence of a regenerative effect 266MeV I LET ~ 50 MeV∙mg -1 ∙cm 2 GaN HEMTs: Heavy Ions irradiation

20. S. Gerardin on behalf of the APOLLO collaboration 20 GaN HEMTs: 60MeV Proton Irradiation 100V devices No SEE We experienced some problems in the testing, mainly due to the intrinsically different structure and behavior of the devices under test  no waveforms are available

21. S. Gerardin on behalf of the APOLLO collaboration 21 GaN HEMTs: Low-energy Protons Effects DC parameters have been measured before and after 1.8 MeV proton irradiation Observed effects include: Increase in gate current Threshold voltage reduction Transconductance drop We investigated damage dependence on proton fluence blocking voltage

22. S. Gerardin on behalf of the APOLLO collaboration 22 GaN HEMTs: Post-rad Gate Current Increase in gate current at all voltages, up to one order of magnitude More pronounced for negative voltage Some room temperature annealing Device exposed to 10 14 p/cm 2 in unbiased conditions

23. S. Gerardin on behalf of the APOLLO collaboration 23 GaN HEMTs: Post-rad Drain Current Unexpected decrease in threshold voltage (1V) Degraded substreshold slope Modest room-temperature annealing Device exposed to 10 14 p/cm 2 in unbiased conditions

24. S. Gerardin on behalf of the APOLLO collaboration 24 GaN HEMTs: Post-rad Drain Current Peak transconductance drop, more than 30% Drain current almost unchanged: threshold voltage reduction offset transconductance degradation Device exposed to 10 14 p/cm 2 in unbiased conditions

25. S. Gerardin on behalf of the APOLLO collaboration 25 GaN HEMTs: Literature Data on Depletion-mode GaN Drain current decrease is typically reported on depletion-mode devices in the literature, in contrast with the behavior of EPC samples Difference due to unknown process steps or introduction of dielectric layers? From Karmakar et al., IEEE- TNS 2004

26. S. Gerardin on behalf of the APOLLO collaboration 26 GaN HEMTs: Drain Current Reduction Compilation of several test data over the course of ten years for GaN HEMTs All devices show decreases in drain From Weaver et al., IEEE-TNS 2012

27. S. Gerardin on behalf of the APOLLO collaboration 27 GaN HEMTs: Discussion Small dependence of degradation on blocking voltage: 40V vs 200V Small dependence of degradation on proton fluence Sample-to-sample variability Possible superimposition of displacement effects and ionization effects Unbiased conditions may not be the worst-case, if ionization effects are present Total ionizing dose data needed to clarify this Degradation only at very high fluence of low-energy protons (worst-case condition for ionization and displacement)

28. S. Gerardin on behalf of the APOLLO collaboration 28 Conclusions SiC and GaN power devices are being extensively tested under different types of radiation, in the framework of the APOLLO R&D collaboration, aiming to use these new technologies for designing power supplies for the future LHC experiments upgrades SiC power MOSFETs are expected to show very good performances in terms of total ionizing dose, but exhibited some issues with SEGR Enhancement-mode GaN transistors displayed considerable hardness to low energy protons, and practical immunity from SEEs Further measurements and irradiations are ongoing, to provide a complete picture of radiation effects in these devices