Higher Gain for PMT: Does it Really Improve SNR?

[xxfzero]

Empirically, photomultiplier tube (PMT) is said to present better signal-to-noise ratio (SNR) than avalanche photodiode(APD). Some materials say that it is attributed to the higher gain and lower dark current of PMT. The benefit of higher gain is confusive for me, since I suppose that noise (of incident photon and from PMT detector) and signal are simultaneously amplified by gain, then why higher gain helps to better SNR?

 

[marcusl]

You are assuming that the gain process adds no noise of its own. That is never true in amplifying devices.

 

[tech99]

My understanding has always been that the electron multiplier action, utilising secondary emission, provides noise-free gain. I suppose the dynodes are cold and so do not emit thermionic electrons. It is sometimes said that individual photons can be detected with a PMT.

 

[ZapperZ]

Empirically, photomultiplier tube (PMT) is said to present better signal-to-noise ratio (SNR) than avalanche photodiode(APD). Some materials say that it is attributed to the higher gain and lower dark current of PMT. The benefit of higher gain is confusive for me, since I suppose that noise (of incident photon and from PMT detector) and signal are simultaneously amplified by gain, then why higher gain helps to better SNR?

This is a bit puzzling. The incident photon is NOT the noise. In fact, that is the SIGNAL. PMT'S are used to detect photons, and some are sensitive all the way down to single-photon level.

The "gain" in a PMT is not simply due to an applied potential across the dynodes. Rather, it is the amount of secondary emission being given off upon each detection or signal. Thus, you can get a higher gain without increasing thermal or field emsision within the PMT.

Zz.

 

[xxfzero]

You are assuming that the gain process adds no noise of its own. That is never true in amplifying devices.

Thank you, Marcusl. Are you talking about excess noise (ratio) of gain of either PMT or APD? Indeed, some material (maybe from website of Hamamatsu Co. , I remember) mentioned that PMT wins in this aspect. But this seems to be a put-in-number problem, while I want to firstly get a qualitative picture of relation between SNR and gain, so I avoid this topic.

 

[xxfzero]

This is a bit puzzling. The incident photon is NOT the noise. In fact, that is the SIGNAL. PMT'S are used to detect photons, and some are sensitive all the way down to single-photon level.

The "gain" in a PMT is not simply due to an applied potential across the dynodes. Rather, it is the amount of secondary emission being given off upon each detection or signal. Thus, you can get a higher gain without increasing thermal or field emsision within the PMT.

Zz.

Thank you ZapperZ. Maybe my expression on noise was unclear. Actually, for simplicity, I consider two kinds of noise: 1) shot noise of incident photon, meaning that even the power of incident light is exactly constant, the number of photon still fluctuates over time, 2) dark durrent of PMT.

If I understand correctly, your explanation means dark current does not increase when applying higher gain. My confusion is exactly on this. Since I consider that electrons due to thermal or field emission (dark current, when no light incidents) from cathode will also be amplified, in other words, generation of secondary electron does not discriminate whether the origin is signal photoemission or thermal/field emission electron, then why gain is said to amplify signal but not noise?

 

[xxfzero]

My understanding has always been that the electron multiplier action, utilising secondary emission, provides noise-free gain. I suppose the dynodes are cold and so do not emit thermionic electrons. It is sometimes said that individual photons can be detected with a PMT.

Thank you, tech99. My confusion is that electrons due to thermal or field emission (dark current, when no light incidents) from cathode will also be amplified, in other words, generation of secondary electron does not discriminate whether the origin is signal photoemission or thermal/field emission electron, then why gain is said to amplify signal but not noise?

 

[ZapperZ]

Thank you ZapperZ. Maybe my expression on noise was unclear. Actually, for simplicity, I consider two kinds of noise: 1) shot noise of incident photon, meaning that even the power of incident light is exactly constant, the number of photon still fluctuates over time, 2) dark durrent of PMT.

If I understand correctly, your explanation means dark current does not increase when applying higher gain. My confusion is exactly on this. Since I consider that electrons due to thermal or field emission (dark current, when no light incidents) from cathode will also be amplified, in other words, generation of secondary electron does not discriminate whether the origin is signal photoemission or thermal/field emission electron, then why gain is said to amplify signal but not noise?

The "gain" is simply amplification of signal per incoming photon. If I use a material with higher secondary emission coefficient, I can increase the amplification WITHOUT increasing the applied voltage. Thus, I did not increase the field-emission current.

Zz.

 

[xxfzero]

The "gain" is simply amplification of signal per incoming photon. If I use a material with higher secondary emission coefficient, I can increase the amplification WITHOUT increasing the applied voltage. Thus, I did not increase the field-emission current.

Zz.

Could you please give a qualitative picture why PMT has a better SNR, with respect to APD? Many thanks.

 

[ZapperZ]

Could you please give a qualitative picture why PMT has a better SNR, with respect to APD? Many thanks.

This is now a different question than you asked in the beginning, and that I was addressing, because you originally asked for why an increase in gain in PMT does not correspond to an increase in noise. My answers were limited to just that.

APD is limited in some sense to the number of electron-hole pairs that can be generated within a semiconductor (in the PN junction). PMT has no such limitation. There are materials that can create 10's of secondary electrons for each incident electron, depending on the incident electron's energy. Those secondary electrons then turn around and bombard the material again to subsequently create even more secondary electrons.

So if you want a higher gain without increasing the internal voltage of the device, coat the dynodes with material that has a higher secondary electron emissivity.

Zz.

 

[tech99]

Thank you, tech99. My confusion is that electrons due to thermal or field emission (dark current, when no light incidents) from cathode will also be amplified, in other words, generation of secondary electron does not discriminate whether the origin is signal photoemission or thermal/field emission electron, then why gain is said to amplify signal but not noise?

Yes, I agree with you that dark current will be amplified the same as the signal, but my understanding is that the amplification process does not add additional noise.

 

[xxfzero]

This is now a different question than you asked in the beginning, and that I was addressing, because you originally asked for why an increase in gain in PMT does not correspond to an increase in noise. My answers were limited to just that.

APD is limited in some sense to the number of electron-hole pairs that can be generated within a semiconductor (in the PN junction). PMT has no such limitation. There are materials that can create 10's of secondary electrons for each incident electron, depending on the incident electron's energy. Those secondary electrons then turn around and bombard the material again to subsequently create even more secondary electrons.

So if you want a higher gain without increasing the internal voltage of the device, coat the dynodes with material that has a higher secondary electron emissivity.

Zz.

I can understand what you addressed that it is possible for PMT to employ higher gain without increasing applied voltage, hence without increasing dark current. Meanwhile for APD, this is limited by numbers of electron-hole pairs that can be generated.

I find one ambiguity in our discussion is probably "origin" of noise due to dark current/ dark count rate. I realize that you attributed it to the dynodes of PMT. In that case, I agree with you. In previous conversation I assumed it is from cathode, with that from dynodes neglected, which I failed to clearly express, I'm sorry. This assumption is because I always use PMT in photo counting style, meaning that the dark counts from dynodes are obviously lower than that from incident photons in terms of electric pulse voltage level, hence filtered by discriminating voltage, then only the dark counts from cathode contribute into noise. That is why I considered the noise is simultaneously amplified together with signal. If I understand correctly?