1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % Converted to ReVTex 4.0
3: % Last modification : 10-12-2002
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5:
6: \documentclass[a4paper,prb,showpacs,superscriptaddress]{revtex4}
7: % RevTex 4.0 format
8:
9: \usepackage{epsfig}
10:
11: \def\baselinestretch{1.5}
12:
13: \textheight=8in
14: \topmargin=20pt
15:
16: \begin{document}
17:
18: \title{Hystersis like behaviour in Thin Films with heating-cooling cycle.}
19:
20:
21: \author{P. Arun}
22: \affiliation{Department of Physics \& Astrophysics, University of
23: Delhi, Delhi - 110 007, India}
24: \email{arunp92@physics.du.ac.in, arunp92@yahoo.co.in, agni@physics.du.ac.in}
25: \author{A.G. Vedeshwar}
26: \affiliation{Department of Physics \& Astrophysics, University of
27: Delhi, Delhi - 110 007, India}
28:
29: \begin{abstract}
30: The expression of temperature distribution along a film thickness is derived
31: and distribution of temperature in the film as the substrate is heated
32: is shown. The variation of film resistance with different substrate temperature
33: is calculated and the existence of temperature gradient along the film
34: thickness with finite thermal conductivity leads to hysteresis like
35: behaviour on heating-cooling the film.
36: \end{abstract}
37:
38: \pacs{73.61; 73.61.G; 81.40.C}
39:
40:
41: \maketitle
42:
43: \section{Introduction}
44:
45: The electrical conductivity measurements are important in characterising
46: conducting or semiconducting materials, both in their thin film and bulk
47: state. It is routinely carried out for various materials. The temperature
48: dependence of resistivity yields information about intrinsic band gap of
49: the material, the activation energy for conduction in films due to grain
50: boundary barrier height or impurity activation energy etc. For the above
51: estimation, the resistance measurement are taken either in the heating or
52: cooling direction of temperature variation. If the system is heated or
53: cooled very slowly, i.e. with same rate of change of temperature, both
54: the heating and cooling cycles coincide. However, considerable difference
55: has been observed when an amorphous film is heated above crystalline
56: transition temperature and cooled back to room temperature. This can
57: be understood as due to structural changes\cite{1, 2, 3}. In such cases after
58: cooling, most of the films do not regain their initial resistance. However
59: some did regain their initial resistance, thus enclosing an area as in
60: hysteresis loops. Hysteresis have been observed in bismuth films even
61: without structural changes\cite{4} where the heating and cooling rates were
62: kept different. This is interesting and since no attempt is made to explain
63: this variation, in this manuscript we attempt to explain the appearance of
64: hysteresis due to non-equilibrium state of the film.
65:
66:
67: \section{Theory}
68:
69: We consider the film to be kept on a copper block which is heated by a
70: heating coil embedded in it. The heating rate is varied by the voltage
71: applied to the heating coil, such that the whole surface of the copper
72: block is having uniform temperature. The film is kept on this copper block
73: resulting in heating from the substrate side (fig 1). The temperature varies
74: along the film thickness with time. The variation of temperature with time
75: and spatial co-ordinates is given by\cite{5}
76: \begin{eqnarray}
77: c_v{\partial T \over \partial t} = \lambda {\partial^2 T \over \partial x^2}
78: \end{eqnarray}
79: where ${\lambda}$ is the thermal conductivity of the film and ${\rm c_v}$ is
80: the specific heat of the film. A solution of this partial differential equation
81: depends on the initial and boundary conditions of the problem. Depending on the
82: initial and boundary conditions solution would be different\cite{6}. The initial
83: condition for the film of thickness 'd' being heated from the substrate side
84: would be given as
85: \begin{eqnarray}
86: T(x=0, t=0) = T_{sub} \\
87: T(x=d, t=0) = T_{sur}
88: \end{eqnarray}
89: while after a long time the film would be uniformly heated, with the
90: surface attaining the same temperature as that of the substrate, i.e.
