betaFeedback

WorksPhysics and Astronomy

Leakage current of high-fluence neutron-irradiated 8" silicon sensors for the CMS Endcap Calorimeter Upgrade

Abstract

The HL-LHC will challenge the detectors with a nearly 10-fold increase in integrated luminosity compared to the previous LHC runs combined, thus the CMS detector will be upgraded to face the higher levels of radiation and the larger amounts of collision data to be collected. The High-Granularity Calorimeter will replace the current endcap calorimeters of the CMS detector. It will facilitate the use of particle-flow calorimetry with its unprecedented transverse and longitudinal readout/trigger segmentation, with more than 6M readout channels. The electromagnetic section as well as the high-radiation regions of the hadronic section of the HGCAL (fluences above 1014 neq./cm210^{14}~n_{eq.}/cm^{2}) will be equipped with silicon pad sensors, covering a total area of 620 m2^2. Fluences up to 1016 neq./cm210^{16}~n_{eq.}/cm^{2} and doses up to 1.5 MGy are expected. The sensors are processed on novel 8" p-type wafers with an active thickness of 300 μmμm, 200 μmμm and 120 μmμm and cut into hexagonal shapes for optimal use of the wafer area and tiling. Each sensor contains several hundred individually read out cells of two sizes (around 0.6 cm2^2 or 1.2 cm2^2). To investigate the radiation-induced bulk damage, the sensors have been irradiated with neutrons at RINSC to fluences between 6.5×1014 neq./cm26.5 \times 10^{14}~n_{eq.}/cm^{2} and 1.4×1016 neq./cm21.4 \times 10^{16}~n_{eq.}/cm^{2}. Electrical characterization results are presented for full sensors, as well as for partial sensors cut from multi-geometry wafers with internal dicing lines on the HV potential within the active sensor area. Leakage current behaviour is investigated for various sensor types and fluence levels, including its temperature dependence. Finally, methods to limit the annealing time of the sensors during irradiation are investigated by analysing the impact of splitting high-fluence irradiations.

*Corresponding author. Email: [email protected] G. Di Guglielmo {sup}22 L. Diehl {sup}16 K. Dilsiz {sup}29 G. G. Dincer {sup}34,16 J. Dittmann {sup}6`

M. Dragicevic {sup}26 D. Du {sup}7 B. Dubinchik {sup}20 S. Dugad {sup}61 F. Dulucq {sup}47 I. Dumanoglu {sup}2,68

B. Duran {sup}32 S. Dutta {sup}38 V. Dutta {sup}15 A. Dychkant {sup}18 M. Dünser {sup}16 F. Ebode Onyie {sup}31

T. Edberg {sup}41 I. T. Ehle {sup}49 A. El Berni {sup}47 F. Elias {sup}40 S. C. Eno {sup}41 E. N. Erdogan {sup}66

B. Erkmen {sup}66 Y. Ershov {sup}20 E. Y. Ertorer {sup}15 S. Extier {sup}47 L. Eychenne {sup}49`

Y. E. Fedar {sup}66 G. Fedi {sup}27 Y. Feng {sup}59 E. Fialova {sup}16 J. P. Figueiredo De Sá Sousa De Almeida {sup}16 B. A. Fontana Santos Alves {sup}49 E. Frahm {sup}42 K. Francis {sup}18 J. Freeman {sup}22`

T. French {sup}16 F. Gaede {sup}19 P. K. Gandhi {sup}22 S. Ganjour {sup}54 A. Garcia-Bellido {sup}53`

F. Gastaldi {sup}49 J. D. Gaytan Villarreal {sup}15 L. Gazi {sup}38 Z. Gecse {sup}22 J. Gehrke {sup}42 H. Gerwig {sup}16

O. Gevin {sup}54 S. Ghosh {sup}62 S. Ghosh {sup}49 K. Gill {sup}16 C. Gillespie {sup}23 S. Gleyzer {sup}3

N. Godinovic {sup}55 P. Goettlicher {sup}19 R. Goff {sup}23 A. Golunov {sup}20 J. D. González Martínez {sup}47`

N. Gorbounov {sup}20 L. Gouskos {sup}13 L. Gray {sup}22 C. Grieco {sup}63 S. Groenroos {sup}25`

D. Groner {sup}16 A. Gruber {sup}16 A. Grummer {sup}22 S. Grönroos {sup}16 D. Guerrero {sup}22`

F. Guilloux {sup}54 Y. Guler {sup}2,69 M. Gumustekin {sup}31 S. Gundacker {sup}26 A. D. Gungordu {sup}31`

K. Guo {sup}45 E. Gurpinar Guler {sup}2,69 H. K. Gutti {sup}22 E. Gülmez {sup}10 B. Hacisahinoglu {sup}32`

Y. Halkin {sup}50 G. Hamilton Ilha Machado {sup}64 H. S. Hare {sup}53 K. Hatakeyama {sup}6

A. H. Heering {sup}46 V. Hegde {sup}6 U. Heintz {sup}13 N. Hinton {sup}13 A. Hinzmann {sup}19`

J. Hirschauer {sup}22 J. Hoff {sup}22 İ. Hos {sup}32,79 B. Hou {sup}7 X. Hou {sup}7 Y. Hou {sup}7 A. Howard {sup}27`

H. Hsieh {sup}58 T. Hsu {sup}58 F. Hummer {sup}34 S. Hung {sup}58 C. Huse {sup}63 M. Imran {sup}30 J. Incandela {sup}63`

E. Iren {sup}66,84 B. Isildak {sup}66 P. S. Jackson {sup}21 W. J. Jackson {sup}42 S. Jain {sup}62`

J. Jaroslavceva {sup}16 A. Jige {sup}63 P. P. Jordano {sup}63 U. Joshi {sup}22 K. Kaadze {sup}33`

V. Kachanov {sup}50 A. Kafizov {sup}35 A. Kafle {sup}18 L. Kalipoliti {sup}49 A. Kallil Tharayil {sup}15`

O. Kaluzinska {sup}16 S. Kamble {sup}28 A. Kaminskiy {sup}6 M. Kandemir {sup}10 M. Kanemura {sup}15`

H. Kanso {sup}9 Y. Kao {sup}58 A. Kapic {sup}48 C. Kapsiak {sup}42 V. Karjavine {sup}20 S. Karmakar {sup}58

A. Karneyeu {sup}46 R. Kaur {sup}3 M. Kaya {sup}10,71 A. Kayis Topaksu {sup}2 B. Kaynak {sup}32`

F. A. Khan {sup}14 A. Khudiakov {sup}50 F. Khuzhaimah {sup}58 J. Kieseler {sup}34 R. S. Kim {sup}23 S. King {sup}45

T. Klijnsma {sup}22 E. G. Kloiber {sup}15 M. Klute {sup}34 Z. Kocak {sup}16 K. R. Kodali {sup}61 K. Koetz {sup}23

T. Kolberg {sup}23 O. B. Kolcu {sup}66,83 J. R. Komaragiri {sup}28,78 M. Komm {sup}19 M. Kovac {sup}56`

H. A. Krause {sup}34 K. Kristiansen {sup}63 A. Kristic {sup}55 M. Krohn {sup}42 B. Kronheim {sup}41`

P. Krueper {sup}27 K. Krüger {sup}19 S. Kulis {sup}16 M. Kumar {sup}54 S. Kumar {sup}62 R. Kumar Verma {sup}25

A. Kunts {sup}50 C. Kuo {sup}17 A. Kurenkov {sup}20 V. Kuryatkov {sup}59 S. Kyre {sup}63 J. Ladenson {sup}13

A. Laffitte {sup}47 P. Lai {sup}7 G. Landsberg {sup}13 J. Langford {sup}27 A. Laudrain {sup}19`

R. Laughlin {sup}23,75 J. Lawhorn {sup}34 O. Le Dortz {sup}49 S. W. Lee {sup}59 A. Lektaouers {sup}52`

D. Lelas {sup}55 M. Leon {sup}16 L. Levchuk {sup}37 D. S. Li {sup}45 J. Li {sup}19 P. Y. Li {sup}23 Y. Li {sup}58 Z. Liang {sup}8

H. Liao {sup}7 C. Lin {sup}17 K. Lin {sup}19 Z. Lin {sup}67 D. Lincoln {sup}22 L. Linssen {sup}16 A. Litomin {sup}50`

G. Liu {sup}7 H. Liu {sup}58 Y. Liu {sup}7 T. Loiseau {sup}16 B. Lopes {sup}16 C. Lu {sup}67 R. Lu {sup}58 P. Lukens {sup}22

