Long-Pass Filter

The long-pass filter (LPF) serves to cut out visible and UV room light. We chose Spin- dler&Hoyer (Gottingen, Germany) RG9 filters of 2 mm thickness that were specially cut to disks of 6.5 mm diameter. There is one filter for each individual channel and they are located between the polymer fibre and MCP-PMT photocathode in order to minimise the detection of stray light. The mechanical holder that positions the polymer fibres and LPFs in front of the detectors is described in section 6.2.1.9 below.

Figure 6-20 shows the measured transmittance of the filter in the region 650-1000 nm. The cut-off (defined here as 50% of the maximum transmittance) is at À^dge ~ 720 nm, and the transmittance is T^a^ ~ 90% at 800 nm and relatively fiat throughout the NIR region of interest. Stray light at wavelengths beyond -870 nm is effectively eliminated by a rapid drop in the MCP-PMT’s quantum efficiency. The filter has been demonstrated to be very effective in a typical laboratory environment, even with bright fluorescent room light present. However, significantly more fR stray radiation is likely to be encountered in a neonatal intensive care unit.

§

0.5 --

650 700 750 800 850 900 950 1000

W avelength [nm]

Figure 6 -2 0 Long-pass filter spectral transmittance.

6.2.1.9 MCP-PMT

We are using a MicroChannel Plate Photomultiplier Tube (MCP-PMT, or MCP) as the detector. It enables single photons to be counted with low dark noise and high sensitivity, and has an excellent time response compared to conventional PMTs. Another common application of MCP-PMTs is the study of fluorescent decay behaviour, where fluorescence lifetimes are very short.

MCP-PMT Basics

Figure 6-21 (a) illustrates the layout of a typical MCP-PMT. The photocathode converts incident light photons into electrons. This process, called photoemission, consists of three stages: (1) absorption of the photon and transfer of its energy to an electron within the photoemissive material, (2) migration of the electron to the surface, and (3) escape of the electron from the material’s surface.

The photon, and hence electron, energy is

he

E = hv = — (6.3)

which corresponds to -1.55 eV at /^ 8 0 0 nm. Some of this energy is lost by electron- electron collisions during the electron migration process, which is why the photocathode must be very thin. For a photoelectron to leave the surface it must have sufficient energy to overcome the inherent potential barrier (work function) of the material. All MCP-PMTs exhibit a long wavelength cut-off because of this minimum energy requirement. The lower wavelength cut-off is normally due to absorption in the window material. The quantum

efficiency of a photon detector is defined as

_ number of photoelectrons emitted ^ ^

number of incident photons

The radiant sensitivity (units mAAV) equals the photocurrent, Iph, produced per unit light

flux, Popt, and is related to the quantum efficiency via the following expression

Figure 6-21 (b) shows the radiant sensitivity for various common photocathode materials. Photoelectrons exiting the cathode are accelerated by a potential along the microchannels (or dynodes in a conventional PMT). They undergo amplification through secondary electron emission resulting from collisions with the channel walls. Many thousand very small diameter tubes (typically <10-50 p.m) each act as an independent electron multiplier. The main advantage over conventional PMT designs is the excellent timing properties. The narrow tubes minimise the Transit Time Spread (TTS) of the electrons in the same way that small diameter optical fibres minimise intermodal dispersion. And because a small TTS results in a fast rise time of the current pulse produced at the anode, this reduces the jitter in the timing discriminator, and hence improves the overall temporal resolution of the system.

1 0 0 0 t o 2 0 0 0 V ty p ic a l Secondary e le c tro n s L ig h t p h o t o n P h o t o - e l e c t r o n P h o to c a th o d e M icro C h an n el p la te s 1000 ... GLASS EXD. RED MULTIALKALI 1 -i-GLASS I iO.I% 0.01 (b) WAyELENGTH (nm) (a)

Figure 6-21 (a) General MCP-PMT layout (reproduced from [Knoll 1989]), and (b) radiant sensitivity of various common photocathode materials (reproduced from [Koyama 1988]).

Hamamatsu R411OU-05MOD MCP-PMTs

The detectors used in this project are ultrafast Hamamatsu Photonics (Hamamatsu City, Japan) R411OU-05MOD MCP-PMTs. They are custom made, and based on the R41 lOU series. We use four MCP-PMTs, each of which is segmented into 8 anodes (cf. schematic in Figure 6-7), therefore providing a total of 32 independent detector channels.

n

R adiant Sensitivity Q u antu m Efficiency

2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0

W avelength [nm]

Figure 6-22 Quantum efficiency and radiant sensitivity of a Hamamatsu R 4 11OU- 05MOD MCP-PMT (S/N ET0002; reproduced from Hamamatsu test sheet).

The photocathode is placed behind a fibre optic plate window, and is made of an extended Mnltialkali material, which ensures that it is sufficiently sensitive in the NIR wavelength region. Figure 6-22 shows the spectral response characteristics (quantum efficiency and radiant sensitivity) of one of our MCP-PMTs. It employs a two-stage dynode structure similar to the one illustrated in Figure 6-21 (a), and has very narrow 10 p.m charmels. Figure 6-23 shows a typical waveform of a single output pulse recorded by Hamamatsu while illuminating the MCP-PMT with multiple photons. The rise time, which includes the finite response of the laser source and detection electronics used in the experiment, is only -210 ps. Hamamatsu estimate the actual transit time spread of electrons in the MCP-PMT to be -45 ps.

> E OO

. 2 [ n s / d i v ]

Figure 6-23 Waveform of the MCP-PMT (S/N ET0002) recorded at V = -3280 V. Rise time = 205 ps, fall time = 317 ps, width = 390 ps. (Reproduced from Hama­ matsu test sheet). See also Figure 6-31 for a density plot of multiple MCP-PMT output pulses.

Table 6-5 summarises the most important specifications.

Model Type

Photocathode Window Dynodes

Max. Supply Voltage

Max. Anode Current (per anode)

Hamamatsu R411OU-05MOD MicroChannel Plate PMT (8 anodes) Multialkali (extended)

Fibre optic

2 stage structure, 10 qm channels -3.5 kV

50 nA (continuous) 200 mA (pulsed peak)

Spectral Response QE=3.6-4.4%, RS=23-30 mA/W (see also Figure 6-22)

Current Gain (@T=-30° C, V -3.3kV ) 10^

Rise Time -210 ps (average)

(see also Figure 6-23)

Estimated TTS -45 ps (average)

Dark Counts (@T=-30°) <100 cps (max.), <10 cps (typ.)

Cross Talk 0.25-0.5%

(full illumination of photocathode)

Table 6-5 Hamamatsu MCP-PMT specifications (averages or ranges of all MCP-PMTs, where appropriate).

Output Pulse Amplitude

It is possible to obtain an order of magnitude estimate of the pulse amplitude for single photon illumination. A single detected photon produces one photoelectron^"^. Given a gain

G, this results in G electrons reaching the anode. Hence the total charge Qtotai produced by a single photon is

Q t o t a i (6.6)

where e is the charge of a single electron. Therefore the average current lavg during a period defined by the transit time spread A trrsis

, G x e

avg ~ . (6.7)

Given the load of the coaxial output cable, Rioad, this corresponds to a voltage Vavg of

G x e “ (6.8) •"TTS For G=10' 6=1.6x10'^^ C tYT^=45 ps

this gives an estimate for the single photon signal pulse amplitude of Vavg=177 mV.

In principle more than one photoelectron can be created by a single photon. However, at NIR wavelengths the photon energy is just sufficient to overcome the work function of the photocathode material, so that a sin g le

However, this number is difficult to verify experimentally, because the finite response of the detection electronics (pre-amplifier, oscilloscope) needs to be taken into account. A measurement of the MCP-PMT pulse after the pre-amplifier stage is shown in Figure 6-31. This measurement indicates a peak voltage of up to about 1.9 V, which is reasonable given the 40 dB gain and finite response of the pre-amplifier.

Count rate

The MCP-PMT count rate is highly linear over many orders of magnitude, but levels off at very high values. This saturation effect is due to the dead time during which charges released in the electron amplification process are being replaced, as well as the finite pulse width caused by the transit time spread. An estimate of the count rate at which saturation occurs can be computed from the maximum continuous anode current lanode, which is specified to be 50 nA for a 5% deviation from linearity. Since the number of electrons at the anode Uanode equals the number of photoelectrons created at the photocathode ricathode

multiplied by the gain G, the anode current can be expressed as ^ a n o d e ^ i P ' c a t h o d e ^ ^ ^ — n ^ ^ c a th o d e

T _ a n o d e '' __ v ' ' c a t h o d e ' ' /

V

(.6.9)

Therefore the rate at which photoelectrons are created, and hence photons detected, is

I ^ c a th o d e

J

= (6.10)

G e At

which is -3x10^ counts per second (cps) for a gain of G=10^. However, this assumes uniform illumination of the whole anode area. For a 50% illumination area, as is approxi­ mately the case for our setup (see also Figure 6-26), the value is -1.5x10^ cps. Section 7.1.2 examines the effects of MCP-PMT and electronic dead times on the system in more detail.

Dark counts

Dark counts arise from spontaneous electron emissions when the thermal energy of the electrons in the photocathode is sufficient to exceed the potential barrier. As with any thermal effect, they increase exponentially with temperature. Hence cooling of the detector dramatically reduces the number of dark counts; in our case from -10^ cps at room temperature to typically ;^10 cps when cooled to -25° C (c.f. Figure 7-1 in section 7.1.1, which discusses sources of random noise in the system). We employ Hamamatsu C2773 MCP-PMT coolers that use Peltier elements to produce a 50° C temperature gradient. Packets of silica gel desiccant have been put into the sealed cooling housing to reduce

humidity and hence condensation. The excess heat is carried away by a water cooling circuit. A specially designed chiller/circulator from Thermal Exchange Ltd. (Leicester, UK) accomplishes this task, and maintains the water temperature at 20-25 °C. The custom-built Peltier element power supplies can be switched on/off by the computer, and the Peltier and cooling water temperatures 4- water flow speed are monitored.

COSMIC RAY PULSE

DARK CURRENT

PULSE_________ SIGNAL PULSE

Threshold

TIME

SIGNAL PULSE + NOISE PULSE