Re-imaging optics & 600 pixels camera

Figure 1. A schematic drawing of lens-antenna coupled optical system for MKIDs camera developed at NAOJ.

The most likely reason is that the sensitivity is limited by stray lights in our measurement system. We thereforeneed more light-tight setup to measure and study the performance difference between MKIDs made of the crystalfilm and the amorphous film.

2.2 Optical System for MKIDs Camera

We adopted lens optical system which is fed by single polarization sensitive double-slot antennas (Fig. 1). TheMKIDs developed at NAOJ are 1/4 λ coplanar-waveguide (CPW) superconducting resonators weakly coupledto a CPW feed line. The opposite side of a resonator to the feed line is a shorted end where the MKID CPWacts as a feed to the double-slot antenna. The lenslets, made of Si, are arrayed and mounted to the backsideof the MKID chip, so that the antennas are placed on geometrical focuses of the leslets. The base design of alenslet is a extended hemispherical lens, and lens array is directly machined by high-speed spindle at MechanicalEngineering (ME) shop at NAOJ. The geometrical parameters of the antenna and the lenslet are adjusted to atarget radiation frequency (150-440 GHz) and bandwidth (10-30 %) by calculating scattering parameter (S11)with HFSS †. The beam patterns of the optical system is also calculated with HFSS to tune the geometricalparameters, so that the beam radiated from the lenslet become symmetric.

The performance of the designed optical system was evaluated with beam measurement at 0.3 K 3He ab-sorption refrigerator, and the measured beam pattern was in good agreement with the simulation.10 OpticalNEP and optical efficiency are also measured with a cold blackbody at temperature between 2 K and 16 K. Theevaluated optical NEP and optical efficiency are 1× 10−14 W/

√Hz and ∼15 %, respectively.9

The reason of these poor optical properties is probably due to the diffusion of quasi-particles generated byincident photons to outside of the MKID’s resonator part. It is required to trap quasi-particles at the mostsensitive part of a MKID, where corresponds to the shorted end in case of 1/4 λ resonator, in order to maintainhigh sensitivity of MKID even under the optically-illuminated condition. This trap can be realized in MKIDswith hybrid design, and good optical NEP and efficiency are demonstrated in Ref. 13.

2.3 Readout Circuit

The fact that a superconductor at T 2Δ) breaks Cooper pair → change in kinetic inductance of resonator

力学インダクタンスはクーパー対の慣性質量に起因するため、超伝導体にEgap＜ hνのエネルギーを持つフォトンが入射しクーパー対を破壊すると力学インダクタンスが変化する。MKIDとはこの力学インダクタンスの変化を読み取る直接検出器である。図 2.2にMKIDおよび等価回路を示す。

図 2.2: (左)MKID (右)MKIDの等価回路 (Barends, 2009)

最初に述べたようにMKIDとはマイクロ波帯で動作する超伝導共振器であり、等価回路から分かるようにキャパシタンスとインダクタンスが並列接続された構造を持つ。また、MKIDでは共振器の長さは読み出しに使用するマイクロ波 (一般には 4 - 8 GHz)の 1/4波長になっており、片側 short , もう片側が openになっていることから長さに対応したマイクロ波で共振を起こす。この時、共振周波数 ω0は

ω0 =2π

4l√

(Lg + Lk)C(2.9)

で表される。ここで、lは共振器の長さ、Lg, Cは超伝導配線が持つインダクタンス、キャパシタンスである。MKIDにEgap＜ hνのエネルギーを持つフォトンが入射しクーパー対を破壊すると力学インダクタンスが変化することから、共振周波数は

ω0 −∆ω0 =2π

4l√

(Lg + Lk + ∆Lk)C(2.10)

となり、図 2.3に示すように低周波側にシフトする。また、図 2.4に示すように short端では電流が最大、電圧が最小であり、open端では電圧が最大、電流が最小となっている。力学インダクタンスの変化に対する感度、つまりMKIDの感度は short端で最大となるため short端でフォトンをたくさん吸収しクーパー対を破壊したい。そのため図 2.5に示すようにMKIDでは short端にアンテナを付けてサブミリ波のフォトンを効率よく吸収するような構造になっている。

18

2 . Noise Measurement

3 . Beam Measurement

4 . Re-imaging Optics

NARUSE et al.: OPTICAL EFFICIENCIES OF LENS-ANTENNA COUPLED KINETIC INDUCTANCE DETECTORS AT 220 GHZ 183

chamber with background pressures of less than 2 10 Pa.Before Al deposition, the Si wafer was immersed in an HFsolution and annealed at 650 C in the UHV chamber to cleanand reconstruct its surface. During the deposition process,the substrate was not directly heated but was warmed by theheat emitted from the K-cell, and was at around 100 C. Thefilm-growth rate was increased gradually from 0.2 to 0.8 s.The residual resistivity ratio (resistivity at 295 K/ resistivity at4.2 K) of the MBE-Al film was about 22. We confirmed thatthe superconducting transition temperature of the epitaxial filmwas 1.21 K by measuring the temperature dependence of themicrowave transmission losses (S21) with a vector networkanalyzer.The Al film was patterned into standard, wave length,

CPW-line resonators with a wet etching process. The CPW-lineresonators had a 3- center line and 2- gaps. The res-onators were capacitively coupled to a feed line. The feed linealso consisted of a CPW line and its center conductor was 12 min width and the gaps were 8 m in width.

IV. MEASUREMENTS

A. Electrical Noise Equivalent Power (NEP)

To evaluate the properties of the resonators, we performedcomplex , amplitude, and phase-noise spectrum measure-ments with a microwave signal generator, an IQ mixer, and a16-bit digitizer. With the exception of the refrigerator, our mea-surement setup was similar to the TU Delft’s system [14]. Thequasiparticle recombination time was estimated from the re-laxation time of the MKID response against LED pulses, theduration of which was around 10 s [15]. A chip containing38 MKIDs without antennas was glued to a sample holder. Thesample box was put on the top load of a 0.1-K dilution refrig-erator and surrounded with magnetic shields. While the sampletemperature was changed from 100 to 500 mK, the resonant fre-quencies, noise spectra, quasiparticle decay time, and qualityfactors of each pixel were measured.THe electrical noise equivalent power of an MKID

is given by [16]

(13)

where represents noise spectrum for the resonator, can befor phase readout or for amplitude readout, is the quasipar-ticle decay time, is the superconducting energy gap, is theoptical modulation frequency, is the efficiency of quasiparticlecreation and is approximately 0.57 [17], for amplitude readout(dissipation readout) or phase readout (frequency readout), and

. represents the quality factor of the MKID, andis the resonant frequency.The amplitude and phase spectra were the lowest at

around 150 mK, as shown in Fig. 5. The resonant propertieswere 3.959 GHz, 170 000, and 450 s. Thesuperconducting energy gap was estimated as 0.18 meV fromthe critical temperature (1.21 K).The sample-bath temperature dependence of the noise spectra

and at 1 kHz are shown in Fig. 6. Because the quasipar-

Fig. 5. NEP spectra in amplitude and phase directions at 150 mK in dark con-ditions. The resonant frequency and quasiparticle lifetime were 3.959 GHz and450 s, respectively.

(a) (b)

Fig. 6. Temperature dependences of (a) amplitude and phase noise and (b) elec-trical-NEP for amplitude and phase readout at 1 kHz in dark conditions.

ticle decay time had a plateau below 0.20 K, for theMKID was saturated below 0.20 K. Stray light entering the res-onators from the coaxial cables and the small gap between thelid and bottom of the sample box could cause the plateau in thedecay time [16]. At high temperature 0.2 K , because theeffect of two level fluctuations in dielectrics located at the sur-face of the aluminum decreased, the excess phase noise was re-duced. On the other hand, the amplitude noise increased withincreasing temperature because of the increment in the numberof quasiparticles in the resonator. Consequently, above 0.25 K,the phase and amplitude were comparable, althoughthe amplitude was lower than the phase at lowtemperature.

B. Optical Measurements

The beam patterns of the lens fed by a double slot antennawith a superconducting resonator were measured with a 0.3-Ksorption cryostat [13]. The measured patterns were in goodagreement with the calculations within the dynamic rangeof the setup. The dynamic range of this measurement wasabout 20 dB and could be limited by the low sensitivity ofthe superconducting resonator because of the relatively highbath temperature. Because the measured beam patterns werecomparable to the calculations, the theoretical patterns wereadapted to estimate the solid angles of the camera in this study.In Section IV-A, the entire device was surrounded by the

sample holder at 0.1 K, but, in this section, the device was ex-posed to blackbody radiation at 8–16 K. The schematic mea-surement configuration for the camera including the optical load

[M. Naruse+2011][M. Naruse+2013]

[S. Ishii+2010]

[T. Nitta+2013]

[S. Sekiguchi, in prep]

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NITTA et al.: BEAM PATTERN MEASUREMENTS OF MILLIMETER-WAVE KID CAMERA 5

Fig. 10. Measurement results of 2D far-eld beam patterns of the extendedhemispherical silicon lens array fed by the double-slot antennas at 440 GHz.The contours were every 2 dB steps.

Fig. 11. Measurement results of 2D far-eld beam patterns of the extendedhemispherical silicon lens array fed by the double-slot antennas at 220 GHz.The contours were every 3 dB steps.

The 1-THz low-pass lter is made from polyethylene sheet con-taining reststrahlen crystals powder [32]. Total transmittance ofthese lters at 220 and 440 GHz band is about 50% and 20%, re-spectively. Temperature of the cold stage was kept at around 300mK and the holding time is about 8 hours. The MKID sample

Fig. 12. Measurement results of 1D far-eld beam patterns at 220 GHz ob-tained from six Al MKIDs. The dot and solid lines show the simulated andthe measured patterns, respectively. Blue lines and pink lines correspond to theE-plane and the H-plane, respectively.

box is connected with 0.86 mm diameter CuNi coaxial cables toa cold low noise amplier (CLNA) on the 4 K stage. The choicefor thin CuNi coaxial cable is to reduce the thermal load on theMKID sample box. The MKID camera was mounted on the 300mK cold stage, as shown in Fig. 7. The beam pattern measure-ments were performed in a magnetic shield room of permalloyto remove the effect of the magnetic eld from the measurementinstruments and terrestrial magnetism.Fig. 8 shows the block diagram of the beam pattern mea-

surement system. External radiation sources for the 220- and440-GHz band are composed of a signal generator (100 kHzf20GHz), frequency multipliers, a power amplier, an attenuator,and a rectangular probe horn.The resonance frequencies of MKIDs are measured from thespectrum taken by a vector network analyzer as shown in

Fig. 9. The shift of resonance frequency and the change of theamplitude of resonances correspond to detected power in anMKID. The detected power is recorded as a function of theposition of the external source, which is controlled by ascanner in front of the vacuum window of the cryostat. The in-cident power to the MKID is adjusted by the gate voltage of thepower amplier. The beam pattern was calculated from the ob-tained data.

C. Results

Typical internal quality factors at 300 mK of 9 pixelsMKID array were around 30000. In the 440 GHz array, morethan 90 pixel resonances out of 102 pixels were measured in 400MHz bandwidth. Figs. 10 and 11 show the measurement resultsof 2D far-eld beam patterns of the extended hemispherical sil-icon lens array fed by the double-slot antennas at 440 GHz and

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NITTA et al.: BEAM PATTERN MEASUREMENTS OF MILLIMETER-WAVE KID CAMERA 5

Fig. 10. Measurement results of 2D far-eld beam patterns of the extendedhemispherical silicon lens array fed by the double-slot antennas at 440 GHz.The contours were every 2 dB steps.

Fig. 11. Measurement results of 2D far-eld beam patterns of the extendedhemispherical silicon lens array fed by the double-slot antennas at 220 GHz.The contours were every 3 dB steps.

The 1-THz low-pass lter is made from polyethylene sheet con-taining reststrahlen crystals powder [32]. Total transmittance ofthese lters at 220 and 440 GHz band is about 50% and 20%, re-spectively. Temperature of the cold stage was kept at around 300mK and the holding time is about 8 hours. The MKID sample

Fig. 12. Measurement results of 1D far-eld beam patterns at 220 GHz ob-tained from six Al MKIDs. The dot and solid lines show the simulated andthe measured patterns, respectively. Blue lines and pink lines correspond to theE-plane and the H-plane, respectively.

box is connected with 0.86 mm diameter CuNi coaxial cables toa cold low noise amplier (CLNA) on the 4 K stage. The choicefor thin CuNi coaxial cable is to reduce the thermal load on theMKID sample box. The MKID camera was mounted on the 300mK cold stage, as shown in Fig. 7. The beam pattern measure-ments were performed in a magnetic shield room of permalloyto remove the effect of the magnetic eld from the measurementinstruments and terrestrial magnetism.Fig. 8 shows the block diagram of the beam pattern mea-

surement system. External radiation sources for the 220- and440-GHz band are composed of a signal generator (100 kHzf20GHz), frequency multipliers, a power amplier, an attenuator,and a rectangular probe horn.The resonance frequencies of MKIDs are measured from thespectrum taken by a vector network analyzer as shown in

Fig. 9. The shift of resonance frequency and the change of theamplitude of resonances correspond to detected power in anMKID. The detected power is recorded as a function of theposition of the external source, which is controlled by ascanner in front of the vacuum window of the cryostat. The in-cident power to the MKID is adjusted by the gate voltage of thepower amplier. The beam pattern was calculated from the ob-tained data.

C. Results

Typical internal quality factors at 300 mK of 9 pixelsMKID array were around 30000. In the 440 GHz array, morethan 90 pixel resonances out of 102 pixels were measured in 400MHz bandwidth. Figs. 10 and 11 show the measurement resultsof 2D far-eld beam patterns of the extended hemispherical sil-icon lens array fed by the double-slot antennas at 440 GHz and

23rd International Symposium on Space Terahertz Technology, Tokyo, 2-4 April 2012 mm for 220 GHz.

B. Fabrication and Evaluation of the Silicon Lens A small diameter lens array has been fabricated with

techniques such as photolithography or laser machining [13]-[15]. Ultra-precision cutting by using ball end mills was tried as another efficient technique to process the high-purity polycrystalline silicon in shape of the lens. The 220 GHz lens diameter and the lens spacing were 4.09 mm and 0.3 mm, respectively. To achieve 0.3 mm in lens spacing and get a surface accuracy in the order of μm, a small diameter end mill was needed. In addition, it required a rotating speed of more than several tens of thousands rpm to keep cutting velocity of end mill. Therefore, a combination of the ultra-precise processing machine, Toshiba ULG-300, and the high-speed spindle, Toshiba ABC-20M, is used for this process. The machining tools are made of TiAlN coated ceramic end mill with radius of 0.5 mm and 0.15 mm. Figure 1 shows the photograph of the 9 pixel 220 GHz silicon lens array under processing, and table 1 shows the processing conditions of the silicon lens array.

A three-dimensional coordinate measuring machine was used to measure the shape of the lenses. The shape error from the designed value was less than  20  μm (Peak-to-Valley). The surface roughness of the top of the lens was also measured by using the non-contact three-dimensional measuring machine. The surface roughness of the top of the lens was around 0.7 μm   in rms. This value is small enough for use in the millimeter wave range.

III. BEAM PATTERN MEASUREMENTS

A. Measurement System The MKID camera was fabricated by using silicon lens

array and Al-based MKIDs. Epitaxial Al(111) film has been grown on a Si(111) wafer by using molecular beam epitaxy [16]. The thickness of the aluminum film was 150 nm. The CPW geometry had a 3 μm-wide central line and 2 μm-wide gaps.

A 3He sorption cooler mounted on a liquid helium cryostat was used for beam pattern measurements of the MKID camera. Temperature of the cold stage was kept at around 300 mK and the holding time is about 8 hours. Figure 2 shows the photograph of the beam pattern measurement system. The millimeter wave signals were radiated from a rectangular probe horn, and scanned around the vacuum window of the cryostat. The response of MKIDs was measured from the S21 spectrum taken by a vector network analyzer.

B. Results Figure 3 shows the measurement results of 2D far-field

beam pattern of the extended hemispherical silicon lens fed by the double slot antenna at 220 GHz. The dynamic range of this measurement system was about 20 dB, which is limited by  the  Al  MKID’s  sensitivity  at  the  measurement  temperature  of 300 mK. The half-power beam width is about 20 degree, and the comparison of the 220 GHz beam pattern between the measurements and the calculations showed good agreement.

TABLE I PROCESSING CONDITIONS OF THE SILICON LENS ARRAY

Rotating speed 40000 [rpm]

Rotation unbalance Below 10 [nm] Cutting feed rate 100-120 [mm/min]

Cutting depth 20-100 [μm]

Fig. 1. The photograph of the 9 pixel 220 GHz silicon lens array under processing.

Fig. 2. The photograph of the beam pattern measurement system.

Fig. 3. The measurement results of 2D far-field beam patterns of the extended hemispherical silicon lens array fed by the double slot antennas at 220 GHz. The contours were every 3 dB steps.

Telescope Focus

(F/# = 6)

M1

Re-imaging Optics

Cryostat

Temperature dependence of electrical-NEP for amplitude and phase readout at 1 kHz in dark conditions.

・ Silicon lens array was directly machined using a high-speed spindle on an ultra-precision

　 machine (Mechanical Engineering shop at ATC)

・ Shape fabrication error : < 50 μm, Surface roughness : < 1μm

・ Lens and antenna geometry were optimized with HFSS

・ Beam patterns were measured by using 0.3 K sorption cooler and external CW source

・ Measured beam patterns of the MKID camera are in good agreement with the calculations

!

・ Re-imaging optics with two dielectric lenses transfers from a telescope focus of F/# = 6 at

　 ambient temperature to a focal plane of F/# = 1 at 100 mK

→ The configuration is overwhelmingly compact compared to reflective optics system

→ It typically provides a large diffraction limited field-of-view

!

✴ High-refractive index lenses

・Thin Lensex) Silicon (n = 3.4), Alumina (n = 3.1)

・Good image quality✴ Cold baffles・Reduction of the stray light

✴ Cryogenic measurement・Stray light power is only 0.2 μW

5 . Observation & Future Plan✴ Nobeyama 45 m Telescope

✴ Antarctic 10 m Terahertz Telescope

・various types of IR blocking filters

→ T = 78% (in-band), 10-10% (out-band)・temperature@focal plane : 100 mK

・Developed camera system with cold re-imaging optics and 600 pixels camera can be installed

　on the M4 focus of the Nobeyama 45 m telescope

✴ Target・Distant galaxies・Unknown sources discovered by

　AKARI satellite

✴ Specification of MKID camera・Frequency : 220 GHz band・Operating temperature : 100 mK

Re-imaging optics of 10 m telescope

850 GHz band 20000 pixels MKID camera

[T. Tsuzuki+2014]

・By extending these technologies, a wide FoV terahertz camera system for the Antarctic 10 m

　terahertz telescope has been designed.

・Optical system provides 1 degree FoV observations for detecting unknown distant galaxies

・850 GHz band 20000 pixels camera with seven hexagonal modules was also designed

!

・Number of pixels : 450・Beam size : ~ 13 “ (D = 22 m)・Field-of-View : ~ 2.5 ‘

・FFTS readout system ( ) with 1 Gsps and 270 MHz bandwidth is used

　for MKID readout

[K. Karatsu+2014]

✴ Noise equivalent power (NEP) of MKID

22

1

, )2(1)2(1,)( resrrqp

A ffNAfSNEP τπτπητδ

θδθ ++&

&'

())*

+

Δ=

−

Noise LifetimeResponsivity

To extend quasi-particle lifetime (τr), single crystal Al films are made on Si wafers by molecular beam epitaxy (MBE).

!・ Al (111) on Si (111) wafer

・ Residual resistivity ratio (RRR) : > 20

!・ NEP was measured by dilution refrigerator

!− Qi ~ 2e+6, τr ~ 500 μs

− Electrical NEP : 6 × 10-18 [W/Hz1/2]

− Optical NEP : ~ 10-15 [W/Hz1/2]

2. M. Naruse et al., J. Low Temp. Phys., 167, 373 (2011).3. M. Naruse et al., IEEE Trans. TST, 3, 180 (2013).5. K. Karatsu et al., J. Low Temp. Phys., 176, 459 (2014). 6. T. Tsuzuki et al., in Proc. SPIE. (2014).

microwave

500 μm