M. Mackenzie {sup}45 C. Madrid {sup}59 A. Magnan {sup}27 F. Magniette {sup}49 A. Mahjoub {sup}49`

D. Mahon {sup}42 G. Majumder {sup}62 V. Makarenko {sup}50 A. Malakhov {sup}20 L. Malgeri {sup}16`

S. Mallios {sup}27 C. Mandoli {sup}61 A. Mankel {sup}59 M. Mannelli {sup}16 M. Manoni {sup}49 J. Mans {sup}42

C. Mantilla {sup}64 G. Martinez {sup}23 C. Massa {sup}49 P. Masterson {sup}63 M. Mattheweman {sup}16`

V. Matveev {sup}20 S. Mayekar {sup}61 I. Mazlov {sup}23 M. Mazza {sup}16 A. Mehta {sup}49 A. Mestvirishvili {sup}29

Y. Miao {sup}45 G. Miele {sup}19 I. R. Mirza {sup}61 S. Moccia {sup}16 G. B. Mohanty {sup}61 F. Monti {sup}16

F. Moortgat {sup}16 M. C. Muehlnikel {sup}16 S. Murthy {sup}15 J. Music {sup}55 Y. Musienko {sup}46,50` J. W. Nelson {sup}57 I. Neutelings {sup}16 N. Nguyen {sup}3 J. Niedziela {sup}19 A. Nikitenko {sup}50 D. Noonan {sup}22 M. Noy {sup}16 K. Nurdan {sup}10,73 S. Obraztsov {sup}49 C. Ochando {sup}49 J. Offermann {sup}13 H. Ogul {sup}29 J. Olsson {sup}23 E. P. Onakpojeruo {sup}31 Y. Onel {sup}29 S. Ozkorucuklu {sup}32 E. Paganis {sup}58 P. Palit {sup}15 C. Palmer {sup}41 R. Pan {sup}67 F. Pantaleo {sup}16 V. Papadimitriou {sup}22 C. Papageorgakis {sup}41 S. Paramesvaran {sup}12 M. M. Paranjpe {sup}41 E. Parker {sup}23 N. Parmar {sup}53 S. Parolia {sup}62 A. G. Parsons {sup}21 P. Parygin {sup}53 N. Pastika {sup}22 M. Paulini {sup}15 C. Paus {sup}43 K. Peñaló Castillo {sup}23 K. Pedro {sup}22 V. Pekic {sup}55 T. Peltola {sup}59 B. Peng {sup}67 A. Perego {sup}16 S. Pereira {sup}61 A. Petrilli {sup}39 T. Pierre-Emile {sup}49 S. K. Podem {sup}62 V. Popov {sup}37 L. Portales {sup}54 O. Potok {sup}32 P. B. Pradeep {sup}27 R. Pramanik {sup}62 D. Primc {sup}16 H. Prosper {sup}23 M. Prvan {sup}55 H. Qu {sup}16 A. Quamesh {sup}34 T. Quast {sup}16 L. Rabour {sup}49 N. Raicevic {sup}48 C. Ramon Alvarez {sup}64 M. A. Rao {sup}30 K. Rapacz {sup}16 W. Redjeb {sup}1 M. Reinecke {sup}19 E. Reinhardt {sup}3 M. Revering {sup}42,81 A. Roberts {sup}15 J. Rohlf {sup}11 P. Rosado {sup}39,16 A. Rose {sup}27 S. Rothman {sup}43 P. K. Rout {sup}17 M. Rovere {sup}16 N. Rubinov {sup}22 P. Rumerio {sup}3 R. Rusack {sup}42 L. Rygaard {sup}19 V. Ryjov {sup}16 S. Sadivnycha {sup}36 M. Ö. Sahin {sup}54 M. A. Sailer {sup}16 U. Sakarya {sup}66 R. Salerno {sup}49 R. Salvatico {sup}16 R. Saradhy {sup}42 M. Saraf {sup}62 K. Sarbandi {sup}16 M. A. Sarkisla {sup}10,74 I. Satyshev {sup}20 N. Saud {sup}30 J. Sauvan {sup}49 G. Schindler {sup}57 A. Schmidt {sup}1 I. Schmidt {sup}29 M. H. Schmitt {sup}45 A. Sculac {sup}55 T. Sculac {sup}56 A. Sedelnikov {sup}50 C. Seez {sup}27 F. Sefkow {sup}19 D. Selivanova {sup}19 V. Sergeychik {sup}50 H. Sert {sup}32 O. Sevinc Kaya {sup}10,72 M. Shahid {sup}30 P. Sharma {sup}62 S. Sharma {sup}51 M. Shelake {sup}61 C. Shen {sup}24 A. Shenai {sup}22 R. Shinde {sup}62 D. Shmygol {sup}50 R. Shukla {sup}27 E. Sicking {sup}16 P. Silva {sup}16 F. Simon {sup}34 C. Simsek {sup}32 E. Simsek {sup}66 Y. Sirois {sup}49 G. Sokmen {sup}49 S. Song {sup}7 Y. Song {sup}49,67 G. Soudais {sup}54 A. Sritharan {sup}19 R. R. St Jacques {sup}33 M. Stamenkovic {sup}13 A. Steen {sup}16 J. Stein {sup}15 J. Strait {sup}22 N. Strobbe {sup}42 X. Su {sup}58 E. Sukhov {sup}20 A. Suleiman {sup}5 D. Sunar Cerci {sup}66 P. Suryadevara {sup}61 K. Swain {sup}62 C. Syal {sup}22 S. Taj {sup}17 B. Tali {sup}2,70 K. Tanay {sup}51 W. Tang {sup}17 A. Tanvir {sup}30 J. Tao {sup}7 T. Tatli {sup}10 R. Taylor {sup}33 Z. C. Taysi {sup}66 G. Teafoe {sup}22 W. Terrill {sup}15 D. Thienpont {sup}47 R. Thomas {sup}61 R. Thomas {sup}27 M. Titov {sup}54 C. Todd {sup}21 E. Todd {sup}23,76 U. G. Tok {sup}2 M. Toms {sup}34 A. Tosun {sup}32,79 J. Troska {sup}16 L. Tsai {sup}58 Z. Tsamalaidze {sup}60 D. Tsionou {sup}58 G. Tsipolitis {sup}4 M. Tsirigoti {sup}16 R. Tu {sup}7 S. N. Tural Polat {sup}66 S. Undleeb {sup}59 L. Urda Gómez {sup}49 E. Usai {sup}3 E. Uslan {sup}2 V. Ustinov {sup}20 A. Uzunian {sup}50 M. Velasco {sup}45 E. Vernazza {sup}16 O. Viahin {sup}36 A. Vijay {sup}28 T. Virdee {sup}27 E. Voirin {sup}22 M. Vojinovic {sup}27 N. Voytishin {sup}20 T. Á. Vámi {sup}63 A. Wade {sup}23,77 D. Walter {sup}16 C. Wang {sup}7 C. Wang {sup}7 F. Wang {sup}7 H. Wang {sup}7 J. Wang {sup}45 K. Wang {sup}67 X. Wang {sup}22 X. Wang {sup}8 Y. Wang {sup}8 Z. Wang {sup}24 E. Wanlin {sup}49 M. Wayne {sup}46 J. Wetzel {sup}29 A. Whitbeck {sup}22,59 R. Wickwire {sup}22 D. Wilmot {sup}63 J. Wilson {sup}6 H. Wu {sup}58 T. Wu {sup}13 M. Xiao {sup}67 H. Yang {sup}7 K. Yaz {sup}66,85 B. Yazici {sup}10,73 Y. Ye {sup}67 T. Yetkin {sup}66,83 R. Yohay {sup}23 T. Yu {sup}7,19 X. Yuan {sup}7 F. Yuksel {sup}66 O. Yuksel {sup}66 I. Yushmanov {sup}19 I. Yusuf {sup}49 A. Zabi {sup}49 A. Zada {sup}7 M. Zalikhа {sup}23 D. Zareckis {sup}21 A. Zarubin {sup}20 P. Zehetner {sup}16 A. Zghiche {sup}7 C. Zhang {sup}7 D. Zhang {sup}63 H. Zhang {sup}7 J. Zhang {sup}7 J. Zhang {sup}7 Z. Zhang {sup}7 J. Zhong {sup}67 Y. Zhou {sup}15 H. Zhu {sup}7 Ç. Zorbilmez {sup}32 E. Zubaroglu {sup}16`

3^{3}The University of Alabama, 500 University Blvd East, Tuscaloosa 35401 AL, USA

4^{4}National Technical University of Athens 28 Oktovriou (Patision) 42, 10682 Athens, Greece

5^{5}University of Bahrain, P.O. Box 32038, Bahrain

6^{6}Baylor University, Waco 76706, TX, USA

7^{7}Institute of High Energy Physics, Chinese Academy of Sciences, 19B Yuruanlu, Shijingshan District, Beijing, China, 100049

8^{8}Tsinghua University, Beijing, 100084, China

9^{9}The Lebanese University, 14 Badaro, Museum, Beirut, Lebanon

10^{10}Boğaziçi University, Bebek, 34342 Istanbul, Türkiye.

11^{11}Boston University, Boston, Massachusetts, USA

12^{12}University of Bristol, Beacon House, Queens Road, Bristol BS8 1QU, UK

13^{13}Brown University, 182 Hope Street, Providence 02912, RI, USA

14^{14}Université Libre de Bruxelles, Boulevard du Triomphe, B-1050 Bruxelles

15^{15}Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh 15213, PA, USA

16^{16}CERN, Espl. des Particules 1, 1211 Geneve 23, Switzerland

17^{17}National Central University, Chung-Li, Taiwan, ROC

18^{18}Northern Illinois University, 1425 W. Lincoln Hwy., DeKalb 60115, IL, USA

19^{19}Deutsches Elektronen-Synchrotron DESY, Notkestr. 85 22607, Hamburg, Germany

20^{20}Authors affiliated with an international laboratory covered by a cooperation agreement with CERN

21^{21}University of Dundee, Nethergate, Dundee, DD1 4HN, Scotland, UK

22^{22}Fermilab, Wilson Road, Batavia 60510, IL, USA

23^{23}Florida State University, 600 W. College Ave., Tallahassee 32306, FL, USA

24^{24}Institute of Modern Physics and Key Laboratory of Nuclear Physics and Ion-beam Application (MOE) - Fudan University

25^{25}University of Helsinki, Gustaf Hällströminkatu 2, 00560 Helsinki, Finland

26Institut für Hochenergiephysik, Nikolsdorfer Gasse 18, 1050 Wien, Vienna, Austria

27Imperial College, Prince Consort Road SW7 2AZ, London, United Kingdom

28Indian Institute of Technology Madras, 60036 Chennai, India

29The University of Iowa, 203 Van Allen Hall, Iowa City, 52242, Iowa, USA

30National Centre for Physics, Quaid-I-Azam University, Islamabad-44000, Pakistan.

31Istanbul Technical University, Maslak, 80625 Istanbul, Türkiye.

32Istanbul University, Beyazıt, 34452 Istanbul, Türkiye.

33Kansas State University, 116 Cardwell Hall, Manhattan, KS 66506, USA.

34Institut für Experimentelle Teilchenphysik, Karlsruher Institut für Technologie, Wolfgang-Gaedestrasse 1, D-76131, Karlsruhe.

35King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia.

36Institute for Scintillation Materials of National Academy of Science of Ukraine, 60 Lenina Ave, 61001 Kharkiv, Ukraine.

