This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights

Nuclear Instruments and Methods in Physics Research A 732 (2013) 109 112 Contents lists available at ScienceDirect Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima A Low Mass On-Chip Readout Scheme for Double-Sided Silicon Strip Detectors C. Irmler a,n, T. Bergauer a, A. Frankenberger a, M. Friedl a, I. Gfall a, T. Higuchi d, A. Ishikawa e, C. Joo h, D.H. Kah g, K.H. Kang g, K.K. Rao f, E. Kato e, G.B. Mohanty f, K. Negishi e, Y. Onuki c, N. Shimizu c, T. Tsuboyama b, M. Valentan a a HEPHY Vienna Institute of High Energy Physics of the Austrian Academy of Sciences, Nikolsdorfer Gasse 18, A-1050 Vienna, Austria b KEK, 1-1 Oho, Tsukuba, Ibaraki 305-0801, Japan c University of Tokyo, Department of Physics, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan d University of Tokyo, Kavli Institute for Physics and Mathematics of the Universe, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8583, Japan e Tohoku University, Department of Physics, Aoba Aramaki Aoba-ku, Sendai 980-8578, Japan f Tata Institute of Fundamental Research, Experimental High Energy Physics Group, Homi Bhabha Road, Mumbai 400 005, India g Kyungpook National University, Department of Physics, 1370 Sankyuk Dong, Buk Gu, Daegu 702-701, South Korea h Seoul National University, High Energy Physics Laboratory, 25-107 Shinlim-dong, Kwanak-gu, Seoul 151-742, South Korea article info Available online 7 June 2013 Keywords: APV25 Silicon strip detector Low mass readout Chip-on-sensor Belle II Silicon vertex detector abstract B-factories like the KEKB in Tsukuba, Japan, operate at relatively low energies and thus require detectors with very low material budget in order to minimize multiple scattering. On the other hand, front-end chips with short shaping time like the APV25 have to be placed as close to the sensor strips as possible to reduce the capacitive load, which mainly determines the noise figure. In order to achieve both minimal material budget and low noise we developed a readout scheme for double-sided silicon detectors, where the APV25 chips are placed on a flexible circuit, which is glued onto the top side of the sensor. The bottom-side strips are connected by two flexible circuits, which are bent around the edge of the sensor. This so-called Origami design will be utilized to build the Silicon Vertex Detector of the Belle II experiment, which will consist of four layers made from ladders with up to five double-sided silicon strip sensors in a row. Each ladder will be supported by two ribs made of a carbon fiber and Airex foam core sandwich. The heat dissipated by the front-end chips will be removed by a highly efficient two-phase CO 2 system. Thanks to the Origami concept, all APV25 chips are aligned in a row and thus can be cooled by a single thin cooling pipe per ladder. We present the concept and the assembly procedure of the Origami chip-on-sensor modules. & 2013 Elsevier B.V. All rights reserved. 1. Introduction B-factories operate at relatively low energies and thus require dedicated particle detectors like the Belle [1] experiment which feature a very low material budget inside the sensitive volume in order to minimize multiple scattering. Belle was located at the asymmetric electron positron collider KEKB in Tsukuba, Japan, which was operated mostly at the ϒð4SÞ resonance. In June 2009 KEKB achieved the world's highest peak luminosity of 2:108 10 34 cm 2 s 1. It has been shut down one year later, after more than 10 years of operation. Since then it is being upgraded to SuperKEKB with a target luminosity of 8 10 35 cm 2 s 1, which is n Corresponding author. Tel.: +43 1 544732838. E-mail address: christian.irmler@oeaw.ac.at (C. Irmler). 40 times the luminosity of its predecessor. Consequently, the Belle detector also requires a significant improvement in order to cope with the higher collision and background rates. In particular the innermost subsystem, the Silicon Vertex Detector (SVD2) was already at its limits in terms of occupancy and dead time. The SVD2 was made from four layers of double-sided silicon detectors (DSSDs), which were read out by VA1TA chips. The future Belle II [2] Silicon Vertex Detector (Belle II SVD) will again consist of four layers of DSSDs, but at higher radii than its predecessor. In its center the Belle II SVD will be complemented by a two-layer DEPFET pixel detector (PXD), which is described in Ref. [3]. Since both subsystems together are parts of the vertex detector (VXD), the SVD layers are numbered from three to six. All sensors of the SVD will be read out by the APV25 [4], the same chip as used in CMS at CERN. The APV25 has a nominal clock frequency of 40 MHz, a peaking time of 50 ns and a 192 cell deep analog pipe 0168-9002/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nima.2013.05.169

110 C. Irmler et al. / Nuclear Instruments and Methods in Physics Research A 732 (2013) 109 112 line and thus provides sufficient readout speed for Belle II at a reasonable dead time of below 1%. Table 1 compares the key parameters of the APV25 and the VA1TA chips. It is obvious that the APV25 is not only faster than the VA1TA, but also has a higher equivalent noise charge (ENC) and thus is more susceptible to noise, mainly caused by capacitive load at its inputs. Hence the APV25 chips have to be placed as close as possible to the sensor strips in order achieve a reasonable signal-to-noise ratio (SNR). On the other hand the material budget in the active volume needs to be minimized to avoid multiple scattering. In Belle this was accomplished by concatenating up to three sensors and placing the readout chips at the edges of the ladders outside the sensitive volume. Unfortunately, such a solution would lead to an unacceptably low SNR in case of the APV25 chip and thus cannot be used for the Belle II SVD ladders. 2. The Origami chip-on-sensor concept In order to overcome these contradictive requirements we developed the so-called Origami chip-on-sensor concept, which allows to place the readout chips directly onto the sensors, while still keeping the amount of material low. In that scheme, the APV25 chips are thinned down to 100 μm thickness and glued onto a three-layer flexible Kapton hybrid. Fig. 1 shows both top and side views of the Origami concept. To keep the drawing simple, only the sensor section of the hybrid board is depicted. In the real design, the Origami flexes are extended towards the edge of the ladder (see Section 3), where connectors and some electronics components are attached to the board. Table 1 Key parameters of VA1TA and APV25 chips. Property VA1TA APV25 Clock frequency 5 MHz 40 MHz Peaking time T p 800 ns 50 ns Analog pipeline None 192 cells Radiation tolerance 200 kgy 41 MGy Equivalent noise charge 180 e + 7.5 e/pf 250 e + 36 e/pf CF sandwich ribs wrapped flex fanout APV25 chips (thinned to 100µm) cooling pipe 3-layer kapton hybrid fanout for n-side (z) APV25 (thinned to 100µm) CF sandwich ribs (mech. support) cooling pipe single-layer flex wrapped to p-side (r-phi) DSSD Kapton Airex DSSD Fig. 1. Top and side views of the Origami chip-on-sensor concept (not to scale). Only the sensor section without the tail of the Origami flex is shown. (a) Top view and (b) side view (cross section). Between the sensor and the hybrid there is a 1 mm thick sheet of Airex, a radiation hard and very lightweight but rigid foam. This foam layer has two functions, the first is to thermally insulate the sensor from the heat dissipating front-end chips. Secondly it acts as an electrical isolator to reduce the capacitive coupling and thus avoid crosstalk between hybrid and sensor strips as well as other unwanted influences from clock and signal lines. The strips on the top side of the sensor are connected to the inner four readout chips by a short pitch adapter, which is glued onto the hybrid in front of the APV25 chips. The strips of the opposite side of the sensor are attached to the outer six APV25 chips by two thin single-layer flex fan-outs, which are bent around the edge of the sensor, hence the name Origami. The APV25 chips have a non-negligible power dissipation and thus need to be cooled. On the Origami hybrid all chips are arranged in a row and thus can be cooled by a single pipe. For the Belle II SVD a highly efficient CO 2 cooling system will be installed which uses stainless steel pipes with an outer diameter of only 1.6 mm and a wall thickness of 100 μm. The mechanical support of the sensors and