91: \begin{eqnarray}
92: T(x, \infty) = T_{sub}\nonumber
93: \end{eqnarray}
94: From the given conditions, we search for a solution of the form
95: \begin{eqnarray}
96: T(x,t) = g(x)h(t)\nonumber
97: \end{eqnarray}
98: Or in other words the variables are separable and thus, solving equation (1) by
99: separable variable method we have the solution
100: \begin{eqnarray}
101: T(x,t)= a + b sin \left ( {\pi x \over 2 d} \right ) e^{-{\pi^2 D t \over 4
102: d^2}}
103: \end{eqnarray}
104: where D is the thermal diffusivity, ${\rm \lambda / c_v }$. Applying the
105: conditions stated in (2) and (3) the above solution may be written as
106: \begin{eqnarray}
107: T(x,t)= T_{sub} - (T_{sub}- T_{sur})sin\left ( \pi x \over 2 d \right )
108: e^{-{\pi^2 D t \over 4 d^2}}
109: \end{eqnarray}
110: Under experimental conditions, where resistivity or resistance is measured as a
111: function of temperature, the substrate temperature would be continuously changing
112: with time i.e. would be time dependent. Hence the above equation can not be
113: used as it is. Numerically, first the surface temperature is calculated for a
114: given substrate temperature at a given instant, along with temperatures along
115: the thickness of the film. The new surface temperature can be plugged back into
116: the expression along with the new substrate temperature. This scenario is valid
117: for film's with moderate thermal conductivity. Too carry out such a numerical
118: calculation we assume the substrate temperature to vary with time during the
119: heating cycle as
120: \begin{eqnarray}
121: T_{sub}(t) = P(1-e^{-Qt})+R\nonumber
122: \end{eqnarray}
123: where 'R' is the room temperature. We have numerically determined the
124: distribution of temperature along the thickness
125: of 1000\AA\, films of varying diffusivity, as described in the last section.
126: Figure 2 shows the variation of temperature along the thickness of three
127: different diffusivities (a) ${\rm 5 \times 10^{-3} \AA^2/sec}$, (b) ${\rm 5
128: \times 10^2 \AA^2/sec}$ and ( c) ${\rm 5 \times 10^3 \AA^2/sec}$.
129: Diffusivity, as already stated is the ratio of thermal conductivity to the
130: specific heat of the material. Comparing three different diffusivity of same
131: thickness implies different materials of same thickness is being studied. If
132: we assume there is not much variation in specific heat, the comparative study
133: is being done for materials of varying thermal conductivity (l). For numerical
134: computation we assume the values of the constants of equation (1) to be
135: ${\rm 360^oC}$, ${\rm 0.00039sec^{-1}}$ and ${\rm 14.5^oC}$ for P, Q
136: and R respectively. The saturation temperature that can be attained would be
137: ${\rm \sim 380^oC}$, which would be very high. Hence, we assume the heater is
138: switched off after 800sec, by which time, the copper block would be at ${\rm
139: \sim 110^oC}$. The family of curves show the spatial distribution of
140: temperature at various given time, namely after (i) 0 sec, (ii) 200sec,
141: (iii) 400sec, (iv) 600sec and (v) 800 sec of heating. As is evident from
142: figure 2(a) the surface remains at room temperature since heat does not spread
143: to the surface though the substrate is getting hotter with time. This is due to
144: the poor thermal conductivity of the film. This difference in surface and
145: substrate temperature decreases with time as can be understood from fig 2(b)
146: and (c).
147:
148:
149: If the heating and cooling is done in vacuum the cooling of the film, i.e.
150: loss of heat can take place by IR radiation losses or by conduction through
151: the substrate side. After the heater is switched off, since the process is in
152: vacuum, the substrate temperature remains constant for an appreciably long time
153: before it starts falling. We assume that the fall in substrate temperature
154: takes place after 200sec from instant that the heater is switched off. Due to
155: the temperature gradient present along the thickness of the film, the surface
156: tends to attain the same temperature as the substrate. Figure 3 shows the
157: temperature along the thickness of the film at various time, between the
158: instant when the heater was switched off to 200 sec after it was switched off.
159: It can be seen from figure 3 (a), the spatial distribution of temperature
160: remains the same even after 200 sec due to the poor thermal conductivity of
161: the film. However, as can be seen from fig 3(b) and (c), with improving thermal
162: conductivity, the film eventually attains equilibrium with time.
163:
164:
165: The variation of the substrates temperature with time is taken as
166: \begin{eqnarray}
167: T_{sub} = Se^{-ut} + R\nonumber
168: \end{eqnarray}
169: where S is the maximum temperature that the substrate had reached in the
170: heating cycle, before the onset of cooling. For computation of the temperature
171: distribution, to maintain assumption that the rate of cooling is very different
172: from the heating rate, 'u' is taken as ${\rm 0.00012sec^{-1}}$. Figure 4 shows
173: the temperature distribution along the length of the film at regular intervals
174: after the onset of cooling. For the film of low thermal conductivity (fig 4 a),
175: the distribution profile is very similar to that of heating cycle as shown in
176: fig 3 (a). However, as exhibited by figure 4 (b) and (c), the profile is
177: different for films with better thermal conductivity, where in some cases
178: (4c, ii-v), the surface is seen to at a higher temperature then the substrate.
179: This immediately suggests there would be some difference in film properties,
180: such as resistance, during heating and cooling cycle.
181:
182: \section{Film Resistance}
183:
184: The film can be thought of an numerous infinitesimal identical thin layers,
185: one on top of the other. All the layers acting as resistive elements with the
186: net resistance of the film being the effect of these resistance appearing in
187: parallel combination. Since the layers are identical, at room temperature all
188: of them have equal value. However, due to the metallic/ semiconducting nature
189: of the film, the resistance of these layers vary with temperature. The variation
190: of resistance with temperature is given as
191: \begin{eqnarray}
192: R = R_o(1+ \alpha T)\nonumber
193: \end{eqnarray}
194: where ${\rm \alpha}$ and ${\rm R_o}$ are the temperature coefficient of
195: temperature (TCR) and the resistance of the identical layer. For the case ${\rm
196: T=0^oC}$, the films resistance would be given as
197: \begin{eqnarray}
198: {1 \over R} = \sum_{i=1}^{i=n} {1 \over R_o} = {n \over R_o}
199: \end{eqnarray}
200: The TCR is positive for metal while it is negative for semiconductors. Since,
201: spatial distribution of temperature was calculated for various substrate
202: temperatures at various instant, the films resistance can be trivially
203: calculated as a function of substrate temperature and time.
204:
205: Figures 5-7 were plotted with data generated assuming the 1000\AA\, thick film
206: to be made up of 10 resistive layers in parallel combination, with each layer
207: to have a room temperature resistance of ${\rm 170 K\Omega}$ and ${\alpha= -0.80
208: \time 10^{-3o}C^{-1}}$. These numerical values are taken from our previous
209: study on ${\rm Sb_2Te_3}$ films \cite{7}. Figure 5 shows the variation of
210: resistance with substrate temperature. As can be seen films with
211: moderate thermal conductivity and those with good thermal conductivity
212: enclose very small area. However, films with intermediate diffusivity enclose
213: large area due to aggravated difference between the heating and cooling cycle.
214:
215: Figure 7 is of interest. The TCR or the variation of resistance with
216: temperature has been calculated for various diffusivity. It is evident that
217: the TCR of good thermally conducting films match the TCR of it's constituent
218: infinitesimal thin layer of which the film is made of. For a mathematical
219: analysis consider the film to be made up of infinite strips of layer, such that
220: each neighbouring layer has a slightly different temperature and inturn a
221: slightly different resistance. The summation sign of equation (6) may then
222: be replaced by an integration sign, hence the net resistance of the film
223: would be given as
224: \begin{eqnarray}
225: {1 \over R} = {1 \over R_o}\int_{i=0}^{n=d/a} {di \over (1 + \alpha
226: T)}\nonumber
227: \end{eqnarray}
228: At an given instant the temperature is given by equation (5), hence the
229: above equation can be re-written as
230: \begin{eqnarray}
231: {R_o \over R} = \int_{0}^{n} {di \over (1 + \alpha T_{sub})
232: -\alpha(T_{sub}-T_{sur})e^{-{\pi^2 D t \over 4 d^2}} sin \left ( {\pi \over 2
233: n}i \right )}
234: \end{eqnarray}
235: For solving the above equation, we substitute
236: \begin{eqnarray}
237: A &=& 1 + \alpha T_{sub}\nonumber \\
238: B &=& \alpha(T_{sub}-T_{sur})e^{-{\pi^2 D t \over 4 d^2}}\nonumber \\
239: x &=& {\pi \over 2 n}i\nonumber
240: \end{eqnarray}
241: Thus, equation (7) can be written as
242: \begin{eqnarray}
243: {\pi R_o \over 2 nR} = \int_{0}^{\pi /2} {di \over A- Bsin \left ( {\pi \over 2
244: n}i \right )}
245: \end{eqnarray}
246: \begin{eqnarray}
247: {\pi R_o \over 2 n R} = {1 \over \sqrt{A^2-B^2}} tan^{-1} \left ( \sqrt{{A+B
248: \over A-B}} \right )
249: \end{eqnarray}
250: As the film's diffusivity increases, the term B becomes smaller and
251: smaller, i.e. tends to zero. The above equation then reduces to
252: \begin{eqnarray}
253: {\pi R_o \over 2 n R} = {1 \over A} tan^{-1}(1)= {\pi \over 2A}\nonumber
254: \end{eqnarray}
255: or
256: \begin{eqnarray}