37NSC Kharkiv Institute of Physics and Technology, 1 Akademichna St., 61108 Kharkiv, Ukraine

38Saha Institute of Nuclear Physics, HBNI, Bidhan Nagar, 700 064 Kolkata, India

39LIP, Avenida Prof. Gama Pinto, n° 2, 1649-003, Lisbon, Portugal

40National Centre for Particle Physics, University of Malaya,Kuala Lumpur 50603,Malaysia

41The University of Maryland, College Park 20742, MD, USA

42The University of Minnesota, 116 Church Street SE, Minneapolis 55405, MN, USA

43MIT, Laboratory for Nuclear Science, 77 Mass Ave, Cambridge, MA 02139, USA

44Nanjing Normal University, 1 Wenyuan Rd., Qixia District, Nanjing, Jiangsu province, P.R. China 210023

45Northwestern University, 2145 Sheridan Rd, Evanston 60208, IL, USA

46University of Notre Dame, Notre Dame 46556, IN, USA

47Laboratoire OMEGA CNRS/IN2P3, Route de Saclay 91128, Ecole Polytechnique, France

48University of Montenegro, Cetinjska br. 2, 81000 Podgorica, Crna Gora, Montenegro 49 Laboratoire Leprince-Ringuet CNRS/IN2P3, Route de Saclay, 91128 Ecole Polytechnique Cedex, France

50 Affiliated with an Institute that was formerly covered by a cooperation agreement with CERN

51 Indian Institute of Science Education and Research, Dr. Homi Bhabha Road 411008, Pune, India

52 Riga Technical University, 6A Kipsalas Street, Riga LV-1048, Latvia

53 University of Rochester, Campus Box 270171, Rochester 14627, NY, USA

54 CEA Paris-Saclay, IRFU, Batiment 141,91191, Gif-Sur-Yvette Paris, France

55 University of Split FESB , R. Boskovica 32, HR-21000, Split, Croatia

56 University of Split, Faculty of Science, R. Boskovica 33, 21000 Split, Croatia

57 Bethel University, 3900 Bethel Drive, St. Paul, MN 55112, USA

58 National Taiwan University, 10617, Taipei, Taiwan

59 Texas Tech University, Department of Physics and Astronomy, Lubbock 79409, TX, USA

60 Georgian Technical University, Kostava str, Tiiblisi, 77, 0160, Georgia.

61 Tata Inst. of Fundamental Research-A, Homi Bhabha Road, Mumbai 400005, India

62 Tata Inst. of Fundamental Research-B, Homi Bhabha Road, Mumbai 400005, India

63 The University of California Santa Barbara, Broida Hall, Santa Barbara 93106, CA, USA

64 University of Virginia 112 Emmet St N, Charlottesville 22903, VA , USA

65 The University of Wisconsin, Madison, WI, USA

66 Yıldız Technical University, Esenler, 34220 Istanbul, Türkiye.

67 Zhejiang University, 866 Yuhangtang Rd, Hangzhou, Zhejiang, China

68 Near East University, Mersin 10, 99138 Nicosia, Türkiye.

69 Konya Technical University, Selçuklu, 42250 Konya, Türkiye.

70 Adıyaman University, Merkez, 02040 Adıyaman, Türkiye.

71 Marmara University, Istanbul, Turkey

72 Milli Savunma University, Naval Academy, Istanbul, Turkey

73 The Science and Technological research Council of Turkey, Informatics and Information Security Research Center, Gebze/Kocaeli, Turkey

74 also at CERN and The Science and Technological research Council of Turkey, Informatics and Information Security Research Center, Gebze/Kocaeli, Turkey

75 U Hamburg, Mittelweg 177, 20148 Hamburg, Germany

76 Indiana University, 107 S. Indiana Avenue, Bloomington, IN 47405

77 University of Michigan, 500 S. State Street, Ann Arbor 48109, MI, USA

78 CFD Research Corporation, 4020 Executive Drive Suite 103, Beavercreek 45430, OH, USA

79 BANGALORE-IISC,CV Raman Rd, Bengaluru, Karnataka 560012, India

80 Istanbul University Cerrahpaşa, Avcılar, 34320 Istanbul, Türkiye.

81 Erzincan Binali Yildirim University, Yalnızbağ, 24002 Erzincan, Türkiye.

82 University of Cambridge (since 2024)

83 Cal Lutheran 60 W. Olsen Road, Thousand Oaks, CA 91360

84 Istinye University, Vadi Kampüsü, 34396 Istanbul, Türkiye.

85 Mimar Sinan Fine Arts University, Şisli, 34480 Istanbul, Türkiye.

86 Turkish-German University, 34820 Istanbul, Türkiye.

E-mail: [email protected]

Introduction

The Large Hadron Collider (LHC) [1] will enter its High-Luminosity phase (HL-LHC) in 2030. It will challenge the detectors with a nearly 10-fold increase in integrated luminosity compared to the previous LHC runs combined [2]. Therefore, the CMS [3] detector will be upgraded to handle the higher levels of radiation and the larger amounts of collision data to be collected. The CMS endcap calorimeter (CE), also known as High-Granularity Calorimeter (HGCAL), will replace the current endcap calorimeters of the CMS detector [4]. It will facilitate the use of particle-flow calorimetry with its unprecedented transverse and longitudinal readout/trigger segmentation, with more than 6M readout channels. The electromagnetic section as well as the high-radiation regions of the hadronic section of the CE (fluences above 1×10141 \times10^{14} 1-MeV neutron-equivalents per square centimetre - neq/cm2^{2}), will be equipped with silicon pad sensors, covering a total area of 620 m2^{2}. Integrated fluences up to 1×10161 \times10^{16} neq/cm2^{2} are expected at the end of its 10-year operation at the HL-LHC. The producer of the silicon sensors is Hamamatsu Photonics K.K.1 The sensors are processed on novel 8” p-type wafers with an active thickness of 300.0 µm, 200.0 µm and 120.0 µm and cut into hexagonal shapes for optimal use of the wafer area and tiling. Each sensor contains several hundred individually read-out cells of two main sizes: around 0.6 cm2^{2} for HD (High Density) sensors and 1.2 cm2^{2} for LD (Low Density) sensors. In order to investigate the radiation-induced bulk damage, the sensors have been irradiated with neutrons at RINSC (Rhode Island Nuclear Science Centre, US) to target fluences between 6.5×10146.5 \times10^{14} neq/cm2^{2} and 1.3×10161.3 \times10^{16} neq/cm2^{2}.

This study includes the dataset analysed in ref. [5] (version 1 campaign) together with the newly analysed version 2 campaign. Compared with version 1, the version 2 sensors have an upgraded layout designed to improve high-voltage (HV) performance and to ensure reliable operation at higher fluence. Accordingly, version 2 targeted a maximum fluence 30.0 % above the 1×10161 \times10^{16} neq/cm2^{2} of version 1, in order to cover the uncertainty in the integrated fluence prediction for the final years of HL-LHC operation. The version 2 campaign also includes additional sensor geometry variants.

The results of this work have been used to optimize the layout of the CE silicon section and to deploy thicker sensors in higher radiation regions. Additionally, the irradiation facility RINSC is investigated for a better understanding and control of the irradiation processes and procedures with emphasis on high-fluence irradiations.

This paper is organized as follows: § 2 describes the version 2 prototypes of the CE silicon pad sensors, irradiated and measured in the campaign conducted after the publication of the previous study. § 3 focuses on the RINSC irradiation facility, describing the process improvements to reduce in-reactor annealing times. § 4 examines the leakage current distribution across the sensor area, with a particular focus on investigating potential effects related to the presence of internal guard rings and high-voltage lines in partial sensors. § 5 deals with the relationship between leakage current and voltage, presenting key observations, including exponential leakage current behaviour under specific conditions. The leakage current-related damage rate coefficient (α\alpha) for various sensor types and fluence levels is presented in § 6, along with an evaluation of the fluence estimation methods applied during irradiation. § 7 presents the temperature dependence of the leakage current to extract the silicon activation energy and identify the primary source of the leakage current. The conclusions of this study are summarized in § 8.

Silicon sensors for the CMS endcap calorimeter upgrade

The previous study [5] described the fabrication and design parameters of the silicon sensors for the CE project, focusing on full sensors from the version 1 prototype campaign. In contrast, the present study analysed version 2 prototypes introduced in 2021. An overview of the sensors examined in both campaigns is provided in Table 1. The table summarizes the sensor characteristics, including layout, thickness, and the delivered fluence (see § 3.2). The term "epi" refers to the epitaxial growth process, while "FZ" stands for float-zone process. The table also lists the material used for the irradiation holder (puck), and provides in-reactor annealing estimates derived from resistance temperature detector (RTD) measurements (see § 3.3).

VersionRound (Part)Thickness [μm]ProcessLayoutTarget fluence [neq_{\mathrm{eq}}/cm2^2]Delivered fluence [neq_{\mathrm{eq}}/cm2^2]APuck materialIn-reactor annealing [min at 60.0 °C]B
FrontCBackAverageD
11300FZfull6.5E+145.0E+14wood9.9-E-
12200FZfull2.5E+151.8E+15wood8.6-E-
13120epifull1.0E+168.1E+15wood2199.3-E-
14200FZfull2.5E+151.8E+15Acrylic-F-E-
15200FZfull2.5E+151.8E+15Acrylic-F-E-
16120epifull1.0E+167.8E+15Acrylic-G-E-
17120epifull2.5E+151.8E+15Acrylic1.3H1.1H1.2
18120epifull5.0E+153.5E+15Acrylic8.47.98.1
19300FZfull1.5E+151.1E+15Acrylic7.56.57.0
110300FZfull1.0E+157.2E+15Acrylic10.19.910.0
111200FZfull2.5E+151.9E+15Acrylic41.69.517.8
21300FZfull1.5E+151.5E+15Acrylic143.76.26.7
22300FZfull2.0E+152.1E+15Acrylic837.53.629.7
23200FZfull4.0E+153.7E+15PEEK503.43.67.1
24120epifull1.0E+169.4E+15PEEK>10000>10000>10000I
25200FZfull4.0E+154.2E+15PEEK48.77.714.0
26200FZfull5.5E+155.2E+15PEEK1028.716.5131.9
27300FZfull2.0E+152.1E+15Acrylic8.38.17.6
28300FZfull1.5E+151.6E+15Acrylic9.96.76.7
29 (I)120epifull1.0E+165.4E+15PEEK126.28.318.7
29 (II)5.0E+15126.2J8.3J18.7J
210 (I)120epifull1.3E+166.5E+15Alu.M88.736.354.8
210 (II)6.6E+15>10000K--
211200FZfull5.5E+154.9E+15Alu.M124.170.980.2
212 (I)120epifull1.3E+166.9E+15Alu.M64.348.753.4
212 (II)6.6E+15121.436.862.6
213200FZpartial5.3E+155.3E+15Alu.M271.339.181.7
214 (I)120epipartial1.0E+164.7E+15Alu.M52.424.530.2
214 (II)4.8E+1534.924.5L30.2L

Table 1. Overview of the irradiation rounds performed at RINSC.

Between version 1 and version 2, several design updates were implemented to improve HV stability. These modifications included changes in the spacing between structures separating individual cells, as well as adjustments to critical HV structures at the sensor edges. The bulk material specifications remained unchanged between the two versions.

Three sensor thicknesses were deployed: 300.0 μm300.0\,\mu\mathrm{m}, 200.0 μm200.0\,\mu\mathrm{m} and 120.0 μm120.0\,\mu\mathrm{m}, as illustrated in Figure 1a. The 300.0 μm300.0\,\mu\mathrm{m} and 200.0 μm200.0\,\mu\mathrm{m} sensors were fabricated using the FZ process, while the 120.0 μm120.0\,\mu\mathrm{m} sensors were produced using the epitaxial process on top of a 180.0 μm180.0\,\mu\mathrm{m} thick handling wafer. Radiation-induced defects in the bulk material generate leakage current, which increases with fluence. Thinner sensors, having a smaller bulk volume, generate less leakage current and exhibit greater radiation hardness compared to thicker sensors [8]. As a result, thinner sensors are deployed in regions with higher radiation exposure, as illustrated in Figure 1b.

Figure 1. a) Sensor cross-sections for different manufacturing processes [9]. b) CMS simulation showing an optimized lateral layout for CE layer 11. c) Overview of LD and HD partial sensor cut types. d) Schematic (not to scale) of the HV protection structures for a full sensor. e) Layouts of LD and HD partial sensor design. f) Schematic (not to scale) of the HV protection structures for a partial sensor (LD Five), which are also present within the sensor’s interior due to the internal dicing lines.

In addition to the full sensors, so-called partial sensors were designed and produced from multi-geometry wafers (MGW) to cover border regions of the CE detector, as depicted in Figure 1b. To maximize detector coverage, partial sensors were designed in various cut types (Figure 1c) and are available in both LD and HD granularities. For this study, HD Bottom and LD Five cut types were selected, due to their ability to remain securely in place within the puck which was originally designed for full-sensor irradiations.

In the context of this study, the main difference between full and partial sensors lies in their design for manufacturing. A full sensor is fabricated from an entire silicon wafer, so the dicing lines, where the wafer is cut to define the sensor edges, are located only at the sensor periphery. To protect the inner sensor cells from the high voltage at the edges, dedicated HV protection structures are implemented exclusively at the dicing lines, as shown schematically in Figure 1d.

The HV ring is located at the sensor edge and is biased to the applied high voltage. Adjacent to the HV ring are two guard rings (GR), which provide the HV protection. The first guard ring (blue) is floating, meaning that it is not connected to any fixed potential. The innermost guard ring (green) is connected to ground potential and shields the inner sensor cells from the high voltage at the sensor edge.

In contrast, partial sensors are manufactured by cutting multiple sensor types from a single wafer (Figure 1c). As a result, internal dicing lines are introduced within the sensor layout, as shown in Figure 1e. To protect the sensor interior from the high voltage present within the sensor area, additional HV protection structures are required.

A schematic representation of the HV protection structures in a partial sensor of type LD Five is shown in Figure 1f. It illustrates how both HV rings and internal GR structures (floating and grounded) are integrated within the sensor interior. A photograph of a sensor of cut type LD Five is provided in Figure 2a, while Figure 2b shows a detailed view of the intersection of HV rings and internal HV protection structures in the inner sensor area.

Figure 2. a) Photograph of an LD Five partial sensor placed in the logistics tray. b) Microscope image showing the internal dicing lines, HV rings, and internal HV protection structures (including the floating GR and grounded GR), on a partial sensor.

Alu. = Aluminum

In addition to the standard (full) cells, CE sensors incorporate other specialized cell types, such as edge cells and calibration cells, which differ in area size. Examples of several cell types are shown in Figure 3 for an HD Top partial sensor. Edge cells ensure complete area coverage of the detector geometry, while calibration cells with their small surface area maintain a sufficient signal-to-noise ratio with minimum ionizing particles until the End of Life of the detector and will be used for energy calibration.

Figure 3. Microscope images of various cell types on an HD Top sensor: a) standard full cell, b) calibration cell, c) edge cells near internal dicing lines, and d) edge cell at outer perimeter. The visible contact marks from previous measurements are highlighted with green circles. They lie within the openings of the sensor passivation layer (visible as brighter regions on the sensor surface), ensuring proper electrical contact. Their positions also enable an estimate of the precision and repeatability of the contacting method.

This study examines whether internal guard rings and HV lines in partial sensors pose a risk of increased leakage currents in the internal sensor cells. This is important because the leakage current must remain within the limits imposed by the detector design, the power supply, and the readout chip specifications. The total sensor leakage current must stay below 10.0 mA [4], and the leakage current per cell must remain below 50.0 µA to satisfy the HGCROC3 requirements [10]. Exceeding these limits can lead to unstable operation and, as observed previously, may even result in sensor damage [5]. For this reason, it is necessary to verify whether partial sensors show an increased risk of such behaviour.

Neutron irradiation at the Rhode Island Nuclear Science Center

Sensors in this work were neutron-irradiated in the Rhode Island Nuclear Science Centre (RINSC). This facility was qualified for the irradiation of the CE sensors, as described in ref. [5].

To position the sensors close to the reactor core, an 8" beam port was used. This port is accessible only when the reactor is off, and the samples were typically removed one day after irradiation. Since the reactor operated at approximately constant power, the delivered fluence could, to first approximation, be controlled by adjusting the irradiation time.

Due to reactor turn-on effects, a fixed irradiation-time uncertainty of 2 min was assumed. Using the fluence (Φ\Phi) scaling factor as described in § 3.2, this corresponds to an absolute fluence uncertainty of 0.93×10140.93 \times10^{14} neq_{\mathrm{eq}}/cm2^2. This contribution was combined in quadrature with a relative uncertainty of 10.0 % associated with the reactor power:

ΔΦ=(0.10⋅Φ)2+(0.93×1014)2,(1)\Delta\Phi= \sqrt{(0.10 \cdot\Phi)^2 + (0.93 \times10^{14})^2}, \tag*{(1)}

The resulting absolute fluence uncertainties are (1.1,5.3,(1.1, 5.3, and 13.5)×1014 neq/cm213.5) \times10^{14}\ \mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2} for irradiation times of 10.9, 113.8, and 292.2 min, respectively. These correspond to relative fluence uncertainties of approximately 20.9 %, 10.1 %, and 10.0 %.

The sensors are placed in so-called pucks and in aluminium cylinders filled with dry ice to limit in-reactor annealing. Schematics shown in Figure 4 indicate the CE silicon sensors marked in green, the dosimetry sensors in grey, and the temperature sensors in black. The aluminium cylinder is filled with dry ice on the side facing the beam port opening.

Schematic of sensor placement within the aluminium transport cylinder at RINSC, showing the puck, silicon sensors, Kapton foil, RTDs, dosimetry sensors, dry ice pellets, antistatic foam, and aluminium cylinder

Figure 4. Schematics (not to scale) of the sensor placement within the aluminium transport cylinder at RINSC, as described in ref. [5]. The positions of the PT1000 RTDs and dosimetry sensors are presented with the approximate dimensions within the puck container.

Procedure adaptations for high-fluence irradiations

Compared to the study presented in ref. [5], this study investigated up to 30 % higher fluences. Achieving higher fluence required keeping the samples in the running reactor for a longer duration, which led to increased sample temperatures. This, in turn, introduced undesired annealing to the sensors, which had to be carefully monitored and controlled.

Several measures were introduced to mitigate in-reactor annealing, as summarized in Table 2. Details of the puck design, including the ventilation holes, were presented in the previous study [5]. Most of these measures were implemented already for the first irradiation round at RINSC. Starting from round 9 of the version 2 campaign, high-fluence irradiation rounds (>1×1016 neq/cm2>1 \times10^{16}\ \mathrm{neq}/\mathrm{cm}^{2}) were split into two parts of equal duration, with dry ice refilled between the two sessions.

Introduced measureIntroduction dateApplied in version, round
Dry ice to cool the cylinder26.08.2020V1-2: all rounds
Ventilation holes26.08.2020V1-2: all rounds
High-fluence round splitting08.03.2022V2: R9, R10, R12, R14
Pre-cooling of the aluminium cylinder18.08.2022V2: R14 second part
Puck material optimizationSee Table 1See Table 1

Table 2. The introduced measures to reduce in-reactor annealing.

The results summarized in Table 1 indicate that splitting high-fluence rounds is an effective approach to reduce in-reactor annealing. An exception occurred in round 10, which, although split, experienced over-annealing during the second part due to limited dry ice availability. This was a logistical constraint specific to that round and does not contradict the overall effectiveness of the splitting strategy demonstrated in the other high-fluence rounds.

Furthermore, several new puck materials were introduced with the aim of improving the thermal conductivity in high-fluence rounds, as shown in Table 1. However, no clear effect of the puck material was observed at similar fluence levels.

A discussion on the potential of pre-cooling of the aluminium cylinder is provided in appendix § 11.

Fluence assessment options

The fluence delivered to the samples during irradiation rounds was estimated in various ways. As discussed earlier in this section, since the reactor operated at approximately constant power, the delivered fluence could be determined by multiplying the irradiation time by a known time-fluence conversion factor. The irradiation time itself was calculated from the reactor’s turn-on/turn-off times, which were recorded to the nearest decimal of a minute for all irradiations.

In the version 1 campaign, an initial conversion factor was used to determine the irradiation time needed to achieve the required target fluence. This factor was based on the fluence assessment using the diodes that were studied for usage in the D0 experiment [11]. However, a comparison with other irradiation facilities revealed that the factor overestimated the delivered fluence.

Between the version 1 and version 2 campaigns the knowledge of the D0 diode thickness was refined: initially assumed to be 240.0 μm\mathrm{\mu m}, it was later determined to be 290.0 μm\mathrm{\mu m}. This discrepancy contributed to the fluence overestimate during the version 1 campaign. Using the updated thickness, the diode-based fluence is typically about 17.0 % lower than previously reported.

Therefore, in the version 2 irradiation campaign, an irradiation time of 21.5 min was used to achieve a fluence of 1×1015 neq/cm21 \times10^{15}\ \mathrm{neq}/\mathrm{cm}^{2}. For consistency, the conversion factor from the version 1 campaign was retrospectively adjusted in this study to match the value applied in version 2. Therefore, this change does not contribute to the results.

In addition, various types of fluence dosimetry sensors were placed inside the puck, both at the front and back, as shown in Figure 4. The dosimetry sensors used in the version 1 campaign were described in ref. [5], while the version 2 campaign introduced additional silicon test structures (diodes) from the CE wafer with an active thickness of 120.0 μm\mathrm{\mu m}. The depletion voltage measurements of the CE test structures were performed at Brown University to assess the depletion voltage, the associated dark current and ultimately the fluence during the irradiations. Since the test structures are also diodes from the same CE wafer as the tested CE sensors, this might have introduced a potential correlation between the measurement and the calibration. The maximum expected correlation was conservatively estimated to be ∼50.0\sim50.0 % when averaging CE- and D0-based fluence estimates (lower and intermediate fluence range: <3×1015 neq/cm2< 3 \times10^{15}\ \mathrm{neq}/\mathrm{cm}^{2}), and to be absent when using iron-foil activation for the fluence determination (high fluence range). However, for the final analysis, the fluence is determined from the irradiation time, as described in § 6.3, and therefore this potential correlation does not affect the fluence values used in this study.

Fluence measurements from the respective dosimetry sensors were averaged, following the procedure described in ref. [5]. It is relevant to mention that the location of the sensors within the puck influences the results of the fluence measurement, due to the varying distance from the dry ice, as well as due to the fluence profile as discussed in § 4. To quantify the scale of position-dependent fluence variations within the puck, the relative median absolute deviation (RMAD = MAD/median) of the fluence values measured across sensor locations was used. After applying the procedure adaptations for high-fluence irradiations described in § 3.1, the RMAD remained below 8.0 % for all campaigns that were not annealed into the reverse-annealing regime (compare Table 1). Overall, the RMAD exceeded 10.0 % only for campaigns 7 and 9 of version 1, reaching 11.0 % and 17.0 % respectively.

In-reactor annealing assessment

For the temperature profile measurement during the irradiation rounds, RTDs were placed inside the puck, as can be seen in Figure 4. There were two sensors placed in the front and one in the back, yet for some rounds there were fewer sensors available due to system failures. If available, the estimation of the temperature experienced by the silicon sensors in the middle of the puck was calculated as an average between the back and front RTD readouts. Starting from round 7 of the version 1 campaign, additional front- and back-side RTDs were introduced, improving the determination of the average sensor temperature and, consequently, the assessment of the equivalent in-reactor annealing time.

An example of a complete temperature profile recording from an irradiation round is shown in Figure 6. The equivalent annealing time at 60.0 ∘C60.0\,^\circ\mathrm{C} was calculated from the average temperature profile using the so-called Hamburg model, as described in ref. [8]. This parametrization is widely used across different fluence levels; however, it was originally developed for n-type sensors and does not directly apply to the p-type sensors foreseen for the CE upgrade.

An example of a temperature profile obtained from an irradiation round at RINSC

Figure 6. An example of a temperature profile obtained from an irradiation round at RINSC, showing averaged measurements at the front, back, and the average of both sides. The distinct phases of irradiation can be inferred from the shape of the profile.

An updated annealing model, validated for the FZ and epitaxial silicon materials used in CE sensors over the relevant fluence range, has recently been published [12]. To estimate the systematic uncertainty associated with the use of the Hamburg model, the equivalent annealing times obtained with the two parametrizations were compared for all irradiation rounds included in the determination of the current-related damage constant.

The largest absolute difference between the two parametrizations was observed for the epitaxial sensor irradiation in version 2, round 14, part 1. For this round, the Hamburg model yields an average equivalent annealing time of 30.2 min, whereas the updated model yields 66.1 min, corresponding to an absolute difference of 35.9 min.

The next-largest difference is 29.3 min for version 2, round 14, part 2, followed by 24.3 min for version 2, round 13. All other FZ irradiations show substantially smaller differences of approximately 8.0 min to 10.0 min.

The equivalent annealing time was strongly impacted by the short time at high temperature during the irradiation, as illustrated in Figure 5. Additionally, because the calculation included waiting times of up to one day post-irradiation, it was also influenced by extended periods at room temperature after the dry ice has sublimated, which could vary depending on the season of the irradiation rounds. The average in-reactor annealing time increased with fluence for rounds irradiated in a single part, as expected.

Figure 5. Annealing time at the sensor position in the reactor versus the maximum temperature measured at the front of the puck during irradiation.

The chosen in-reactor annealing measurement method introduced uncertainties in estimating the equivalent annealing time for the CE silicon sensors. These uncertainties were caused by missing back-side measurements in some rounds and significant discrepancies between the front and back temperature readings in others, as shown in Table 1. The impact of these discrepancies has been previously studied using leakage current versus annealing time measurements for CE diodes, as discussed in ref. [13].

Leakage current profiles across sensors

The system to perform the electrical characterization of the irradiated sensors as well as the characterization procedure was described in detail in ref. [5]. The measurement was done using a probe- and switch-card system called ARRAY [14]. New probe cards were designed and fabricated for partial LD and HD multigeometry wafers. They have been tested on non-irradiated partial sensors first. Pins of the probe cards that were not used for the partial sensor type measured were left open by contacting them to Kapton foil, placed at the probe-station chuck next to the partial sensor, as can be seen in Figure 7. Except for the measurement discussed in § 7, all sensor measurements were performed on a chuck set to −40.0 ∘C-40.0\,^{\circ}\mathrm{C}. The chuck (C200-40 model, manufactured by ATT Systems) is specified by the manufacturer to have a temperature accuracy and stability of ±0.1 ∘C\pm0.1\,^{\circ}\mathrm{C} [15]. The horizontal temperature uniformity of the chuck surface was taken into account following the procedure described in ref. [5]. The typical uncertainty of the voltage measurements is below 2%2\%, while the uncertainty of the leakage current is below 0.1%0.1\%. In all plots in this study, these uncertainties are smaller than the symbol size.

A low-density CE silicon partial sensor of type LD Five connected to the ARRAY System

Figure 7. A low-density CE silicon partial sensor of type LD Five before connecting to the switch- and probe-card of the non-temperature controlled ARRAY System at CERN. The missing sensor area is covered with Kapton foil. The arrangement is equivalent to the one utilized in the measurements of the irradiated sensors.

Voltage dependence of leakage current profiles

Figure 8 presents the per-cell leakage currents, interpolated to an effective bias voltage of 600V600\mathrm{V}. The cell leakage current was normalized by the cell volume, where the n-implant area was used to define the cell area. As examples, results from two irradiation rounds with partial sensors and one high-fluence round with a full sensor were selected. These results confirm the presence of similar leakage current profiles across the surfaces of sensors from the same irradiation round, consistent with the observations reported in ref. [5].

Per-cell volume-normalized leakage current profiles across three representative irradiation rounds

Figure 8. Per-cell volume-normalized leakage currents, interpolated to an effective bias voltage of 600 V, for three representative irradiation rounds. The first and second columns show profiles for partial sensors.

The leakage current profiles could have two sources: a fluence gradient within the reactor, or an annealing time variation caused by temperature differences across the sensor, resulting from uneven cooling during the irradiation.

To investigate the origin more closely, the leakage current distribution across the sensor was analysed, taking into account the differences between puck materials and the effects of split versus single rounds (see § 3.1). This is motivated by the fact that the puck materials differ in their thermal conductivity, which could lead to variations in the cooling homogeneity. The analysis shows no clear correlation between the puck material and the leakage current distribution across the sensors, as shown in Figure 9. This suggests that the observed leakage current profiles are likely due to the fluence profile in the reactor rather than the annealing profile.

Figure 9. Leakage current dispersion (RMAD, compare § 3.2) versus fluence for different puck materials.

The spatial distribution of the leakage current I(x,y)I(x,y) was modelled using a two-dimensional elliptical Gaussian function with rotation, defined over the sensor cell coordinates xcellx_{\mathrm{cell}} and ycelly_{\mathrm{cell}}, as illustrated in Figure 10a. The function takes the form:

Figure 10. a) Volume-normalized per-cell leakage current distribution displayed as a 2D heatmap at an effective bias voltage of 600 V for a representative partial hexagonal LD sensor after 84 minutes of annealing. The plot includes 2D Gaussian fits of the leakage current distribution, with red dots marking selected cells of different shapes aligned along an iso-fluence line. The coordinate system is referenced to the centre of a full sensor. b) IV curves for selected cells indicated in (a) for the partial LD sensor. c) Normalized IV curves for cells of different shapes for the LD partial sensor shown in (a), d) for an HD partial sensor, e) for an LD full sensor, and f) for an HD full sensor. The normalization is performed with respect to the leakage current of a standard-shaped cell at each voltage point. All leakage currents are scaled to the respective cell volume.

I(xcell,ycell)=I0⋅exp⁡(−[a(xcell−x0)2+2b(xcell−x0)(ycell−y0)+c(ycell−y0)2])(2)I(x_{\mathrm{cell}},y_{\mathrm{cell}})=I_0\cdot\exp\left(-\left[a(x_{\mathrm{cell}}-x_0)^2+2b(x_{\mathrm{cell}}-x_0)(y_{\mathrm{cell}}-y_0)+c(y_{\mathrm{cell}}-y_0)^2\right]\right) \tag*{(2)}

where (x0,y0)(x_0,y_0) is the centre of the distribution, I0I_0 is the amplitude, σx\sigma_x and σy\sigma_y are the widths along the principal axes, and φ\varphi is the rotation angle of the ellipse. The parameters a, b, c are derived from σx\sigma_x, σy\sigma_y and φ\varphi. A fit to the measured data yielded a reduced χ2\chi^2 value close to 1. This supports the assumption of a smooth, continuous variation of the leakage current, which was used to define "iso-fluence/annealing" contours across the sensor surface, indicated by the blue lines in Figure 10a. These "iso-fluence/annealing" lines represent constant leakage current level on the sensor, and they indicate hypothetical contours across the sensor area where cells experience comparable fluence or annealing conditions, inferred from similar leakage current values.

To numerically verify whether different cell types across the sensors exhibit similar leakage current behaviour with increasing voltage, the volume-normalized IV curves were compared for cells from both full and partial sensors. The comparison was performed using cells positioned approximately along the "iso-fluence/annealing" lines, as illustrated in Figure 10a. Where applicable, cells of different geometries were chosen for the comparison, as shown in Figure 10b, to further test whether sensor layout affects current behaviour.

Discussion of results for partial sensors

As shown in Figure 8, the leakage current profiles are smooth for both the partial and full sensors at high fluence. In particular, for the partial sensors (first two columns), no instabilities related to the sensor layouts or the cell vicinity to the internal HV protection structures are observed (see § 2). Moreover, it can be observed that the leakage current maximum occurs within the sensor area, not in the cut-out region. This is relevant because the cells at maximum leakage current are used in the analysis in § 5 and § 6 for both partial and full sensors. This observation supports the use of partial sensors for that analysis.

For quantitative comparison, the volume-normalized per-cell leakage current (ρl\rho_{\mathrm{l}}) was further normalized at each voltage point to the leakage current of a standard-shaped reference cell (ρl std\rho_{\mathrm{l\,std}}).

As shown in Figure 10c to Figure 10f, the IV curves agree within 8 % across different cell types and sensor variants. Moreover, the current variation among differently shaped cells remains stable as the bias voltage increases, with the largest observed deviation being 7 % for an HD full sensor (see Figure 10f).

