1. Introduction

The multi-pixel photon counter (MPPC) is a novel semiconductor (Si) photon detector [1, 2, 3] and became commercially available since 2007. It consists of micro avalanche photo-diode (APD) pixels in a ~ mm2 surface area [4]. Such a photon-counting detector can be extremely useful in vast applications such as quantum optics, quantum information science, and atomic and molecular physics [5, 6, 7]. Geiger-mode avalanche photodiodes for three-dimensional imaging [8], operation of silicon single photon avalanche diodes at cryogenic temperature [9], and photon-number-resolving detector with 10 bits of resolution with 1024 pixels at infrared wavelength [10] have been addressed recently. There are, however, only a few published references reporting characteristics of such a new Si-based MPPC in the visible region of the spectrum as an efficient fluorescence detector, as far as the authors aware, where large photon detection dynamic range plays an important role [11, 12]. In this paper, we present an efficient fluorescence detector made of a temperature-stabilized Si-based MPPC at temperatures down to 5 °C to extend the application area of the multi-pixel photon counter into the fields of laser spectroscopy. We show that effective cooling of the device by means of a compact thermo-electric cooler (TEC) brings several advantages, such as high gain, low dark noise rate, and thus high signal-to-noise ratio in the efficient fluorescence detection at 398.9 nm of the ¹ S ₀↔¹ P ₁ transition of ytterbium (Yb) atoms in an effusive atomic beam. We present also a comparison of the fluorescence detection efficiencies between the MPPC device and a side-on photo-multiplier (PM) tube with known gain positioned at the symmetric location from the Yb atomic beam.

2. Experiments

We used a Si-based MPPC that has 1,600 pixels in 1-mm² area (Hamamatsu S10362-11-025U) in this experiment. Each pixel in the MPPC works in the limited Geiger mode with a reverse bias voltage around 70 V, which is a few volts above the breakdown voltage [1, 2, 3, 4, 11, 12]. If a photon hits a pixel and produces a photo-electron, the photo-electron induces a Geiger avalanche. Since one pixel saturates with one avalanche, the multi-pixel structure is necessary in order to count the number of photons. The avalanche signals from all the pixels are summed up and read out as a signal so that we may achieve a large dynamic range in photon detection. Since the response time of each pixel to detect single photon is on the order of 10 ns which is limited by the quenching register in each pixel [4], we can in principle detect 1,600 pixels/10 ns=1.6×10¹¹ photons/s without photo-current saturation. In this paper, we exploit this huge dynamic range of single-photon detection capability of the MPPC for the efficient fluorescence detection.

Figure 1 shows the schematic diagram of our temperature-stabilized Si-based MPPC module. This set up does not require a current-to-voltage converter (pre-amplifier), rather a 1 MΩ load resister was used to measure the photon detection signal as a voltage signal. The metal body of the MPPC is mounted on one side of a TEC which has a hole at the center with 5-mm diameter. The other side of the TEC is thermally contacted to the anodized aluminium cap to effectively dissipate excess heat when we decrease the temperature of the MPPC down to 5 °C. A current-limiting resister R=10 kΩ and a surge capacitor C=0.1 µF to supply a bias voltage V _bias were wired on the Cu board which was connected electronically to the ground. The temperature of the MPPC was actively stabilized by using a TEC (I _TEC) on which the MPPC was mounted. We used a home-made temperature controller based on an active servo-control technique [13]. With the single-stage TEC we were able to control the temperature of the MPPC down to 5 °C with a temperature stability of ±0.005 °C. Finally, the MPPC module was filled with dry N₂ gas and sealed with a slide glass (SG) to prevent the water condensing, which has no anti-reflection coating on both sides.

 

Fig. 1. (Color online). Schematic of a temperature-stabilized MPPC module. SG; slide glass, Th; thermistor, TEC; thermo-electric cooler, R; current-limiting resister, C; capacitor, Cu; copper board, V_bias; bias voltage, V_sig; signal; I_TEC; TEC current, R_Th; thermistor resistance.

