Detection

1D - Detection

1D-detection was first done with a photo-multiplier Hamamatsu H5783P.

Later, a new hybrid detector connecting a semiconductor and a standard photo-multiplier device from Becker and Hickl (HPM100-40) [Hic] was inte-grated in the setup.

Detectors, especially photo-multipliers, using accelerated photo electrons to reach a sufficiently high signal level, exhibit a characteristic response func-tion. As the used B&H ToF-card registrates 1024 time slots corresponding to 23,3 ns in most of the experiments, one channel represents 22 ps. So, the pulses from the fs-laser can be treated as being delta-like for the card. Us-ing the Time-of-Flight setup without a sample and a highly attenuated laser pulse, one can therefore easily measure the response function of the whole

system, including the response of the detector as well as the response of the photodiode. This response function needs also to be known for deconvoluting the data of a sample, so that the curves can be compared to the theoretical prediction being calculated assuming a delta-like input pulse.

0 2 4 6 8 1 0 1 2 1 4

1 E - 6 1 E - 5 1 E - 4 1 E - 3 0 , 0 1 0 , 1

1

counts (normalised)

t i m e ( n s )

P M H 5 7 8 3 P ( H a m a m a t s u )

h y b r i d P M ( B & H )

Figure 4.4: Left-hand side: comparison of the response of the two 1D photon detection devices used in the experiments. The pulses are shifted for better visibility of the pulse edges. Right-hand side: the upper part shows the direct view onto the detector area through the opening for the sample holder and the housing of the Hamamatsu photomultiplier, the lower part shows the BH-HPM100-40 with the tube as holding fixture for the sample holder.

Fig. 4.4 shows the measured response curves for both detectors. One can see a typical second ”nose” in the semi-log plot for the Hamamatsu photo-multiplier, which makes deconvolution of the data with the response to the laser pulse complicated, since it increases noise and is cause for artifacts. The hybrid detector shows no second nose and a smaller width of the response.

Using the new detector, the data can be deconvoluted much better and with less noise. Additionally, this detector is more sensitive to wavelengths larger than 600 nm as shown in fig. 4.5.

Fitting the pulse shapes with a Gaussian, one can extract a FWHM of 0,18 ns for the Hamamatsu photomultiplier and 0,11 ns for the BH-HPM100-40. The overall time resolution cannot be better than the response width of the setup to the reference pulse. When deconvoluting the data with the pulse shape, artificial high frequency noise is generated. We eliminate this noise by applying a low pass filter, which can equivalently be described as applying a moving average over the data. Unless stated otherwise, all 1D ToF data sets shown in this work are filtered with a moving average over 8 channels, which is the FWHM of the pulse width of the Hamamatsu photomultiplier.

Because of the measurement technology used, one has to make sure, that

4.1 Time-of-Flight Setups Setups

HPM-100-40

Absence of afterpulsing improves dynamic range of fluorescence decay measurements

Left: Fluorescence decay recorded with conventional PMT. The background is dominated by afterpulsing. Middle: The only source of background in the HPM is thermal emission of the photocathode. The dynamic range is substantially increased. Right: The lower background yields improved lifetime accuracy and lifetime contrast in FLIM measurements.

Fluorescence correlation measurements are free of afterpulsing peak

Left: Autocorrelation of continuous light signal of 10 kHz count rate, conventional GaAsP PMT. Middle: Autocorrelation of continuous light signal of 10 kHz count rate, HPM-100 module. The curve is flat down to the dead time of the TCSPC module. Right: FCS curve of fluorescein solution, HPM-100 module. The red curve is a fit with one triplet time and one diffusion time. bh DCS-120 confocal FLIM system, laser 473 nm.

Dark count rate vs. temperature Detection quantum efficiency vs. wavelength Typical values and range of variation APD voltage 95% of maximum

10 20

Wavelength Range 300 nm to 730 nm

Detector Quantum efficiency, at 500 nm 45%

Dark Count rate, Tcase = 22°C 560 s^-1

Cathode Diameter 3 mm

TCSPC IRF width (Transit Time Spread) 120 ps, FWHM

Single Electron Response Width 850 ps, FWHM

Single Electron Response Amplitude 50 mV, Vapd 95% of Vmax

Output Polarity negative

Output Impedance 50 Ω

Max. Count Rate (Continuous) > 10 MHz

Overload shutdown at >15 MHz

Detector Signal Output Connector SMA

Power Supply (from DCC-100 Card) + 12 V, +5 V, -12V

Dimensions (width x height x depth) 60 mm x 90 mm x 170 mm

Optical Adapters C-Mount, DCS-120, LSM 710 NDD port

HPM-100 Conventional PMT

DCS-120 with HPM-100

Conventional PMT _HPM-100

HPM-100

Figure 4.5: Spectral response of the two detectors used. Left-hand side:

