X-ray Detectors: What are the Needs?

Size: px

Start display at page:

Download "X-ray Detectors: What are the Needs?"

Junior Gilbert
6 years ago
Views:

1 X-ray Detectors: What are the Needs? Sol M. Gruner Physics Dept. & Cornell High Energy Synchrotron Source (CHESS) Ithaca, NY

Intelligent Direct- Detection Phosphor-Coupled CCDs Image Plates X-ray Film

2 simplified view of the Evolution of Imaging Synchrotron X-ray Detectors (Similar considerations apply to non-imaging detectors) Direct-Detection Intelligent Direct- Detection Phosphor-Coupled CCDs Image Plates X-ray Film What are the needs? What is feasible in the next decade, given sufficient R&D? 2

3 What can t we do? Where are the needs? High x-ray energy (>30 kev) imaging w/ high efficiency and micron resolution. High count-rate, wide solid angle spectroscopy with high energy resolution, especially for very hard x-rays. X-ray imaging with high sensitivity, wide dynamic range intelligent pixel detectors where pixels are microns across. X-ray imaging with intelligent pixel detectors where the intelligence is in firmware on-line analysis. Large format x-ray pixel detectors of many types, but with edgeless sensors. High dynamic range (1 to 10 7 x-rays), large format (> 2k x 2k) integrating detectors framing at a few hundred Hz for LCLS and other low-duty cycle XFELs a few MHz for European XFEL and NGLS. Solid-state detectors capable of counting at >> 1GHz/pix. Imaging pixel array detectors for very hard x-rays. Yet there are approaches for each of these that promise feasibility on the 10-year time scale, given sufficient R&D. 3

4 What follows are some thoughts on just a few aspects of what might be feasible on a decade time scale, given sufficient R&D and a consistent decade-long strategy. My purpose in presenting these is to provoke discussion, not to be comprehensive. We ll hear from the experts during this workshop. The workshop goal is a comprehensive roadmap to guide detector development. 4

Basic Physics: Direct Detection in Silicon Si is a superb x-ray to electrical signal converter. @ 10 kev, in < 1 ns, radius of e-h cloud ~ 1 micron. Number e-h pairs: E x-ray / 3.

5 Basic Physics: Direct Detection in Silicon Si is a superb x-ray to electrical signal 10 kev, in < 1 ns, radius of e-h cloud ~ 1 micron. Number e-h pairs: E x-ray / 3.65 ev 10 kev yields 10000/3.65 = 2740 ± 20. ΔE = ±3.65 x 20 ev = 146 ev width Charge collection time sets limit on photon counting makes photon counters impractical for most XFEL experiments um, ~20 ns Width of charge cloud at collection electrode sets limit on spatial resolution and charge sharing between pixels. 5

6 simplified view of the Evolution of Imaging Synchrotron X-ray Detectors (Similar considerations apply to non-imaging detectors) Direct-Detection Phosphor-Coupled CCDs Image Plates X-ray Film Intelligent Direct- Detection What is feasible in the next decade, given sufficient R&D? Consider: Pixel size & complexity Spatial resolution Time resolution Analog dynamic range 6

7 simplified view of the Evolution of Imaging Synchrotron X-ray Detectors (Similar considerations apply to non-imaging detectors) Direct-Detection Phosphor-Coupled CCDs Image Plates X-ray Film Intelligent Direct- Detection What is feasible in the next decade, given sufficient R&D? Consider: Pixel size & complexity Spatial resolution Time resolution Analog dynamic range 7

3D-ICs based on Silicon-on-Insulator (SOI) Wafers Small prototypes of VIPs are extendable to sizes of 1024 1024 pixels, bearing the actual needs of the application.

8 3D-ICs based on Silicon-on-Insulator (SOI) Wafers Small prototypes of VIPs are extendable to sizes of pixels, bearing the actual needs of the application. The top tier contains a gated charge integrator, a single ended AC-coupled offset corrected discriminator with capacitively injected threshold, an analog memory for reference sample, an analog memory for post discriminator sample, a pulse generator for time stamping lock and hit information lock, a receiving part of test-charge injection capacitance and a bonding pad to the detector. The intermediate tier features an analog memory cell for time stamping (distributed voltage ramp), a 7-bit SRAM-like digital time stamping memory with output enable control to read on the same lines on which time ticks in Gray code are distributed. The bottom tier hosts the sparsification system: token propagation logic, wiredor line access logic for X-line/Y-line of a hit pixel address generator, test-charge injection logic and a peripheral serialization and output part. From: Deptuch et al, FERMILAB-CONF PPD This is ~200 transistor level of complexity in 20 um pixel, typical of today s intelligent PADs. Deptuch et al, FERMILAB-PUB ppd On a decade time scale pixels with reasonable levels of complexity and 10 20um pixel sizes are feasible. 8

9 simplified view of the Evolution of Imaging Synchrotron X-ray Detectors (Similar considerations apply to non-imaging detectors) Direct-Detection Phosphor-Coupled CCDs Image Plates X-ray Film Intelligent Direct- Detection What is feasible in the next decade, given sufficient R&D? Consider: Pixel size & complexity Spatial resolution Time resolution Analog dynamic range 9

10 High Spatial Resolution Using Doped Garnets Single crystal YAG:Ce and GGG:Eu screens with doped layers microns thick are commercially available (e.g., ESRF; laser vendors). Present spatial resolutions of ~0.7um are available with reasonable efficiencies at up to ~20 kev. The wavelength of light and photoelectron emission will likely limit this to small digit improvements, at best. High x-ray energy stopping power is limited by doped layer. A new idea is needed. From: Koch at al., J. Opt. Soc. Am. A 15 (1998)

11 simplified view of the Evolution of Imaging Synchrotron X-ray Detectors (Similar considerations apply to non-imaging detectors) Direct-Detection Phosphor-Coupled CCDs Image Plates X-ray Film Intelligent Direct- Detection What is feasible in the next decade, given sufficient R&D? Consider: Pixel size & complexity Spatial resolution Time resolution Analog dynamic range 11

Time Resolution ~500 um ~20 ns few ns using Si Parker et al., IEEE Trans. Nucl. Sci. 58 (2011) 404. Z. Li, Nucl. Instr. and Meth. A (2011), doi:10.1016/j.nima.2011.05.

12 Time Resolution ~500 um ~20 ns few ns using Si Parker et al., IEEE Trans. Nucl. Sci. 58 (2011) 404. Z. Li, Nucl. Instr. and Meth. A (2011), doi: /j.nima D silicon sensors capable of a few ns response are in advanced R&D. On a decade time scale, use of exotic semiconductors and few hundred ps response may be feasible. 12

13 Time Resolution: Use Nanopillars From: Chuang et al., NANO Letters 11 (2011) 385 This is an LED, but they also report on Avalanche Photodiodes (APD) A dense forest of nanopillar APDs are in principle capable of few ps response. With sufficient R&D fill factors of ~25% may become feasible. Readout electronics then become limiting. 13

14 Frame Time Considerations: Front-end amplifier settling time. Time to transfer data to off-asic digital memory. Parallelize! KECK PAD Noise Parameter Minimum exposure time Target Value < 0.5 x-ray/pixel/accumulation <150 ns for 12-bit imaging Capacitor well depth Nonlinearity (% full well) < 0.2% x-rays Diode conversion layer 500 μm thick Si Number of capacitor wells/pix 8 Full chip frame time 1 msec/frame, e.g., 8 msec for 8 capacitors Radiation lifetime Pixel size Detector chip format Dark current > 50 Mrad at detector 8 kev 150 μm on a side, or 128 x 128 pixels per IC 2 x 4 chips = 256 x 512 pixels 2 x-rays/pix/sec Koerner & Gruner, J. Synchro. Rad. 18 (2011) 157. < 150 ns for 12 bit settling shown. Equivalent to ~ kev x-rays. Faster for fewer bits. A few bits in 10 s of ns should be feasible. 14

15 On the 10 year time scale, detectors of large format (>10 6 pixels), wide dynamic range (> kev x-rays/pix/frame), frame rates of ~100ns, and frame depths of hundreds of frames are likely feasible. If the dynamic range is reduced to ~10 s of x-rays/pix/frame, frame rates can likely fall to a few 10 s of ns. 15

16 Warnings! 1. Difference between feasibility and reality: $ and a consistent, long-term strategy. The U.S. is falling behind, and will continue to do so unless a consistent long-term strategy is adopted. 2. No one detector will have all the characteristics discussed. Many approaches and detectors will be needed. 16

17 END 17

Implementation of A Nanosecond Time-resolved APD Detector System for NRS Experiment in HEPS-TF

Implementation of A Nanosecond Time-resolved APD Detector System for NRS Experiment in HEPS-TF LI Zhen-jie a ; MA Yi-chao c ; LI Qiu-ju a ; LIU Peng a ; CHANG Jin-fan b ; ZHOU Yang-fan a * a Beijing Synchrotron