Download Full Size | PDF

In order to compare the fluorescence detection efficiencies between the temperature-stabilized MPPC and a side-on photo-multiplier tube (Hamamatsu H9370-03) which has a known gain, we have positioned two detectors at the symmetric locations from a Yb effusive beam as shown in Fig. 2. The collimated effusive Yb atomic beam was generated in a ultra-high vacuum chamber at the temperature of 430±1 °C so that the longitudinal velocity v of the Yb atomic beam was well defined by the modified Boltzmann velocity distribution function. We used an ion pump with a 75 l/s pumping speed to pump the ultra-high vacuum chamber down to 4×10^-6 Pa. Table 1 shows various parameters of the MPPC and the PM tube used in the experiment to calculate the fluorescence detection efficiency. We used a home-made extended-cavity diode laser (ECDL) to excite the ¹ S ₀↔¹ P ₁ electric dipole transition at 398.9 nm for the measurement of photon detection efficiency of the MPPC [14].

We first calculate detected photon numbers N_p[photons/s·sr] by the PM tube of which detection area, gain, and anode radiant sensitivity (RS) are 3.7×13 mm², 1×10⁵ at the control voltage of 0.58 V, and 1 V/nW, respectively, as listed in Table 1. Since we know the atomic flux of the ¹⁷⁴Yb isotope F ₁₇₄ at the oven temperature of 430 °C in a collimated atomic beam as F ₁₇₄=4.4×10¹⁰ atoms/s, the detected photons can be calculated as

where d/v is the time taken by the atoms in the laser beam, R[photons/s] in Eq. (1) is the fluorescence rate from the ¹ P ¹ state, and ΩPM is the detector solid-angle/4π. The fluorescence rate R[photons/s] is given by

where Γ=2π×28 MHz is the decay rate of the excited-state, ρee is the excited-state population, and s=I/I_s is the saturation parameter. I and I_s=58 mW/cm² are the laser intensity and saturation intensity of the ¹ S ₀↔¹ P ₁ transition, respectively. From the detected photon number, we can also calculate the signal voltage V_PM[V] from the PM tube module as V_PM=N_p×l×h̄ ω×RS=1.3 V with A_D=3.7×13 mm², where l=1- cumulative loss =0.99 including the antireflection-coated vacuum window and RS is the anode radiant sensitivity of the PM tube at 398.9 nm.

 

Fig. 2. (Color online). Schematic diagram of an experimental set up for fluorescence detection. ECDL; extended-cavity diode laser, PMT; photo-multiplier tube, MPPC; multi-pixel photon counter module.

Download Full Size | PDF

We estimate now the signal voltage V_MPPC[V] of the MPPC module on a 1 MΩ load impedance by assuming that the fluorescent photons emitted uniformly over 4π solid-angle. By the same analogy with the PM tube signal V_PM, V_MPPC[V] can be estimated as

where $F_{174}\times \left(\frac{d}{v}\right)\times R\times {\Omega }_{\mathrm{MPPC}}\times l\times T_r=4.9\times {10}^6$ is the detected photons/s·sr by the MPPC module, Ω_MPPC is the detector solid-angle/4π, T_r=0.91 is the transmittance of the bare slide glass in Fig. 1, andR_L is the load impedance. Note here that due to a low fill factor of 0.31, the photon detection efficiency (PDE) of the MPPC is only 0.25. In Eq. (3), the gain term contains the conversion factor between detected photon numbers and electron bunches in the MPPC module. Since the gain of the MPPC depends on the bias voltage V_bias and the operation temperature T, we estimate roughly the signal amplitude with parameters in Table 1 at the gain

Table 1. Parameters of the MPPC [4], the PM tube, and other conditions.

View Table

of 1×10⁵, which is the same as that of the PM tube, and results in V_MPPC=700 mV on an 1 MΩ load. In these estimations, we subtracted the background signal due to the ambient photons that is clearly seen in Fig. 3. In order to compare fluorescence detection efficiencies between the two devices, we consider the effective signal amplitude per unit detector area which is defined by

where i=MPPC,PM and A_D is the detector area. From Table 1 and the previous estimations, we have V^PM_eff=27 mV/mm² and V^MPPC_eff=700 mV/mm², thus we can conclude that the MPPC module may have a 25-fold large effective signal amplitude per unit area compared to that of the PM tube at the same gain of 1×10⁵ (see Fig. 3).

Figure 3 shows the effective signal amplitude per unit detector area measured by our temperature-stabilized Si-based MPPC module and the PM tube without averaging versus frequency detuning for various operating parameters. Each fluorescence peak corresponds to the Yb isotopes and for the fermionic isotopes the quantum number F due to the hyperfine structure are shown in parenthesis [15]. From the bottom the signals correspond to the PM tube with a control voltage (CV) of 0.58 V and gain of 1×10 ⁵, the MPPC at T=5 °C and V_bias=69.6 V, the MPPC at T=20 °C and V_bias=70.8 V, and the MPPC at T=5 °C and V_bias=71.6 V with gain of 1.2×10⁵ (see later), respectively. As mentioned above, since the detector area of the PM tube is 48 times larger than the MPPC, the voltage signal of the PM tube at the gain of 1×10⁵ is only two times larger than that of the MPPC at T=5 °C and V_bias=71.6 V, thus the effective signal magnitude of the MPPC is 25 times larger than that of the PM tube. At these parameters the gain of the MPPC was actually measured to be 1.2×10⁵ from Eq. (3), which is in good agreement with the manufactures value [4]. We note that the signal fluctuation of the PM tube at the frequency of no fluorescence is actually ~3 times larger than that of the MPPC at a similar gain, e.g. at T=5 °C and V_bias=71.6 V, due to the low operation temperature of the MPPC which causes the reduction of the dark noise. Note also that the average level of the signal back ground increases as increasing the gain of the device due to ambient photons.

We finally investigate the bias voltage dependence of the effective signal amplitude V^MPPC_eff corresponding to the ¹⁷⁴Yb isotope for different values of MPPC operation temperature as shown in Fig. 4. Error bars include the statistical standard deviation and atom number fluctuation due to oven temperature instability (±1 °C) during the experiment. For comparison, the PM tube signal V^PM_eff at the gain of 1×10⁵ is also shown in Fig. 4. As the operation temperature increases the signal amplitude decreases due to the decrease of the gain at the constant bias voltage, which means that the generated number of electrons in the Geiger mode decreases caused by the reduced lattice vibration in the Si crystal. Therefore, the probability that carriers may strike the crystal also decreases before the accelerated carrier energy has become large enough, and make it difficult for ionization to occur [1, 2, 3, 4, 11, 12]. Moreover, as the temperature decreases, the gain at a fixed reverse bias increases, which is confirmed by using the Eq. (3) and Fig. 4 in a full agreement with the MPPC characteristics at the single photon level measured in Refs. [4, 11, 12], showing that the dynamic range of our temperature-stabilized Si-based MPPC module is over 10⁶ for single photon detection by 1,600 APD pixels. We expect that the dynamic range for photon detection of the MPPC may increase in principle up to 10 ¹¹, which is limited by pixel numbers and quenching time of each pixel. From Fig. 4 we can also infer the temperature dependence of the breakdown voltage V_BD over which the avalanche of electron bunches occurs. V_BD decreases as the temperature decreases due to the same reason as the gain increase when the operation temperature decreases, for example from V_BD=69.2 V at T=25 °C to V_BD=68.6 V at T=5 °C.

 

Fig. 3. (Color online). Effective signal amplitudes per unit detector area V_eff measured by the MPPC module and the PM tube without average versus frequency detuning for various operating parameters (see text). Each fluorescence peak corresponds to the Yb isotopes and for the fermionic isotopes the quantum number F due to the hyperfine structure are shown in parenthesis.

Download Full Size | PDF

3. Conclusions

In conclusion, we have presented a temperature-stabilized Si-based multi-pixel photon counter with 1600 pixels in a 1-mm² area, of which temperature could be lowered down to 5 °C, at the level of incident photon numbers over 1×10⁶/s at 398.9 nm. The temperature dependence of the effective signal amplitudes as a function of bias voltage and temperature was measured and compared to that of the PM tube with known gain and showed a 25-fold larger detection efficiency per unit area and a dynamic range of over 10⁶. The noble Si-based multi-pixel photon counter thus can not only be useful for the effective single photon detection [1, 2, 3, 4, 11, 12], but also for the effective fluorescence detection with large incident photons, where the detected signal increases linearly depending on the incident photon numbers in principle up to 10¹¹ photons/s. We anticipate thus that the MPPC would be an essential device in varies fields of quantum optics and atomic spectroscopy experiments [5, 6, 7].

 

Fig. 4. (Color online). Bias voltage dependence of the effective signal amplitude V^MPPC_eff of the MPPC corresponding to the ¹⁷⁴Yb isotope for different values of operation temperature between 25 °C and 5 °C in a 5 °C step. Error bars include the statistical standard deviation and atom number fluctuation due to oven temperature stability. For comparison, the PM tube signal V^PM_eff at the gain of 1×10⁵ is also shown. Lines between data points are for a guideline only.

Download Full Size | PDF