INVITED PAPER
Special Section on Superconducting Signal Processing TechnologiesProposal of a Compact Neutron Di
ffraction System with a
Single-Flux-Quantum Signal Processor
Akira FUJIMAKI†a), Member, Isao NAKANISHI†, Shigeyuki MIYAJIMA†, Student Members, Kohei ARAI††, Yukio AKITA††, and Takekazu ISHIDA††, Nonmembers
SUMMARY We propose a neutron diffractometer system based on MgB2thin film detectors and an SFQ signal processor. Small dimensions
of MgB2 thin film detectors and high processing capability of the single
flux quantum (SFQ) circuits enable us to handle several thousand or more detectors in a cryocooler, leading to a very compact system. In addition, the system can provide many diffraction patterns for different kinetic energies simultaneously. Kinetic energy is determined for individual neutrons by means of the time-of-flight method by using SFQ time-to-digital convert-ers (TDCs). Digital outputs of the TDCs are multiplexed in time domain and sent to room-temperature electronics with reduced number of cables. A dual-input SFQ signal processor including TDCs and a multiplexer has been successfully demonstrated with a time resolution of 20 ns and power consumption of 400μW. These values show high feasibility of the neutron diffraction system proposed here.
key words: neutron, diffraction, single flux quantum, time-to-digital con-verter, time-of-flight
1. Introduction
A neutron diffraction system is a strong tool for determining the atomic and magnetic structure of a material. In partic-ular, the system has high sensitivity to light atoms such as hydrogen. The system is expected to be applied to drug-discovery research, because affinity for biological bodies is affected by the positions of hydrogen atoms inside a drug. The mechanism of superconductivity in oxide superconduc-tors has not been revealed, while it is strongly related to the magnetic structures of the materials. Analyses based on neutron diffraction would help us to reveal the mechanism.
In recent years, several neutron sources with high flux density have been developed and have been easily accessi-ble to researchers [1], [2]. At present,3He gas tube detectors
are used for detecting neutrons in any application because of high efficiency and high neutron-to-gamma flux ratio. How-ever, gas tube detectors have typical dimensions of a few centimeters, so that detector array systems like a diffrac-tometer become bulky. Response times and relaxation times of the gas tube detectors are of the order of 10 ms. These two kinds of times lead to very low counting rate, and users are obliged to spend relatively long time for obtaining ex-perimental data.
Manuscript received August 23, 2010. Manuscript revised December 24, 2010.
†The authors are with the Department of Quantum
Engineer-ing, Nagoya University, Nagoya-shi, 464-8603 Japan.
††The authors are with the Department of Physics and
Electron-ics, Osaka Prefecture University, Sakai-shi, 599-8531 Japan. a) E-mail: [email protected]
DOI: 10.1587/transele.E94.C.254
In 2007, superconducting neutron detectors made of
10B-enriched MgB
2 thin films were demonstrated [3]. The
detectors would use a non-equilibrium phenomenon caused by the nuclear reactions with energy of 2.3 MeV between
10B in MgB
2and neutrons. A part of the energy is released
to the superconducting MgB2film. The detectors have a
me-ander line of MgB2and a hot spot is created inside the line
when neutrons come to the detector. The temperature of the detector stage on which MgB2detectors are mounted needs
to be set close to the transition temperature of an MgB2film
for detecting neutrons effectively [3]. Typically, meander lines are confined to an area of 200μm × 200 μm. The re-sponse time was measured to be smaller than 2 ns. These features make it possible to construct compact and fast de-tector array systems.
A neutron diffraction system is one of the most promis-ing applications based on detector array systems. The sys-tems will be made up of more than 1000 detectors. How-ever, this leads to increased number of cables with which the detectors are connected to room-temperature electron-ics. Large heat inflow through these cables is unavoidable, so that the temperature of the detector stage becomes un-stable in a conventional way. Thus, a low temperature mul-tiplexing technique for the detector outputs is essential for reducing the number of the cables and heat inflow.
Single-flux-quantum (SFQ) circuits operate at a high frequency with very low power consumption [4], [5]. This feature is most appropriate for making a multiplexer work-ing at low temperature. In fact, an SFQ multiplexer for a detector application has been studied so far [6].
In this paper, we propose a compact neutron diffraction system composed of MgB2 thin film detectors and an SFQ
signal processor. We describe configuration of the system and the operating principle of the processor. We also report preliminary results for an SFQ signal processor.
2. Configuration of the Neutron Diffraction System
Figure 1 shows a block diagram of our neutron diffraction system including a neutron source. A start signal is pro-vided repeatedly to the neutron source and all the time-to-digital converters (TDCs) from a room-temperature gener-ator. Triggered by each start signal, the source generates a flux of cold neutrons with various energies. These cold neutrons move to a detector array through a sample to be measured. Diffraction occurs in this process and a diffrac-Copyright c 2011 The Institute of Electronics, Information and Communication Engineers
Fig. 1 Configuration of a neutron diffraction system based on MgB2
thin film detectors and an SFQ signal processor. Kinetic energies are de-termined for individual neutrons at each detector chain by means of the TOF method. Introduction of a time domain multiplexer enables remark-able reduction in the number of cremark-ables between the SFQ stage and room-temperature electronics.
tion pattern is obtained by counting the number of incident neutrons having a certain kinetic energy in each detector.
Kinetic energy of a neutron detected at each detector can be determined by means of the time-of-flight (TOF) technique in our system, while general diffraction systems use filters for choosing a preset kinetic energy or wave-length. Our system has capability to provide diffraction pat-terns for various kinetic energies simultaneously. This is one of the unique features in our system. Introduction of fast-response MgB2 detectors and high-speed SFQ signal
processing enable us to shorten the total time required for data acquisition and data processing.
As shown in Fig. 1, the system has multiple detector chains. Each chain is composed of a detector, a compara-tor, and a TDC. If a neutron comes to the deteccompara-tor, a current flowing on the detector is reduced because a part of the me-ander line becomes resistive. The comparator picks up a change of the current and generates an SFQ pulse serving as a stop signal. TOF is measured at the TDC by counting clocks generated at an embedded SFQ clock generator for the duration between a start signal and the corresponding stop signal. Note that all the detector chains work indepen-dently, though the start signal is distributed to all the TDCs simultaneously.
m-bits parallel outputs of the TDCs are multiplexed in
time domain and sent to room-temperature electronics in-cluding a demultiplexer. This procedure is repeated with the period of the start signals Tneutron. The room-temperature
electronics calculates kinetic energies for all the detected neutrons and shows diffraction patterns, in other words, the number of detected neutrons for each kinetic energy and for each detector.
In physical meaning, time resolution of the TDCs is limited by fluctuation of response time of MgB2thin film
de-tectors or timing jitter of SFQ circuits. Thus, the TDC itself is able to have the time resolution of much smaller than 1 ns. In a practical neutron diffraction system, the time resolution is determined by time spread of neutron fluxes that are
gen-erated by a start signal. Japan Proton Accelerator Research Complex (J-PARC) has a representative neutron source. In the source, the spread is around 1μs, while Tneutronis 40 ms.
Thus, dynamic range of 16 bits (m = 16) is enough for an actual system.
The number of detectors accommodated by the system
n is roughly expressed as n= BTneutron
m ,
where B is smallest band width in the system. Usually, B is determined by the throughput of the room-temperature elec-tronics in which real-time signal processing is executed. Our system prototype described later has B of 2.5 MHz, which corresponds to the throughput of a field programmable gate array (FPGA) used for the demultiplexer. If we assume
m= 16 and Tneutron= 40 ms, n becomes approximately 6000
in our system prototype. This means that heat inflow is sup-pressed remarkably through the cables because we use only 16 cables for 6000 detectors.
3. System Prototype
We have started to develop the system prototype. The aim of the development is to show potential of a system having a configuration indicated in Fig. 1. The development includes the installation of an SFQ signal processor to the SFQ stage and the installation of MgB2thin film detectors to the
detec-tor stage. We use a Gifford-Mac Mahon (GM) cryocooler for refrigerating these two stages. The temperature of the SFQ stage is kept as low as possible, while that of the detec-tor stage is controlled precisely to be a temperature which is just below the transition temperature of the MgB2thin film
detectors. The typical temperature of the SFQ stage and the detector stage are 4 K and 27 K, respectively.
Figure 2 shows a photograph of surrounding structures of the cold head in the GM cryocooler, though the cold head located at the center of the photograph is difficult to see. The cold head is thermally connected to the SFQ stage. The detector stage is fixed to the SFQ stage with stainless-steel pillars that have low thermal conduction.
A detector probe is installed below the detector stage as shown in Fig. 2, and houses a sapphire substrate on which MgB2 thin film detectors are formed. Several heaters and
temperature sensors are installed around the detector probe to control the operating temperature of the MgB2 thin film
detectors. A scanner system based on a pulsed laser [7] is able to be installed in front of the detectors. The scan-ner is used for examining each detector before irradiation of neutrons because the response of the detectors to the pulsed laser is proven to be similar to the response to a neutron.
The SFQ probe is set on the SFQ stage. The direction of the installation is the opposite side to the detector probe in order to avoid exposure of radiation. The SFQ probe is covered by double magnetic shield cans made of a kind of Permalloy. This enables us to reduce residual magnetic field at the probe. In addition, ferromagnetic materials are
elimi-nated from the inside of the shield cans and their neighbor-hoods.
As described above, impulse-like outputs of the detec-tors operating around 27 K have to be sent to the compara-tors of the SFQ circuits operating at 4 K with little distortion. In other words, we need both broad band widths and low thermal conduction for cables with which the two probes are connected. Considering lengths or attenuations of the ca-bles, we employ CuNi coaxial semi-rigid cables. A similar situation occurs between the probes and room-temperature electronics. We need longer cables in this situation. Thus we use brass coaxial semi-rigid cables having higher elec-tric conductivity than CuNi.
These coaxial cables are connected properly through the RF connectors mounted on the RF connector plate cooled down to the same temperature as that of the SFQ stage. DC bias lines are also provided to the two probes through connectors mounted on the DC connector plate. This enables us to do maintenance easily and to change con-nections if necessary.
Several other techniques are introduced to our system prototype for reducing noise including crosstalk, for sup-pressing the effect of vibration of the GM cryocooler, for widening band widths in signal lines, and for increasing testability. For example, several kinds of low-pass filters are used for signal lines or bias lines in order to suppress mixture of noises. In addition, all the setting values such as bias currents are remotely-operable, which is required in an actual experiment of neutron irradiation.
System noise is estimated by comparing this cryo-cooler-based system and a conventional system. The con-ventional system is composed of a liquid helium bath,
Fig. 2 Microphotograph of surrounding structures of the cold head in our GM cryocooler. The SFQ stage is kept around 4 K, while the temperature of the detector stage is controlled to be just below transition temperatures of the detectors. These structures are designed from many aspects such as noise, broad-band signal transmission, and thermal conduction.
low noise amplifiers, and an oscilloscope. We measured power spectrum of the same SFQ analog-to-digital converter (ADC). The ADC used here has very high current sensitiv-ity [8], so that the ADC works as a good monitor of system noise. Noise power in the system prototype with a frequency band of 1 kHz–5 MHz is increased by 3 dB compared to that in the conventional system placed inside of an electromag-netic shielded room. We think from this estimation that our system prototype is applicable to an actual experiment of neutron irradiation.
4. Demonstration of SFQ Signal Processor
We also started to develop SFQ signal processors. Figure 3 shows the block diagram of a dual-input SFQ signal pro-cessor developed in this study. The dual-input SFQ signal processor is composed of three identical quasi-one-junction SQUIDs (QOSs), two TDCs, and a multiplexer. The QOSs are used as comparators with a single threshold [6], though they have periodic response to magnitude of an input signal. The QOSs are designed so as to readout detector outputs through relatively long CuNi cables. Current sensitivity of the QOSs is measured to be a fewμA, which is sufficient for picking up the detector outputs.
In this study, all the inputs for generating start signals and stop signals are provided from external signal sources. For simplicity, we supply a continuous sinusoidal wave to QOS 0 and also supply the same wave having a phase shift of π to QOS A and QOS B as the inputs. An embedded SFQ ring oscillator generates clocks with a frequency of 12.8 GHz. The clocks are provided to the QOSs. QOS 0 generates an SFQ pulse when the magnitude of the input ex-ceeds the threshold of the QOS 0. This pulse serves as a start signal common to TDC A and TDC B. QOS A and QOS B generate a stop signal for TDC A and TDC B, respectively.
The main component of the TDCs is a 4-bit counter. The clock generator CLK A and CLK B generate 50-MHz clocks which are made by dividing master clocks with a fre-quency fmof 12.8-GHz. TDC A counts 50-MHz clocks of
CLK A between a start signal from QOS 0 and a stop signal
Fig. 3 Block diagram of the dual-input SFQ signal processor. QOS 0 generates a common SFQ start signal, while QOS A and QOS B generate individual stop signals for TDC A and TDC B, respectively. By using CLK A and CLK B which have the same frequency with different phases, time-domain multiplexing is achieved in a conventional confluence buffer.
Fig. 4 Microphotograph of a dual-input SFQ signal processor. The pro-cessor was fabricated with the SRL standard process based on 2.5-kA/cm2 Nb/AlOx/Nb Josephson integrated circuit technology.
from QOS A. TDC B also counts clocks. Note that CLK B generates a clock with a time delay of 10 ns after CLK A generates a clock. Digital outputs of TDC B are sent to the multiplexer 10-ns later than those of TDC A. As a result, a conventional confluence buffer can behave as a multiplexer, which executes multiplexing in time domain.
Dynamic range of 4-bits for the TDCs is insufficient for an actual diffraction system. However, it is very easy to expand the dynamic range because the counters are made up of conventional SFQ toggle-flip-flops (TFFs). Time res-olution is set to be 20 ns in the TDCs. This time resres-olution is the best value under the restriction of our signal source used here as described later. Note that the time resolution is easy to be adjusted to the specification of an actual neutron source, because the clock generators CLK A and CLK B are binary frequency dividers composed of the same TFFs.
Figure 4 shows a microphotograph of a dual-input SFQ signal processor. The processors were fabricated with the SRL standard process [9] based on 2.5-kA/cm2 Nb/AlOx/Nb Josephson integrated circuit technology. The
SFQ components except the QOSs were designed using the CONNECT top-down design tools [10]. The total number of Josephson junctions was 1252, and the whole processor occupied 1.8 mm × 1.8 mm.
Figure 5 displays an example of a set of waveforms in a test experiment. SFQ processors are examined in the conventional system described above. Output waveforms of corresponding SFQ/DC converters are observed for CLK A, CLK B, and Bit 0. Thus, an SFQ pulse is generated both at each rising edge and at each falling edge in the waveforms. These signals are amplified by 46 dB with very low noise semiconductor amplifiers with band width of 120 MHz, and recorded in a digital oscilloscope. The repetition rates in CLK A and CLK B are set to be about 50 MHz by adjusting bias current supplied to the embedded ring oscillator. We found from these waveforms that an SFQ pulse was
gener-Fig. 5 Waveforms in an experiment for a dual-input SFQ signal proces-sor. CLK B generates an SFQ pulse with a time delay of 10 ns after CLK A generates a pulse. The repetition rates of CLK A and CLK B are 50 MHz. The waveform of ‘Bit 0’ means the LSB of TDC A is “1” and that of TDC B is “1.” This result is obtained for counting clocks for the duration TOF indicated in the figure. The waveform of ‘Input S’ is a sinusoidal wave pro-vided to QOS 0. The same waves are provide to QOS A and QOS B with a phase shift ofπ.
ated 10 ns later at CLK B than at CLK A.
We show a sinusoidal waveform of Input S and the threshold of QOS 0 at the bottom of Fig. 5. Here, Input S is a current supplied to QOS 0 as shown in Fig. 3. When the magnitude of Input S exceeds the threshold at the rising edge, a start signal is sent to TDC A and TDC B. The same wave with a phase shift ofπ is provided to both Input A and Input B. Thus, a stop signal for QOS A and that of QOS B are generated at the timing when the magnitude crosses the threshold at the falling edge. The threshold of QOS A and QOS B are the same and indicated in Fig. 5. In the case shown in Fig. 5, 15 clocks are provided to both TDC A and TDC B during a period between the start signal and the stop signal. Thus, 4-bit parallel digital outputs of “1111” would be obtained. The waveform ‘Bit 0’ indicated in Fig. 5 shows that TDC A outputs “1” as the least significant bit after the stop signal is generated. TDC B also outputs “1” 10 ns later, Both TDCs output “1” for each bit. We successfully demon-strated the measurement of the time difference and the time-division multiplexing. In this experiment, dual-input SFQ signal processor consumed 400μW.
Figure 6 shows measured time difference as a function of input time difference determined by period of the sinu-soidal wave. The linear dependence was obtained. This means that our TDCs worked correctly. We also observe the spread of the data of±20 ns. This is caused by a time
Fig. 6 Measured time difference as a function of input time difference in a TDC.
jitter of 20 ns in our signal source used here.
As mentioned in Sect. 2, our diffraction system has ca-pability to accommodate 6000 detectors with dynamic range of 16 bits. If Tneutron is 40 ms, clocks with a frequency of
2.1 MHz (= fm/6000) to be distributed to each TDC can be
generated for 6000 TDCs by using the master clocks with a frequency of 12.8 GHz. A 16-bit serial read/write memory should be added to each TDC for handling many data. Mul-tiplexing is executed by reading serial data stored in memo-ries in an ascending order with the master clocks [11], and then serial/parallel conversion is done. Power consumption can be estimated to be 250μW for a single input. We can easily reduce the power consumption to 25μW/input by de-creasing a value of a bias resistance, because power is con-sumed mostly at bias resistors [12]. From the point of view of the power consumption, SFQ signal processor consumes only 150 mW for 6000 detectors. This value is still accept-able for refrigeration by our GM cryocooler, though the ac-tual number of detector chains accommodated in the system is limited to around 20 at present because of insufficient uni-formity in characteristics of detectors and small integration level of SFQ circuits.
5. Conclusion
We described a concept of our neutron diffractometer sys-tem based on MgB2 thin film detectors and an SFQ
sig-nal processor. The system has high potential to accommo-date several thousand or more detectors in a commercially-available cryocooler. The total system would be very com-pact compared to a system based on conventional gas tube neutron detectors. In our system, kinetic energy is deter-mined for individual neutrons detected at the detectors by measuring TOF with SFQ-TDCs. Mapping of the number of the detected neutrons for each kinetic energy corresponds to a diffraction pattern. An SFQ signal processor including SFQ-TDCs and a time-domain multiplexer can handle out-puts of a large number of detectors in a cryocooler because the SFQ circuits have the nature of low power
consump-tion as well as the high-speed nature. A dual-input SFQ signal processor has been successfully demonstrated with a time resolution of 20 ns and powerconsumption of 400μW. These values show high feasibility of our neutron diffraction system.
Acknowledgment
This development was supported by SENTAN, JST. Na-tional Institute of Advanced Industrial Science and Technol-ogy partially contributed to the circuit fabrication.
References
[1] K. Oikawa, F. Maekawa, M. Harada, T. Kai, S. Meigo, Y. Kasugai, M. Ooi, K. Sakai, M. Teshigawara, S. Hasegawa, M. Futakawa, Y. Ikeda, and N. Watanabe, “Design and application of NOBORU — Neutron beam line for observation and research use at J-PARC,” Nucl. Instr. and Meth. A, vol.589, no.2, pp.310–317, May 2008. [2] W. Braeutigam, O. Felden, R. Maier, S. Martin, A. Schnase,
G. Stassen, and E. Zaplatine, “Design considerations for the linac system of the ESS,” Nucl. Instr. and Meth. B, vol.161-163, pp.1148– 1153, March 2000.
[3] T. Ishida, M. Nishikawa, Y. Fujita, S. Okayasu, M. Katagiri, K. Satoh, T. Yotsuya, H. Shimakage, S. Miki, Z. Wang, M. Machida, T. Kano, and M. Kato, “Superconducting MgB2thin film detector
for neutrons,” J. Low Temp. Phys., vol.151, no.3-4, pp.1074–1079, Feb. 2008.
[4] K. Nakajima, Y. Onodera, and Y. Ogawa, “Logic design of Joseph-son network,” J. Appl. Phys., vol.47, no.4, pp.1620–1627, April 1976.
[5] K.K. Likharev and V.K. Semenov, “RSFQ logic/memory fam-ily: A new Josephson-junction technology for sub-terahertz-clock-frequency digital systems,” IEEE Trans. Appl. Supercond., vol.1, no.1, pp.3–28, March 1991.
[6] A. Bozbey, S. Miyajima, H. Akaike, and A. Fujimaki, “Single-flux-quantum circuit based readout system for detector arrays by using time-to-digital conversion,” IEEE Trans. Appl., Supercond., vol.19, no.3, pp.509–513, June 2009.
[7] T. Ishida, M. Nishikawa, S. Miki, H. Shimakage, Z. Wang, K. Satoh, T. Yotsuya, M. Machida, and M. Kato, “Superconducting radiation detector by using a microfabricated MgB2 meander line,” Physica
C, vol.460-462, pp.618–619, Sept. 2007.
[8] A. Fujimaki, Y. Nishido, and A. Sekiya, “High-resolution analog-to-digital converters toward software-defined-radio receivers,” IEICE Trans. Electron., vol.E89-C, no.2, pp.113–118, Feb. 2006. [9] S. Nagasawa, Y. Hashimoto, H. Numata, and S. Tahara, “A 380 ps,
9.5 mW Josephson 4-Kbit RAM operated at a high bit yield,” IEEE Trans. Appl. Supercond., vol.5, no.2, pp.2447–2452, Jan. 1995. [10] S. Yorozu, Y. Kameda, H. Terai, A. Fujimaki, T. Yamada, and
S. Tahara, “A single flux quantum standard logic cell library,” Phys-ica C, vol.378-381, pp.1471–1474, Sept. 2002.
[11] A. Kirichenko, S. Sarwana, O. Mukhanov, I. Vernik, Y. Zhang, J. Kang, and J.M. Vogt, “RSFQ time digitizing system,” IEEE Trans. Appl. Supercond., vol.11, no.1, pp.978–981, March 2001. [12] Y. Yamanashi, T. Nishigai, and N. Yoshikawa, “Study of LR-loading
technique for low-power single-flux-quantum circuits,” IEEE Trans. Appl. Supercond., vol.17, no.2, pp.150–153, June 2007.
Akira Fujimaki received the B.E., M.E., and Dr. Eng. degrees from Tohoku University, Sendai, Japan, in 1982, 1984, and 1987, respec-tively. He was a Visiting Assistant Research Engineer at the University of California, Berke-ley, in 1987. Since 1988, he has been work-ing on superconductor devices and circuits at the School of Engineering, Nagoya University, Nagoya, Japan, where he is currently a pro-fessor. His current research interests include single-flux-quantum circuits and their applica-tions based on low- and high-temperature superconductors.
Isao Nakanishi received the B.E. degree in electronic engineering from Nagoya University, Nagoya, Japan, in 2009. He is currently work-ing toward the M.E. degree at the same univer-sity. His research interests include the single-flux-quantum circuits and their applications. He is a member of the JSAP.
Shigeyuki Miyajima received the B.E. and M.E. degrees in electronic engineering and quantum engineering from Nagoya University, Nagoya, Japan, in 2007 and 2009, respectively. He is currently working toward Ph.D. degree at the same university. His research interests in-clude the single-flux-quantum circuits and their applications. He is a member of the Japan Soci-ety of Applied Physics.
Kohei Arai received the B.E. and M.E. de-grees from Osaka Prefecture University, Sakai, Japan, in 2008 and 2010, respectively. He has been working on superconductivity detectors. He is an engineer in Seiko Epson Corporation.
Yukio Akita received the B.E. from Osaka Prefecture University, Sakai, Japan, in 2009. He is a graduate student at Osaka Prefecture Uni-versity, and is currently studying on the super-conducting detector.
Takekazu Ishida received the B.E. from Tohoku University, Sendai, Japan, in 1976. He received the M.E. and Ph.D. degrees from Kyoto University in 1978 and 1982, respectively. He has been working on material properties of su-perconductors for years. He proposed a new idea to produce a neutron detector using a new superconductor MgB2in 2001. He is currently
a professor of Department of Physics and Elec-tronics at Osaka Prefecture University. His re-search interests are nanofabrication of supercon-ductor, its potential application as superconducting devices and magnetic properties of anisotropic superconductors.