257: {nR \over R_o} = 1+\alpha T\nonumber
258: \end{eqnarray}
259: on re-arranging
260: \begin{eqnarray}
261: R = {R_o \over n}(1+\alpha T)\nonumber
262: \end{eqnarray}
263: using the equation showing rise in temperature with time, we have
264: \begin{eqnarray}
265: {dR \over dt} = {R_o \over n}(\alpha PQe^{-Qt})\nonumber
266: \end{eqnarray}
267: and
268: \begin{eqnarray}
269: {dT \over dt} = PQe^{-Qt}\nonumber
270: \end{eqnarray}
271: Thus the film's TCR works out as
272: \begin{eqnarray}
273: TCR_{film} = {n \over R_o}{dR \over dT} = \alpha\nonumber
274: \end{eqnarray}
275: The mathematics show that not only does a good thermally conducting film's TCR
276: match that of the infinitesimal thin layer of which the film is made of, it
277: is also independent of the rate of heating/ cooling. It can be inferred that
278: the TCR of the film's with lower thermal conductivity would show dependence on
279: the rate of heating and cooling. This can be seen from figure 8. Figure 8 shows
280: the effect the rate of heating would have on the slope, and
281: inturn the TCR. The data was calculated in the same manner as
282: discussed in the previous sections. While Figure 8A exhibits the variation
283: of resistance with temperature for a poor thermally conducting film
284: (D=500\AA/sec), Figure 8B is for a good conducting film (D=5000\AA/sec).
285: Three curves are present in both figures, each for different heating rates,
286: namely (i) ${\rm 3.6 \times 10^{-3o}C/sec}$, (ii) ${\rm 72 ^oC/sec}$ and
287: (iii) ${\rm 216 ^oC/sec}$. All three curves coincide for the conducting film.
288: However, in the figure 8A, where a low thermal conducting film the curves do
289: not coincide and their slopes are different. Thus, the TCR values would depend
290: on the rate of heating and cooling. An interesting feature is that the
291: resistances at various temperatures of a poor conducting film measured at
292: very low heating rates match those of a good conducting film being heated
293: rapidly.
294:
295: \section{Conclusions}
296: The electrical studies of thin films are usually done by heating the sample
297: and measuring resistance/ resistivity with temperature. Though, the
298: measurements are to be done after the film has attained a steady
299: temperature, usually the measurement is done as the film is being heated or
300: cooled. As discussed in the article, if the film has a finite thermal
301: conductivity (i.e. it is not metallic), one essentially is making
302: measurement in non-equilibrium conditions. Thus, parameters like TCR etc.
303: computed is not only material dependent but depends on conditions of the
304: experiment, e.g. the rate of heating or cooling. It is essentially due to
305: this non-equilibrium measurement that leads to a loop like formation due to
306: the heating-cooling cycle.
307:
308: \begin{acknowledgments}
309: The authors would like to acknowledge the contributions of Prof. S. R.
310: Choudhury, Pankaj Tyagi and Naveen Gaur.
311: \end{acknowledgments}
312:
313: \pagebreak
314: \begin{thebibliography}{99}
315:
316: \bibitem{1} V. Damodara Das and D. Karunakaran, Phys. Rev. B., 39 (1989) 10
317: 872.
318:
319: \bibitem{2} V. Damodara Das and P. Gopal Ganesan, Solid State Commun., 106
320: (1998) 315.
321:
322: \bibitem{3} V. Damodara Das and S. Selvaraj, J. Appl. Phys., 83 (1993) 3696.
323:
324: \bibitem{4} K. Jayachandran and C. S. Menon, Pramana, 50 (1998) 221.
325:
326: \bibitem{5} H. S. Carslaw and J. C. Jaeger, "Conduction of Heat in Solids",
327: (Oxford Univ. Press, Oxford 1954).
328:
329: \bibitem{6} S. M. Sze, "Semiconductor Devices, Physics and Technology", (Wiley,
330: New York1993).
331:
332: \bibitem{7} P. Arun and A. G. Vedeshwar, presented for publication in
333: Physica B.
334:
335: \end{thebibliography}
336:
337:
338: \pagebreak
339:
340:
341: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
342: % Figure inclusion (at the end of paper)
343: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
344: \begin{figure}
345: \begin{center}
346: \epsfig{file=fig1.ps,width=5 in}
347: \caption{ Direction of heat flow and initial condition of temperature on both
348: surfaces of the film.
