Passivation of Si(Li) detectors operated above cryogenic temperatures for space-based applications
Nathan Saffold, Field Rogers, Mengjiao Xiao, Radhika Bhatt, Tyler Erjavec, Hideyuki Fuke, Charles J. Hailey, Masayoshi Kozai, Derik Kraych, Evan Martinez, Cianci Melo-Carrillo, Kerstin Perez, Chelsea Rodriguez, Yuki Shimizu, Brian Smallshaw
PPassivation of Si(Li) detectors operated above cryogenictemperatures for space-based applications
N. Sa ff old a,1 , F. Rogers b , M. Xiao b , R. Bhatt b , T. Erjavec b , H. Fuke c , C. J. Hailey a ,M. Kozai c , D. Kraych a , E. Martinez a , C. Melo-Carrillo a , K. Perez b , C. Rodriguez a ,Y. Shimizu d , B. Smallshaw a a Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027 b Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139 c Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency (ISAS / JAXA), Sagamihara,Kanagawa 252-5210, Japan d Kanagawa University, Yokohama, Kanagawa 221-8686, Japan
Abstract
This work evaluates the viability of polyimide and parylene-C for passivation of lithium-driftedsilicon (Si(Li)) detectors. The passivated Si(Li) detectors will form the particle tracker and X-raydetector of the General Antiparticle Spectrometer (GAPS) experiment, a balloon-borne experi-ment optimized to detect cosmic antideuterons produced in dark matter annihilations or decays.Successful passivation coatings were achieved by thermally curing polyimides, and the optimizedcoatings form an excellent barrier against humidity and organic contamination. The passivatedSi(Li) detectors deliver (cid:46) −
100 keV X-rays while op-erating at temperatures of −
35 to −
45 °C. This is the first reported successful passivation ofSi(Li)-based X-ray detectors operated above cryogenic temperatures.
Keywords:
Semiconductor detectors, Particle tracking detectors, X-ray detectors, Passivation,Cosmic rays, Dark Matter
1. Introduction
A key technical challenge in lithium-drifted silicon (Si(Li)) detector fabrication is passi-vation to protect against environmental contaminants. The General Antiparticle Spectrometer(GAPS) experiment has previously reported on Si(Li) detectors fabricated in-house [1] and to-gether with Shimadzu Corp [2, 3]. Here, we report on the selection and testing of a passivationmethod to ensure the long-term stability of Si(Li) detectors for the GAPS experiment. GAPSis a balloon-borne instrument optimized to detect low-energy ( < / nucleon) cosmic rayantinuclei. Several well-motivated dark matter (DM) models allow annihilation or decay intoStandard Model particles, that would produce a flux of low-energy antinuclei [4, 5, 6, 7]. At lowenergies, secondary production of antideuterons and antihelium from cosmic ray interactionswith the interstellar medium is suppressed, making low-energy antideuterons and antihelium a Corresponding author
Email: nas2173@columbia
Preprint submitted to Nuclear Instruments and Methods in Physics Research A February 12, 2021 a r X i v : . [ a s t r o - ph . I M ] F e b smoking gun” signal of dark matter [8, 9]. Using data from three Antarctic long duration bal-loon (LDB) flights, GAPS will set leading limits on the antideuteron and antihelium flux (at lowenergies) [10] and extend the antiproton spectrum to low energies (E < ∼ ∼
15 m combined surface area of scintillator. The particletracker consists of ten 1.6 × planes of Si(Li) detectors, with each layer separated with10 cm spacing. A large instrument is necessary to have high acceptance to cosmic ray antinuclei,but presents some experimental di ffi culties. Due to the large volume of the instrument, it isimpossible to fly a pressure vessel or cryostat within the weight constraints of an LDB flight.Therefore, the Si(Li) detector array is cooled to relatively high operational temperatures, between −
35 and −
45 °C, using a novel heat pipe system outlined in [12]. The Si(Li) detector array servesas both the target material to slow down the incoming antiparticle and the detector to measure anincoming particle’s d E / d x and the products of exotic atom de-excitation and annihilation. Figure 1: Exposed view of GAPS payload. Plastic scintillator paddles (black) comprise the outer TOF ‘umbrella’ andinner TOF ‘cube.’ The inner TOF cube surrounds the tracker, housing 10 layers of 144 detectors. Each detector module(gray) houses four Si(Li) detectors in 2x2 array. Structural supports are also shown.
The GAPS detection scheme exploits exotic atom physics to identify incoming antinuclei. Alow-energy antiparticle first passes through the TOF, which measures its velocity and d E / d x andprovides a high-speed trigger. It then traverses the Si(Li) detector tracking system, undergoingd E / d x losses until it is captured by an atomic nucleus in a Si(Li) detector or aluminum support,forming an exotic atom. The exotic atom de-excites, emitting Auger electrons and X-rays, andultimately annihilates, producing pions and protons [13, 14, 15]. The annihilation vertex providesa unique signature to discriminate antinuclei from baryonic cosmic rays. The stopping depth,d E / d x , energies of the X-rays produced during the de-excitation, and multiplicity of pions andprotons emerging from the annihilation vertex are used to distinguish di ff erent antinuclei species.This work reports on the R&D e ff ort to develop and select a passivation method for the GAPSSi(Li) detectors. Si(Li) passivation techniques and GAPS requirements for selecting a passiva-tion method are reviewed in Sec. 2. Adhesion and thermal testing of passivation candidates aredescribed in Sec. 3. In Sec. 4, we outline the noise testing conducted to assess passivated detectorperformance. Accelerated lifetime testing and long-term monitoring of passivated detectors are2eported in Sec. 5. Finally, conclusions and future prospects are presented in Sec. 6.
2. Si(Li) Passivation Review & Requirements
The GAPS Si(Li) fabrication method was developed in collaboration with Shimadzu Corpo-ration, and further details on the fabrication technique and process yield are presented in [2]. Thegeometry of the GAPS Si(Li) detectors is shown in Figure 2. In order to distinguish de-excitationX-rays from di ff erent antinuclei, the GAPS Si(Li) detectors must deliver (cid:46) −
100 keV X-rays. The LDB power and thermal constraints require the detectorsto be operable at a bias voltage of 250 V, at the relatively high temperatures of −
35 to −
45 °C.The unpassivated detectors delivered the requisite noise performance, as demonstrated in [3].However, these measurements were made in a humidity-controlled lab environment and the de-tectors were cleaned immediately before testing. Detector performance is sensitive to the surfacestate of the exposed silicon of the grooves (Fig. 2,
Figure 2:
Top:
Photograph of passivated 8-strip GAPS flight detector with ruler for scale. If left unpassivated, thegrooves segmenting the strips and guard ring are susceptible to contamination that can degrade detector performance.
