physics0306153/pn.tex
1: \documentclass[aps, sort&compress, longtable, square, floatfix]{revtex4}
2: %\usepackage[pdftex]{graphicx}
3: \usepackage[dvips]{graphicx}
4: \usepackage{times}
5: 
6: \begin{document}
7: 
8: \title{The Pedagogy of the p-n Junction: Diffusion or Drift?}
9: 
10: \author{S.O. Kasap}
11: \affiliation{Department of Electrical Engineering, University of Saskatchewan
12: Saskatoon S7N 0W0, Canada}
13: 
14: \author{C. Tannous}
15: \affiliation{Laboratoire de Magn\'etisme de Bretagne,  CNRS/UMR 6135,
16: 6 avenue Le Gorgeu BP:  809,  29285 Brest CEDEX,  France}
17: 
18: \begin{abstract}
19: The majority of current textbooks on device physics at the undergraduate 
20: level derive the diode equation based on the diffusion of injected minority 
21: carriers. Generally the drift of the majority carriers, or the extent of 
22: drift, is not discussed and the importance of drift in the presence of a 
23: field in the neutral regions is almost totally ignored. The assumptions of 
24: zero field in the neutral regions and conduction by minority carrier 
25: diffusion lead to a number of pedagogical problems and paradoxes for the 
26: student. The purpose of this paper is to address the 
27: pedagogical problems and paradoxes apparent in the current treatment of 
28: conduction in the pn junction as it appears in the majority of texts.
29: \end{abstract}
30: 
31: \maketitle
32: 
33: \section{INTRODUCTION AND BACKGROUND}
34: 
35: The pn junction theory forms an integral part all physical electronics 
36: courses. At both undergraduate and graduate levels, the conventional 
37: analysis of the pn junction device under forward bias conditions follows 
38: closely Shockley's original treatment [1] in which the diode equation is 
39: derived based on the injection and diffusion of minority carriers. There 
40: are, however, a number of paradoxes in the treatment of the subject matter 
41: in the present textbooks (see below) which tends to mislead the students. 
42: Under zero and forward bias conditions, the pn junction displays a number
43: of characteristic features with particular reference 
44: to carrier concentrations, exposed ionized dopants in the space charge layer 
45: (SCL) or the depletion layer, internal field, $E(x)$ in the SCL, and the 
46: built-in voltage V$_{0}$. In a typical undergraduate course the arguments 
47: used in driving the diode equation follow a sequence of simplifying assumptions:
48: 
49: (a) The depletion region or the SCL (space charge layer) has a much higher resistance than the 
50: neutral regions  so that the applied voltage 
51: drops across the depletion region. There is therefore no field in the 
52: neutral regions. 
53: 
54: (b) The applied forward bias reduces the built-in potential V$_{0}$ and allows the 
55: diffusion and hence injection of minority carriers. 
56: From assumption (a), the law of the junction gives the injected minority 
57: carrier concentration in terms of the applied voltage. For example for holes 
58: injected into the n-side,
59: 
60: \begin{equation}
61: {\rm{p}}_{\rm{n}} {\rm{(0)}} = {\rm{p}}_{{\rm{n0}}} {\rm{exp(}}\frac{{{\rm{eV}}}}{{{\rm{kT}}}}{\rm{)}}
62: \label{eq1}
63: \end{equation}
64: 
65: where p$_{n}$(0) is the hole concentration just outside the depletion 
66: region, at the origin x=0, p$_{n0}$ is the equilibrium hole 
67: concentration in the n-region, p$_{n0}$=n$_{i}^{2}$/N$_{d}$, V is the 
68: applied bias, and the other symbols have their usual meanings.
69: 
70: (c) Since the electric field in the neutral regions is assumed to be zero, 
71: the continuity equation for the minority carriers is greatly simplified and 
72: becomes analytically tractable even at the junior undergraduate level. For 
73: holes in the n-region, under steady state conditions, $\partial 
74: $p$_{n}$/$\partial $t=0 and the continuity equation is simply
75: 
76: \begin{equation}
77:  -  \frac{{\rm{1}}}{{\rm{e}}}\frac{{{\rm{dJ}}_{{\rm{hn}}} }}{{{\rm{dx}}}}  - 
78:  \frac{{{\rm{p}}_{\rm{n}} }}{{\tau _{{\rm{hn}}} }} = 0
79: \label{eq2}
80: \end{equation}
81: 
82: where J$_{hn}$ is the hole current density and $\tau _{hn}$ is the 
83: minority carrier (hole) recombination time both in the n-region. It is 
84: tacitly assumed in eq.(\ref{eq2}) that the minority carrier concentration is much 
85: less than the equilibrium majority carrier concentration so that a constant 
86: minority carrier lifetime can be defined which is independent of the 
87: majority carrier concentration, n$_{n}$. 
88: 
89: The hole current density however is simply the diffusion component as the 
90: electric field is assumed to be zero,
91: 
92: \begin{equation}
93: J_{hn}  =  - eD_{hn} \frac{{d{\rm{p}}_{\rm{n}} }}{{{\rm{dx}}}}
94: \label{eq3}
95: \end{equation}
96: 
97: where D$_{hn}$ is the diffusion coefficient of the minority carriers (holes) 
98: in the n-region. 
99: 
100: Substituting eq. (\ref{eq3}) into eq. (\ref{eq2}), leads to
101: 
102: \begin{equation}
103: D_{hn} \frac{{{\rm{d}}^{\rm{2}} {\rm{p}}_{\rm{n}} }}{{{\rm{dx}}^{\rm{2}} }} 
104:  -  \frac{{{\rm{p}}_{\rm{n}} }}{{\tau _{{\rm{hn}}} }} = 0
105: \label{eq4}
106: \end{equation}
107: 
108: and solving eq.(\ref{eq4}) for p$_{n}$ one obtains, for a long diode,
109: 
110: \begin{equation}
111: \Delta {\rm{p}}_{\rm{n}} {\rm{(x)}} = \Delta {\rm{p}}_{\rm{n}} 
112: {\rm{(0)exp(-}}\frac{{\rm{x}}}{{{\rm{L}}_{{\rm{hn}}} }}{\rm{)}}
113: \label{eq5}
114: \end{equation}
115: 
116: where $\Delta $p$_{n}$=p$_{n}$-p$_{n0}$ is the excess minority carrier 
117: concentration and x is measured from just outside the depletion region. 
118: The long diode assumption means that the length of the 
119: neutral region, l$_{n}$, is much greater than the minority carrier diffusion 
120: length, L$_{hn}$, defined as $\sqrt{D_{hn} \tau_{hn}}$. The long 
121: diode assumption however is not necessary for the derivation of the diode 
122: equation; it serves to simply the solution of eq.(\ref{eq4}) to a single 
123: exponential rather than a hyperbolic sine function.
124: 
125: With the minority carrier concentration given as in eq.(\ref{eq5}), from eq.(\ref{eq3}) 
126: the hole current density is
127: 
128: \begin{equation}
129: J_{hn} {\rm{(x)}} = \frac{{eD_{hn} \Delta {\rm{p}}_{\rm{n}} (0)}}{{{\rm{L}}_{{\rm{hn}}} }}{\rm{exp(-}}\frac{{\rm{x}}}{{{\rm{L}}_{{\rm{hn}}} }}{\rm{)}}
130: \label{eq6}
131: \end{equation}
132: 
133: There is obviously a similar minority carrier diffusion current density in 
134: the p-region for the injected electrons, i.e.
135: 
136: \begin{equation}
137: J_{ep} {\rm{(x)}} = \frac{{eD_{ep} \Delta {\rm{n}}_{\rm{p}} (0)}}{{{\rm{L}}_{{\rm{ep}}} }}{\rm{exp(-}}\frac{{{\rm{x}}}}{{{\rm{L}}_{{\rm{ep}}} }}{\rm{)}}
138: \label{eq7}
139: \end{equation}
140: 
141: where D$_{ep}$ and L$_{ep}$ are the electron diffusion coefficient and 
142: diffusion length, $\Delta $n$_{p}$ is the excess electrons concentration all 
143: in the p-region, and x' is distance measured away from the depletion region 
144: in the p-side. 
145: 
146: (d) It is assumed that the depletion region is so narrow that the currents 
147: do not vary across this region. Then the majority carrier current at x=0 is 
148: the same as the minority carrier current at x'=0. Similarly the majority 
149: carrier current at x'=0 is the same as the minority carrier current at x=0. 