349: }
350: \label{fig:1}
351: \end{center}
352: \end{figure}
353:
354: \begin{figure}
355: \begin{center}
356: \epsfig{file=fig2.ps,height=7.5in, width=4 in}
357: \caption{Variation of temperature along the thickness of a 1000\AA\, thick
358: film of different diffusivity (a) ${\rm 5 \times 10^{-3}\AA/sec}$,
359: (b) ${\rm 5 \times 10^{2}\AA/sec}$ and (c) ${\rm 5 \times 10^{3}\AA/sec}$ after
360: (i) 0 sec, (ii) 200 sec, (iii) 400 sec, (iv) 600 sec
361: and (v) 800 seconds of substrate heating.}
362: \label{fig:2}
363: \end{center}
364: \end{figure}
365:
366: \begin{figure}
367: \begin{center}
368: \epsfig{file=fig3.ps,height=6.75in, width=4 in}
369: \caption{Variation of temperature along the thickness of a 1000\AA\, thick
370: film of different diffusivity (a) ${\rm 5 \times 10^{-3}\AA/sec}$,
371: (b) ${\rm 5 \times 10^{2}\AA/sec}$ and (c) ${\rm 5 \times 10^{3}\AA/sec}$ after
372: (i) 0 sec, (ii) 40 sec, (iii) 80 sec, (iv) 120 sec, (v) 160 sec
373: and (vi) 200 seconds after the source of heating was switched off.}
374: \label{fig:3}
375: \end{center}
376: \end{figure}
377:
378:
379: \begin{figure}
380: \begin{center}
381: \epsfig{file=fig4.ps, height=7.5in, width=4.2in}
382: \caption{ Variation of temperature along the thickness of a 1000\AA\, thick
383: film of different diffusivity (a) ${\rm 5 \times 10^{-3}\AA/sec}$,
384: (b) ${\rm 5 \times 10^{2}\AA/sec}$ and (c) ${\rm 5 \times 10^{3}\AA/sec}$ after
385: (i) 8000 sec, (ii) 16000 sec, (iii) 24000 sec, (iv) 32000 sec and (v) 40000
386: seconds after
387: the setting in of the film's cooling.}
388: \label{fig:5}
389: \end{center}
390: \end{figure}
391:
392: \begin{figure}
393: \begin{center}
394: \epsfig{file=fig5.ps,width=5.5 in}
395: \caption{
396: Hysteresis loops formed in film resistance with the heating -cooling cycle.
397: The calculations were done for film thickness of 1000\AA\, and diffusivity (i)
398: ${\rm 5 \times 10^{-3} \AA^2/sec}$, (ii) ${\rm 50 \AA^2/sec}$, (iii) ${\rm 5
399: \times 10^2 \AA^2/sec}$ and (iv) ${\rm 5 \times 10^3 \AA^2/sec}$.
400: }
401: \label{fig:5}
402: \end{center}
403: \end{figure}
404:
405: \begin{figure}
406: \begin{center}
407: \epsfig{file=fig6.ps,width=5 in}
408: \caption{ The variation in the area enclosed by loops formed during the
409: resistance variation with temperature during heating-cooling cycles. The
410: variation is due to the difference in the films diffusitivity. }
411: \label{fig:6}
412: \end{center}
413: \end{figure}
414:
415:
416: \begin{figure}
417: \begin{center}
418: \epsfig{file=fig7.ps,width=5 in}
419: \caption{ Computed TCR for films of different diffustivity, where the films
420: are assumed to be of same thickness and made up of numerous layers, with all
421: the layers having the same TCR.}
422: \label{fig:7}
423: \end{center}
424: \end{figure}
425:
426: \begin{figure}
427: \begin{center}
428: \epsfig{file=fig8.ps,width=5 in}
429: \caption{ Figure exhibits the change of resistance of with temperature for
430: (A) a poor thermal conducting film and (D= ${\rm 5 \times 10^{2}\AA/sec}$)
431: (B) a good thermally conducting film (D= ${\rm 5 \times 10^{3}\AA/sec}$).
432: The heating rates were maintained different (i) ${\rm 3.6 \times
433: 10^{-3o}C/sec}$, (ii) ${\rm 72 ^oC/sec}$ and (iii) ${\rm 216 ^oC/sec}$.}
434: \label{fig:8}
435: \end{center}
436: \end{figure}
437: \end{document}
438: