Reliability aspects of a radiation detector fabricated by post-processing a standard CMOS chip Cora Salm * , Victor M. Blanco Carballo, Joost Melai, Jurriaan Schmitz MESA+ Institute for Nanotechnology, University of Twente, Semiconductor Components Group, P.O. Box 217, 7500 AE Enschede, The Netherlands article info Article history: Received 28 June 2008 Available online 10 August 2008 abstract This paper describes various reliability concerns of the newly developed INGRID detector. This radiation detector is fabricated by waferscale CMOS post-processing; fresh detectors show excellent performance. Since the microsystems will be used unpackaged they are susceptible to all kinds of environmental con- ditions. The device passed tests of micro-ESD, radiation hardness, dielectric strength; but humidity tests show one weakness of SU-8 as a structural material. Already after 1 day of exposure to a humid condition the structural integrity, as measured by a shear stress test, is dramatically lowered. Dry storage of these devices is therefore a necessity. KMPR photoresist shows promising results as an alternative structural material. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction We have recently shown the 3D integration of a miniaturized gas-gain radiation detector on CMOS chips [1]. For radiation imag- ing applications gas-gain grids are commonly used. These grids are punctured metal membranes suspended some micrometers over an anode plane, inside a gas volume, filled with for example a he- lium/isobutane mixture. Ionizing radiation (e.g. a cosmic ray parti- cle) that crosses the gas volume over the grid can liberate electrons that are then driven towards the anode by an electric field. The high electric field (100 kV/cm) between grid and anode causes an electron avalanche in this region and hence an exponential in- crease in the number of free electrons that reach the anode. A charge sensitive amplifier connected to the anode records arrival time, position and pulse height. When a microchip is used as the anode [2,3] the signal is picked up directly at the origin, reaching very high sensitivity (Fig. 1). The chip in this figure has an array of (charge collecting) bond pads each connected to a pre-amplifier and buffer. With a manually mounted grid, misalignment between holes in the grid and the sensing array on the chip leads to Moiré effects. These are overcome, by processing the punctured grid on top of the chip through wafer post-processing. This paper briefly describes micro-ESD, a failure mechanism known from the separately mounted predecessor of the microsys- tem that has been tackled during the development phase of this integrated microsystem. The bulk of this paper focuses on the deg- radation of these microsystems before its first use. These microsys- tems will be used without a package and are thus susceptible to degradation due to environmental conditions such as high humidity. 2. Materials and processing details The process details of the INtegrated Grid (INGRID) can be found in [1] (and references therein). In this study we used alumi- num as material for both anode and punctured grid. The 55 lm high support pillars are made of SU-8, a negative tone photoresist commonly used in MEMS fabrication [4]. We use SU-8_50, since this can be spun in the thickness range suitable for our microsys- tems. SU-8 processing takes place below 95 °C making it suitable for CMOS post-processing; and developed SU-8 is radiation hard [5]. Note that multiple thinner layers, using e.g. SU-8_5 or SU-8_10, may also be spin coated to make a thicker layer and subsequently expose the film. The composition of the different kinds of SU-8 is different, mainly the amount of solvents that determine the viscos- ity. For each kind of SU-8 and each application the process flow has to be optimized. The impact of using a different kind of SU-8 on the reliability of the INGRID microsystem was NOT part of this study. In Section 5 we report initial studies on detectors made with alternative electrode or electrode covering materials and KMPR as functional material to replace SU-8. 3. Micro-ESD A problem associated with all gas-filled proportional chambers is sparking. When an electron avalanche reaches Raether’s limit [6], it may evolve into a discharge. We assume that the density and energy of the participating electrons becomes high, forming 0026-2714/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.microrel.2008.06.038 * Corresponding author. Tel.: +31 534892648; fax: +31 534891034. E-mail address: c.salm@utwente.nl (C. Salm). Microelectronics Reliability 48 (2008) 1139–1143 Contents lists available at ScienceDirect Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel