Particulars, Advanced measurement System



In this document we will try to give you an insight on how our TCT system is constructed. This shall give you some basis in your decision what particular configuration do you want to purchase from us. Majority of our customers decide to do it the easy way (buying a full system from us) but there are some that are a bit adventurous and want to do it the hard and in some cases more expensive way (buying just bits and parts from us and constructing the whole apparatus by themselves). Whatever you decide to do is fine whit us and I can assure you that we will try to help as much as possible. After all we at Particulars are just a bunch of scientists and it goes without saying that our designs are completely open for everybody to copy.


So how does a TCT look like? The basic design can be seen in the following picture.


As you can see this is a pretty classical approach. We supply HV to the sample via Bias-T which keeps the line between DUT and AMP tuned to 50 Ω over wide frequency range. Since the amount of charge in the detector is ferly large one can use a rather noisy wide band amplifier. In this way not only the charge collected but also current signal shape can be measured from which all sorts of parameters including field shape, carrier velocities, trapping times can be deduced. The ability to focus the laser beam and to move the device under test makes life even more interesting as you can make your measurement position sensitive.


We have seen different approaches in the past. They vary on how to deliver HV to the sample, some use charge sensitive amps etc. However based on our experience the above described approach gives the most informative dataset which can be later on analyzed in one or another way.



The next picture illustrates how does it look in real life.


So let us take a look on each individual component:


1.)  LASER


Our light source comes in two slightly different configurations. In the so-called "open" (top left) one the light from the laser diode is simply collimated before it is released. We use this laser in non-focusable application and in our educational set-up. In the case of the "pigtailed" configuration the light is fed into a single mode laser (core diameter around 6 μm). Since we use this on for the focusable system the core diameter is quite important since it influences the spot size at the end.


The core of the laser we produce is a ferly innovative driver, which produces relatively fast current pulses. As you can see from the attached image the light pulse delivered from it can go as low as 400 ps FWHM.



The laser can be operated like a normal device meaning that with each laser lease it delivers a trigger out signal. One can also trigger the lease with an external trigger pulse.


However to give a device a bit of versatility we have attached a LPC 1768 ARM Cortex microcontroller to the laser control board which can be programed by the computer via USB with a relatively simple user interface This gives the user the possibility to:


         change pulse duration from 400 ps to 4000 ps

         change pulse frequency from 50 Hz to 1 MHz

         program a desired sequence of pulses (1024 bit)


The last option is illustrated in the following picture where a train of "pump" laser pulses (used to change the state of electron system in the device under test) is followed by a "probe" pulse, which measures the response. "Probe" pulse is of course accompanied with trigger out signal to synchronize the data acquisition.



Our driver is used with variety of laser diodes ranging from 1 mW to 100 mW CV power. We now use 640 nm and 1064 nm diodes as those two wavelengths play a major role in TCT measurements. While 640 nm light pulse deposits all of its energy on the surface (penetration length ~ 3 μm) the 1064 nm one produces charge in detector in a minimum ionizing particle like fashion.







The particulars AM-01 amp is a RF amplifier with a (0.01-2000) MHz bandwidth. The amplification can be set to 35 dBm and 55 dBm. While it is advised for this to be done before shipment a skillful engineer can by all means do it itself.


When designing it we have placed a particular attention to a low frequency margin, a band widely ignored by RF engineers since it plays no role in RF signal transmission. However if a Fourier analysis is performed on a typical detector response one can see that low frequency part of the spectrum is widely populated. Using an amp with low cut-off in the order of 100 KHz or above can therefore result in signal differentiation and an appearance of the undershot in the signal, which is certain circumstances unavoidable, but at least one must try to make it as low as possible.



3.)  BIAS-T


BT-01 is a standard three port network used for setting DC bias point of device under test while not disturbing the rest of the RF read-out chain. It has a 0.01 MHz to 2000 MHz bandwidth with a reasonable 50 Ω impedance match over the entire band.


There are however two, from the TCT measurement point of view, important differences from the other devices readily available on the market. It can withstand DC voltages up to 2000 V. The fact that this kind of voltage span is rarely (if at all) used in RF devices explains the lack of this kind of Bias-Ts on the market. Secondly it has a leakage current in the order of few nA. In the RF field current bleeding in the range of μA is disregarded, as it is irrelevant for the rest of the devices in chain. However if one wants to measure I-V curve while performing a TCT voltage scan current leakage in the Bias-T must be kept low.





F-01 is a higher order low pass filter. When placed between DC source and DC input of the Bias-T it strongly suppresses the high frequency interferences that are normally superimposed on a DC voltage that comes from a voltage generator. DC voltage sources (no matter the price no matter the brand) normally suffer from serious pollution of DC voltage with random RF interference (our credits to honorable exceptions to this rule like Wentzel electronic). Since the voltage is connected directly to the DUT a path for the RF interferences is open trough bias-T to the amp disturbing in this way the measurement. It is therefore essential to suppress them at the source.







Cooler consists of water cooled block and a Peltier element. If one circulates the water through the block the back plane of the Peltier is kept at relatively low and constant temperature allowing the active element to cool the DUT down to -20oC. Heating up to 80oC is also possible in case things like accelerated annealing is needed.






The software for Data Acquisition with Particulars Scanning TCT gives user a possibility to take and store large amount of waveforms usually associated with scanning TCT technique. Together with a laser control software (and a separate temperature controller if it requires a computer control at all) it gives the user a total control of the system. Automatic variations of the parameters like position, bias voltage and temperature etc. creates a multidimensional matrix of acquired waveforms that can be analyzed off-line.


Details about the package can be found at:


But having a nice data set will not improve your publication rating all by it self. Sooner or later you will have to do some analysis. To help you with that we have created a ROOT based analysis library for TCT measurements called TCTAnalyse. TCTAnalyse is a shared library (.dll under Windows .sl under Linux), which is dedicated to analysis of the TCT measurements taken with Particulars TCT setups. It is based on ROOT in the sense that it relays heavily on its class libraries for all aspects of operation (visualization, IO, user interface...). The class library can be used to build executable code, but the primary use is within the root interpreter (CINT), where programs are executed in the form of macros.


TCTAnalyse Referenc Guide as well as download links can be found at:





As an example a functional diagram of single column of a 3D detector (column size 10 μm, pitch 80μm) is illustrated on the right where the color of each pixel represent the charge collected from this particular point. Needless to say this is just one of the possibilities of data representation in TCTAnalyse.