150: Thus the total current density is
151: 
152: \begin{equation}
153: J = J_{hn} (0) + J_{ep} (0)
154: \label{eq8}
155: \end{equation}
156: 
157: 
158: and using the law of the junction for $\Delta $p$_{n}$(0) and $\Delta 
159: $n$_{p}$(0) from eq.\ref{eq1} one obtains the diode equation,
160: 
161: \begin{equation}
162: J = \frac{{eD_{hn} \Delta {\rm{p}}_{\rm{n}} (0)}}{{{\rm{L}}_{{\rm{hn}}} }} 
163: + \frac{{eD_{ep} \Delta {\rm{n}}_{\rm{p}} (0)}}{{{\rm{L}}_{{\rm{ep}}} }}
164: \label{eq9}
165: \end{equation}
166: 
167: or 
168: 
169: 
170: \begin{equation}
171: J = e\left[ {\frac{{D_{hn} p_{n0} }}{{{\rm{L}}_{{\rm{hn}}} }} + 
172: \frac{{D_{ep} n_{p0} }}{{{\rm{L}}_{{\rm{ep}}} }}} \right] 
173: \left( {{\rm{exp}}(\frac{{eV}}{{kT}})  -  1} \right)
174: \label{eq10}
175: \end{equation}
176: 
177: 
178: This is the general diode equation found in the majority of textbooks which 
179: follow the above sequence of steps, either through tacitly or explicitly 
180: stated assumptions in (a) to (d). Many texts, for simplicity, consider 
181: either the p$^{ + }$n or the n$^{ + }$p junction. For the p$^{ + }$n 
182: junction, $N_{A} >> N_{D}$ and  eq.(11) becomes,
183: 
184: \begin{equation}
185: J = en_i^2 \left[ {\frac{{D_{hn} }}{{{\rm{L}}_{{\rm{hn}}} N_D }} + 
186: \frac{{D_{ep} }}{{{\rm{L}}_{{\rm{ep}}} N_A }}} \right] 
187: \left( {{\rm{exp}}(\frac{{eV}}{{kT}})  -  1} \right)
188: \label{eq11}
189: \end{equation}
190: 
191: The assumption stated explicitly in (d) is commonly overlooked in many 
192: current textbooks on the subject and is one of the key assumptions in the 
193: derivation of equation (\ref{eq2}) as discussed, for example, by Moloney \cite{moloney}.
194: 
195: Equation (\ref{eq2}) gives the impression that the conduction process in the
196:  p$^{ + }$n junction diode is the diffusion of injected holes in the n-region. 
197: The above steps invariably lead to a conclusion for the student that it is the 
198: minority carrier diffusion which constitutes the forward current.
199: 
200: Close examination of the above steps in the derivation exposes a number of 
201: serious pedagogical paradoxes and problems for the student and the 
202: instructor. The diode equation in eq. \ref{eq11} is so entrenched in our teaching 
203: of the pn junction that it has been used to design many simple but fruitful 
204: laboratory experiments as reported in various journals.
205: 
206: \section{PEDAGOGICAL PROBLEMS}
207: 
208: The conventional undergraduate level treatment in Section I leads to a 
209: number of pedagogical problems and paradoxes. We cite those we have 
210: encountered frequently in two undergraduate classes during the treatment of 
211: the forward biased long diode:
212: 
213: (a) If there is no field in the neutral region then there can be no net 
214: charge at any point in this region inasmuch as $dE/dx=0$. Then the excess 
215: majority carrier concentration should follow the decay of the excess 
216: minority carrier concentration, $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x), 
217: which then follows eq. \ref{eq5}. But the gradients of $\Delta 
218: $n$_{n}$(x)=$\Delta $p$_{n}$(x) along x must be the same as well. Therefore 
219: majority carriers diffuse towards the right as well and since in silicon 
220: $D_{e} > D_{h} $, and electrons are negatively charged the net current is 
221: actually in the reverse direction. The current must be in the opposite 
222: direction to the applied voltage!
223: 
224: (b) The minority carrier current, which is due to diffusion, decays with x. 
225: Since the total current must be constant, the majority carrier current must 
226: increase with x. However, there is no field in the neutral region which 
227: means that electrons must diffuse. From the first paradox above this 
228: diffusion can not make up for the decay in the hole current. 
229: 
230: (c) Far away from the depletion region, both the hole and electron 
231: concentrations are almost uniform. If there is no electric field, then the 
232: current, due to diffusion, must vanish. How is it that the current stays 
233: constant in the neutral region (indeed in the whole device)?
234: 
235: (d) The absence of an electric field in the neutral region means that 
236: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x). But when holes are injected into 
237: the n-region they recombine with electrons so that , intuitively, the 
238: majority carrier concentration should decrease not increase.
239: 
240: The concept of field free neutral regions is so deeply rooted in the present 
241: treatment that many authors do indeed show the excess majority carrier 
242: concentration increasing towards the junction like the excess minority 
243: carrier concentration. This fact alone is contrary to student intuition 
244: that, if anything, the excess majority concentration should remain uniform. 
245: It is clear that the instructor has a responsibility to clear this paradox. 
246: It is interesting to note that a large number of authors sketch the carrier 
247: concentration profiles with the majority carrier concentration shown as 
248: uniform whereas others show the excess majority carrier profile following 
249: the excess minority carrier concentration profile as indicated in Table I 
250: for a survey of a large number of books on the subject. The differences in 
251: the diagrams only add confusion to the student's understanding. Few authors 
252: allude to the presence of the electric field and majority carrier drift to 
253: overcome some of the problems listed above. There is however no satisfactory 
254: treatment in the majority of the textbooks used in English speaking 
255: countries we have examined as illustrated in Table I. In most books the 
256: field in the neutral region is totally neglected in the treatment. The 
257: result, we believe, is pedagogical paradoxes and a student who is confused. 
258: It seems that only some of the early books published in the sixties 
259: consider the need and the effects of the field in the neutral regions in 
260: their discussion of conduction in the forward biased pn junction. 
261: 
262: \section{DIFFUSION OR DRIFT?}
263: 
264: For simplicity we consider a long p$^{ + }$n junction diode. Equation (\ref{eq2}) 
265: describes its conventional current-voltage characteristics. We use the 
266: parameters listed in Table 2 which represent "typical" parameters for a p$^{ 
267: + }$n junction Si diode albeit classroom parameters. We also assume small 
268: injection so that $p_{n0} << n_{n0}$ (or N$_{D})$. The latter assumption 
269: means that the minority carrier recombination time remains a useful 
270: parameter . For forward bias we take the voltage across the diode to be 
271: typically 0.55V. The depletion region extends essentially into the n-side 
272: and its width W, is much shorter than the hole diffusion length in this 
273: region as indicated in Table II. Similarly the relatively tiny extension of 
274: the depletion width into the p$^{ + }$region is much shorter than the 
275: electron diffusion length there. Lengths of the neutral regions are taken to 
276: be about ten times their minority carrier diffusion lengths to represent a 
277: "long-diode", i.e. l$_{n}$=10L$_{h}$ and l$_{p}$=10L$_{e}$.
278: 
279: 
280: \begin{equation}
281: J = \frac{{eD_{hn} n_i^2 }}{{{\rm{L}}_{{\rm{hn}}} N_D }}\left( 
282: {{\rm{exp}}(\frac{{eV}}{{kT}})  -  1} \right)
283: \label{eq12}
284: \end{equation}
285: 
286: 
287: The first attempt to overcome the problems listed in Section II is to allow 
288: some of the applied voltage, a small fraction of it, to drop across the 
289: neutral n-region of the p$^{ + }$-n junction. This is easily accepted by the 
290: student since the neutral regions must have some finite resistance even 
291: though much smaller than the depletion region. This means that the law of 
292: the junction remains approximately valid. What is the field in the n-region?
293: 
294: The total current through the p$^{ + }$n junction must be continuous. This 
295: means that at any point in the n-region,
296: 
297: 
298: \begin{equation}
299: J =  - eD_{hn} \frac{{d\Delta p_n }}{{dx}} + e\mu _{hn} p_n E_n  + 
300: eD_{en} \frac{{d\Delta n_n }}{{dx}} + e\mu _{en} n_n E_n =  constant   \\
301: \label{eq13}
302: \end{equation}
303: 
304: 
305: where E$_{n}$ is the field in the n-region at x. Initially E$_{n}$ is 
306: assumed to be small but finite. 
307: 
308: Since $n_{n} >> p_{n}$ , $n_{n} \approx N_{D}$ (small injection), and 
309: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) the above equation simplifies to,
310: 
311: 
312: \begin{equation}
313: J = e(D_{en}  - D_{hn} )\frac{{d\Delta p_n }}{{dx}} + e\mu _{en} N_D E_n(x) 
314: \label{eq14}
315: \end{equation}
316: 
317: The requirement of an internal field is quite transparent from eq. (\ref{eq3}). 