Bottom:
Diagram of 8-strip GAPS detector cross section. 1) The top hat geometry is defined using Ultrasonic ImpactGrinding (UIG) to remove the top perimeter of the Si wafer, leaving a ∼ ∼ ∼ µ m di ff used n + layer. 3) Li ions drifted into p-type bulk to create compensated active volume. 4) The Siin the top hat ‘brim’ remains uncompensated. Electrodes consist of ∼
20 nm of Ni (5) and ∼
100 nm of Au (6). ∼ ∼ GAPS requires a passivation candidate that provides a barrier to environmental contaminantsand is robust to thermal cycling and mechanical shock. The passivation coating must protect thedetector from its ambient environment and prevent the deleterious e ff ects of surface contami-nation which can produce high leakage current and 1 / f noise. The passivation process must beperformed at low enough temperatures to avoid Li di ff usion e ff ects. Specifically, the Li in the n + ff ectively shunt thestrips. Furthermore, the cure cycle must not cause significant decompensation of the positive Liions in the p-type bulk, which could cause poor charge collection from the detector’s active vol-ume and increase the voltage required to deplete and operate the detector [16]. The passivationprocedure must be adaptable to the geometrical constraints of GAPS, and be routinely appliedby technicians to passivate the > layers are typically obtained at high temperatures( T > was not explored. SiN, TaN, and TiN are often used as passivation coatings forSi substrates, and are typically produced using chemical vapor deposition (CVD), atomic layerdeposition (ALD), or sputtering. These materials were not explored because the deposition tech-niques are not suitable for this work. CVD and ALD of SiN, TaN, and TiN typically require highsubstrate temperatures ( >
300 °C), and sputtering is di ffi cult to confine to the bare Si surfaceswith high reproducibility [17, 18, 19]. Another common method for passivation of silicon detec-tors is silicon monoxide. However, high temperature treatments are often necessary to produceoptimal SiO films. Furthermore, previous studies indicate that devices passivated with SiO havehigher leakage currents and additional 1 / f noise [20]. Thus SiO was not explored. Previous workdemonstrated successful passivation of Si(Li) detectors using hydrogenated amorphous silicon( α -Si:H) [21]. α -Si:H is typically deposited by RF sputtering onto the detector. Amorphoussilicon passivation was not explored because 1) the deposition process is complicated and re-producibility is poor, 2) the heat involved in the RF sputtering process can have an e ff ect on Licompensation, and 3) GAPS operates Si(Li) detectors at higher temperatures than this previousstudy, and in this temperature range α -Si coated detectors have reported higher leakage currentcharacteristics than bare detectors [21].Based on previous work, we chose to focus passivation R&D on polymers, specifically poly-imides and Parylene-C [22]. Previous work reports successful passivation of Si(Li) detectorsusing polyimides [22, 23] and Parylene-C [22]. However, these detectors were operated at muchlower temperatures than the GAPS operating temperature, where di ff erent noise componentsdominate. This work evaluates the viability of polyimide (PI) and parylene-C for passivation ofSi(Li) detectors operated well above cryogenic temperatures.
3. Mechanical Testing
In order for a passivation method to be viable for the GAPS experiment, the passivationcoating must adhere to the Si surfaces. Polymer coatings typically fail due to cracking and / ordelamination, and these failures are directly related to the state of stress in the coating. Thestress is due to a combination of the material properties of the coating, the processing conditionsused to produce the coating, and the environment. If the stress exceeds the ultimate strengthof the coating, the coating fails by cracking. If the stored energy in the coating exceeds thework of adhesion to the substrate, the coating can delaminate. Furthermore, a mismatch inthe coe ffi cient of thermal expansion (CTE) between the polymer and the substrate can lead todelamination / cracking during temperature cycling [24].In order to test the mechanical properties of the coatings, adhesion and thermal testing wasperformed. Adhesion performance was assessed using a 180° pull test, following the ASTM43359 standard (see Sec. 3.3.1). Thermal testing consisted of cycling the detector between roomtemperature and −
50 °C, and visually inspecting the polymer coatings under a microscope (seeSec. 3.4.1).
Test-grade Si wafers , were prepared for adhesion testing and thermal cycling studies. Foradhesion testing studies, adherence to the ASTM D3359 standard required applying the polymerto a planar wafer surface. For thermal cycling studies, it was desired to apply the polymerto a sample geometry that is analogous to a detector, specifically to grooves that have beenchemically polished to smoothness. Therefore, grooves were cut into thermal cycling samplesusing Ultrasonic Impact Grinding (UIG) before cleaning, etching, and applying the polymer (seeTable 1). Thermal Sample UIG protocol:
Using UIG, 1 mm wide, 350 µ m deep grooves were cut intowafers. The groove depth and width were chosen to be analogous to the grooves segmentingthe GAPS Si(Li) detectors. After UIG, the samples were cleaned ultrasonically in ACS-gradehexane to remove any wax and abrasive slurry from the UIG process. After ultrasonic cleaningwith hexane, the following sample preparation was performed. Thermal & Adhesion Test Sample Preparation:
A 3-step cleaning process was performed onall wafers, consisting of ultrasonic cleanings in ACS-grade acetone, methanol and DI water. Allsamples were etched in an HNA (20 mL 49% Hydrofluoric Acid, 35 mL 60% Nitric Acid, 55 mLGlacial Acetic Acid) solution for 10 minutes. The etching process chemically polishes the Sisurfaces, in particular the groove surfaces, which are left rough from the UIG process. Withthe etchant formulation used, a 10 minute etch was su ffi cient to produce the smooth and glassygroove surfaces typical of the GAPS Si(Li) detectors [2]. These smooth surfaces presented anadhesion challenge, as rougher surfaces are typically better for polymer adhesion. After cleaningand etching, Polyimide or parylene-C was applied to the wafers. Adhesion and thermal testingwas performed for two polyimides and Parylene-C, with and without an adhesion promoter. Table 1: Sample preparation and cleaning protocols for adhesion and thermal testing samples. 3-step cleaning entailsultrasonic cleaning a wafer in acetone, methanol, and DI for 5 minutes each, followed by drying with N2. Passivationcoatings were applied to samples after etching.
Sample Code UIG Hexane 3-Step Clean EtchAdhesion (cid:55) (cid:55) (cid:51) (cid:51)
Thermal (cid:51) (cid:51) (cid:51) (cid:51)
Two polyimide (PI) precursors, VTEC PI-1388 and Ube U-Varnish-S, were used for the fol-lowing studies. VTEC PI-1388 was selected due to its relatively low cure temperature ( ∼
250 °Cfor full imidization), while Ube U-Varnish-S was selected because its CTE is a close match to sil-icon ( ∼ / °C). PI was applied by painting a polyimide precursor onto the wafer surface, andcuring the polyimide in a Vulcan 3-1750 furnace. Each polyimide precursor is manufactured by Test-grade Si wafers were procured from Addison Engineering and Wafer World ≥
250 °C) are desired to fully drive out the solventand imidize the PI, but are not feasible for Si(Li) substrates because of high lithium mobility insilicon. To avoid movement of Li in the n + layer and compensated region, we did not use curetemperatures exceeding 210 °C.The following processing parameters were varied between PI samples: cure temperature,cure time, heating rate, use of silane adhesion promoter, and dilution of polyimide (see Table 2).These processing parameters were tuned to produce PI coatings robust to mechanical and thermalstresses. Previous Si(Li) passivation literature suggested ‘soft-baking’ the polyimide at 120 °Cfor 25 minutes [22, 23], which drives most of the solvent from the PI precursor, but leaves the de-gree of imidization relatively low [22]. This soft-bake was used as a baseline cure condition, butfurther optimization was necessary based on adhesion and thermal testing results. ‘Rapid curing’,by placing a substrate with PI into a pre-heated oven set to the cure temperature, was comparedto ‘slow curing’, by placing the substrate with PI into an oven at room temperature, and rampingthe temperature to the cure temperature with a specified heating rate. For slow curing, a heatingrate of 5 °C / min was used. This heating rate was chosen based on a previous study that found thatheating rates <
10 °C / min lead to a higher degree of solvent removal and imidization [25]. To testthe e ff ect of silane on polyimide adhesion, γ -Aminopropyltriethoxysilane (APS), was applied tosome etched wafers before applying the PI. A 0.1% (v / v) solution of APS was prepared in DIwater and mixed for 1 hour on a hot plate with a magnetic stir bar. The APS solution was thenapplied by hand to the clean wafer surfaces, and baked on a hot plate at 85 °C for 30 minutes.This APS application protocol was based on a previous study, which demonstrated that APS in-creased the adhesion strength of the PI-silicon interface by a factor of 25 [24]. The PI precursorwas either applied to the substrate ‘neat’ as it arrived from the manufacturer, or in a 1:1 dilutionof PI precursor and pure NMP. Diluting the PI precursor solution decreases its viscosity, andenables application with a pipette. Parylene-C was applied in a vapor deposition chamber (SCS Labcoter 2, PDS 2010). Thedeposition process conformally coats the substrate with a Parylene-C film. For a given depositionchamber, the resulting film thickness is proportional to the mass of the dimer that is vaporized,so that for the chamber employed in this study 1 g of dimer resulted in a 1 µ m thick coating.For all parylene-C samples, 5 g of dimer was used to produce 5 µ m thick coatings. For somesamples, a silane adhesion promoter (A-174) was applied before parylene-C deposition, to assessthe silane’s impact on coating adhesion. For adhesion and thermal testing studies, samples wereconformally coated.In preparation for noise testing, a method to mask the readout strips was developed. Selec-tive deposition by shadow masking and surface priming was attempted. For shadow masking,wafers were wedged in between two aluminum plates with machined cut-outs in the shape of thedetector strips. After parylene-C deposition, the masks were cut away from the wafer using asharp blade. For surface priming, a Micro-90 solution was painted onto the strip surface beforeparylene-C deposition. Micro-90 (Mfg: Cole-Parmer) is a soap-like solution that inhibits theParylene-C adhesion to silicon, enabling us to peel the Parylene-C from the selectively primedsurfaces [26]. After vapor deposition, the Parylene-C was mechanically removed from the elec-6 able 2: PI application parameters were optimized to provide a coating with good adhesion and thermal properties.Initial samples (Sample code 1A) were ‘rapidly cured’ by placing them in an oven pre-heated to the cure temperature;however, it was quickly determined that ‘slow curing’ by ramping the oven temperature from room temperature to thecure temperature yielded better coatings. The final PI passivation protocol is highlighted in gray. Sample Code Cure Temperature Cure Time Heating Rate APS Dilute PI1A 120 °C 25 mins None (cid:55) (cid:55)
1B 120 °C 25 mins 5 °C / min (cid:55) (cid:55)
2A 180 °C 10 mins 5 °C / min (cid:55) (cid:55)
2B 180 °C 10 mins 5 °C / min (cid:51) (cid:55)
2C 180 °C 10 mins 5 °C / min (cid:55) (cid:51)
2D 180 °C 10 mins 5 °C / min (cid:51) (cid:51)
3A 180 °C 25 mins 5 °C / min (cid:55) (cid:55)
3B 180 °C 25 mins 5 °C / min (cid:51) (cid:55)
3C 180 °C 25 mins 5 °C / min (cid:55) (cid:51)
3D 180 °C 25 mins 5 °C / min (cid:51) (cid:51)
4A 210 °C 60 mins 5 °C / min (cid:55) (cid:51)
4B 210 °C 60 mins 5 °C / min (cid:51) (cid:51) trodes by scraping and pulling with electrostatic discharge safe polyvinylidene fluoride tippedtweezers and the Micro-90 was cleaned from the detector surface using a methanol-soaked swab. & Results3.3.1. Adhesion Testing Method & Success Criteria
Adhesion testing was performed on test samples using an ASTM D3359 cross hatch adhesiontest [27]. Using a razor blade, an X-shape cut was notched into the polymer coating. Elcometer99 tape was pressed and smoothed down onto the coating surface, on top of the X-cut. Within90 ±
30 seconds of applying the tape, the tape was pulled o ff , pulling it back upon itself at a 180°angle. After pulling, the coating was inspected for blemishes, and the tape was inspected forresidue from the coating. The degree of polymer removal due to the pull test is graded on a 0-5scale (5 indicates no polymer removed, 0 indicates a majority of the polymer was removed). Inorder to pass adhesion testing, a coating had to pass the 180° pull test with a score greater thanfour. No di ff erence in adhesion strength was noted between neat and dilute samples of a givenPI. PI samples without APS failed all adhesion tests, while PI samples with an APS pre-coatingpassed all adhesion tests, for both neat and dilute PI-1388 and U-Varnish-S. Without an APSadhesion layer, VTEC PI-1388 demonstrated stronger adhesion properties than Ube U-Varnish-S, as U-Varnish-S samples demonstrated a higher degree of PI removal after adhesion testing,regardless of dilution. All samples prepared with the final PI application protocol (Sample Code4B) passed adhesion tests with a grade of 5.Conformally coated parylene-C samples passed all adhesion tests, with and without a silanebase layer. Thus, for noise testing, parylene-C samples were not prepared with a silane pre-coating, while PI samples were primed with an APS / DI solution.7 .4. Thermal Testing & Results3.4.1. Thermal Cycling Testing Method & Success Criteria
Samples were thermal cycled in a custom testing setup that consisted of dry ice and EPS foaminsulation. The testing setup was assembled to cycle the samples between room temperature and −
50 °C, with a ramp rate < / min. This temperature profile is consistent with the cooling usedduring laboratory calibration and expected during the LDB flight. During thermal cycling, thewafers were kept in polypropylene carrying cases, to avoid condensation on the wafer surfaceswhen opening the chamber. After each thermal cycle, the coatings were inspected visually andunder a microscope for cracking and delamination. A sample was required to survive twelvethermal cyclings without exhibiting delamination or cracking to be deemed successful. Parylene-C coatings were extremely robust to thermal stresses, and did not demonstrate anydelamination or cracking through 12 thermal cycles.Initially, PI samples were cured at 120 °C for 25 minutes (Sample Codes 1A-1B), based onprevious successful passivation work [22, 23]. However, these samples exhibited delaminationand cracking through successive thermal cyclings, and the cure temperature was subsequentlyincreased to 180 °C (Code 2A-3D). PI applied ‘neat’ to the grooves demonstrated poor repro-ducibility, as the neat PI precursor had a high contact angle with the Si surface and aggregatedduring curing, leaving bare silicon surfaces exposed. Therefore, dilute PI application was se-lected for noise testing. Rapidly cured samples, cured in an oven pre-heated to the cure temper-ature, demonstrated higher failure rates than slow cured samples because slow curing promotesbetter solvent removal and prevents thermal shock in the coating [25]. The PI samples primedwith APS and slow cured at ≥
180 °C for ≥
10 mins (Sample Codes 2D, 3D, 4B) were robust tothermal cycling (see Fig. 3).
Figure 3:
Left / Center:
Microscope images of grooves of thermal test sample before (left) and after (center) two thermalcycles. After two thermal cycles, spots appeared at bottom of groove. These spots are indicative of delamination of thePI from the silicon surface, due to CTE mismatch. This sample was cured at 120 °C for 25 minutes, a cure cycle thatwas found to be insu ffi cient to produce a robust coating. Right:
Microscopic images of a PI coated groove, passivatedusing the final procedure. No delamination or cracking was observed after twelve thermal cyclings in PI coatings usingthe final passivation procedure.
4. Noise Testing
After tuning the application protocols to produce coatings with acceptable adhesion and ther-mal properties (see Sec. 3), the passivation coatings were applied to Si(Li) detectors, and detector8oise performance was assessed before and after passivation. The main requirement is that GAPSSi(Li) detectors must provide (cid:46)
A detector’s grooves and top hat were cleaned before passivation. The standard cleaningprotocol consists of applying methanol to the tip of a cleanroom swab, and gently swabbing thedetector’s bare silicon surfaces. This cleaning protocol removes any particulate contaminationfrom the surfaces and sets the surface state of the silicon to be lightly n-type. The cleaningmust be performed in a low-humidity environment ( <
10% relative humidity), specifically a N2purged glove box, to yield consistently good results. Detectors fabricated in-house at ColumbiaUniversity (TDxxxx) and at Shimadzu (Shxxxx) were passivated and tested. There are somedi ff erences in the fabrication protocols [1, 2]. Typically, the in-house detectors were used todemonstrate proof of concept, whereas the success criteria were ultimately applied to Shimadzudetectors.Parylene-C and PI passivation coatings were applied to in-house detectors as proof of con-cept. Parylene-C was applied using Micro-90 to mask the electrodes and enable readout (seeSec. 4.4.1). PI was applied using an APS adhesion promoter and a diluted VTEC PI-1388 poly-imide precursor solution. For initial testing purposes, the APS and PI cure conditions werevaried to produce optimal passivated detector noise performance. The minimal cure temperatureand time that produced coatings with acceptable adhesion and thermal properties were 85 °C for30 minutes for APS curing, and 180 °C for 10 minutes for PI curing (see Sec. 3.2.2). These cureconditions were compared to higher temperature cures that were chosen to exceed the boilingpoint of the solvent in the APS and PI precursor solution. To improve the solvent removal anddegree of PI imidization, PI was cured at 210 °C for 1 hour, and APS was cured at 110 °C for20 minutes. Based on the initial testing results, where lower temperature cures resulted in de-graded leakage current and energy resolution (see Sec. 4.4.1), the higher temperature cures wereused for all subsequent PI passivated detectors.The final PI passivation protocol is as follows: • Mix 0.1% (v / v) solution of APS in de-ionized water for 1 hour • Apply APS solution to detector grooves and top hat using pipette • Bake detector in open glass petri dish on hotplate at 110 °C for 20 minutes • Let detector cool, mix 1:1 dilution of PI precursor in NMP by hand with teflon applicator • Degas PI in rough vacuum for 10 minutes to remove bubbles • Apply PI precursor solution to detector grooves and top hat using pipette • Cure in oven at 210 °C set point temperature for 1 hour with 5 °C / min heating rate • After 1 hour at 210 °C set point, prop oven open to decrease temperature gradually • When oven temperature reaches 70 °C ( ∼
40 minutes), remove detector from oven andplace in dry boxTemperature testing was conducted to ensure that the substrate temperature reached 110 °Cduring the APS cure and >
204 °C during the PI cure. A dummy substrate was placed in the oven9nd on the hot plate used for curing passivation coatings. Thermocouples were used measure thesubstrate’s surface temperature. A MicroDAQ USB-TEMP temperature data acquisition modulewas used to log the substrate temperature at each second during the cure cycles. During APScuring, the substrate equilibrated to ∼
110 °C after 10 minutes. In the oven, the substrate reached ∼
210 °C at the end of the 1 hour cure cycle, and the temperature ramp rate was kept below5 °C / min to avoid thermally shocking the PI coating. Noise testing was performed at MIT. To assess the success of a passivation coating, roomtemperature leakage current and cold ( ∼ –37 °C) spectral measurements were performed on a de-tector before and after applying the passivation coating. Per-strip leakage current was measureddirectly using a Keithley 487 Picoammeter / Voltage source with all other strips and guard ringgrounded. The Keithley 487 ramped the voltage applied to the p-side from 0 V to −
400 V in 25 Vincrements while measuring the resulting leakage current from a given strip on the n + side.Energy resolution measurements were performed in a custom vacuum testing chamber out-lined in [3] or in a SUN EC13 Temperature Chamber. Both chambers were cooled using LN2; thecustom chamber was pumped to ∼ −
35 °C > T > −
45 °C. In thechambers, the detectors were uniformly irradiated with a 100 µ Ci Am radioactive source. Dur-ing operation, the p-side of a detector was biased to −
250 V using a Tennelac 953 HV supply inthe vacuum chamber, and a CAEN N1471 in the SUN chamber. The signal was readout from then + side by a custom 8-channel discrete-component charge-sensitive preamplifier board, whichwas pressure mounted to the strips via spring-loaded pins. In the vacuum chamber, signal fromone preamplifier was processed by a Canberra 2020 Spectroscopy Amplifier at various peakingtimes and digitized by an Ortec Ametek Easy MCA module. In the SUN chamber, signals areshaped and digitized using a CAEN N6725 digitizer.A noise model is used to characterize each detector, identify each noise source that con-tributes to the overall energy resolution, and determine if the noise arises from intrinsic detectorperformance or from the readout chain [3, 29]. The equivalent noise charge (ENC) from the de-tector and readout chain, and the FWHM energy resolution can be estimated as follows [16, 30]: ENC = (cid:32) qI leak + kTR p (cid:33) τ F i + kT (cid:32) R s + g m (cid:33) C tot τ F ν + A f C tot F ν f (1) FWHM = . (cid:15) ENCq (2)In Eqs. (1) and (2), q is the fundamental charge, k is the Boltzmann constant, and (cid:15) isthe ionization energy of silicon (3.6 eV per electron-hole pair). R p , g m , F i , F ν , and F ν f arefixed parameters specific to preamplifier and shaping amplifier used in the readout chain and areoutlined in [3].The most relevant parameters in this study are I leak , C tot , A f , and R s , which are fit to charac-terize the noise performance of the detector. I leak is the temperature-dependent per-strip leakagecurrent. The total input capacitance ( C tot ) is the sum of all parallel capacitances including theindividual strip capacitance ( C det ), the capacitance of the FET input stage ( C FET ∼
10 pF), theinterelectrode capacitance of adjacent strips and the grounded guard ring ( C int ), and any straycapacitance ( C stray ∼ R s is the sum of all series resistances that can arise from the detector10nd preamplifier mounting. A f is the coe ffi cient of 1 / f noise which may arise from surface e ff ectsor (ideally just) preamplifier noise.Energy resolution was measured as a function of peaking time for detectors at a given op-erational temperature before and after applying a passivation coating. The peaking time ( τ ) vs.energy resolution (FWHM) data is used with Eq. (2) to find a best fit for I leak , C tot , A f , and R s while keeping the other noise model parameters fixed. Since, C tot , A f , and R s are degenerate,they cannot be fit simultaneously, so an iterative approach is used, which is described in [3]. Thefitted values for each strip are compared before and after passivation, and we assess the successcriteria based on the overall energy resolution and fit parameters.The following success criteria were imposed on passivated detector energy resolution andnoise model fit parameters:1. FWHM energy resolution at optimal peaking time (cid:46) I leak (cid:46) A f (cid:46) . × − V .The fitted I leak and A f cuto ff s were chosen based on unpassivated detector measurements per-formed in [3]. Before passivating Shimadzu flight detectors, in-house fabricated detectors were passivatedand tested to validate a passivation method. Parylene-C was applied to an in-house fabricatedSi(Li) detector, TD0087. After applying the parylene-C coating using selective deposition withMicro-90, the detector’s leakage current saturated the picoammeter’s current limit (2.5 mA) at ∼
30 V. Since the leakage current from the active area was < µ A during this test, it indicated alarge leakage current through the guard ring due the passivation coating. Based on this result andconcerns about the reproducibility of the coating method, Parylene-C was not further explored.An in-house fabricated Si(Li) detector, TD0093, was passivated with PI cured at 180 °C for10 minutes. After passivation, the detector’s leakage current was an order of magnitude higherthan its pre-passivation value at room temperature, −
36 °C, and −
48 °C. Re-baking the detec-tor at 180 °C for 25 minutes reduced the detector leakage current to its pre-passivation values.Therefore, we determined that the elevated leakage current was caused by an insu ffi cient curecycle that did not drive out the solvent and imidize the PI precursor, and all subsequent detectorspassivated with PI were cured at 210 °C for 1 hour. The heating involved in this cure cycle didnot significantly a ff ect the detector’s lithium distribution, and did not degrade detector energyresolution when operated at 250 V. We note that the voltage required to deplete and operate thedetector increases from ∼
80 V before passivation, to ∼
150 V after passivation; however, this isacceptable since the detectors will be operated at 250 V.To isolate the e ff ects of the APS application procedure, APS was applied to an eight-stripShimadzu detector, Sh0075, following the protocol used for thermal and adhesion samples. Af-ter curing at 85 °C for 30 minutes, the detector’s spectral performance degraded due to 1 / f noise(see Fig. 4). Re-heating the detector at 110 °C for 20 minutes improved the detector’s spec-tral performance so that no degradation was noted compared to pre-passivation measurements.Therefore, the 85 °C cure was deemed insu ffi cient to dry the APS solution, and APS was curedat 110 °C for all subsequent passivations. 11 s )24681020 F W H M E n e r g y R e s o l u t i o n ( k e V ) leak = 1.9 [nA], A f = 0.6 × 10 [V ]B: I leak = 2.2 [nA], A f = 0.7 × 10 [V ]C: I leak = 2.3 [nA], A f = 0.6 × 10 [V ] 12-20-2018 - Post-SilaneA: I leak = 2.7 [nA], A f = 3.1 × 10 [V ]B: I leak = 3.4 [nA], A f = 4.3 × 10 [V ]C: I leak = 2.2 [nA], A f = 5.7 × 10 [V ]01-25-2019 - Post-BakeA: I leak = 2.1 [nA], A f = 1.1 × 10 [V ]B: I leak = 2.2 [nA], A f = 1.4 × 10 [V ]C: I leak = 2.6 [nA], A f = 1.1 × 10 [V ]
30 35 40 45 50 55 60 65 70
Energy [keV] R a t e [ H z k e V ] Pre-Silane DataPre-Silane Best-FitGaussian FWHM: 3.0 ± 0.1 keVCompton EdgePost-Silane DataPost-Silane Best-FitGaussian FWHM: 4.4 ± 0.1 keVCompton EdgePost-Bake DataPost-Bake Best-FitGaussian FWHM: 3.1 ± 0.1 keVCompton Edge
Figure 4:
Left:
Energy resolution (FWHM) as a function of peaking time for three strips of the 8-strip detector Sh0075,measured at ∼ –37 °C. Baseline measurements were taken after cleaning the detector strips as outlined in Sec. 4.2. Post-silane measurements were taken after applying the APS adhesion promoter to the detector’s grooves and top hat andcuring on a hot plate at 85 °C for 30 minutes. After this APS cure cycle, the detector’s energy resolution degraded and1 / f noise component increased. After re-baking the detector at 110 °C, the energy resolution and 1 / f noise componentrecovered to pre-passivation levels. Right:
Corresponding spectra for strip A of Sh0075 at the optimal peaking time(4 µ s). Each spectra shows a photopeak and a low-energy tail. GEANT4 simulations have confirmed that the low-energytail is due to X-rays scattered from material in the testing chamber and are not due to charge trapping. The data is fit to aGaussian (dash-dotted) plus an error function (dotted) as discussed in [3]. Pre-silane and post-bake measurements wereperformed using a preamplifier with higher attenuation than the preamplifier used for post-silane measurements, so thepost-silane count rate was scaled to make the di ff erence in spectral shape more evident. Detectors passivated with the final PI passivation protocol demonstrated good leakage currentcharacteristics and indicated no degradation in X-ray energy resolution. The optimized passiva-tion procedure was applied to a batch of 12 eight-strip flight detectors. Given time-constraints,these detectors were not tested for energy resolution before passivation, and not all strips weremeasured after passivation. At least four strips per detector were randomly sampled for post-passivation energy resolution measurements. Each measured detector strip had (cid:46) / f noise (see Fig. 5). A small fraction of de-tector strips demonstrated fitted leakage current above the acceptance criteria, but were deemedsuccessful based on their energy resolution and 1 / f noise.12 s )24681020 F W H M E n e r g y R e s o l u t i o n ( k e V ) A: I leak = 1.4 [nA], A f = 0.6 × 10 [V ]C: I leak = 2.1 [nA], A f = 0.7 × 10 [V ]E: I leak = 2.0 [nA], A f = 0.6 × 10 [V ]G: I leak = 1.3 [nA], A f = 0.7 × 10 [V ]A: I leak = 1.4 [nA], A f = 0.6 × 10 [V ]C: I leak = 2.1 [nA], A f = 0.7 × 10 [V ]E: I leak = 2.0 [nA], A f = 0.6 × 10 [V ]G: I leak = 1.3 [nA], A f = 0.7 × 10 [V ] 2.0 2.5 3.0 3.5 4.0 4.5 5.0Measured Energy Resolution FWHM (keV)05101520 N u m b e r o f S t r i p s Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223 V )05101520 N u m b e r o f S t r i p s Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223 N u m b e r o f S t r i p s Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223Sh0222Sh0256Sh0258Sh0253Sh0254Sh0333 Sh0334Sh0336Sh0178Sh0210Sh0221Sh0223
Figure 5:
Top Left:
Energy Resolution (FWHM) as a function of peaking time for strips A, C, E, and G of Sh0221, aGAPS flight detector after passivation with VTEC PI-1388 polyimide using final protocol (see Sec. 4.2). Measurementswere performed at −
37 °C in the vacuum testing set-up outlined in Sec. 4.3.
Top Right:
Energy resolution (FWHM) atthe optimal peaking time after passivation for each measured strip of the 12 eight-strip flight detectors, using optimizedpassivation procedure.
Bottom Left:
Best-fit A f component (V ) of the 1 / f noise for the strips of these detectors. BottomRight:
Best-fit leakage current (nA) for the strips of these detectors.
5. E ff ectiveness in Protecting Detectors GAPS is scheduled for three LDB science flights. Therefore, the mission’s success relieson passivated detector performance remaining stable for several years. Even with a passivationcoating, precautions are taken to ensure that detectors have minimal exposure to humidity andorganics. For long-term storage, detectors are stored in a vacuum sealed antistatic bag withdesiccant in a freezer, which has been demonstrated to produce an environment with <
5% rel-ative humidity (RH) at −
20 °C. During routine calibration and processing, lab spaces will bemaintained at <
30% RH at room temperature. During integration, the modules will be purgedcontinuously with N2 to mitigate humidity and organic outgassing.Accelerated lifetime testing was conducted to assess the PI passivated detector’s robustness tocontamination from humidity (see Sec. 5.2) and organic materials (see Sec. 5.3) used in detectorassemblies. Furthermore, a long-term detector performance monitoring program is ongoing totrack detector performance over time (see Sec. 5.4).
Acceleration factors were computed by examining water vapor barrier penetration. The ac-celeration factor for humidity exposures is proportional to the number of water molecules hitting13he detector surface at a given temperature and humidity, and can be expressed as [31]: a = hP vw ( T ) h P vw ( T ) e − EpR ( T − T ) (3)where P vw ( T ) is the water saturation vapor pressure, E p is the activation energy for water di ff usioninto the polymer film, and R is the gas constant. T and h are the temperature (Kelvin) and RHunder test conditions, while T and h are the temperature (K) and RH during normal field use. E p (cid:39)
11 kJ / mol was used to compute the humidity acceleration factor, based on an empiricalmeasurement of water di ff usion into polymer films [32].The success criteria for accelerated humidity exposures was motivated by the foreseen pro-cessing and storage conditions. To be deemed successful, a detector was required to undergohumidity exposures equivalent to 14 days at 50% RH. This exposure is equivalent to ∼ ∼
30% RH at room temperature), and ∼
20 years in a desiccated bag in a freezer at ∼ –20 °C. Accelerated humidity exposures were performed by placing a detector in an airtight chamberwith a small water dish. The entire chamber was heated to 60 °C for 3 hours, and the humidityin the chamber increased to ∼
80% RH. Each exposure is equivalent to ∼ Three detectors fabricated in-house were subjected to humidity exposures. Two detectors(TD0090 and TD0093) were passivated following the passivation protocol (see Sec. 4.4.2) be-fore being exposed to humidity. One detector (TD0094) was left unpassivated as control. De-tector room temperature leakage current was measured before and after exposure to humidity.The passivated detectors demonstrated no degradation in leakage current, while the unpassivateddetector’s leakage current increased significantly (see Fig. 6). L e a k a g e C u rr e n t ( u A ) TD0094TD0093TD0090Pre-ExposurePost 1 ExposurePost 2 ExposuresPost 3 ExposuresPost 4 ExposuresPre-ExposurePost 1 ExposurePost 2 ExposuresPost 3 ExposuresPost 4 Exposures
Figure 6: Room temperature leakage current measurement of TD0090 (passivated), TD0093 (passivated), and TD0094(unpassivated) through successive humidity exposures. Solid lines indicate leakage current before exposure to humidity,whereas dashed and dotted lines indicate successive humidity exposures. In this test, the total accelerated exposure wascomparable to 7 days at 50% RH during field use, following Eq. (3). The passivated detector performance was stablethrough all exposures, while the unpassivated detector leakage current degraded significantly.
Once integrated, the detector modules consist of four eight-strip Si(Li) detectors, an ASICboard, and fluorosilicone and G10 detector retaining parts fastened into an aluminum frame. Themodules are sealed on both sides with an aluminized polypropylene window. In this study, detec-tors were exposed to outgassing from organics from materials that are constituent in a detectormodule.Assuming di ff usion as the dominant source of outgassing, the outgassed material at a time t is given by: ∆ m ( t , T ) = f m m (cid:114) tt r e EaR ( Tr − T ) (4)where m is the initial mass of the organic, E a is the activation energy of the organic material, R is the gas constant, and f m is the fractional mass loss at a reference time ( t r ) and temperature( T r ) [33]. Since we are mainly concerned with finding an acceleration factor, we note that thetotal mass loss (TML) ∝ t e − EaRT , where T and t are the accelerated exposure temperature andtime during lab testing, and T and t are the temperature and time of exposure during normalfield use. An acceleration factor for the lab exposures to organics can be found using the scalingrelation: a ≡ t t = e EaR ( T − T ) (5)In this study, detectors were exposed to a FR-4 circuit board, G10, fluorosilicone, and vacuumgrease that will be used to install and seal the detector modules. The activation energies forthese materials are not well measured, so to compute an acceleration factor, E a =
10 kJ / mol wasused, based on the typical activation energy for di ff usion driven outgassing [33]. To mitigateoutgassing, the modules are equipped for N2 purge, however N2 purge is not always feasible.The success criteria was motivated by the approximate amount of time that detector surfaces willbe exposed to organics when not being purged. Detectors were exposed to organics equivalent to > Accelerated organics exposures were performed by placing a detector in a chamber next to ahot plate, and heating the organic material on a hot plate to increase its TML. For each exposure,the organic material was heated to ∼
70 °C for 6 hours, an equivalent exposure of ∼
30 days,using Eq. (5). Before and after each exposure, the detector’s room temperature leakage currentwas measured. After achieving an equivalent exposure > .3.3. Testing Two passivated 8-strip detectors, Sh0079 and Sh0161, were selected for exposure to organ-ics. Both detectors demonstrated no change in room temperature leakage current characteristicsthrough exposures to organics equivalent to > / f noise was observed (see Fig. 7). s )24681020 F W H M E n e r g y R e s o l u t i o n ( k e V ) leak = 1.2 [nA], A f = 0.6 × 10 [V ]C: I leak = 1.1 [nA], A f = 0.6 × 10 [V ]E: I leak = 1.3 [nA], A f = 0.6 × 10 [V ]G: I leak = 1.2 [nA], A f = 0.6 × 10 [V ]01-22-2019B: I leak = 1.2 [nA], A f = 0.6 × 10 [V ]C: I leak = 1.1 [nA], A f = 0.6 × 10 [V ]E: I leak = 1.3 [nA], A f = 0.6 × 10 [V ]G: I leak = 1.2 [nA], A f = 0.6 × 10 [V ] 1 10Peaking Time ( s )24681020 F W H M E n e r g y R e s o l u t i o n ( k e V ) B: I leak = 1.3 [nA], A f = 0.7 × 10 [V ]C: I leak = 1.3 [nA], A f = 1.0 × 10 [V ]E: I leak = 1.4 [nA], A f = 1.2 × 10 [V ]G: I leak = 1.4 [nA], A f = 1.6 × 10 [V ] B: I leak = 1.3 [nA], A f = 0.7 × 10 [V ]C: I leak = 1.3 [nA], A f = 1.0 × 10 [V ]E: I leak = 1.4 [nA], A f = 1.2 × 10 [V ]G: I leak = 1.4 [nA], A f = 1.6 × 10 [V ] Figure 7:
Left:
Energy resolution (FWHM) as a function of peaking time for strips B, C, E, and G of Sh0079, an eight-strip flight detector. Measurements were performed at ∼ –35 °C immediately after passivation. Right:
Energy resolution(FWHM) as a function of peaking time for same strips of Sh0079, after organics exposures equivalent to 6 months offield use. No degradation in noise performance was observed, and the fit parameters of the noise model were consistent.
A long-term monitoring program is ongoing to ensure detector stability. Over the courseof a year, the energy resolution of passivated GAPS flight detectors was measured in the SUNchamber. Between measurements, detectors were stored in a dry box or in a vacuum-sealed bagwith desiccant in a freezer at ∼ –20 °C. No detector degradation has been noted over the courseof twelve months (see Fig. 8). S h 0 2 0 0 _ A S h 0 2 0 0 _ B S h 0 2 0 0 _ C S h 0 2 0 0 _ D S h 0 2 0 0 _ E S h 0 2 0 0 _ F S h 0 2 0 0 _ G S h 0 2 0 0 _ H S h 0 2 2 0 _ A S h 0 2 2 0 _ B S h 0 2 2 0 _ C S h 0 2 2 0 _ D S h 0 2 2 0 _ E S h 0 2 2 0 _ F S h 0 2 2 0 _ G S h 0 2 2 0 _ H S h 0 2 5 4 _ A S h 0 2 5 4 _ B S h 0 2 5 4 _ C S h 0 2 5 4 _ D S h 0 2 5 4 _ E S h 0 2 5 4 _ F S h 0 2 5 4 _ G S h 0 2 5 4 _ H S h 0 2 5 3 _ A S H 0 2 5 3 _ B S h 0 2 5 3 _ C S h 0 2 5 3 _ D S h 0 2 5 3 _ E S h 0 2 5 3 _ F S h 0 2 5 3 _ G S h 0 2 5 3 _ H
X-ray FWHM [keV]
T e s t i n g D a t e
S h 0 2 1 9 _ A S h 0 2 1 9 _ B S h 0 2 1 9 _ D S h 0 2 1 9 _ E S h 0 2 1 9 _ F S h 0 2 1 9 _ G S h 0 2 1 9 _ H
Figure 8: Energy resolution (FWHM) at optimal peaking time for strips of five GAPS flight detectors, measured ondi ff erent dates as part of long-term detector stability monitoring program. Passivated Si(Li) detectors demonstrate stableperformance over year-long timescales.
16n conjunction with accelerated lifetime testing, the preliminary results of long-term monitor-ing are promising. They indicate that storing the passivated GAPS flight detectors in a vacuumsealed bag with desiccant in a freezer at ∼ –20 °C is su ffi cient to maintain stable detector per-formance for the lifetime of the GAPS experiment. Moreover, it is not necessary to store theseSi(Li) detectors under bias to mitigate Li re-distribution, which simplifies the long-term storagescheme needed to maintain detector performance. More details will be presented elsewhere.
6. Conclusion
We have demonstrated that Si(Li) surfaces can be successfully passivated for detectors oper-ated at temperatures as high as −
35 °C. Polyimide and parylene-C were explored as passivationcandidates. While parylene-C coatings demonstrated acceptable thermal and adhesion proper-ties, excessive leakage current and poor deposition reproducibility made it unfit for large-scalepassivation. After optimizing the application protocol, polyimide was selected as the passivationcoating for GAPS Si(Li) detectors.The PI passivated detector performance has been demonstrated to meet the requirement of ≤ −
35 °C to −
45 °C). The passivation protocol is easy to apply by technicians, andthe resulting passivation coating provides a robust barrier to humidity, organic, and particulatecontamination. Passivated detector performance is stable over year-long time scales. This proto-col is being employed to passivate 1440 flight detectors for the GAPS experiment, the first 491of which have been passivated using the selected protocol. These passivated Si(Li) detectors willform the first large-area silicon detector array with X-ray spectral capabilities operated abovecryogenic temperatures at high altitudes.
7. Acknowledgements
We thank SUMCO Corp. and Shimadzu Corp. for their cooperation in detector development.We also thank the GAPS collaboration for their consultation and support. This work is partiallyfunded by the NASA APRA program (Grant Nos. NNX17AB44G and NNX17AB46G). K.Perez and M. Xiao receive support from the Heising-Simons Foundation. F. Rogers is supportedby the NSF Graduate Research Fellowship (Grant No. 1122374). This work is partly supportedin Japan by JAXA / ISAS Small Science Program FY2017. H. Fuke receives support from JSPSKAKENHI grants JP26707015, JP17H01136, and JP19H05198 and from the Mitsubishi Founda-tion. M. Kozai receives support from JSPS KAKENHI grants JP17K14313 and JP20K14505. Y.Shimizu receives support from JSPS KAKENHI grant JP20K04002 and Sumitomo Foundationgrant.
ReferencesReferences [1] K. Perez, et al., Fabrication of low-cost, large-area prototype Si(Li) detectors for the GAPS experiment, NuclearInstruments and Methods in Physics Research A 905 (2018) 12–21. arXiv:1807.07912 , doi:10.1016/j.nima.2018.07.024 .[2] M. Kozai, et al., Developing a mass-production model of large-area Si(Li) detectors with high operating tem-peratures, Nuclear Instruments and Methods in Physics Research A 947 (2019) 2695. arXiv:1906.05577 , doi:10.1016/j.nima.2019.162695 .
3] F. Rogers, M. Xiao, K. Perez, S. Boggs, T. Erjavec, L. Fabris, H. Fuke, C. Hailey, M. Kozai, A. Lowell, N. Madden,M. Manghisoni, S. McBride, V. Re, E. Riceputi, N. Sa ff old, Y. Shimizu, Large-area Si(Li) detectors for X-rayspectrometry and particle tracking in the GAPS experiment, Journal of Instrumentation 14 (10) (2019) P10009–P10009. doi:10.1088/1748-0221/14/10/p10009 .[4] F. Donato, N. Fornengo, P. Salati, Antideuterons as a signature of supersymmetric dark matter, Phys. Rev. D 62(2000) 043003. doi:10.1103/PhysRevD.62.043003 .[5] F. Donato, Antiprotons in cosmic rays from neutralino annihilation, Nuclear Physics B - Proceedings Supplements138 (2005) 303 – 306. doi:https://doi.org/10.1016/j.nuclphysbps.2004.11.068 .[6] R. Duperray, B. Baret, D. Maurin, G. Boudoul, A. Barrau, L. Derome, K. Protasov, M. Bu´enerd, Flux of lightantimatter nuclei near earth, induced by cosmic rays in the galaxy and in the atmosphere, Phys. Rev. D 71 (2005)083013. doi:10.1103/PhysRevD.71.083013 .[7] T. Aramaki, S. Boggs, S. Bufalino, L. Dal, P. von Doetinchem, F. Donato, N. Fornengo, H. Fuke, M. Grefe,C. Hailey, B. Hamilton, A. Ibarra, J. Mitchell, I. Mognet, R. Ong, R. Pereira, K. Perez, A. Putze, A. Raklev,P. Salati, M. Sasaki, G. Tarle, A. Urbano, A. Vittino, S. Wild, W. Xue, K. Yoshimura, Review of the theoreticaland experimental status of dark matter identification with cosmic-ray antideuterons, Physics Reports 618 (2016) 1– 37. doi:https://doi.org/10.1016/j.physrep.2016.01.002 .[8] K. Mori, C. J. Hailey, E. A. Baltz, W. W. Craig, M. Kamionkowski, W. T. Serber, P. Ullio, A Novel AntimatterDetector Based on X-Ray Deexcitation of Exotic Atoms, The Astrophysical Journal 566 (1) (2002) 604–616. doi:10.1086/338057 .[9] C. J. Hailey, An indirect search for dark matter using antideuterons: the GAPS experiment, New Journal of Physics11 (10) (2009) 105022. doi:10.1088/1367-2630/11/10/105022 .[10] P. von Doetinchem, T. Aramaki, N. Bando, S. E. Boggs, H. Fuke, F. H. Gahbauer, C. J. Hailey, J. E. Koglin,S. A. I. Mognet, N. Madden, S. Okazaki, R. A. Ong, K. M. Perez, T. Yoshida, J. Zweerink, The flight of theGAPS prototype experiment, Astroparticle Physics 54 (2014) 93–109. arXiv:1307.3538 , doi:10.1016/j.astropartphys.2013.11.009 .[11] T. Aramaki, S. Boggs, P. von Doetinchem, H. Fuke, C. Hailey, S. Mognet, R. Ong, K. Perez, J. Zweerink, Potentialfor precision measurement of low-energy antiprotons with gaps for dark matter and primordial black hole physics,Astroparticle Physics 59 (2014) 12 – 17. doi:https://doi.org/10.1016/j.astropartphys.2014.03.011 .[12] S. Okazaki, H. Fuke, H. Ogawa, Y. Miyazaki, K. Takahashi, N. Yamada, Meter-scale multi-loop capillary heat pipe,Applied Thermal Engineering 141 (2018) 20 – 28. doi:https://doi.org/10.1016/j.applthermaleng.2018.05.116 .[13] T. Aramaki, S. K. Chan, W. W. Craig, L. Fabris, F. Gahbauer, C. J. Hailey, J. E. Koglin, N. Madden,K. Mori, H. T. Yu, K. P. Ziock, A measurement of atomic X-ray yields in exotic atoms and implicationsfor an antideuteron-based dark matter search, Astroparticle Physics 49 (2013) 52–62. arXiv:1303.3871 , doi:10.1016/j.astropartphys.2013.08.003 .[14] F. J. Hartmann, Exotic Atom Cascade Processes in Atoms with Z >
2, Springer US, Boston, MA, 1990, pp. 127–139. doi:10.1007/978-1-4899-3701-8_12 .[15] D. Gotta, Precision spectroscopy of light exotic atoms, Prog. Part. Nucl. Phys. 52 (2004) 133–195. doi:10.1016/j.ppnp.2003.09.003 .[16] F. Goulding, Semiconductor detectors for nuclear spectrometry, I, Nuclear Instruments and Methods 43 (1966)1–54.[17] X. Meng, Y.-C. Byun, H. S. Kim, J. S. Lee, A. T. Lucero, L. Cheng, J. Kim, Atomic Layer Deposition of SiliconNitride Thin Films: A Review of Recent Progress, Challenges, and Outlooks, Materials 9 (12). doi:10.3390/ma9121007 .[18] O. H. Kim, Atomic layer deposition of gallium nitride and tantalum nitride, Ph.D. thesis, Doctoral DissertationsAvailable from Proquest (2009).[19] A. E. Kaloyeros, F. A. Jov´e, J. Go ff , B. Arkles, Review—Silicon Nitride and Silicon Nitride-Rich Thin FilmTechnologies: Trends in Deposition Techniques and Related Applications, ECS Journal of Solid State Science andTechnology 6 (10) (2017) P691–P714. doi:10.1149/2.0011710jss .[20] W. L. Hansen, E. E. Haller, G. S. Hubbard, Protective surface coatings on semiconductor nuclear radiation detec-tors, IEEE Transactions on Nuclear Science 27 (1) (1980) 247–251. doi:10.1109/TNS.1980.4330834 .[21] J. T. Walton, R. H. Pehl, Y. K. Wong, C. P. Cork, Si(li) x-ray detectors with amorphous silicon passivation, IEEETransactions on Nuclear Science 31 (1) (1984) 331–335. doi:10.1109/TNS.1984.4333270 .[22] M. Jantunen, S. A. Audet, Surface passivated Si(Li) detectors for an X-ray detector array, Nuclear Instruments andMethods in Physics Research A 353 (1994) 89–92. doi:10.1016/0168-9002(94)91609-8 .[23] C. S. Rossington, P. M. Fine, N. W. Madden, Large area, low capacitance si(li) detectors for high rate x-rayapplications, IEEE Transactions on Nuclear Science 40 (4) (1993) 354–359. doi:10.1109/23.256579 .[24] K. C. Sheth, Stress, mechanical and thermal characterization of anisotropic polyimide thin films and coatings,Ph.D. thesis, UMass Amherst, Doctoral Dissertations Available from Proquest, AAI9709653 (1996).
25] T.-C. J. Hsu, Z.-L. Liu, Solvent e ff ect on the curing of polyimide resins, Journal of Applied Polymer Science46 (10) (1992) 1821–1833. doi:10.1002/app.1992.070461014 .[26] B. J. Kim, E. Meng, Micromachining of parylene c for biomems, Polymers for Advanced Technologies 27 (5)(2016) 564–576. doi:10.1002/pat.3729 .[27] A. International, ASTM D3359-17 Standard Test Methods for Rating Adhesion by Tape Test, Standard, ASTMInternational, West Conshohocken, PA (2017). doi:https://doi.org/10.1520/D3359-17 .[28] V. Scotti, A. Boiano, L. Fabris, M. Manghisoni, G. Osteria, E. Riceputi, F. Perfetto, V. Re, G. Zampa, Front-endelectronics for the GAPS tracker, PoS ICRC2019 (2020) 136. arXiv:1909.01682 .[29] T. Aramaki, S. K. Chan, C. J. Hailey, P. A. Kaplan, T. Krings, N. Madden, D. Proti´c, C. Ross, Development oflarge format Si(Li) detectors for the GAPS dark matter experiment, Nuclear Instruments and Methods in PhysicsResearch A 682 (2012) 90–96. doi:10.1016/j.nima.2012.04.054 .[30] H. Spieler, Semiconductor Detector Systems, Series on Semiconductor Science and Technology, OUP Oxford,2005.[31] W. Nelson, Accelerated Testing: Statistical Models, Test Plans, and Data Analysis, Wiley series in probability andmathematical statistics: Applied probability and statistics, John Wiley & Sons, 1990.[32] T. Ogawa, T. Nagata, Y. Hamada, Determination of di ff usion coe ffi cient of water in polymer films by tga, Journalof Applied Polymer Science 50 (6) (1993) 981–987. doi:10.1002/app.1993.070500607 .[33] A. C. Tribble, B. Boyadjian, J. Davis, J. Ha ff ner, E. McCullough, Contamination control engineering design guide-lines for the aerospace community, NASA Contractor Report 4740, Rockwell International Corporation (May1996).ner, E. McCullough, Contamination control engineering design guide-lines for the aerospace community, NASA Contractor Report 4740, Rockwell International Corporation (May1996).