IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 6, DECEMBER 2004 3069 A New Technique for the Investigation of Deep Levels on Irradiated Silicon Based on the Lazarus Effect Pedro Rato Mendes, Member, IEEE, Maria C. Abreu, Vladimir Eremin, Zheng Li, Tapio O. Niinikoski, Sónia Rodrigues, Patrick Sousa, and Elena Verbitskaya Abstract—A new technique for the investigation of deep levels on irradiated silicon by measuring the charge collection efficiency (CCE) of samples from 220 down to 90 K is presented here. The temperature and time dependencies of the CCE have been mea- sured with unprecedented precision and resolution for standard and oxygenated silicon diodes, and the data obtained have been analyzed in the framework of the Lazarus effect and polarization models, extracting information about the radiation-induced deep levels in the materials. Results are presented and discussed in terms of these models and what can be inferred from them when applied to experimental data. I. INTRODUCTION T HE present and near future challenges in experimental high-energy physics involve building large detectors for high luminosity colliders such as the LHC at CERN (good examples are the ATLAS [1] and CMS [2] experiments). The extremely high radiation levels expected close to the interaction point at these collider experiments call for tracking detectors with very good radiation resistance, capable of operating throughout the experiment’s lifetime of ten years, accumulating fluences up to equivalent 1-MeV neutrons per square centimeter (hereafter noted eq.n/cm ). After the observation of the so-called Lazarus effect back in 1998, namely the recovery of the charge collection efficiency (CCE) of heavily irradiated silicon at cryogenic temperatures [3], the RD39 Collaboration at CERN demonstrated the feasibility and operation of cryo- genic radiation-hard silicon detectors, also contributing to the understanding of the physics of irradiated silicon at low tem- peratures [4]. The growing quantity and quality of experimental Manuscript received November 15, 2003; revised August 8, 2004 and September 22, 2004. This work was supported in part by the Por- tuguese Foundation for Science and Technology (FCT) under Contract POCTI/FNU/43681/2001. P. Rato Mendes is with LIP Lisbon, 1000-149 Lisbon, Portugal (e-mail: rato@lip.pt). M. C. Abreu is with LIP Lisbon and University of Algarve, FCT, 8000-117 Faro, Portugal (e-mail: mabreu@ualg.pt). V. Eremin is with Ioffe Physico-Technical Institute, 194021 St. Petersburg, Russia (e-mail: Vladimir.Eremin@pop.ioffe.rssi.ru). Z. Li is with Brookhaven National Laboratory, Upton, NY 11973-5000 USA (e-mail: zhengl@bnl.gov). T. O. Niinikoski is with CERN, CH-1211 Geneva, Switzerland (e-mail: Tapio.Niinikoski@cern.ch). S. Rodrigues is with LIP Lisbon, 1000-149 Lisbon, Portugal (e-mail: ro- driguessonia@ualg.pt). P. Sousa is with LIP Lisbon, 1000-149 Lisbon, Portugal, and also with the University of Algarve, FCT, 8000-117 Faro, Portugal (e-mail: pe- sousa@ualg.pt). E. Verbitskaya is with Ioffe Physico-Technical Institute, 194021 St. Peters- burg, Russia (e-mail: Elena.Verbitskaya@pop.ioffe.rssi.ru). Digital Object Identifier 10.1109/TNS.2004.839077 TABLE I SAMPLES AND IRRADIATIONS data, together with theoretical developments in the modeling of radiation damaging mechanisms, have led to the double junction [5], detector polarization [6], and Lazarus effect [7] models which, considered together, explain the behavior of observed data. These models also showed that both the time and temperature dependencies of CCE at low temperatures can yield information about material parameters such as activation energies, cross sections, and densities of radiation induced deep levels, with more consistent results than those given by standard methods like deep level transient spectroscopy (DLTS) or transient current technique (TCT) [4]–[7]. In order to explore the potential of deep level investigation based on the CCE measurement at low temperatures in a more systematic way, the time and temperature dependencies of the CCE of standard and oxygenated silicon diodes have been measured from 220 to 95 K at different bias voltages (both reverse and forward) and the data has been analyzed in the framework of the detector polarization and Lazarus effect mode. Although these models are only applicable for diodes under reverse bias after space charge sign inversion (SCSI), data on forward bias operation is also presented to provide an extended discussion of the CCE behavior at low temperatures. II. EXPERIMENTAL SETUP AND METHODS The time and temperature dependencies of the CCE of heavily irradiated samples have been measured using the simple cryostat at the LIP laboratory in the University of Algarve, Portugal, described in [8]. All the samples used in this work are made of high resistivity (4–6 k cm) - m-thick square silicon diodes 7 7 mm, manufactured by Brookhaven National Laboratory (BNL). Two of the samples studied (1031-3 and 1031-2, hereafter denoted STD1 and STD2, re- spectively) are standard diodes irradiated with 24 GeV protons at CERN, Geneva, Switzerland, while the other two (997-8 and 997-9, hereafter denoted OXY1 and OXY2, respectively) are oxygenated diodes (93 h in oxygen at 1200 C, oxygen concentration not measured) irradiated with 192 MeV pions at PSI, Zurich, Switzerland. Sample labels and irradiation data can be found in Table I with all fluences expressed in equivalent 0018-9499/04$20.00 © 2004 IEEE