318: The minority concentration gradient is negative and $D_{en} > D_{hn}$ which 
319: means that the first term in eq.(\ref{eq3}) is negative so that current is in the 
320: negative direction. Unless there is an internal field drifting the majority 
321: carriers it is not possible to obtain a positive current.
322: 
323: The pedagogic development at this point must make use of the excess minority 
324: carrier concentration in eq.\ref{eq5}. If the field is indeed sufficiently small 
325: it may be assumed that the excess minority carrier profile, $\Delta 
326: $p$_{n}$(x), is still given by eq.\ref{eq5}. The validity of this assumption will 
327: be demonstrated below with an illustrative example. One can also assume that 
328: eq. (\ref{eq2}) can still be used to describe the total diode current. Then from 
329: eqs. \ref{eq5}, \ref{eq2} and \ref{eq3}, one can obtain the field E$_{n}$ in the n-region 
330: 
331: 
332: \begin{equation}
333: E_n(x) = \left( {\frac{n_i }{N_D }} \right)^2\left( {\frac{kT}{eL_{hn} }} 
334: \right)\frac{1}{b_n }\exp (\frac{eV}{kT})\left[ {1 + (b_n - 1)\exp ( - 
335: \frac{x}{L_{hn} })} \right]
336: \label{eq15}
337: \end{equation}
338: 
339: where we have used the definition b$_{n}=\mu _{en}$/$\mu _{hn}$=D$_{en}$/D$_{hn}(>1) $.
340:  Equation (\ref{eq4}) describes the field outside 
341: the SCL in the so-called neutral region that is needed to maintain the diode 
342: current. In the n-region the field increases towards the SCL. The increase in the field is 
343: required to make up for the negative electron diffusion current. Far away 
344: from the depletion region the current is maintained by a constant field of 
345: magnitude,
346: 
347: \begin{equation}
348: E_{n}(x)  = \left( {\frac{{n_i }}{{N_D }}} \right)^2 \left( {\frac{{kT}}{{eL_{hn} }}}
349:  \right)\frac{1}{{b_n }}\exp (\frac{{eV}}{{kT}})\left[ {1 +
350:  (b_n  - 1)\exp ( - \frac{x}{{L_{hn} }})} \right]
351: \label{eq16}
352: \end{equation}
353: 
354: An interesting feature is that the magnitude of the field increases 
355: exponentially with the applied voltage contrary to student intuition based 
356: on the applied voltage simply dividing between the resistance of the 
357: depletion region and the resistance of the neutral region.
358: 
359: With the field given in eq.\ref{eq15} a paradox mentioned in Section II develops 
360: in that E varies spatially across the neutral region so that $dE/dx$ is not 
361: zero. Gauss equation in point form (or the Poisson equation) in the 
362: n-region is:
363: 
364: \begin{equation}
365: E_{n}(x \mapsto \infty)  = \left( {\frac{{n_i }}{{N_D }}} \right)^2 
366: \left( {\frac{{kT}}{{eL_{hn} }}} \right)\frac{1}{{b_n }}\exp (\frac{{eV}}{{kT}})
367: \label{eq17}
368: \end{equation}
369: 
370: 
371: \begin{equation}
372: \frac{\varepsilon }{e}\frac{{dE_n }}{{dx}} = \Delta p_n (x)-\Delta n_n (x)
373: \label{eq18}
374: \end{equation}
375: 
376: where $\varepsilon $ is the total permittivity of Si (=$\varepsilon 
377: _{o}\varepsilon _{r})$.
378: 
379: Since $\Delta $p$_{n}$(x) is determined by eq.\ref{eq5}, the excess majority 
380: carrier concentration is:
381: 
382: \begin{equation}
383: \Delta n_n (x) = \Delta p_n (x)\left[ {1 - \frac{{\frac{\varepsilon }{e}
384: \frac{{dE_n }}{{dx}}}}{{\Delta p_n (x)}}} \right] = \Delta p_n (x)
385: \left[ {1 + \frac{{\varepsilon (b_n  - 1)kT}}{{e^2 b_n N_D L_{hn}^2 }}} \right]
386: \label{eq19}
387: \end{equation}
388: 
389: 
390: Substituting typical values for $\varepsilon $, b$_{n}$, N$_{D}$, L$_{hn}$ 
391: from Table II into eq.(\ref{eq2}) shows that the second term is $\sim $4.1x10$^{ - 
392: 6}$. Thus $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) and the charge neutrality 
393: condition for all practical purposes remains valid. We have found the 
394: requirement of $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) to be somewhat 
395: contrary to student intuition. This is further exasperated by many texts 
396: showing the majority carrier concentration uniform in the neutral regions 
397: (see Table I) which misleads the student. Qualitatively, the injected holes 
398: into the n-region disturb the charge neutrality and set-up a field here 
399: which then drives the electrons towards the SCL until a steady state is 
400: reached between electron drift and diffusion. Thus the increase in the 
401: majority carrier concentration towards the SCL is due to the driving effect 
402: of the field, E$_{n}$, even though it appears at first that $\Delta $n$_{n}$ 
403: should decrease towards theSCL as injected holes recombine with electrons 
404: 
405: Once the field in the n-region is given as eq.\ref{eq15}, the student can readily 
406: calculate the various contributions to the total current density using 
407: eq.\ref{eq13}. The magnitudes of the various current components (majority carrier diffusion
408:  and drift, and minority carrier diffusion and 
409: drift) and their directions 
410: are listed in Table II. In general, the \textit{drift of 
411: the majority carriers is the most significant contribution to the pn junction
412: diode current.} How is then the 
413: diffusion terminology comes to appear in explaining the diode current even 
414: though the biggest contribution is drift? 
415: 
416: Given that the depletion region is very narrow and that recombination in 
417: this region is negligibly small due to the very small concentrations of 
418: carriers, then in the steady state one must have 
419: $\frac{\partial J_e }{\partial x} = 0$ and the electron current J$_{e}$ 
420: must be constant through 
421: the SCL. But, electron drift at x'=0, i.e minority carrier drift, is 
422: negligible and the electron current there is primarily a diffusion current 
423: just like the hole current in the n-region. Thus the total electron current 
424: at x=0 must equal to the electron diffusion current at x'=0. Similarly the 
425: hole diffusion current at x=0 is equal to the total hole current at x'=0 
426: which is essentially by drift. It is apparent that by evaluating the 
427: \textit{minority carrier diffusion current just outside depletion region we 
428: are indirectly determining the total majority carrier current at the other 
429: side of the depletion region}. This is a subtle point that seems to have 
430: been short circuited in the 
431: majority of the texts. Consequently, the diode equation stated in equation 
432: (10) is only valid if the SCL width is much shorter than the minority 
433: carrier diffusion length in that region
434: 
435: An important paradox that must be addressed by the instructor is the much 
436: cherished minority carrier concentration profile stated in eq.\ref{eq5} for a 
437: long diode. Equation \ref{eq5} is the solution of the continuity equation in the 
438: absence of an electric field. As a first step one can assume a constant 
439: field, E$_{n}$, in the neutral region to examine its effects on the excess 
440: minority carrier concentration profile. In the presence of a constant field, 
441: the general continuity equation in eq.\ref{eq2} leads to 
442: 
443: \begin{equation}
444: D_{hn} \frac{{{\rm{d}}^{\rm{2}} {\rm{p}}_{\rm{n}} }}{{{\rm{dx}}^{\rm{2}} }} - 
445:  \mu _{hn} E_n \frac{{{\rm{dp}}_{\rm{n}} }}{{{\rm{dx}}}} - 
446:  \frac{{{\rm{p}}_{\rm{n}} }}{{\tau _{{\rm{hn}}} }} = 0
447: \label{eq20}
448: \end{equation}
449: 
450: Since this is a linear differential equation, the undergraduate student can 
451: readily solve it or accept its solution by substituting the solution into 
452: eq.(\ref{eq3}). For a long neutral region, the solution is:
453: 
454: \begin{equation}
455: \Delta {\rm{p}}_{\rm{n}} {\rm{(x)}} = \Delta {\rm{p}}_{\rm{n}} 
456: {\rm{(0)exp(-}}\frac{{\rm{x}}}{{{\rm{a}}_{\rm{n}} }}{\rm{)}}
457: \label{eq21}
458: \end{equation}
459: 
460: where a$_{n}$ is a "length constant" which can be determined by substituting 
461: eq.(\ref{eq4}) into (\ref{eq3});
462: 
463: \begin{equation}
464: {\rm{a}}_{\rm{n}}^{\rm{2}}  -  (\tau _{{\rm{hn}}} \mu _{hn} 
465: E_n {\rm{)a}}_{\rm{n}}  -  (D_{hn} \tau _{{\rm{hn}}} ) = 0
466: \label{eq22}
467: \end{equation}
468: 
469: Solving this quadratic equation we find,
470: 
471: \begin{equation}
472: {\rm{a}}_{\rm{n}}  =  \sqrt {(D_{hn} \tau _{{\rm{hn}}} )} 
473: \left[ {\sqrt {\alpha _n  + 1}  + \alpha _n } \right] = 
474: L_{hn} \left[ {\sqrt {\alpha _n  + 1}  + \alpha _n } \right]
475: \label{eq23}
476: \end{equation}
477: 
478: where $\alpha _{n}$ is defined as
479: 
480: \begin{equation}
481: \alpha _n  =  \frac{{\frac{1}{2}\tau _{{\rm{hn}}} 
482: \mu _{hn} E_n }}{{L_{hn} }} = 
483: \frac{{\frac{1}{2}L_{drift} }}{{L_{hn} }} = 
484: \frac{1}{2}\frac{{Schubweg}}{{Diffusion \hspace{1mm} Length}}
485: \label{eq24}
486: \end{equation}
487: 
488: The parameter $\alpha_{n}$ represents the comparative effect of drift to 
489: diffusion since L$_{drift}=^{1}$/$_{2}\tau _{hn}\mu _{hn}$E$_{n}$ 
490: is the so-called Schubweg of the minority carriers, distance drifted before 
491: recombination. If the field is small $\alpha  \quad _{n }$will be small and in 
492: the limit of zero field, E$_{n} \sim 0$, $a_{n}\sim L_{hn}$ and the
493:  theory approaches the conventional zero-field treatment. 
494: At a forward bias of 0.55V, the field is maximum at x=0, and using this 
495: maximum value one finds $\alpha _{n}$=0.00106 and 
496: a$_{n}$=(1.0011)L$_{hn}$. At V=0.65V, a$_{n}$=(1.05)L$_{hn}$ and a$_{n}$ is 
497: still very close to L$_{hn }$(within 5{\%}) even though the injected hole 
498: concentration is now only ten times smaller than the equilibrium majority 
499: concentration which sets the limit of small injection. Although the solution 
500: in eq.(\ref{eq3}) does not apply when the field is non-uniform as in eq.\ref{eq15}, it 
501: does nonetheless provide convenient means for the student to examine the 
502: possible effect of the field on the excess minority carrier profile. 
503: 
504: The field in the p-region can be similarly derived. The total current in the 
505: p$^{ + }$ region is
506: 
507: 
508: \begin{equation}
509: J = eD_{ep} \frac{{d\Delta n_n }}{{dx}} + e\mu _{ep} n_p E_p  - 
510: eD_{hp} \frac{{d\Delta p_p }}{{dx}} + e\mu _{hp} p_p E_p  =  constant  \\
511: \label{eq25}
512: \end{equation}
513: 
514: 
515: 
516: Since the total current must be constant and assigned to be described by 
517: eq. \ref{eq12}, using the corresponding version of eq. \ref{eq5} for minority and 
518: majority carrier excess concentration in the p-region one can derive
519: 
520: 
521: \begin{equation}
522: E_p(x)  = \exp (\frac{{eV}}{{kT}})\left( {\frac{{n_i^2 }}{{N_D N_A }}} \right)\left( {\frac{{kT}}{{eL_{hn} }}} \right)\left( {\frac{{\mu _{hn} }}{{\mu _{hp} }}} \right)\left[ {1 - C_p \exp ( - \frac{{x'}}{{L_{ep} }})} \right]
523: \label{eq26}
524: \end{equation}
525: 
526: 
527: where C$_{p}$ is defined as
528: 
529: \begin{equation}
530: C_p  = \left( {b_p  - 1)} \right)\left( {\frac{{\mu _{hp} }}{{\mu _{hn} }}} \right)
531: \left( {\frac{{L_{hn} }}{{L_{ep} }}} \right)\left( {\frac{{N_D }}{{N_A }}} \right)
532: \label{eq27}
533: \end{equation}
534: 
535: in which b$_{p }=\mu _{ep}$/$\mu _{hp}$ is the electron to hole drift 
536: mobility ratio in the p-region. We assumed that, as usual under forward 
537: bias, $V>>kT/e $. Equation (\ref{eq9}) shows that the field is minimum right at the 
538: SCL, x'=0, and increases exponentially to a constant value away from the 
539: junction. Substituting typical values from Table 
540: II shows that $E_{p}<<E_{n}$. When the two fields are compared, one finds  
541: that E$_{n}$ is at least three orders of magntiude greater 
542: than E$_{p}$. In fact E$_{p}$ is almost unifrom in the p$^{ + }$-region. 
543: Most interestingly and importantly, even though E$_{p}$ is even smaller than 
544: E$_{n}$, its effect is most significant. One readily can calculate the 
545: contributions of each term to the current density in the p-region from eq. \ref{eq24}.
546: The values at the SCL are listed in Table II where it is apparent that 
547: the current is carried almost totally by \textit{majority carrier drift}. 
548: A distinctly different behavior 
549: in the p$^{ + }$-region from that observed in the n-region is the fact that 
550: the majority carrier diffusion is insignificant and that minority carrier 
551: diffusion, though larger than majority diffusion, is some three orders of 
552: magnitude smaller than majority carrier drift. From the above discussion for 
553: conduction on the n-side it is clear that in deriving the diode equation we 
554: are calculating the hole (majority) drift current in the p$^{ + }$-region by 
555: evaluating the hole (minority) diffusion current in the n-region simply 
556: because the total hole current does not change through the SCL as long as 
557: the latter is thinner than the hole diffusion length. 
558: 
559: It is always useful for the student to reconfirm that the majority of the 
560: voltage drops across the depletion region by evaluating the voltage drop 
561: across the neutral regions. If V$_{n}$ is the voltage drop across the 
562: n-region then
563: 
564: \begin{equation}
565: V_n  = \int_0^{l_n } {E_n dx} 
566: \label{eq28}
567: \end{equation}
568: 
569: or
570: 
571: \begin{equation}
572: V_n  = \left( {\frac{{n_i }}{{N_D }}} \right)^2 \left( {\frac{{kT}}{{eL_{hn} }}}
573:  \right)\frac{1}{{b_n }}\left( {l_n  + (b_n  - 1)L_{hn} } \right)\exp (\frac{{eV}}{{kT}})
574: \label{eq29}
575: \end{equation}
576: 
577: Equation (\ref{eq2}) shows that the voltage drops increases exponentially with the 
578: applied bias contrary to an intuitive guess. Using typical values, at 
579: V=0.55V, V$_{n}$ is 0.00168V, whereas at V=0.6V, V$_{n}$ is 0.019V and the 
580: injected hole concentration in the n-region is about 11{\%} of N$_{D}$ which 
581: is the limit of small injection. At V=0.65V, V$_{n}$ becomes 0.121V which is 
582: quite significant but at this bias voltage the injected hole concentration 
583: is no longer small compared with N$_{D}$. There is a clear indication that 
584: as the voltage across the diode increases more and more of the applied 
585: voltage drops across the neutral regions which deteriorates the law of the 
586: junction. It is not difficult to show that since $E_{p}<<E_{n}$ and 
587: $l_{p} << l_{n}$, the voltage drop across the p$^{ + }$-region is orders of 
588: magnitude smaller than V$_{n}$.
589: 
590: \section{NUMERICAL DESCRIPTION OF THE PN JUNCTION:}
591: While the previous sections discussed the Diffusion/Drift approximation problem,
592: this section is oriented towards an \textit{exact} description of the PN
593: junction without making any assumptions. \\  
594: 
595: Starting from the constitutive system of equations \cite{polak}:
596: 
597: \begin{equation}
598: \frac{d\psi }{dx} = - E
599: \end{equation}
600: 
601: \begin{equation}
602: J_n = e\left[ {D_n } \right.\frac{dn}{dx} - \mu _n n\frac{d\psi }{dx}\left. 
