Re: [Rollei] selenium/silicon cells

["Gene Johnson" <genej2 >]

Wow,

Emmanuel, I bow humbly in your general direction.  I will save this message
for that day when I can comprehend it more fully!

Gene
- ----- Original Message -----
From: <bigler  
To: <rollei  
Sent: Wednesday, November 06, 2002 2:14 AM
Subject: Re: [Rollei] selenium/silicon cells


> From Siu Fai:
>
> >..my limited knowledge in solid state physics does not explain the
> >logarithmic relationship.characteristics
>
> No need of a complex solid ste physics course to understand this if
> you admit the equivalent electrical model.
>
> A good tutorial :
>
> http://usa.hamamatsu.com/cmp-detectors/photodiodes/diodes.htm#Equivalent
Circuit
>
>
> Well I would not be surprised if a selenium cell has a current-voltage
> characteristics under various light fluxes similar to a silicon
> photodiode. After all a selenium cell is in fact a metal-semiconductor
> junction.
>
> For precise photometric measurements, silicon photodiodes are usually
> reverse-polarized, then you get an extremely linear relationship in a
> range of 1:10,000 beween the reverse current and the light flux.
> Reverse polarized, the cell has an extremely high impedance. Used
> un-polarized, as a generator, you get a solar cell with a non linear
> current vs light response except for very small series resistors. I do
> not know for selenium cells, but the simplest model for the silicon
> photodiode is simply an exponential curve for which the zero asymptote
> is shifted downward linearly with respect to the incident light flux.
>
> I = Is*(exp(eV/kT) -1) - Iflux
>
> where Iflux is proportional to the incident light flux :
> Iflux ~ k*flux, where k is a sensitivity constant that varies with the
> incident wavelength.
>
> Reverse polarized V<0 you get very quickly, even for small values of V:
>
> I = -Is -Iflux, but since Is is extremely small (few nA), eventually
> this reduces to I ~ -Iflux, hence the extremily good linearity of the
> reverse current vs the incident flux.
>
> Used as a generator V>0 you get the non linear behavior. With a null
> series resistor, V=0, you get I = -Iflux and a linear behavior. Well
> impossible to actually get V=0 with a classical am-meter readout, but
> connected to an op-amp, you can do it "virtually" with a so-called
> current-to-voltage converter.
>
> For a non-zero series resistor, you can solve graphically the current
> vs photon flux if you assume that Iflux = k*flux. For an infinite
> series resistor or connected to an op-amp mouned as a "voltage
> follower", I = 0, you get the logarithmic response :
>
> V = (kT/e) Log (1+Iflux/Is).
>
> So for any series resistor you get a behavior which is anywhere
> in-between a linear and a logarithmic response : this makes a lot a
> various behaviors possible !! With the classical am-meter readout, you
> are always non linear since there is always an equivalent resistor for
> the coil (not speaking of the "critical resistor" required for proper
> damping of the needle).
>
> This explains very simply that, at least for the silicon photodiode,
> whether you add a small or a large series resistor you get either a
> linear or log response in a V-I diagram.
>
> I would assume that a selenium cell behaves quite similarly although
> the apparent "universality" of the function eV/kT makes difficult to
> believe that the same expression would be true for all kinds of
> junctions whether they are PN silicon, PN germanium of
> semiconductor-metal.
>
> --
> Emmanuel BIGLER
> <bigler  

------------------------------