Strange bipolar current pulses in PIN diode with beta source

The other day I decided to start working on the MIP calibration of the TCT and so I installed one of the PIN diodes we have. To do the MIP calibration basically we have to compare the collected charge when the device is excited by a MIP particle, e.g. a beta electron, and the collected charge when the device is excited by the laser in the TCT. Hopefully I will soon write a detailed post on this procedure, but today it is not what I want to write about.

The strange positive pulses

When I started collecting signals coming out of the PIN diode and exciting it with beta electrons from a Sr-90 source, I noticed something strange that I had not seen before with any of the LGADs I have been playing with. So here there is a picture of the setup:

Picture of the radioactive source (gray cylinder) on top of the PIN diode (below the protective metallic square cap).

We can see the radioactive source on top of the board in which the PIN diode is installed. Everything in the setup is very standard compared to many measurements I have made before with LGADs instead of PIN diodes. So I already have some experience with this on how to connect things, how to place them, how to set up the oscilloscope, etc.

So I proceeded with the acquisition of events and I got this:

After seeing these signals I was quite surprised, because I was expecting only negative pulses, as usual. However, as can be seen, there are also positive pulses.

How can there be positive pulses together with negative pulses coming out of the same device?!?!?!?!

The way a charged particle produces a pulse of current in a silicon detector is depicted below:

Illustration of how a charged particle produces a pulse of current in a semiconductor detector.

The silicon device is polarized in such a way that there is a uniform electric field. When the particle passes through the silicon, it produces e⁻h⁺ (electron-hole) pairs and the electric field does not allow them to recombine. Instead they drift towards the terminals and they produce a current that we measure. In the depicted example above the current would be negative. Sow, how is it possible that a MIP particle can produce a current in the opposite direction? Are the e⁻h⁺ pairs drifting in the opposite direction than that ruled by the electric field? Is dark energy behind this?

An explanation for the positive pulses

Fortunately there was a simpler explanation pointed out by my colleague Riccardo del Burgo: The positive peaks may be due to carriers collected by the guard ring instead of the readout pad.

If we go into more detail our electrical circuit actually look like this:

The whole (simplified) circuit taking into account the silicon detector, the circuit implemented in the board (see here), the oscilloscope and the bias voltage source.

When the particle produces the e⁻h⁺ pairs they drift towards the surface and the oscilloscope measures the current through the resistor, as shown. Note that this current will be negative, as things were drawn. So we expect to see a negative pulse of voltage in the oscilloscope. Below there is a microscope picture of the PIN diode showing the guard ring and the readout pad. There can also be seen the two wire bondings connecting them.

Microscope picture of the PIN diode.

So, how can the oscilloscope measure a current following in the opposite direction? Consider a particle hitting the detector in the guard ring, as shown below (this is totally possible since the particles from the beta source are emitted in random directions):

In this case the charge carriers will be collected by the guard ring instead of the readout pad. In principle, and this is just circuit analysis, this will cause the current to flow like this:

My understanding of the positive current pulses.

Now we take into consideration the following facts:

  • The cables that connect the bias voltage source (the cables through which I2 goes) are quite long, I would say more than 1 meter long. This creates some parasitic resistance/inductance that affect to I2.
  • There is a certain parasitic capacitance between the readout pad and the bottom electrode (I draw it in light blue), and I guess it should be the dominant capacitance in the device given its dimensions (almost 1 mm² the surface of the readout pad and around 50 µm height).

Before the impact of the MIP particle the parasitic capacitor is charged by the bias voltage source. When the MIP impinges, the current I1 is established and it will induce currents I2 and I3. Because I2 is subjected to long cables it is a “slow current”. For I3, in turn, it is the opposite situation: The parasitic capacitor is already charged and ready to discharge, and it is “very close to the action” (in fact it is inside the device itself). Thus, I3 wins and this is what the oscilloscope sees as a positive current pulse.

It is true that if this is the explanation, there should be a negative pulse following the positive one to charge back the parasitic capacitor within the device. And this current is not seen in the oscilloscope (see the screenshots before). Maybe the re-charge of the parasitic capacitor is slow enough that it is not appreciable in the time scale I was measuring? I don’t know.

Verifying the hypothesis using the TCT

If the previous explanation is correct, then shining the TCT close to the guard ring should produce a similar effect. Thus, we should expect something like this:

Expected type of pulses by shining the TCT pulsed laser in different regions of the PIN diode, according to the previous explanation.

The blue square in the region number 1 is a window in the metal such that the light from the laser can reach the silicon. So if the TCT is shined through this window be are basically producing the carriers in the middle of the readout pad and the current through R should be negative. If the laser is shined in the region number 3 the electrons will be collected by the guard ring and the current should be positive. In region 2 something in between should happen.

Since the device was already installed in the TCT, I just moved the stages to those regions and observed how the signal looked like. The results are below:

This is in agreement with Riccardo’s observation, so the origin of the positive peaks is now understood!

Why this does not happen with LGAD devices?

Well, it happens. But if so, why have I never observed these positive peaks while working with LGADs before? The reason has to do with the internal gain of the LGADs. I repeated the process of shining the TCT in the different regions but now with an LGAD, and the results were these:

As can be seen the region 1 has a “big” negative pulse. In region 2 the pulse is still negative, but since it is in the edge of the gain layer there is almost no gain. In region 3 there is a positive pulse, indeed, but it is affected by no gain at all. Thus it is much smaller than pulses in region 1.

When using an LGAD with the beta source this happens in the same way. The small positive peaks, thus, are just hidden by the noise and never seen.