603: \right]
604: \end{equation}
605: 
606: 
607: \begin{equation}
608: J_p = - e\left[ {D_p } \right.\frac{dp}{dx} + \mu _p p\frac{d\psi 
609: }{dx}\left. \right]
610: \end{equation}
611: 
612: 
613: \begin{equation}
614: \frac{1}{e}\frac{dJ_n }{dx} = R(n,p) + G(x)
615: \end{equation}
616: 
617: 
618: \begin{equation}
619: \frac{1}{e}\frac{dJ_p }{dx} = - R(n,p) - G(x)
620: \end{equation}
621: 
622: with the recombination term of the Shockley-Read-Hall form:
623: 
624: \begin{equation}
625: R(n,p) = \frac{(np - n_i^2 )}{T_1 n + T_2 p + T_3 }
626: \end{equation}
627: 
628: 
629: where T$_{1}$, T$_{2 }$and T$_{3 }$are time constants. Typically for Silicon 
630: T$_{1}=\tau _{p}$= 10$^{ - 5}$ sec,$_{ }$T$_{2}=\tau _{n}$= 10$^{ 
631: - 5}$ sec$_{ }$and T$_{3}$= ($\tau _{p }+\tau _{n})$n$_{i}$ \cite{ringhofer}.
632: 
633: In order to solve the boundary value problem associated with the above 
634: system (when a voltage is applied to the PN junction) we transform it into a 
635: new hybrid system of first-order (Current and carrier density equations) and 
636: one second-order differential equation (Poisson equation).
637: 
638: The mathematical/numerical reasons for performing this transformation reside 
639: in the fact the above system is a "singularly singular perturbed problem" 
640: \cite{mark84,ascher}. Many algorithms \cite{ascher,ascher83} have been 
641: developed in order to deal with this difficulty stemming from several facts:
642: \begin{enumerate}
643: 
644: \item $\psi $, n and p are fast variables in comparison with E, J$_{n}$ and 
645: J$_{p}$ \cite{ringhofer,mock,please}.
646: 
647: \item Near the limits of the depletion layer the values of n and p change by 
648: several orders of magnitude making the space-charge zone a double boundary 
649: layer. This difficulty is of the same type as the one encountered in 
650: Hydro/Aerodynamics where the fluid velocity field changes by several orders 
651: of magnitude near an obstacle.
652: 
653: Recognizing the difficulty due to the presence of the space-charge layer, a 
654: standard way to find a valid solution is to treat the boundary layer 
655: separately from the rest of the diode. In spite of the success of this 
656: approach \cite{mari,arand}, one might feel uneasy about this artificial 
657: dichotomy and rather tackle the problem with new powerful mathematical/numerical
658: methods that handle the layer and the rest of the device on the same equal footing. 
659: 
660: \item When the above system is rewritten explicitly in terms of the Poisson 
661: equation as we will do below, the second spatial derivative of the electric 
662: potential is multiplied by a very small number $\lambda ^{2}$ 
663: ($\lambda \sim 10^{-4} \mbox{to} 10^{-3})$. Actually, this is the reason the 
664: problem is called singularly perturbed: the solution with $\lambda=0$ 
665: is entirely different from the solution with $\lambda $ finite 
666: but small \cite{ascher}.
667: 
668: \end{enumerate}
669: 
670: One of the early algorithms aimed at circumventing the above difficulties is 
671: the Scharfetter-Gummel algorithm \cite{scharfetter}. The latter attempts at segregating 
672: the fast/slow variables by integrating out the fast variables over some 
673: small interval while holding the slow variables constant over that same 
674: interval. The Scharfetter-Gummel algorithm leads to a spatial exponential 
675: discretization that will alleviate for the rapid variation of the fast 
676: variables.
677: 
678: Many variants of the Scharfetter-Gummel algorithm have been developed \cite{he} 
679: in order to cure some of its shortcomings which generally are numerical 
680: oscillations and crosswind effects. These lead to a loss of accuracy of the 
681: solution and sometimes preclude convergence towards the solution.
682: 
683: We decided not to use the Gummel algorithm or any of its variants but rather 
684: to tackle the problem from the singular perturbation point of view since 
685: this approach is more rigorous and leads to a better understanding and 
686: control of the instability problem encountered in the semiconductor system of 
687: equations. 
688: 
689: We first transform the system in the following dimensionless two-point 
690: boundary value problem with no generation processes considered:
691: 
692: \[
693: \frac{dn}{dx} = C_1 J_n + n\frac{d\psi }{dx}
694: \]
695: 
696: 
697: \[
698: \frac{dp}{dx} = - C_2 J_p - p\frac{d\psi }{dx}
699: \]
700: 
701: 
702: \[
703: \frac{dJ_n }{dx} = C_3 \frac{(np - 1)}{n + \tau _1 p + \tau _2 }
704: \]
705: 
706: 
707: \[
708: \frac{dJ_p }{dx} = - C_3 \frac{(np - 1)}{n + \tau _1 p + \tau _2 }
709: \]
710: 
711: 
712: \[
713: \frac{d^2\psi }{dx^2} = C_4 (n - p + N_D )
714: \]
715: 
716: The constants C$_{1}$,$_{ }$C$_{2}$,$_{ }$C$_{3}$ and C$_{4}$ are given by:
717: 
718: \[
719: C_1 = \frac{J_0 L_D }{en_i D_n }, C_2 = \frac{J_0 L_D }{en_i D_p }, C_3 = \frac{e L_D n_i }{J_0 T_1 },  
720: C_4 = \frac{ e L_D^2 n_i }{\epsilon_s U_T }
721: \]
722: 
723: where the Debye length L$_{D}$ is given by $L_D=\sqrt{\frac{k_B T \epsilon_S}{n_i e^2}}$ and the scaling current
724: $J_0=\frac{n_i \mu_n k_B T}{L_D}$. 
725: 
726: The thermal voltage U$_{T}$=k$_{B}$T/e (T is the temperature and k$_{B}$ is 
727: Boltzmann constant). The time constants are now 
728: $\tau_{1}$=T$_{2}$/T$_{1}$ and $\tau_{2}$=T$_{3}$/(T$_{2}$n$_{i})$.
729: 
730: The above system is now in the appropriate form to integrate with a powerful 
731: B-spline collocation based algorithm specifically tailored for two point 
732: singularly perturbed boundary value problems: COLSYS \cite{ascher,mark86,ascher83}.
733: The algorithm is based on a controllable meshing technique \cite{ascher83,mark83} of the 
734: boundary layer which will lead essentially to the damping of the existing
735: singularities. The layer-damping mesh, being exponential in nature, 
736: encompasses the Scharfetter-Gummel case and can be shown rigorously to have 
737: the form:
738: 
739: \[
740: h_{i}=h_{i-1} \exp(\alpha h_{i-1}/\lambda )
741: \]
742: 
743: where $h_{i}$ is the i-th mesh point, $\alpha $ is a constant related to the 
744: required accuracy and the nature of the collocation and $\lambda {\rm g}$s 
745: the singular value parameter.
746: 
747: Previously, Markowich et al. \cite{mark84} tackled this problem from the same angle 
748: but they solved the symmetric diode case with one boundary-layer at the 
749: origin.
750: 
751: In this work we tackle the non-symmetric case where a double-boundary layer 
752: is present around the origin starting from very accurate initial conditions.
753: 
754: Varying the applied bias by steps of 0.1V we calculate the carrier density 
755: profiles, the potential and the electric field.
756: 
757: The drift and diffusion current density profiles are also obtained. Several 
758: tests are used in order to check the validity of the solution obtained. The 
759: first test is an accuracy test whereby we require a given accuracy and check 
760: whether the criterion is met. The next test is based on the requirement of 
761: convergence: the collocation builds a non-linear set of equations that has 
762: to be solved iteratively. The additional tests are the independant checks of 
763: the constancy of the current densities locally on each side of the junction 
764: and globally over the entire junction. The tests are shown in the current 
765: density profile figures.
766: 
767: The final test we use is the approximate validity of Shockley's equation. 
768: Varying the voltage, we obtain the IV characteristics of the junction and we 
769: compare it to the Shockley's case. Since we have used the Shockley-Read-Hall 
770: recombination term all over the PN junction we do not expect the Shockley's 
771: case but rather the general form:
772: 
773: \[
774: I=I_{S} [\exp(eV/\eta k_{B}T)-1]
775: \]
776: 
777: The comparison of the obtained IV characteristic to the Shockley formula is 
778: displayed in Fig. 1. The calculated characteristic falls between the two 
779: Shockley curves $\eta=1.1$ and  $\eta=1.2$ in a finite current interval.
780: This means, the  general Shockley formula is not valid, within the singular 
781: perturbation approach, for arbitrary current values.
782: 
783: \begin{figure}[!h]
784: \begin{center}
785: %\scalebox{0.3}{\includegraphics[angle=-90]{fig1.pdf}}
786: \scalebox{0.3}{\includegraphics[angle=-90]{shockley.eps}}
787: \end{center}
788:   \caption{IV characteristics obtained for ther PN junction and Shockley approximation
789: with  $\eta=1.1$ and  $\eta=1.2$.}
790: \label{fig1}
791: \end{figure}
792: 
793: 
794: \section{Conclusions}
795: The physics of the PN junction is gaining back interest with new developments in the
796: area of nanoelectronics specially in the area of spintronics where one has to account for 
797: the spin of the carriers in addition to their charge. The usual approximations that are valid
798: and successful in the description of the PN junction physics at the micron scale must be entirely
799: reviewed and adapted to the nano scale. The diffusion/drift approximation as well as the
800: nature of the singularities of the problem have been reviewed and reformulated in a way such that
801: the underlying assumptions are revealed with their consequences.  
802: 
803: 
804: \begin{thebibliography}{00}
805: \bibitem{moloney} M. J. Moloney, Justifying the simple diode equation, Am. J. 
806: Phys.\textbf{ 54}, 914-916.
807: \bibitem{allison} J. Allison, Electronic Engineering Semiconductors and Devices, Second 
808: Edition (McGraw-Hill Book Company, London, 1990), Ch. 7
809: \bibitem{carroll} J.E. Carroll, Physical Models for Semiconductor Devices, (Edward Arnold 
810: Publishers Ltd, London, 1974), Ch. 4.
811: \bibitem{colclaser} R.A. Colclaser and S. Diehl-Nagle, Materials and Devices for Electrical 
812: Engineers and Physicists (McGraw-Hill, New York, 1985), Ch. 4
813: \bibitem{elliott} R.J. Elliot and A.F. Gibson, An Introduction to Solid State Physics (The Macmiilan Press Ltd., London, 1982), Ch. 9.
814: \bibitem{ferendeci} A.M. Ferendeci, Physical Foundations of Solid State and Electron Devices 
815: (McGraw-Hill, New York, 1991), Ch. 8
816: \bibitem{fraser} D.A. Fraser, The Physics of Semiconductor Devices, Fourth Edition (Oxford University Press, Oxford, 1986), Ch. 3
817: \bibitem{goodge} M. Goodge, Semiconductor Device Technology (Macmillan, Publishers Ltd., London, 1983) Ch. 1
818: \bibitem{gibbons} J.F. Gibbons, Semiconductor Electronics (McGraw-Hill Book Company, New York, 
819: 1966; out of print), Ch.6
820: \bibitem{gray67} P.E. Gray and C. Searle, Electronic Principles (John Wiley and Sons, Newy 
821: York, 1967; out of print), Ch.4.
822: \bibitem{gray64} P.E. Gray, D. DeWitt, A.R. Boothroyd, J.F. Gibbons, Physical Electronics and 
823: Circuit Models of Transistors (John Wiley and Sons, Newy York, 1964; out of 
824: print), Ch.2 and Appendix B.
825: \bibitem{grove} A.S. Grove, Physics and Technology of Semicopnductor Devices (John Wiley and Sons, Newy York, 1967; out of print), Ch. 3.
826: \bibitem{lonngren} K. Lonngren, Introduction to Physical Electronics (Allyn and Bacon Inc, Boston, 1988), Ch. 6
827: \bibitem{navon} D.H. Navon, Semiconductor Microdevices and and Materials (Holt, Rinehart and 
828: Winston, New York, 1986) Ch. 6
829: \bibitem{neamen} D.A. Neamen, Semiconductor Physics and Devices (Irwin, Boston, 1992), Ch. 7
830: \bibitem{pulfrey} D.L. Pulfrey and N.G. Tarr, Introduction to Microelectronic Devices 
831: (Prentice Hall, Englewood Cliffs, NJ, 1989) Ch.6
832: \bibitem{seymour} J. Seymour, Electronic Devices and Components (Pitman Publishing Ltd., 
833: London, 1981), Ch. 3
834: \bibitem{shur} M. Shur, Physics of Semiconductor Devices (Prentice Hall, Englewood Cliffs, 
835: NJ, 1990) Ch. 2
836: \bibitem{solymar} L. Solymar and D. Walsh Lectures on the Electrical Properties of Materials, 
837: Fourth Edition (Oxford University Press, Oxford, 1988), Ch.9
838: \bibitem{streetman} B.G. Streetman, Solid State Electronic Devices, Third Edition 
839: \bibitem{yang} E. S. Yang, Fundamentals of Semicodnuctor Devices (McGraw-Hill, New York, 
840: 1978), Ch. 4
841: \bibitem{valdes} L.B. Valdes, The Physical Theory of Transistors (McGraw-Hill Book Company, 
842: US, 1961; out of print), Ch.9
843: \bibitem{ziel} A. van der Ziel, Solid State Physical Electronics, Second Edition 
844: (Prentice-Hall Inc., Englewood, NJ, 1968; out of print) Ch. 15.
845: \bibitem{fwang} F.F.Y. Wang, Introduction to Solid State Electronics Second Edition, (North 
846: Holland - Elsevier Science B.V., Amsterdam, 1989), Ch. 14.
847: \bibitem{swang} S. Wang, Fundamentals of Semiconductor Theory and Device Physics (Prentice 
848: Hall, Englewood Cliffs, NJ, 1989) Ch. 2
849: \bibitem{zambuto} M. Zambuto, Semiconductor Devices (McGraw-Hill, New York, 1989), Ch. 6.
850: \bibitem{sze85} S.M. Sze, Semiconductor Devices (John Wiley and Sons, Newy York, 1985), Ch.3.
851: \bibitem{sze81} S.M. Sze, Physics of Semiconductor Devices, Second Edition (John Wiley and 
852: Sons, New York, 1981), Ch. 2
853: \bibitem{tyagi} M.S. Tyagi, Introduction to Semiconductor Materials and Devices (John Wiley 
854: and Sons, New York, 1991), Ch. 7
855: \bibitem{polak} S.J Polak, C. Den Heijer, W.H.A. Schilders and P. Markowich: 
856: "Semiconductor Device Modeling from the numerical point of view" Int. J. for 
857: Num. Methods in Engineering, Vol 24, 763-838 (1987).
858: \bibitem{ringhofer} C. A. Ringhofer and C. Schmeiser: "A modified Gummel Method for the 
859: basic Semiconductor Equations" IEEE Transactions CAD Vol7, 251-253 (1988).
860: \bibitem{mock} M. S. Mock:"On the convergence of Gummel's numerical algorithm" 
861: Solid-State Electronics, Vol 15, 1-4 (1972).
862: \bibitem{please} C. P. Please:"An analysis of Semiconductor PN junctions" IMA Journal 
863: of Applied Mathematics, Vol 28, 301-318 (1982).
864: \bibitem{scharfetter} D. L. Scharfetter and H.K. Gummel:"Large-Signal Anlysis of a Silicon 
865: Read Diode Oscillator" IEEE Transactions ED-16, 64-77 (1969).
866: \bibitem{he} Y. He and G. Cao:"A generalized Scharfetter-Gummel method to eliminate 
867: Cross-Wind effects" IEEE Transactions CAD-10, 1579-1582 (1991).
868: \bibitem{mari} A. De Mari:"An accurate numerical steady-state one dimensional 
869: solution of the PN junction"Solid State Electron ics, Vol 11, 33-58 (1968).
870: \bibitem{arand} V. Arandjelovic:"Accurate Numerical Steady-State solutions for a 
871: diffused one dimensional junction diode", Solid State Electron ics, Vol 13, 
872: 865-871 (1970).
873: \bibitem{mark84} P. Markowich and C. A. Ringhofer:"A Singularly Perturbed boundary 
874: value problem modelling a semiconductor device" SIAM J. Appl. Math. Vol 44, 
875: 231-256 (1984).
876: \bibitem{ascher} U.M. Ascher, R.M. Mattheij and R. D. Russel:"Numerical Solution of 
877: Boundary Value Problems for Ordinary Differential Equations", Prentice-Hall 
878: (Englewood Cliffs).
879: \bibitem{mark86} P. Markowitch and C. Schmeiser: "Uniform asymptotic representation of 
880: solutions of the basic semiconductor-device equations" IMA Journal of 
881: Applied Mathematics, Vol 36, 43-57 (1986).
882: \bibitem{ascher83} U.M. Ascher and R. Weiss:"Collocation for singular perturbation 
883: problems I: First-order systems with constant coefficients" SIAM J. Num. 
884: Analysis Vol 20, 537-557 (1983).
885: \bibitem{mark83} P. Markowich and C. A. Ringhofer:"Collocation methods for boundary 
886: value problems on 'long' intervals" Math. Comp. Vol 40, 123-150 (1983).
887: \end{thebibliography}
888: 
889: \newpage
890: 
891: \section{Tables}
892: 
893: \begin{table}[ht!]
894: \begin{tabular}
895: {|p{86pt}|p{50pt}|p{93pt}|p{86pt}|p{108pt}|p{108pt}|}
896: \hline
897: AUTHOR& 
898: REF.& 
899: SEMIQUANTITIVE DRIFT ANALYSIS& 
900: FIELD IN NEUTRAL REGIONS& 
901: MAJORITY CARRIER CONCENTRATION& 
902: COMMENT \\
903: \hline
904: Zambuto \cite{zambuto}& 
905: Ch. 6& 
906: Qualitative& 
907: Not discussed& 
908: Not shown& 
909: Undergraduate \\
910: \hline
911: Yang \cite{yang}& 
912: Ch. 4& 
913: Qualitative& 
914: Not discussed& 
915: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x)& 
916: Undergraduate \\
917: \hline
918: Elliott and Gibson \cite{elliott}& 
919: Ch. 9& 
920: None& 
921: Not discussed& 
922: n$_{n}$ shown constant& 
923: Undergraduate \\
924: \hline
925: Fraser (UK)\cite{fraser}& 
926: Ch. 3& 
927: None& 
928: Not discussed& 
929: n$_{n}$ shown constant& 
930: Undergraduate \\
931: \hline
932: Seymour \cite{seymour}& 
933: Ch. 3& 
934: None& 
935: Not discussed& 
936: Not shown& 
937: Undergraduate \\
938: \hline
939: Streetman \cite{streetman}& 
940: Ch. 5& 
941: Some (Example 5.4)& 
942: Mentioned but not discussed& 
943: Not shown& 
944: Undergraduate \\
945: \hline
946: Valdes \cite{valdes}& 
947: Ch. 9& 
948: Yes& 
949: Discussed& 
950: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x)& 
951: Out of print / Undergraduate \\
952: \hline
953: Sze \cite{sze85} & 
954: Ch. 3& 
955: None& 
956: Not discussed & 
957: Not clear & 
958: Undergraduate \\
959: \hline
960: Carroll (UK) \cite{carroll}& 
961: Ch. 4& 
962: None& 
963: Mentioned but not discussed & 
964: Not clear& 
965: Undergraduate \\
966: \hline
967: Solymar and Walsh (UK) \cite{solymar}& 
968: Ch.9& 
969: None& 
970: Not discussed& 
971: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) implied& 
972: Undergraduate \\
973: \hline
974: Shur \cite{shur}& 
975: Ch. 2& 
976: Some (Fig. 2.3-8)& 
977: Not discussed& 
978: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) implied& 
979: Senior UG / Graduate level \\
980: \hline
981: Pulfrey and Tarr (Canada) \cite{pulfrey}& 
982: Ch. 6& 
983: None& 
984: Not discussed& 
985: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) implied but n$_{n}$=constant in diagrams& 
986: Undergraduate \\
987: \hline
988: Colclaser and Diehl-Nagle \cite{colclaser}& 
989: Ch. 7& 
990: None& 
991: Not discussed& 
992: Not mentioned and not shown& 
993: Undergraduate \\
994: \hline
995: Navon \cite{navon} & 
996: Ch. 6& 
997: None& 
998: Not discussed& 
999: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) in diagram & 
1000: Undergraduate \\
1001: \hline
1002: Sze \cite{sze81}& 
1003: Ch. 2& 
1004: None& 
1005: Not discussed& 
1006: Not clear& 
1007: Senior Undergraduate / Graduate \\
1008: \hline
1009: Wang  \cite{swang}& 
1010: Ch. 14& 
1011: None& 
1012: Not discussed& 
1013: Inferred& 
1014: Senior Undergraduate \\
1015: \hline
1016: Goodge (UK) \cite{goodge}& 
1017: Ch.1& 
1018: None& 
1019: Not discussed& 
1020: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) in diagram& 
1021: Undergraduate \\
1022: \hline
1023: Grove \cite{grove}  & 
1024: Ch.3& 
1025: None& 
1026: Not discussed& 
1027: Not clear & 
1028: Out of print / Undergraduate \\
1029: \hline
1030: Van Der Ziel \cite{ziel}& 
1031: Ch. 15& 
1032: None& 
1033: Not discussed& 
1034: Not clear & 
1035: Out of print / Undergraduate \\
1036: \hline
1037: Lonngren \cite{lonngren}& 
1038: Ch. 6& 
1039: None& 
1040: Not discussed& 
1041: Not discussed& 
1042: Undergraduate \\
1043: \hline
1044: Tyagi \cite{tyagi}& 
1045: Ch. 7& 
1046: Qualitative& 
1047: Mentioned but not discussed& 
1048: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) (Fig.7.3)& 
1049: Undergraduate \\
1050: \hline
1051: Gibbons \cite{gibbons}& 
1052: Ch. 6& 
1053: Yes & 
1054: Discussed& 
1055: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x)& 
1056: Out of print / Undergraduate \\
1057: \hline
1058: Ferendeci \cite{ferendeci}& 
1059: Ch. 8& 
1060: None& 
1061: None& 
1062: Not clear& 
1063: Undergraduate \\
1064: \hline
1065: Allison \cite{allison}& 
1066: Ch. 7& 
1067: None& 
1068: None& 
1069: Not clear& 
1070: Undergraduate \\
1071: \hline
1072: Neamen \cite{neamen}& 
1073: Ch. 8& 
1074: Some& 
1075: E estimated& 
1076: Not clear& 
1077: Undergraduate \\
1078: \hline
1079: F. Wang \cite{fwang}& 
1080: Ch. 8& 
1081: None& 
1082: None& 
1083: Not clear& 
1084:  \\
1085: \hline
1086: Gray et al. \cite{gray64}& 
1087: Ch.2 and App. B. & 
1088: Some& 
1089: Yes& 
1090: $\Delta $n$_{n}$(x)=$\Delta $p$_{n}$(x) in diagram& 
1091: Out of print / Undergraduate \\
1092: \hline
1093: \end{tabular}
1094: \caption{Treatment of the pn junction in a selection of the textbooks suitable for a 
1095: physical electronics (solid state electronic devices) course.}
1096: \end{table}
1097: \label{tab1}
1098: 
1099: \squeezetable
1100: 
1101: \begin{table}
1102: \begin{tabular}{|p{216pt}|p{194pt}|p{252pt}|}
1103: \hline
1104: \textbf{PROPERTY / PARAMETER}& 
1105: \textbf{TYPICAL VALUE}& 
1106: \textbf{COMMENT} \\
1107: \hline
1108: Permittivity& 
1109: 11.9& 
1110:  \\
1111: \hline
1112: Intrinsic concentration, ni, cm$^{ - 3}$& 
1113: 1.5$\times $10$^{10}$& 
1114:  \\
1115: \hline
1116: Donor concentration, N$_{D}$, cm$^{ - 3}$& 
1117: 5$\times $10$^{15}$& 
1118:  \\
1119: \hline
1120: Acceptor concentration, N$_{A}$, cm$^{ - 3}$& 
1121: 10$^{19}$& 
1122: p$^{ + }$-n junction \\
1123: \hline
1124: Equilibrium hole concentration \par in n-region, p$_{n0}$, cm$^{ - 3}$.& 
1125: 4.50$\times $10$^{4} $& 
1126:  \\
1127: \hline
1128: Equilibrium electron concentration  \par in p-region, n$_{p0}$, cm$^{ - 3}$.& 
1129: 22.5 \quad & 
1130:  \\
1131: \hline
1132: Hole recombination time in n-region, $\tau _{hn}$, s& 
1133: $\tau _{hn} = \frac{5\times 10^{ - 7}}{1 + 2\times 10^{ - 17}N_D }$ \par $\tau _{hn} = 4.54\times 10^{ - 7}$& 
1134: $\tau _{en }=\tau _{hn}$ \\
1135: \hline
1136: Electron recombination time in p-region, $\tau _{ep, }$s& 
1137: $\tau _{ep} = \frac{5\times 10^{ - 7}}{1 + 2\times 10^{ - 17}N_A }$ \par $\tau _{ep} = 2.49\times 10^{ - 9}$& 
1138: $\tau _{hp }=\tau _{ep}$ \\
1139: \hline
1140: Electron drift mobility in the n-region,  \par cm$^{2}$V$^{ - 1}$s$^{ - 1}$.& 
1141: $\mu _{en} = 88 + \frac{1252}{1 + 6.984\times 10^{ - 18}N_D }$ \par $\mu _{en} = 1298$& 
1142: $b_n = \frac{\mu _{en} }{\mu _{hn} } = 2.86$ \\
1143: \hline
1144: Hole drift mobility in n-region,  \par cm$^{2}$V$^{ - 1}$s$^{ - 1}$ &  $\mu _{hn}  = 54.3 + \frac{{407}}{{1 + 3.745 \times 10^{ - 18} N_D }} $ 
1145: \par $\mu _{hn}  = 453.8$ & \\
1146: \hline
1147: Electron drift mobility in the p-region,  \par cm$^{2}$V$^{ - 1}$s$^{ - 1}$.& $\mu _{ep}  = 88 + \frac{{1252}}{{1 + 6.984 \times 10^{ - 18} N_A }}$
1148: \par $\mu _{ep}  = 105.7$  & $b_p = \frac{\mu _{ep} }{\mu _{hp} } = 1.63$ \\
1149: \hline
1150: Hole drift mobility in p-region,  \par cm$^{2}$V$^{ - 1}$s$^{-1}$.&  $\mu _{hp}  = 54.3 + \frac{{407}}{{1 + 3.745 \times 10^{ - 18} N_A }}$
1151:  \par $\mu _{hp}  = 64.9 $ &  \\
1152: \hline
1153: Electron diffusion coefficient in n-region, D$_{en}$, cm$^{2}$.s$^{ - 1}$&  33.55&   $ \frac{D_{en}}{\mu _{en}}=  \frac{kT}{e}$\\
1154: \hline
1155: Hole diffusion coefficient in n-region, D$_{hn}$,  \par cm$^{2}$.s$^{ - 1}$& 
1156: 11.73&  \\
1157: \hline
1158: Electron diffusion coefficient in p-region, D$_{ep}$, cm$^{2}$.s$^{ - 1}$& 
1159: 2.73& 
1160:  \\
1161: \hline
1162: Hole diffusion coefficient in p-region, D$_{hp}$, cm$^{2}$.s$^{ - 1}$& 
1163: 1.68& 
1164:  \\
1165: \hline
1166: Electron diffusion length in n-region, L$_{en}$, cm& 
1167: 3.90$\times $10$^{ - 3}$&   L$_{en}$=$\sqrt{D_{en}\tau_{en}}$
1168:  \\
1169: \hline
1170: Hole diffusion length in n-region, L$_{hn}$, cm& 
1171: 2.31$\times $10$^{ - 3}$& 
1172:  \\
1173: \hline
1174: Electron diffusion length in p-region, L$_{ep}$, cm& 
1175: 8.24$\times $10$^{ - 5}$ & 
1176:  \\
1177: \hline
1178: Hole diffusion length in p-region, L$_{hp}$, cm& 
1179: 6.46$\times $10$^{ - 5}$& 
1180:  \\
1181: \hline
1182: Builtin potential, V$_{bi}$, V& 
1183: 0.854& 
1184:  \\
1185: \hline
1186: E$_{bi }$, V.cm$^{ - 1}$& 
1187: 3.60$\times $10$^{4}$& 
1188: No bias \\
1189: \hline
1190: Width, W, of depletion region, cm& 
1191: 4.741$\times $10$^{ - 5}$& 
1192: On n-side. Much shorter than hole diffusion length on n-side. \\
1193: \hline
1194: Width, W$_{n}$, of depletion region in n-side, cm& 
1195: 4.739$\times $10$^{ - 5}$& 
1196: Much shorter than hole diffusion length in n-region. \\
1197: \hline
1198: Width, W$_{p}$ of depletion region in p-side, cm& 
1199: 2.369$\times $10$^{ - 8}$& 
1200: Much shorter than electron diffusion length in p-region \\
1201: \hline
1202: Length of n-side& 
1203: 2.31$\times $10$^{ - 2}$& 
1204: 10L$_{h}$ Long diode \\
1205: \hline
1206: Length of p$^{ + }$-side& 
1207: 8.24$\times $10$^{ - 4}$& 
1208: 10L$_{e }$Long diode \\
1209: \hline
1210: \textbf{FORWARD BIAS, V}& 
1211: \textbf{0.55}& \\
1212: \hline
1213: Built-in electric field, V.cm$^{ - 1}$& 
1214: 2.15$\times $10$^{4}$& 
1215: Smaller than zero bias case \\
1216: \hline
1217: Width, W, of depletion region at 0.55V, cm& 
1218: 2.83$\times $10$^{ - 5}$& 
1219: On n-side. Narrower under forward bias. Much shorter than hole diffusion length on n-side. \\
1220: \hline
1221: p$_{n}$(0), injected hole concentration at x=0& 
1222: 7.81$\times $10$^{13}$& 
1223:  \\
1224: \hline
1225: n$_{p}$(0), injected electron concentration at x'=0& 
1226: 3.91$\times $10$^{10}$& 
1227:  \\
1228: \hline
1229: p$_{n}$(0)/N$_{D}$& 
1230: 0.0156& 
1231: 1.56{\%}, small injection. At V=0.60V, this becomes 11{\%} \\
1232: \hline
1233: n$_{p}$(0)/N$_{A}$& 
1234: 3.91$\times $10$^{ - 9}$& 
1235: Extremely small injection \\
1236: \hline
1237: J$_{0h}$, A.cm$^{ - 2}$& 
1238: 3.663$\times $10$^{ - 11}$& 
1239:  \\
1240: \hline
1241: J$_{0e}$, A.cm$^{ - 2}$& 
1242: 1.195$\times $10$^{ - 13}$& 
1243:   $J_{0e} << J_{0h}$ \\
1244: \hline
1245: J$_{0}$, A.cm$^{ - 2}$& 
1246: 3.675$\times $10$^{ - 13}$& 
1247: $ \approx $J$_{0h}$ \\
1248: \hline
1249: J$_{0.55}$, A.cm$^{ - 2}$& 
1250: 0.0638& 
1251: $J=J_{0}[\exp(V/kT)-1]$ \\
1252: \hline
1253: E$_{n\times }$, V.cm$^{ - 1}$& 
1254: 0.0613& 
1255: Field in n-region far away from junction \\
1256: \hline
1257: E$_{nmax}$, V.cm$^{ - 1}$& 
1258: 0.1749& 
1259: Field just outside SCL \\
1260: \hline
1261: E$_{nmax}$/E$_{bi }$& 
1262: 8.13$\times $10$^{ - 6}$& 
1263:  \\
1264: \hline
1265: V$_{n }$at V=0.55V& 
1266: 0.00168& 
1267: Very small, V$_{n }$is 0.3{\%} of bias \\
1268: \hline
1269: V$_{n }$at V=0.60V& 
1270: 0.0116& 
1271: V$_{n }$is 12.3{\%} of bias  \\
1272: \hline
1273: E$_{pmax}$ = E$_{p\times }$, V.cm$^{ - 1}$& 
1274: 6.1$\times $10$^{ - 4}$& 
1275: Field far away from junction \\
1276: \hline
1277: E$_{pmax}$/E$_{bi }$& 
1278: 2.8$\times $10$^{ - 8}$& 
1279: Extremely small \\
1280: \hline
1281: V$_{p}$& 
1282: 5.04$\times $10$^{ - 7}$& 
1283: Extremely small, $ V_{p}<< V_{n} << V $ \\
1284: \hline
1285: \textbf{JUST OUTSIDE SCL ON n-SIDE, x=0}& & \\
1286: \hline
1287: Majority drift current, A.cm$^{ - 2}$& 
1288: 0.1820 & 
1289: Largest magnitude in positive direction \\
1290: \hline
1291: Majority diffusion current, A.cm$^{ - 2}$& 
1292: -0.1818& 
1293: Opposite direction, about the same magnitude as majority drift \\
1294: \hline
1295: Minority drift current, A.cm$^{ - 2}$& 
1296: 0.001& 
1297: Smallest magnitude \\
1298: \hline
1299: Minority diffusion current, A.cm$^{ - 2}$& 
1300: 0.06359& 
1301: About a third of the magnitude of majority drift. \\
1302: \hline
1303: \textbf{JUST OUTSIDE SCL ON p-SIDE, x'=0}& & \\
1304: \hline
1305: Majority drift current, A.cm$^{ - 2}$& 
1306: 0.06351& 
1307: Largest magnitude. Dominates conduction. \\
1308: \hline
1309: Majority diffusion current, A.cm$^{ - 2}$& 
1310: -0.000891& 
1311: Opposite direction \\
1312: \hline
1313: Minority drift current, A.cm$^{ - 2}$& 
1314: 2.33$\times $10$^{-19}$& 
1315: Smallest magnitude-virtually zero \\
1316: \hline
1317: Minority diffusion current, A.cm$^{ - 2}$& 
1318: 0.00255& 
1319: Next largest magnitude \\
1320: \hline
1321: \end{tabular}
1322: \caption{Properties of the p$^{ + }$-n junction }
1323: \label{tab2}
1324: \end{table}
1325: 
1326: \end{document}
1327: