Does anyone have to hand a link to post-2000 documentation for the diode model ?
For example, the poorly-named “recombination current parameter” ISR is significant in many models for commercial diodes, and describes an added current of the form
ISR { exp( e V / NR KB T) - 1} | 1 - V/VJ |**M
(found by matching Qspice, and LTspice, output curves) with NR typically near 2.
So ISR is the coefficient of a current with the voltage-dependence shape that I would expect from the contribution from thermal carrier generation within the depletion region around the junction. Having this shape is pretty important for describing the current at moderate reverse bias. (The positive-voltage shape is ugly and unphysical around VJ, but here the usual IS term dominates in typical models.)
The 1991 “SPICE Diode Model” from Howard T Russell does not address this, nor does the LTspice or Qspice help files. Intriguingly, The Qspice helpfile says “Recombination current . . supported so that conventional industry diode models run without modification” hinting that the author hesitated to support ISR for some reason.
I am currently searching through the NGspice source, hoping this component of the diode model may have appeared before NGspice and Qspice split in the spice family tree.
The term involving ISR is not in berkeley spice 3f5, nor Ngspice, nor Xyce (from Sandia National Labs in the US). It seems to have appeared first in PSpice, then other implementors supported it for the sake of compatibility.
For example, Altium make public their interface to implementation of Berkeley spice 3f5 (Diode Model | Online Documentation for Altium Products) where under “PSPICE support” they document that they accept parameters ISR and NR.
The PSPICE reference guide confirms the form (with some softening of the sharp zero at V=VJ) but gives no references to physical justification.
The most accessible link I have for explaining the relevant physics is from Prof Dr Helmut Föll in Kiel Semiconductors in the section “Generation Currents from the Space Charge Region” where the space-charge region is what I would call the depletion region. The width of the depletion region increases with reverse bias, giving the factor ( 1 - V/VJ )**M .
ISR and NR are basically PSpice-proprietary model parameters implemented so that people would make diode models that only ran in PSpice in the interest of selling more PSpice licenses. QSPICE needs to have the model parameter implemented in order to be able to run 3rd party models, but I rarely use them to author SPICE diode models myself. The equations are documented in the 2nd Edition of Semiconductor Device Modeling with SPICE by Giuseppe Massobrio and Paolo Antognetti, McGraw Hill, 1993.
SPICEdiodeModel.pdf is the 1991 paper by Howard T. Russell, where Fig. 3.6 illustrates a discrepancy between SPICE at the time and a real diode. The discrepancy in ‘region I’ of forward voltage between 0 and 100 mV is filled in by the term involving ISR and NR.
Russell’s ‘Spice Diode Model’ Fig. 3.7 illustrates a larger discrepancy with reverse bias. The where the simple diode equation gives current that quickly saturates at −Is. But the depletion region grows with stronger reverse voltage, so the contribution from electron-hole pairs generated in the depletion region increases (moderately) with reverse bias.
Just the word ‘recombination’ used to describe the ISR component confused me, because in an I-V curve it shows up mostly in reverse bias where that component corresponds physically to generation of carriers in the depletion region.
To be fair, the PSPICE manual does refer to Grove “Physics of Semiconductor Devices” (1967) that probably does describe the physics behind the ISR term, somewhere.
A common textbook today, Sze and Lee’s “Semiconductor Devices”, describes in §3.4.2 the two components as
IS component: “diffusion current”
ISR component: “current from generation and recombination in the depletion region”
but then later when they talk about the forward-biased case they shorten to “recombination current”