hybrids is realized by two carbon fiber reinforced ribs with a 3 mm thick Airex core, which are glued onto the bottom side of the sensors. This sandwich structure ensures sufficient mechanical strength but also the lowest possible contribution to the material budget. Taking all the components into account we calculated an average radiation length of only 0.548% X 0 [2] for one ladder. The feasibility of the Origami concept has been proven by building several prototype modules [5]. Results from beam tests at CERN have shown that for the Belle II DSSD a cluster SNR between 12 and 18 can be achieved, depending on pitch and length of the strips. Altogether, the Origami concept is a reasonable compromise between signal-to-noise ratio and material budget. 3. Belle II SVD ladder design The Belle II SVD will consist of four layers of DSSDs at radii of 38, 80, 104 and 135 mm. All detectors will be made from 6 in. wafers and arranged in ladders of 2, 3, 4 and 5 sensors. In order to optimize the incidence angle in the forward region, a slanted design with trapezoidal sensors will be implemented in the outer three layers. More details can be found in Ref. [6], here we focus on the utilization of the Origami concept for the central sensors of the Belle II SVD ladders. Fig. 2 shows an exploded view of the outermost layer. It consists of four rectangular sensors and one slanted, trapezoidal DSSD in the very forward region. While the two edge sensors will be read out by conventional printed circuit boards at the edges of the ladder, the Origami concept will be used for the remaining three inner sensors. Three different designs of Origami flexes are required. All of them are almost identical from the electrical point of view but differ in length and shape, driven by mechanical constraints of the SVD. According to the sensor location on the ladder and the counting order of the z-coordinate of Belle II, the different types are named z, ce and +z, where ce is an abbreviation for center. While the +z, which has the most complex shape, is only required in layer 6, the two other types will also be used in layer 5, while layer 4 only needs z. The design of all three types is already final and the series production of the flexes has started recently. Moreover, prototypes of the ce and z versions were successfully tested by building a module consisting of two DSSDs in June 2012. Unfortunately it is not possible to build a complete Belle II SVD ladder by composing individual Origami modules, because the tail of the ce flex is routed to the backward edge of the ladder underneath the z. In fact a dedicated set of jigs and a complex

Origami +z Origami ce Origami -z Author's personal copy C. Irmler et al. / Nuclear Instruments and Methods in Physics Research A 732 (2013) 109 112 111 Airex Pitch adapters and readout PCB of the forward edge sensor Pitch adapters for bottom strips DSSDs CF ribs Pitch adapters and readout PCB of the backward edge sensor Fig. 2. Exploded view of a Belle II SVD layer 6 ladder, showing all the components of the Origami design. procedure are required to build the ladders, including several alternating steps of gluing, wire bonding and testing. Attaching of Origami pitch adapters Place DSSDs onto the assembly bench and align them 4. Assembly procedure In this section we give an overview of the assembly procedure of a Belle II SVD Origami ladder. A more detailed discussion of each step, in particular the assembly of a prototype Origami module consisting of 2 DSSDs, is given in Ref. [7]. The SVD ladder in principle consists of three sections: the backward sensor, the Origami sensors in the middle part and the trapezoidal forward sensor. Accordingly, the assembly procedure can be divided into four sub-procedures: the backward part assembly, the forward part assembly, the Origami part assembly and finally some common tasks like gluing the sensors onto the ribs. Fig. 3 shows a rough overview of the ladder assembly task flow. To clarify the notation of the sensor sides, it has to be mentioned that DSSDs made from n-type bulk will be used in Belle II. Their p-side strips are on the sensor side which faces the beam pipe. Hence, with respect to the orientation of the Origami concept and the ladder design, the p-side is the bottom-side and the n-side is the top-side of the DSSD. To assemble a ladder, a number of vacuum jigs is required to place, pickup, align, glue and wire bond the components. The alignment of all these jigs is ensured by linear bushings and precision pins. An overview of the jigs and their primary purpose is given in Table 2. In the following description of the procedure, the jigs are numbered according to the first column of this table. At first, the two pitch-adapters, which are intended to connect the bottom-side strips of the Origami sensors to the APV25 chips, are glued onto the p-side and connected by wire bonding. For that, three jigs are required: the sensor jig (1) and two pitch adapter (PA) jigs (2). By providing three sets of them, this can be done for all the Origami sensors at once. Then these sensors are placed p-side down onto the so-called assembly bench (4), which is a complex jig sitting on the assembly base (3). The bench has three individual vacuum chambers, one for each of the Origami sensors and is used to safely carry the inner three sensors during the assembly of the Origami part. The assembly base (3) is intended to align the jigs for the backward (6) and forward sensors (7) as well as the rib jig (15) to the assembly bench by linear bushings and precision pins. Fig. 4 shows the above mentioned jigs (4, 6, 7) attached to the assembly base (3) together with five sensors, which are already placed at their designated positions. For the test assembly we used dummy sensors made from Aluminum. In the next step the coordinate measurement machine and the small xyθ stage (5), depicted in front of the bench, are used to align the sensors individually. For that, the xyθ stage is placed onto a sensor time Origami assembly Backward assembly Forwardassembly Glue forward and backward sensors onto ribs Attach Origami sensors to ribs Fig. 3. Assembly flow of a Belle II SVD layer 6 ladder. Table 2 List of the ladder assembly jigs and their main applications (PA¼pitch adapter). No. Name Purpose 1 Sensor jig Fix sensors to attach bottom-side PAs 2 PA jigs (two designs) Align and glue bottom-side PAs 3 Assembly base Align jigs to each other 4 Assembly bench Carry Origami sensor 5 xyθ stage Align the sensors 6 Forward sensor inlay Carry forward sensor 7 Backward sensor inlay Carry backward sensor 8 Forward PA jig Attach forward pitch adapter 9 Backward PA jig Attach backward pitch adapter 10 Forward jig Glue forward sensor onto ribs 11 Backward jig Glue backward sensor onto ribs 12 Airex jig Align and attach Airex sheet 13 Origami alignment jig Align Origami flexes 14 Origami jigs Pick up and glue Origami flexes 15 Rib jig Mount and align ribs Fig. 4. The assembly bench (4) on the assembly base (3), with dummy sensors on it, placed on a coordinate measurement machine. The xyθ stage (5) in front of it will be used to individually align each sensor. Numbers in brackets refer to column one of Table 2.

112 C. Irmler et al. / Nuclear Instruments and Methods in Physics Research A 732 (2013) 109 112 in order to slightly lift it up. While the alignment marks of the sensor are observed by the coordinate measurement machine, the sensor can be moved and rotated until it has reached its target position. By repeating this for all sensors, they can be aligned within a few minutes, resulting in a precision below 5 μm. Once the sensors are aligned, the assembly bench (4) carrying the Origami sensors is lifted. Thereafter also the forward and backward sensor jigs (6, 7) can be removed from the assembly base (3). From now on the further assembly of the backward part, the Origami part and the forward part can be done independently. In case of the two edge sensors, the subsequent steps are attaching and wire bonding of the n-side pitch adapters as well as connecting them to the edge hybrid. Then the sensors are flipped by picking them up with other jigs (8, 9). Afterwards the p-side pitch adapters are glued onto DSSDs and hybrids and are finally connected by wire bonding. As next step for the Origami sensors, the Airex sheet is aligned on the so-called Airex jig (12). After glue is dispensed onto the foam, the jig is placed bottom-up onto the assembly bench (4) so that the Airex is attached to the sensors. Once the glue is cured, the Airex jig (12) is removed from the assembly bench (4). Then the Origami flexes, which are already equipped with electronics components and APV25 chips, are glued on top of the Airex sheet one by one. According to Fig. 2 the order to attach the three flexes is ce, z and finally +z. Atfirst, the hybrid is placed onto the so-called Origami alignment jig (13) where it is aligned by small reference holes in the flex and pins. Thereafter the flex is lifted by the Origami jig (14), which has grooves at the locations of the APV25 chips in order to prevent damage to the chips and the already applied wire bonds. While the Origami alignment jig (13) can be used for all three types of flexes, a dedicated pickup jig (14) for each hybrid design is required. Once the flex is picked up, glue is dispensed onto its bottom surface followed by placing the jig onto the bench (4), so that the Origami flex is glued onto the Airex sheet. After all three flexes are attached, we perform wire bonding between the n-side strips and the corresponding pitch adapters located at the top side of the hybrids. Naturally, an optical inspection of the result is performed after each of the steps. However, after wire bonding of the n-side strips, the inspection should be done carefully, because in the next step the pitch adapters emerging from the bottom sides of the sensors are bent around the edges and glued onto the Origami flexes in front of the APV25 chips. Afterwards, some of the n-side wire bonds are covered by these pitch adapters and thus cannot be repaired later. As shown in Fig. 5 the flexes are bent using a micropositioner equipped with a custom vacuum nozzle. Thanks to this tool, it is easy to precisely align the flexes to the bond pads of the APV25 chips. Thereafter, wire bonding between APV25 chips and the pitch adapters is performed, followed by an electrical test of the Origami part. Then, the assembly bench (4), which still carries the Origami sensors, is lifted and put aside. The ladder support structure, consisting of the ribs and two mount blocks, is assembled and mounted onto a rib jig (15), which is already placed on the assembly base (3). Now the jigs number 10 and 11 are used to Fig. 5. A micro-positioner equipped with a custom vacuum head is used to bend and glue the bottom-side pitch adapters onto the Origami flex. Fig. 6. A layer 6 dummy ladder of the Belle II SVD. The ladder is still mounted onto the rib jig (15). glue both the backward and the forward part onto the ribs, finally followed by attaching the central Origami part. Recently, the whole procedure and most of the required jigs were verified in a dummy assembly. Although both the jigs and the procedure still need some minor improvements, we succeeded to build a full layer 6 dummy ladder, depicted in Fig. 6. 5. Summary The Origami chip-on-sensor concept is a low mass readout scheme for double-sided silicon strip detectors. In this scheme the readout chips for both sides of the sensor are located on a single flex hybrid, which is glued directly onto the sensors. Thanks to the lightweight design it is a tolerable compromise between reasonable SNR and low material budget and will be utilized to build the ladders of the Belle II SVD. The procedure to do this is almost finalized and has recently been verified during a dummy ladder production. References [1] A. Abashian, et al., The Belle Detector, Nuclear Instruments and Methods in Physics Research A 479 (2002) 117. [2] Z. Dolezal, S. Uno, (Eds.), Belle II Technical Design Report, KEK Report 2010-1, 2010 arxiv:1011.0352v1 [physics.ins-det], /http://xxx.lanl.gov/pdf/ 1011.0352v1S. [3] C. Marinas, M. Vos, Nuclear Instruments and Methods in Physics Research A 650 (2011) 59. [4] M. French, et al., Nuclear Instruments and Methods in Physics Research A 466 (2001) 359. [5] C. Irmler, et al., Construction and Performance of a Double-Sided Silicon Detector Module Using the Origami Concept, TWEPP-09, CERN-2009-006, 2009, pp. 211 215, /http://cdsweb.cern.ch/record/1234895/files/p211.pdfs. [6] M. Friedl, et al., The Belle II Silicon Vertex Detector, in this issue. [7] C. Irmler, et al., Journal of Instrumentation, http://dx.doi.org/10.1088/ 1748-0221/8/01/C01014, in press.