H5783P (None Type) taken from [Hama]. Right-hand side: BH-HPM100-40 taken from [Hic].

for each pulse impinging onto the sample, at most one photon is emitted from the detected surface of the sample, because if there were two photons, only the first one would be detected and therefore counted to the overall time distribution, leading to a bias for shorter times corresponding to shorter path lengths. To avoid this effect, the usual count rate of the detection unit should be lower than 1/10 of the incident pulse rate. On the other hand the count rate should be sufficiently high to keep a good signal-to-noise ratio, because the dark count of any detector is a fixed rate independent of the illuminated intensity under normal circumstances. Typically, we try to reach count rates of the order of 3·10⁴Hz for the 1D-Time-of-Flight experiments whereas the typical repetition rate of the laser sources are in the range of several 10⁶Hz, therefore ensuring, that roughly each 100^th pulse will lead to a measurable event. The count rate should, in any case, not sink below 10³Hz to maintain a reasonable signal-to-noise ratio.

2D - Detection

While the 1D time-of-flight allows for a very accurate intensity profile of the overall transmitted photons over almost five decades, the 2D-setup can be used to directly observe the photon cloud expansion on the exit surface of the samples. In order to achieve time resolved 2D single-photon detection, a commercially available system from LaVision is used. It consists of an

ultra-focusing lens

sample

High Rate Intensifier CCD camera

lens

pulsed beam

Figure 4.6: Sketch of the setup for the 2D Time-of-Flight, using the fs-system as light source. The HRI works as image intensifier and ultrafast shutter with gating time of 1 ns and synchronization step size of 250 ps.

fast switchable high rate intensifier (HRI - PicoStar) with GaAsP as active photo cathode and a Peltier-cooled 16 bit CCD camera with 512 x 512 pixels for detecting the intensified image of the fluorescent screen of the HRI. The HRI works as a single photon intensifier with a maximal quantum efficiency of 40,6 % at 590 nm and can be gated with an opening time down to 1 ns, synchronizable to the laser pulse with a time step size of 250 ps. In contrast to the 1D detection, it would be possibly hazardous to the HRI to measure a direct laser pulse for reference reasons, as the risk of damaging the intensifier tube is high if only a very small region of the photo cathode is illuminated, thus making deconvolution of the recorded pictures impossible to do. The setup including the fs-laser system is sketched in fig. 4.6.

The laser beam is focused onto the sample surface with a lens (f = 200 mm). The backside of the sample is imaged onto the photo cathode of the HRI via a reversed Schneider-Kreuznach 25 mm lens with k = 0,8.

The HRI then intensifies the image with a maximal gain of 36000 during its gating time. The 16 bit CCD camera integrates the intensified pictures for a fixed exposure time over 400 ms (representing roughly 2.5 million experiment runs). In contrast to the 1D experiment, where the Time-of-Flight distribu-tion is measured, here the arrival time of the photons is selected by the HRI, thus we do not have to worry about any biasing effects as we had to take care for in the 1D setup. This means, as long as we do not drive the HRI tube into saturation, we can use the maximal available laser power without attenuation.

A time series of these pictures indeed proves that we can watch the photon cloud, having diffused through a multiple scattering slab, increasing over time as being depicted in fig. 4.7.

For measuring the magnification of the imaging system, a picture of an illuminated mm-graph paper, put at the sample’s back position, is taken and evaluated.

4.1 Time-of-Flight Setups Setups

Figure 4.7: Series of pictures taken with the camera (from left to right, from top to bottom) of a weakly localizing R104 sample. The pictures show how the light cloud exiting a multiple scattering sample evolves. The time difference between two consecutive pictures is 1 ns. Color coding is blue for low intensity, red for high intensity, but is normalized to the total intensity for each picture as can be seen in the noise in the pictures later in time.

Stop Signal Detection

In the 1D experiments described in this thesis, a 2%/98%-beam splitter is used for generating the synced stop signal: the 98%-beam is used for illumi-nating the sample, whereas the 2%-beam is sent to a fast photodiode S5972 from Hamamatsu [Hamb] according to fig. 4.1. The signal of the photodiode is then led over a long delay line to the sync connector of the Time-of-Flight

measuring card. This system can be used with any pulsed laser system, because the stop signal is directly extracted from the laser beam.

In contrast, the 2D setup with the HRI gets its laser synced signal in-directly by the TTL trigger output of the control electronics of the pulse picker, which is used in the fs-laser-system. The pulse picker again gets his sync signal by a fast photo diode which is integrated in the TiSa laser. Since the needed laser power can only be provided by the fs-system this is no draw-back. In principle, by converting the output of a photo diode signal to TTL signals, one could also use any other kind of pulsed laser system.