This result demonstrates that differently shaped cells, including those near the guard ring for full sensors and the internal guard ring for partial sensors, exhibit stable leakage current behaviour with increasing bias voltage. This stability, regardless of cell geometry or proximity to HV protection structures, proves the robustness of the sensor design and its uniform response to applied bias voltage.

Leakage current dependence on voltage

Results on leakage current behaviour

In this study, the term "current maximum" refers to the average leakage current of the three full cells showing the highest leakage current within each sensor. This region corresponds to the area of maximum neutron fluence, and the resulting value is denoted as IcellI_{\mathrm{cell}} at fluence max in the plots.

Most sensors exhibited a typical diode-like leakage current-voltage (IV) dependence, as shown by the coloured graphs in Figure 11a. Overall, 31 of the 36 measured sensors displayed stable leakage current profiles. However, during the initial irradiation rounds with fluences exceeding 1×1016 neq/cm21 \times10^{16}\,\mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2}, extreme conditions were reached, with temperatures as high as 125.0 ∘C125.0\,^{\circ}\mathrm{C}. These conditions resulted in significant in-reactor annealing, equivalent to approximately 10000 minutes at 60.0 ∘C60.0\,^{\circ}\mathrm{C}. One example is round 4 of the version 2 campaign, listed in Table 1.

Example IV curves for sensors from the version 2 irradiation campaign

Figure 11. Example IV curves for sensors from the version 2 irradiation campaign. a) Per-cell currents in current maximum. b) Total current as seen by the power supply.

Sensors exposed to such prolonged annealing (far into the reverse annealing regime [6, 7]) and high fluences displayed a different IV behaviour. Specifically, the leakage current showed an exponential rise, as shown by the grey graphs in Figure 11a. This effect is discussed in more details in § 5.2.

The total current measured by the power supply, when scaled by sensor volume, aligns well with the cell currents in terms of magnitude (see Figure 11b). After scaling, IV curves cluster together for sensors of the same thickness, suggesting that the leakage current is predominantly influenced by bulk effects rather than surface phenomena.

Analysis of exponential increase in leakage current

The exponential rise in leakage current observed in some highly irradiated sensors after prolonged in-reactor annealing suggests the presence of a field-dependent current enhancement mechanism. The ARRAY probe-card system used for 8"-sensor characterization does not allow charge collection measurements with external particle or laser sources. Therefore, a direct experimental investigation of the microscopic origin of the exponential current increase on the full sensors was beyond the scope of this work.

Dedicated diode test structures were included in the wafer layout, specifically to characterize bulk radiation-damage properties that cannot be measured directly on the full sensors [12, 16]. The diode test structures were fabricated on the same wafers as the 8" sensors and therefore share the same active thickness, silicon material, and fabrication process. Complementary measurements on these diode test structures showed a simultaneous increase in leakage current and charge collection after prolonged annealing, consistent with charge multiplication [12]. These measurements therefore provide indirect experimental support for charge multiplication as one possible explanation for the exponential leakage-current increase observed in the full sensors.

For CE sensors, the high-fluence irradiation rounds in which the irradiation was divided into two parts to limit the cumulative in-reactor annealing time, including version 2 rounds 12 and 14, did not exhibit the exponential leakage-current increase. As discussed in § 3.1, round 10 was an exception because limited dry-ice availability during the second irradiation period resulted in additional annealing.

Figure 11b further shows that the total leakage current measured at the power supply exhibits the same exponential high-voltage behaviour as the individual-cell leakage currents for sensors exposed to high fluence and prolonged annealing. This behaviour indicates that the effect is not limited to individual cells but applies to the entire sensor. When cumulative in-reactor annealing was reduced by splitting the irradiation into multiple steps, both the per-cell and total-current IV characteristics recovered their stable diode-like behaviour.

These results indicate that limiting cumulative in-reactor annealing, for example by splitting high-fluence irradiation rounds, is an effective strategy to suppress the exponential high-voltage increase in leakage current. Minimizing excessive annealing and its resulting effects is crucial for accurately assessing the full sensor performance in the detector, as the sensors will not experience annealing times as prolonged as those observed during the irradiations at RINSC.

Leakage current dependence on fluence

Cell current at fluence maximum

The sensor leakage current contributes directly to the electronic noise and power dissipation of the detector [4]. This section focuses on sensors irradiated to a maximum target fluence of 1.3×1016 neq/cm21.3 \times10^{16}\,\mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2}, which is 30.0% above the expected End-of-Life fluence for the HL-LHC (see § 1). This provides a safety margin for evaluating sensor performance under extreme radiation conditions.

The leakage currents of the CE sensors were scaled to relevant temperatures for easier assessment using the temperature dependence:

I(Tref)=I(T)⋅(TrefT)2⋅exp⁡(Eband2kB(1T−1Tref)),(3)I(T_{\mathrm{ref}}) = I(T) \cdot\left(\frac{T_{\mathrm{ref}}}{T}\right)^2 \cdot\exp\left(\frac{E_{\mathrm{band}}}{2k_{\mathrm{B}}}\left(\frac{1}{T}-\frac{1}{T_{\mathrm{ref}}}\right)\right), \tag*{(3)}

where Eband=1.21E_{\mathrm{band}} = 1.21 eV is the effective bandgap energy [17], and kBk_{\mathrm{B}} is the Boltzmann constant. This relation models the temperature dependence of the leakage current in silicon and is widely used in high-energy physics [6]. Its validity for CE sensors over the relevant temperature range has been confirmed in [5] for fluences around 1×10151 \times10^{15} neq_{\mathrm{eq}}/cm2^2. The accuracy of this scaling method at higher fluences, such as 1×10161 \times10^{16} neq_{\mathrm{eq}}/cm2^2, is further discussed in appendix § 12.

Table 3 summarizes the leakage current estimates at both −35.0 °C and −30.0 °C, for cells at current maximum and the total current of the sensor as seen by the power supply. The target temperature of −35.0 °C corresponds to the planned CE operating condition. However, measurements suggest that certain regions of the sensors may reach temperatures as high as −30.0 °C, due to thermal resistance in the cooling setup as well as heat generated by components on the front-end PCB layer. Therefore, the scenario of sensor cells operating at −30.0 °C was also evaluated.

TemperatureVbias=600.0 VV_{\mathrm{bias}} = 600.0\ \mathrm{V}Vbias=800.0 VV_{\mathrm{bias}} = 800.0\ \mathrm{V}
IcellI_{\mathrm{cell}} at fluence max [μA\mu\mathrm{A}]ItotalI_{\mathrm{total}} [mA]IcellI_{\mathrm{cell}} at fluence max [μA\mu\mathrm{A}]ItotalI_{\mathrm{total}} [mA]
−35.0 ∘C-35.0\ ^\circ\mathrm{C}13.75.4518.457.26
−30.0 ∘C-30.0\ ^\circ\mathrm{C}26.2110.4335.2713.88
Spec. Limit (Per Cell)50–50–
Spec. Limit (Total)–10–10

Table 3. Estimated per-cell and total leakage current for different CE detector operating scenarios, for a sensor irradiated to a target fluence of 1.3×1016 neq/cm21.3 \times10^{16}\ \mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2} (version 2 Round 12).

While the per-cell leakage current remains within specification under all conditions, the total sensor current exceeds the design limit at −30.0 °C, emphasizing the need for effective detector cooling. It is also worth noting that partial sensors produce lower total leakage currents due to their reduced number of cells.

Current-related damage factor

The current-related damage factor (α\alpha) quantifies the proportionality between the radiation-induced increase in leakage current (ΔI\Delta I), normalized to the sensor volume (VV), and the particle fluence (Φ\Phi). The α\alpha parameter is independent of the type, resistivity, and impurity content of the silicon material used [6, 7]. It is defined as:

α=ΔIV⋅Φ(4)\alpha= \frac{\Delta I}{V \cdot\Phi} \tag*{(4)}

Table 4 presents the α\alpha values obtained in this study scaled to different temperatures for comparison with the reference values from the previous studies on silicon sensors, which were also included in the table. The α\alpha factor was extracted from the leakage current data through a fit. The leakage current for this analysis was determined by averaging the three highest-leakage current cells within each sensor. The data presented corresponded to samples annealed for a total equivalent of (80±20)(80 \pm20) min at 60.0 ∘C60.0\ ^\circ\mathrm{C}, unless stated otherwise.

Table 4: Comparison of the α\alpha values obtained in this study, scaled to different temperatures using the normalization relation presented in eq. (3), with reference α\alpha values reported in other studies. For the CE sensor results, the first quoted uncertainty is statistical, and the second is systematic (see text for details).

Dataα−35.0 ∘C×10−19 A/cm\alpha_{-35.0\,^\circ\mathrm{C}} \times10^{-19}\ \mathrm{A/cm}α−20.0 ∘C×10−19 A/cm\alpha_{-20.0\,^\circ\mathrm{C}} \times10^{-19}\ \mathrm{A/cm}α20.0 ∘C×10−17 A/cm\alpha_{20.0\,^\circ\mathrm{C}} \times10^{-17}\ \mathrm{A/cm}
CE sensors, version 1A,B1.1±0.0±0.21.1 \pm0.0 \pm0.27.4±0.3±1.57.4 \pm0.3 \pm1.54.4±0.2±0.94.4 \pm0.2 \pm0.9
CE sensors, version 2A,B1.3±0.0±0.31.3 \pm0.0 \pm0.38.6±0.2±1.78.6 \pm0.2 \pm1.75.1±0.1±1.05.1 \pm0.1 \pm1.0
CE sensors, combinedA,B,C1.3±0.0±0.31.3 \pm0.0 \pm0.38.2±0.2±1.68.2 \pm0.2 \pm1.64.7±0.1±1.04.7 \pm0.1 \pm1.0
CE sensors, total currentA,B,C1.2±0.0±0.21.2 \pm0.0 \pm0.27.5±0.2±1.57.5 \pm0.2 \pm1.54.4±0.1±0.94.4 \pm0.1 \pm0.9
CE diodes (JSI) [13]D,E6.8±0.1±0.76.8 \pm0.1 \pm0.7
p-type sensors [6, 7]E,F3.99±0.033.99 \pm0.03

A Irradiated at RINSC. B Measured at the bias voltage 600.0 V and at fluence maximum. C “Combined” refers to a fit using CE sensor data from both version 1 and version 2 campaigns. D Irradiated at JSI in Ljubljana. E Measured at the depletion voltage. F Irradiated mostly at Physikalisch-Technische Bundesanstalt (PTB) in Braunschweig.

Table 4. Comparison of the α\alpha values obtained in this study, scaled to different temperatures using the normalization relation presented in eq. (3), with reference α\alpha values reported in other studies. For the CE sensor results, the first quoted uncertainty is statistical, and the second is systematic (see text for details).

To extend the dataset analysed in this study, results from version 1 CE sensors campaign were included alongside those from the version 2 campaign. However, the α\alpha value obtained in this study is not directly comparable to the value reported for version 1 sensors in ref. [5], as a different fluence estimation method has since been used (see § 3.2). Therefore, the α\alpha value of the version 1 campaign was recalculated in this study using the fluence derived from the irradiation time and is presented in table 4.

The version 2 campaign shows an α\alpha value approximately 16 % higher than that of the version 1 campaign at −20.0 ∘C-20.0\ ^\circ\mathrm{C}. The available measurements do not allow the origin of this discrepancy to be determined conclusively. A possible contributing factor may be an incomplete estimation of the annealing time in parts of the version 1 campaign, however, this cannot be verified from the available data. Therefore, this difference was conservatively incorporated into the systematic uncertainty. The total systematic uncertainty on α\alpha was conservatively estimated to be 20 %. After this adjustment, the results from the two campaigns are consistent within the total uncertainties, and the datasets were combined for the final fit shown in Figure 12a.

Figure 12. a) Cell leakage current at current maximum versus delivered fluence, based on combined data from the version 1 and version 2 campaigns. b) Total leakage current as seen by the power supply versus delivered fluence.

In table 4, the first quoted uncertainty is statistical, and the second is systematic. The statistical uncertainty includes contributions from the fit, fluence estimation, annealing-time estimation, temperature uncertainty during the IV measurements, and sensor thickness variation. The systematic uncertainty accounts for the observed discrepancy between the version 1 and version 2 campaigns, as well as facility-dependent effects related to irradiation conditions and fluence determination. The total uncertainties are obtained by adding the individual contributions in quadrature within each category.

The combined CE sensor dataset irradiated at RINSC exhibits α\alpha values that are systematically higher than those of CE diodes (21 %) and other p-type sensors (18 %) irradiated at other facilities, as shown in Table 4. Variations in neutron flux and exact fluence estimation between different facilities cannot be neglected. Since JSI is a reference facility, the difference in α\alpha value between the CE sensors irradiated at RINSC and the CE diodes irradiated at JSI was included in the estimation of the systematic uncertainty of the result. This uncertainty accounts for irradiation-related effects at RINSC and the limited understanding of the annealing process. Also, the reference values were obtained at the depletion voltage (measured at the frequency of 10.0 kHz), while the bias voltage of 600.0 V was utilized in this study. The depletion voltage is lower than the mentioned bias voltage even for the samples of 300.0 μ\mum thickness [13].

The volume-normalized total current versus fluence is compatible with the cell current taken at the fluence maximum, as shown in Figure 12b. The slightly reduced α\alpha value is related to the on average lower fluence exposure of the full sensor with respect to the cells in fluence maximum. If the average current of the full cells is used instead of the maximum, an α\alpha value of 7.5±0.2±1.5×10−197.5 \pm0.2 \pm1.5 \times10^{-19} A/cm is obtained, which is closer to the α\alpha value for the total current.

Validation of fluence assessment procedure

For the α\alpha factor analysis, the use of an accurate fluence value is crucial. This study compares three fluence estimation methods: (i) fluence estimation from dosimetry sensors, (ii) target fluence, and (iii) fluence estimation from irradiation time. These methods were introduced in § 3.2.

In the version 1 campaign analysis reported in ref. [5], fluence was estimated using dosimetry objects. However, the number and type of dosimetry sensors varied between campaigns (and even between rounds within a campaign) making this method unsuitable for reliable comparison. As discussed in § 3.2, the target fluence is derived from a time conversion factor, which was modified between campaigns. For example, the version 2 campaign involved longer irradiation times for the same target fluence as version 1, leading to inaccurate target fluence estimates for some version 1 rounds. This inconsistency excludes target fluence as a reliable option for fluence assessment. Since the reactor was operated at stable power during irradiation (see § 3.2), the fluence estimation based on delivered irradiation time, calculated from the reactor power curves, is both reliable and reproducible across campaigns.

These considerations are further supported by a comparison of the fitted leakage currents and extracted α\alpha values for each estimation method. As shown in Figure 13, a systematic offset between the version 1 and version 2 campaigns is observed when using dosimetry-based fluence estimates. An even larger offset is seen when using the target fluence method. In contrast, when fluence is estimated from delivered irradiation time, the fit lines from both campaigns align more closely, and the extracted α\alpha values agree within their respective uncertainties. Therefore, the analyses in this work use fluence values based on delivered irradiation time, applying a consistent conversion factor to both campaigns.

Figure 13. Comparison of per-cell leakage current at sensor current maximum using three fluence evaluation methods.

Additionally, it is worth mentioning that Figure 13 also illustrates that the current-related damage rates α−20.0 ∘C\alpha_{-20.0\,^\circ\mathrm{C}} for version 1 sensors range from 5.3×10−195.3 \times10^{-19} A/cm to 7.4×10−197.4 \times10^{-19} A/cm, while for version 2 sensors, the range is narrower, from 8.4×10−198.4 \times10^{-19} A/cm to 8.6×10−198.6 \times10^{-19} A/cm. This suggests that fluence assessment methods have improved in the version 2 campaign, but still retain some degree of uncertainty. These improvements can be attributed to better sensor placement, an increased number of dosimetry objects, and an optimized time conversion factor (see § 3.2).

Leakage current versus temperature

Extraction of activation energy

The temperature dependence of the leakage current in neutron-irradiated CE sensors was investigated in a narrow interval between −40.0 ∘C-40.0\,^\circ\mathrm{C} and −36.0 ∘C-36.0\,^\circ\mathrm{C}. This range was selected based on practical limitations of the measurement setup. The lower limit of −40.0 ∘C-40.0\,^\circ\mathrm{C} corresponds to the lowest temperature reachable by the cold chuck [15], while measurements above −36.0 ∘C-36.0\,^\circ\mathrm{C} posed a risk of exceeding the safe operational current limits of the measurement setup, and some measurement points were not recorded specifically at the bias voltage of 800.0 V, as seen in Figure 14a. Within the accessible range, the temperature was incremented in 1.0 °C steps to ensure precise extraction of the activation energy.

Figure 14. a) Per cell (number 374) leakage current scaled with cell volume. Cell 374 is a full cell of a partial sensor of high density granularity and 120.0 μm\mu\mathrm{m} thickness, irradiated to a delivered fluence of 9.5×10159.5 \times10^{15} neq/cm2\mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2} (version 2 Round 14). b) Arrhenius curves at different interpolated bias voltages for the same cell of the same sensor as in a).

The activation energy (EAE_{\mathrm{A}}) was extracted by analysing the temperature dependence of the leakage current density, following an Arrhenius-type behaviour, as illustrated in Figure 14b. For sensors irradiated to various fluences, activation energies in the range 0.58–0.63 eV were measured. These values are consistent with previous findings reported in ref. [18].

Comparison with Shockley Read Hall theory

The Shockley Read Hall (SRH) theory describes the temperature-leakage current dependence I(T)I(T) for bulk leakage current as follows:

I(T)≈T2exp⁡(−EA2kBT).(5)I(T) \approx T^{2} \exp\left(-\frac{E_{\mathrm{A}}}{2k_{\mathrm{B}}T}\right). \tag*{(5)}

This theoretical model, derived from the migration rate of defects in a material [8], helps distinguish between leakage current contributions from bulk recombination centres and surface effects, such as charge trapping in the silicon oxide layer [19].

As shown in Figure 14a, the leakage current increases with temperature, as expected from the Shockley-Read-Hall model [19, 20]. The activation energy values extracted in § 7.1 correspond to an equivalent band gap energy range of 1.16–1.26 eV, using the relation:

Eband=2EA,(6)E_{\mathrm{band}} = 2E_{\mathrm{A}}, \tag*{(6)}

as described in ref. [21]. This range aligns closely with the widely accepted band gap energy value of Eband=1.21E_{\mathrm{band}} = 1.21 eV, as also noted in appendix § 12.

The agreement between the measured activation energies and the SRH prediction indicates that bulk recombination centres, rather than surface defects, are the dominant source of leakage current in neutron-irradiated CE sensors. The absence of significant deviations from the SRH model further suggests that contributions from surface-related leakage mechanisms (e.g., silicon oxide charge trapping) are negligible.

Summary and outlook

This study investigates the behaviour of CE silicon sensors exposed to fluences in range between 5×10145 \times10^{14} neq/cm2\mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2} and 1.3×10161.3 \times10^{16} neq/cm2\mathrm{n}_{\mathrm{eq}}/\mathrm{cm}^{2}. The results show that the leakage current profiles are smooth for all sensors, with no instabilities related to sensor layouts or proximity to internal HV structures in partial sensors.

Most sensors exhibit a typical diode-like IV dependence. However, sensors exposed to prolonged annealing and high fluences developed a pronounced exponential increase in leakage current at high voltages. This behaviour was mitigated by splitting high-fluence irradiation rounds to limit cumulative in-reactor annealing, demonstrating that controlling the annealing history suppresses the exponential increase in leakage current and preserves stable diode-like IV characteristics.

The measured leakage current at a fluence approximately 30 % above the expected HL-LHC End-of-Life fluence confirms that CE sensors perform as expected under operational conditions. At the planned operating temperature of −35.0 ∘C-35.0\,^\circ\mathrm{C}, both the per-cell and total sensor leakage currents remain within the limits set by the detector design and the readout chip. However, when assuming a higher sensor temperature of −30.0 ∘C-30.0\,^\circ\mathrm{C}, the total leakage current would exceed the specification limit, showing the importance of efficient cooling to ensure full compliance at extreme fluences.

The leakage current increases with fluence, following the expected trend. We observed that partial sensors with internal HV lines exhibit leakage current behaviour consistent with that of full sensors. Additionally, HD sensors irradiated up to 1.3×1016 neq/cm21.3 \times10^{16}\ \mathrm{n_{eq}/cm^2} follow the anticipated leakage current trend.

The current-related damage factor values obtained in the previous and current irradiation campaigns are consistent within the total uncertainties, justifying their combination into a single dataset. A comparison of single-cell and total sensor currents from both full and partial sensors reveals consistent behaviour across all sensor types. However, the consolidated CE sensor α\alpha values are systematically higher than those reported in the literature, likely due to variations in neutron flux, fluence estimation across facilities, and bias voltage differences.

Finally, this work presents the first measurement of the temperature dependence of the IV characteristics of CE silicon sensors after high-fluence neutron irradiation. The measured activation energies are consistent with theoretical expectations, indicating that bulk recombination centres, rather than surface defects, dominate the leakage current in irradiated CE silicon sensors.

Neutron irradiation at RINSC has proven essential for assessing the bulk radiation hardness of large-area silicon sensors for the CE detector. For future neutron irradiations at RINSC, we recommend deploying additional temperature sensors to improve the monitoring of sensor annealing and to minimize uncertainties in the assessment of in-reactor annealing. Specifically, we suggest placing at least two RTDs on the front and two on the back of the puck, ideally positioned directly adjacent to the silicon sensors. Further improvements could be achieved by directly measuring the horizontal temperature distribution across the sensor area, for which an array of temperature sensors distributed across the wafer is proposed.

Additionally, ensuring consistent sensor orientation (rotation) relative to the reactor during irradiations is recommended. This would facilitate a more accurate assessment of annealing and fluence profiles, particularly for high-fluence rounds (>1.0×1016 neq/cm2>1.0 \times10^{16}\ \mathrm{n_{eq}/cm^2}) that are split into multiple irradiation parts.

Acknowledgments

We thank the staff at the Rhode Island Nuclear Science Center for their support during the preparation and execution of the neutron irradiation of the CE silicon pad sensor prototypes. The EP-DT and the former EP-LCD group at CERN have developed essential infrastructure, such as the ARRAY system including the associated data acquisition software, and have co-financed the acquisition of the utilised cold-chuck probe station at CERN. Without their input, this R&D milestone towards the realisation of this novel calorimeter would not have been possible. We thank Zachary Zawisza for his analysis of the leakage-current temperature dependence. We are thankful for the technical and administrative support at CERN and at other CMS institutes and thank the staff for their contributions to the success of the CMS effort. We acknowledge the enduring support provided by the following funding agencies and laboratories: BMBWF and FWF (Austria); CERN; CAS, MoST, and NSFC (China); MSES and CSF (Croatia); CEA, CNRS/IN2P3 and P2IO LabEx (ANR10-LABX-0038) (France); SRNSF (Georgia); BMBF, DFG, and HGF (Germany); GSRT (Greece); DAE and DST (India); MES (Latvia); MOE and UM (Malaysia); MOS (Montenegro); PAEC (Pakistan); FCT (Portugal); JINR (Dubna); MON, RosAtom, RAS, RFBR, and NRC KI (Russia); MoST (Taipei); ThEP Center, IPST, STAR, and NSTDA (Thailand); TUBITAK and TENMAK (Turkey); STFC (United Kingdom); and DOE (USA).

Appendix

Pre-cooling of the aluminium cylinder

Before the second half of round 14 in the version 2 campaign, the aluminium cylinder was kept in a freezer due to radiation safety concerns, as it still showed residual radioactivity from the first half of the irradiation. This raised a discussion about whether pre-cooling could reduce in-reactor annealing. According to Table 1, pre-cooling reduced the equivalent time at 60.0 °C by 31 % in the front measurement. However, a comparable reduction of 34 % was observed in round 12 without pre-cooling. Given the significant effort required for pre-cooling, including an extra trip to the reactor, and the lack of substantial impact on in-reactor annealing, this method is not recommended for future campaigns.

Leakage current temperature scaling

Since the leakage current I(Tref)I(T_{\mathrm{ref}}) in silicon sensors is strongly temperature-dependent, it must be scaled to a common reference temperature (TrefT_{\mathrm{ref}}) to ensure consistent comparison across different measurement conditions. This correction is applied using an exponential scaling law derived from the temperature dependence of the leakage current in silicon [19]:

I(Tref)=I(T)⋅c(T,Tref),(7)I(T_{\mathrm{ref}}) = I(T) \cdot c(T,T_{\mathrm{ref}}), \tag*{(7)}

where the scaling factor c(T,Tref)c(T,T_{\mathrm{ref}}) is given by:

c(T,Tref)=(TrefT)2⋅exp⁡(Eband2kB(1T−1Tref)).(8)c(T,T_{\mathrm{ref}}) = \left(\frac{T_{\mathrm{ref}}}{T}\right)^2 \cdot\exp\left(\frac{E_{\mathrm{band}}}{2k_{\mathrm{B}}}\left(\frac{1}{T}-\frac{1}{T_{\mathrm{ref}}}\right)\right). \tag*{(8)}

Here, Eband=1.21E_{\mathrm{band}} = 1.21 eV is the effective silicon bandgap energy [17], and kBk_{\mathrm{B}} is the Boltzmann constant. This expression is widely adopted in the high-energy physics community for leakage current normalization [6].

The validity of this scaling law for CE silicon sensors at moderate fluences has been confirmed in [5]. To evaluate its accuracy at high fluences (e.g., 1×10161 \times10^{16} neq_{\mathrm{eq}}/cm2^2), scaled leakage current values were compared to direct measurements taken over a narrow temperature range between −40.0 °C and −36.0 °C.

The comparison in Table 5 confirms that the leakage current increases with temperature, as expected [20], and that the scaling law remains accurate within this temperature interval. At 600.0 V and −36.0 °C, the average absolute deviation between measured and scaled values was approximately 0.5 µA at the cell level and 0.1 mA for the total current. ref-14* continuation is skipped because it belongs to the main bibliography.

SensorVoltageType−40.0 °C−39.0 °C−38.0 °C−37.0 °C−36.0 °C
V2, R14600 VIcellI_{\mathrm{cell}}, measured [µA]5.836.557.488.389.53
V2, R14600 VIcellI_{\mathrm{cell}}, scaled [µA]-6.697.678.7710.03
V2, R14600 VIcellI_{\mathrm{cell}}, difference [µA]--0.14-0.19-0.39-0.50
V2, R14800 VIcellI_{\mathrm{cell}}, measured [µA]7.408.329.49--
V2, R14800 VIcellI_{\mathrm{cell}}, scaled [µA]-8.499.7211.1312.72
V2, R14800 VIcellI_{\mathrm{cell}}, difference [µA]--0.17-0.23--
V2, R14600 VItotalI_{\mathrm{total}}, measured [mA]1.371.551.761.972.25
V2, R14600 VItotalI_{\mathrm{total}}, scaled [mA]-1.571.802.062.36
V2, R14600 VItotalI_{\mathrm{total}}, difference [mA]--0.02-0.04-0.09-0.11
V2, R14800 VItotalI_{\mathrm{total}}, measured [mA]1.741.962.242.502.85
V2, R14800 VItotalI_{\mathrm{total}}, scaled [mA]-2.002.292.623.00
V2, R14800 VItotalI_{\mathrm{total}}, difference [mA]--0.04-0.05-0.12-0.15

Table 5. Per-cell leakage current measurements and scaled results for CE sensors across temperatures at two voltage settings.

References

  1. [1]L. Evans and P. Bryant, LHC Machine, JINST 3 (2008) S08001.
  2. [2]B. Alonso and others, High-Luminosity Large Hadron Collider (HL-LHC): Technical design report, CERN Yellow Reports: Monographs, CERN-2020-010 (2020), DOI:10.23731/CYRM-2020-0010.
  3. [3]CMS Collaboration, The CMS Experiment at the CERN LHC, JINST 3 (2008) S08004.
  4. [4]CMS Collaboration, The Phase-2 Upgrade of the CMS Endcap Calorimeter, CERN-LHCC-2017-023, CERN, Geneva (2017), DOI:10.17181/CERN.IV8M.1JY2.DOI
  5. [5]CMS HGCAL collaboration, Neutron irradiation and electrical characterisation of the first 8” silicon pad sensor prototypes for the CMS calorimeter endcap upgrade, JINST 18 (2023) P08024 [2209.10159].DOI
  6. [6]M. Moll, Displacement Damage in Silicon Detectors for High Energy Physics, IEEE Transactions on Nuclear Science 65 (2018) 1561.DOI
  7. [7]CERN-ROSE/RD48 collaboration, Leakage current of hadron irradiated silicon detectors - material dependence, Nucl. Instrum. Meth. A 426 (1999) 87.
  8. [8]M. Moll, Radiation damage in silicon particle detectors: Microscopic defects and macroscopic properties, Ph.D. thesis, Hamburg University, 1999, DESY-THESIS-1999-040.
  9. [9]P. Paulitsch, The silicon sensors for the High Granularity Calorimeter of CMS, Nucl. Instrum. Meth. A 978 (2020) 164428 [2002.11449].DOI
  10. [10]F. Bouyjou et al., HGCROC3: the front-end readout ASIC for the CMS High Granularity Calorimeter, Journal of Instrumentation 17 (2022) C03015.
  11. [11]D0 Collaboration, D0 Run IIB Upgrade Technical Design Report (2002).
  12. [12]L. Diehl et al., Temperature dependence of the long-term annealing behavior of neutron irradiated diodes from 8-inch p-type silicon wafers, JINST 21 (2026) P03034.
  13. [13]J. Kieseler et al., Isothermal annealing of radiation defects in silicon bulk material of diodes from 8” silicon wafers, JINST 18 (2023) P09010 [2211.04849].
  14. [14]E. Brondolin et al., ARRAY: An Open Source, Modular and Probe-Card based System with Integrated Switching Matrix for Characterisation of Large Area Silicon Pad Sensors, Nucl. Instrum. Meth. A 940 (2019) 168 [1903.10262].
  15. [15]ATT Systems, “Vacuum chucks.” https://www.att-systems.com/products/#chucks.
  16. [16]O.A. Kałuzińska et al., Annealing behaviour of charge collection of neutron irradiated diodes from 8-inch p-type silicon wafers, JINST 20 (2025) P06054.DOI
  17. [17]I. Dawson and others, Radiation effects in the LHC experiments: Impact on detector performance and operation, CERN Yellow Reports: Monographs, CERN-2021-001 (2021), DOI:10.23731/CYRM-2021-001.
  18. [18]V. Obreja and A. Obreja, Activation energy values from the temperature dependence of silicon PN junction reverse current and its origin, Physica Status Solidi (A) 207 (2010) 1252.DOI
  19. [19]A. Chilingarov, Temperature dependence of the current generated in Si bulk, JINST 8 (2013) P10003.DOI
  20. [20]H. Spieler, Semiconductor Detector Systems, Semiconductor Science and Technology, Oxford University Press, Oxford (2005).DOI
  21. [21]A. Grove, Physics and Technology of Semiconductor Devices, Wiley, Oxford (1967).

Paper details

vlnc:W_rac634x2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents