cond-mat0505145/trm.tex
1: %\documentclass[twocolumn,aps,showkeys,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: \documentclass[preprint,aps,showkeys,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: %\documentclass[preprint,aps]{revtex4}
4: %\documentclass[preprint,aps,draft]{revtex4}
5: %\documentclass[prb]{revtex4}% Physical Review B
6: 
7: \usepackage{graphicx}
8: \usepackage{dcolumn}
9: \usepackage{bm}
10: 
11: %\nofiles
12: 
13: 
14: \begin{document}
15: 
16: %\preprint{APS/123-QED}
17: 
18: \title{Relaxation of thermo-remanent magnetization in Fe-Cr GMR multilayers}
19: 
20: \author{R. S. Patel}
21: \author{A. K. Majumdar}
22:  \email{akm@iitk.ac.in}
23: \affiliation{Department of Physics, Indian Institute of Technology,
24: Kanpur-208 016, India}
25: 
26: \author{A. K. Nigam}
27: \affiliation{Tata Institute of Fundamental Research, Homi Bhabha Road,
28: Mumbai-400 005, India}
29: 
30: 
31: \date{\today}
32:              
33: 
34: \begin{abstract}
35: 
36: The time decay of the thermo-remanent magnetization (TRM) in Fe-Cr
37: giant magnetoresistive (GMR) multilayers has been investigated.
38: The magnetization in these multilayers relaxes as a function of time after
39: being cooled in a small magnetic field of 100 Oe to a low temperature
40: and then the magnetic field is switched off. Low-field
41: ($<$ 500 Oe) magnetization studies of these samples have shown
42: hysteresis. This
43: spin-glass-like behavior may originate from structural imperfections at
44: the interfaces and in the bulk. We find that the magnetization
45: relaxation is logarithmic. Here the magnetic viscosity
46: is found to increase first with increasing temperature, then it reaches
47: a maximum around T$_g$, and then it decreases with increasing temperature.
48: This behavior is different from that of conventional spin glasses
49: where the logarithmic creep rate is observed to increase with
50: temperature. Power law also gives good fits and it is better than the logarithmic
51: fit at higher temperatures. The dynamical
52: effects of these multilayers are related to the 
53: relaxation of thermally blocked superparamagnetic grains and 
54: magnetic domains in the film layers.
55: 
56: \end{abstract}
57: 
58: \pacs{75.70.Cn, 75.50.Lk, 76.60.Es, 75.10.Nr}
59:                              
60: \keywords{Thermoremanence, Magnetization, Spin glass, Fe-Cr multilayers}
61: 
62: \maketitle
63: 
64: 
65: \section{INTRODUCTION}
66: In Fe-Cr GMR multilayers the ferromagnetic (FM)
67: Fe layers are exchange coupled through the non-magnetic 
68: Cr spacer layers. An antiferromagnetic arrangement of the Fe layers is 
69: engineered by varying Cr spacer layer thickness. With varying Cr 
70: thickness successive Fe layers show oscillatory antiferromagnetic 
71: (AFM) and FM couplings but with decreasing peak coupling strength.
72: It is believed that in an external magnetic field $H>H_{sat}$, the magnetization
73: in the Fe layers are aligned as in a ferromagnetic material but in 
74: the absence of the magnetic field they are in antiferromagnetic configuration.
75: Our study is focussed at finding the magnetic relaxation of these multilayers
76: in low external magnetic fields ($\sim$100 Oe).
77: 
78: 
79: Thermoremanent magnetization is a thermally activated process. When the 
80: applied magnetic field is removed the magnetization tries to approach 
81: the remanent magnetization value in order to minimise the energy of 
82: the system. A magnetic system has dipolar energy, anisotropy energy, 
83: and exchange energy. In general, ferromagnetic materials do not 
84: respond to the magnetic fields immediately. There is a time lag between 
85: the application/withdrawl of the magnetic field and the response of magnetization 
86: to the field. This phenomenon is called ``magnetic viscosity''
87: \cite{Street:1949}. Folks and Street \cite{Folks:1994} described
88: this time lag by domain processes which progress through states of metastable
89: equilibrium to a stable state. The major domain processes are:
90: \begin{enumerate}
91: \item coherent or incoherent rotation of magnetization in single domain particles,
92: \item pinning and unpinning of domain boundary walls, and
93: \item nucleation of domains of reverse magnetization.
94: \end{enumerate}
95: 
96: 
97: 
98: Coherent rotation of magnetization vector and the
99: Bloch-wall formation are primary consequences of lowering energy after 
100: the external field is applied/removed. Thermal agitation plays
101: an important role in transition from metastable to stable
102: states. One may ask what are the time scales to achieve stable 
103: states for paramagnetic, FM, and AFM materials. What is the role of magnetic
104: interaction and crystal structure in the time dependence of magnetization? 
105: 
106: 
107: Dahlberg et al. \cite{Dahlberg:1994} explained the time 
108: dependence as a consequence of interactions or couplings. The 
109: interaction between relaxing spins, the dipole coupling, drives the
110: system from an initial state towards the equilibrium state. They attributed the 
111: strong time dependence of the magnetization in Co-Cr films to the 
112: demagnetization in the film.
113: 
114: 
115: Sinha \cite{Sinha:1996} et al. found that in $Fe_{80-x}Ni_xCr_{20}$ ($x$ = 30,
116: FM phase) M(t) fits well to the power law decay. They found that with the increase
117: of wait time, the exponent becomes smaller. For $x$ =14 (AF phase) also,
118: power law decay describes M(t) quite well and the 
119: exponent decreases with increasing temperature.
120: Chamberlin \cite{Chamberlin:1991} studied EuS which is considered to be an
121: ideal Heisenberg FM system with a Curie temperature $\sim 16.6$ K.
122: They found that the plots of remanent magnetization
123: versus logarithm of time show negative curvature for $T<T_B$(=17.75 K) and 
124: exhibits an S-shaped curve for $T>T_B$ . Ulrich et al.
125: \cite{Ulrich:2003} studied the magnetic relaxation of single
126: domain ferromagnetic particles below the blocking temperature. 
127: They found that for all particle densities the relaxation decays following
128: a power law, with density-dependent exponent and a temperature-dependent
129: prefactor. They used Monte Carlo simulations to study the 
130: influence of dipolar interactions and polydispersion on the magnetic 
131: relaxation of single-domain FM particles below the blocking temperature.
132: They concluded (i) stretched exponential decay at low densities, (ii) a power
133: law decay at intermediate densities, and (iii) relaxation toward a non-
134: vanishing remanent magnetization at high densities.
135: 
136: 
137: 
138: In an earlier study of $Cu_{100-x}Mn_x(x$ = 76, 83 and both in AF phase) we found 
139: \cite{Patel:2002} that the power law
140: decay is better at lower temperatures. The fits at a given temperature 
141: improve with stronger long-range antiferromagnetic order (AF1 structure).
142: Leighton and Schuller \cite{Leighton:2001} used time-dependent magnetization
143: measurements to probe the asymmetry in the magnetization reversal mechanism
144: in exchange-biased $MnF_2/Fe$ bilayers. They found that on one side of the loop,
145: coherent rotation of magnetization plays an important role while on the other side
146: the domain nucleation. The time dependence of the magnetization on the
147: coherent rotation side of the loop had a form that is consistent with a small 
148: distribution of energy barriers. The side of the loop characterised by domain
149: nucleation and propagation shows a logarithmic time dependence with a field
150: dependent viscosity. Similar investigation was also carried by
151: Fullerton and Bader \cite{Fullerton:1996} on Fe/FeSi multilayers.
152: 
153: 
154: Chamberlin et al. \cite{Chamberlin:1984} studied the time decay of the
155: thermoremanent magnetization in 1.0 \% CuMn and 2.6 \% AgMn spin glasses.
156: They found it to be a stretched exponential function. Till that time no data had been
157: published supporting an algebraic decay (power law) for magnetization.
158: Chamberlin also \cite{Chamberlin:1984b} found that the effect of wait-
159: time $t_w$ can be empirically characterised as an exponential decrease
160: of the relaxation frequency with increasing wait time. If the sample is 
161: allowed to remain in the field-cooled state long enough before the field 
162: is removed, then the magnetization will not relax: a spin glass can have 
163: a permanent magnetization in zero field. Chubykalo and Gonzalez
164: \cite{Chubykalo:1999} simulated the relaxational behavior of
165: Co-Ni multilayers with different layer thickness. Their simulations
166: had shown that the thermally activated demagnetization process in 
167: Co-Ni multilayers does not occur according to the Arrhenius kinetics.
168: Panagiotopoulos et al. \cite{Panagioto:2002} found that the
169: magnetic relaxation follows the ln(t) behavior at 5 K in La-Ca-Mn-O
170: FM/AF multilayer with $T_B$ = 70 K. In FM films that exhibit a wide
171: range of energy barriers the magnetization time decay follows a ln(t)
172: behavior only below a blocking or freezing temperature, resulting from
173: the superposition of many exponential decays with different magnetic
174: relaxation times. Chen et al. \cite{Chen:2003} found that the magnetization 
175: relaxation of a $[Co_{80}Fe_{20}(1.4 nm)/Al_2O_3(3 nm)]_{10}$ sample
176: obeys a power-law decay of the thermoremanent magnetic moment with
177: an asymptotic remanence when starting from a completely relaxed FC state.
178: 
179: 
180: Street and Brown \cite{Street:1994} described two types of mechanisms that are
181: responsible for the time dependence in ferromagnetic materials known as 
182: ``diffusion'' and ``fluctuation'' after-effect or viscosity. Chantrell et al.
183: \cite{Chantrell:1994} gave a phenomenological theory based on the intrinsic
184: energy barrier to explain the form of time dependence of the magnetization.
185: The slow relaxation is related to the irreversible magnetic behavior via a 
186: fictitious fluctuation field $H_f$ which itself determines a quantity called
187: the activation volume $V_{act}$. Both $H_f$ and $V_{act}$ are related to
188: the magnetization reversal process. For granular materials $V_{act}$ is generally
189: smaller than the grain size. A simple model for the time dependence of the switching 
190: field in magnetic recording media by Sharrock \cite{Sharrock:1994} accounts for thermally
191: assisted crossing of an energy barrier whose height is reduced by the applied field. Goodman
192: et al. \cite{Goodman:2000} studied a system with a narrow distribution of energy barriers.
193: In such systems, at fields less than the coercive field, an accelerating variation of magnetization
194: with log of time is seen and a decelerating behavior of field above the coercivity.
195: In the coercive field region an `S' shaped variation is observed.
196: 
197: 
198: 
199: 
200: \section{EXPERIMENTAL DETAILS}
201: Our samples were grown on Si substrates by ion beam sputter deposition
202: technique using Xe ion at 900 V with beam current of 20 mA and 
203: 1100 V with beam current of 30 mA. 
204: The typical structures are Si/Cr(50 \AA)/[Fe(20 \AA)/Cr(t \AA)]$\times$
205: 30/Cr(50 -t \AA). Samples 1 - 4 have t = 6, 8, 10 and 12 \AA, respectively. 
206: Samples 1 and 2 are deposited at 900 V and samples 3 and 4 are at 
207: 1100 V. Our multilayers show a GMR $(=((\rho(H,T)-\rho(0,T))/\rho(0,T))
208: \times 100 \%)$ of $\approx$ 1, 33, 20 and 21 \% at 4.2 K for an 
209: external field of $\approx$ 1 tesla ($H_{sat}$). These samples are 
210: well characterised and the details have been 
211: given elsewhere\cite{Lannon:2002}. 
212: All the experiments were done with a Quantum Design superconducting 
213: quantum interference device (SQUID) magnetometer(MPMS). A magnetic 
214: field of 100 Oe is applied in the plane of the multilayer samples
215: at 300 K and the sample is cooled down to the measuring temperature. After
216: the temperature is stabilised, the magnetic field is set to zero and
217: M(T) measurements were started and continued till 12,000 s.
218: 
219: 
220: 
221: 
222: \section{THEORY}
223: Chamberlin \cite{Chamberlin:1994} summarised the time dependence 
224: of magnetization in terms of different mathematical expressions as follows:
225: 
226: \begin{enumerate}
227: \item The most popular empirical expression for characterising amorphous
228: materials has been the Kohlrausch-Williams-Watts stretched exponential
229: $M(t) \propto exp(-(t/\tau)^\beta) $.
230: \item For crystals, the Curie-von Schweidler power law $M(t) \propto 
231: t^{-\beta}$.
232: \item For magnetic aftereffects, the Ne\`{e}l logarithmic time 
233: dependence $M(t) \propto log(t/\tau)$ is popular.
234: \end{enumerate}
235: When the applied magnetic field is removed, the magnetization takes 
236: finite time to cross the energy barrier for reversal. If the energy 
237: barriers are identical then the magnetic moment
238: $M$ is given by \cite{Leighton:2001}
239: 
240: \begin{equation}
241: \label{eq:exp}
242: M(t) = A + B \ exp(-t/\tau_0),
243: \end{equation}
244: 
245: \noindent
246: where
247: 
248: \begin{equation}
249: \frac{1}{\tau_0} = f_0 \ exp\left(\frac{E_A}{k_BT}\right),  \nonumber
250: \end{equation}
251: $A$ and $B$ are constants, $t$ represents
252: the elapsed time, $f_0$ is the attempt frequency, $E_A$ is the
253: activation energy, and T is the 
254: measurement temperature. When there is a distribution of barriers, then
255: 
256: \begin{equation}
257: \label{eq:ln}
258: M(t) = M_0 - S \ ln(t).
259: \end{equation}
260: 
261: Here S is called the ``magnetic viscosity" and $M_0$ is a constant 
262: at a given measuring field. S is expected to reach a peak value near
263: the coercive field $H_c$ where the rate of change of the moment with 
264: time reaches the maximum. The intermediate case with a small 
265: distribution of barrier heights is more complicated to quantify.
266: 
267: Power-law behavior is predicted by scaling theories for domain growth 
268: and internal dynamics \cite{Chamberlin:1991}. Calculations based on
269: Sherrington-Kirkpatrick mean-field model suggest that the magnetization
270: decays algebraically as
271: 
272: 
273: \begin{equation}
274: \label{eq:power}
275: M(t) = M_0 \ t^{-\beta}.
276: \end{equation}
277: 
278: 
279: Other treatments yield Kolrausch-Williams-Watts stretched-exponential
280: relaxation which is common in spin-glass like systems 
281: \cite{Chamberlin:1994} given by
282: 
283: 
284: \begin{equation}
285: \label{eq:strexp}
286: M(t) = M_0 \ exp\left( -\left( \frac{t}{\tau} \right)^n \right).
287: \end{equation}
288: 
289: When measured in a
290: small static field, the temperature dependence of the magnetization of a
291: spin glass changes abruptly at the glass transition temperature ($T_g$).
292: Above $T_g$ the magnetization obeys the Curie-Weiss law, attributable 
293: to weakly interacting paramagnetic spins whereas below $T_g$ the 
294: magnetization is cooling history dependent. If it is field cooled then the
295: magnetization is nearly independent of temperature below $T_g$; if it is zero-field 
296: cooled then the magnetization increases with increasing temperature.
297: 
298: 
299: \section{RESULTS AND DISCUSSION}
300: Structure-wise our samples are a good combination of crystalline 
301: layers of ferromagnetic Fe and antiferromagnetic Cr and the polycrystalline interfaces.
302: So these samples provide a complicated combination of crystalline layers 
303: and polycrystalline interfaces consisting of ferromagnetic, 
304: antiferromagnetic and spin-glass like structures as shown schematically
305: in Fig. \ref{fig:sample}.  In this figure black and grey spheres represent Cr and Fe atoms,
306: respectively. Directions of the arrow show the direction of spin alignment.
307: Typical estimates of the relative strengths of the Fe-Fe, Fe-Cr, and Cr-Cr 
308: couplings are 1 : -0.3 : -0.18 \cite{Shi:1997} and 1 : -0.55 : 
309: -0.3 \cite{Berger:1997}. In other words, Fe-Fe coupling is FM and Fe-Cr
310: \& Cr-Cr couplings are AF. Fe-Fe coupling is two/three times stronger than Fe-Cr coupling.
311: `?' in Fig. \ref{fig:sample} represents uncertainty of spin direction(frustration) at that site. For a particular direction,
312: either it will be violating the FM Fe-Fe coupling or the AF Fe-Cr coupling. As the Fe-Fe coupling
313: is stronger than the Fe-Cr and Cr-Cr coupling most of the frustration occurs at Cr sites (black spheres).
314: Attempts to explain magnetic relaxation with mathematical 
315: expressions began a century back. We have made a similar attempt to understand
316: the response of the combined system for small magnetic fields. Although
317: Chamberlin\cite{Chamberlin:1994} argued that these empirical expressions
318: are simple mathematical formulae that give generally good agreement with
319: a wide variety of measurements but demonstrating agreement with these 
320: formulae tells nothing about the mechanism of response.
321: 
322: 
323: 
324: For our analysis, we have used a standard non-linear least-squares-fit
325: program. Here $\chi^2$ is defined as
326: \begin{equation}
327: \chi^2 = \frac{1}{N} \sum_{i=1}^{N} 
328: \frac{(M_{i-measured}-M_{i-fitted})^2}{M_{i-mean}^2}.
329: \end{equation}
330: 
331: In our multilayer
332: system, the ferromagnetic Fe layers have in-plane magnetization but as 
333: they are not super-lattices, there will be some domain 
334: like structures in the ferromagnetic plane with distribution of
335: domain volume. Thus we expect that there is a
336: distribution of barriers for which logarithmic relaxation of magnetization is generally found.
337: So we first tried the logarithmic fit in the form of 
338: Eq. (~\ref{eq:ln}). For our analysis we have taken data from 100 s to
339: 12,000 s. We have made measurements with a SQUID magnetometer and it 
340: takes $\sim$ 10 s to make one measurement and a much longer time to 
341: achieve stability and equilibrium condition after removal of the 
342: magnetic field. So we have not analysed the data for t $<$ 100 s.
343: 
344: 
345: 
346: 
347: Logarithmic relaxation gives excellent fits for samples 2 - 4 with correlation coefficients
348: $\sim$ 0.997 and the normalised $\chi^2$ consistent with the experimental 
349: resolution of one part in $10^4$. For sample 1,
350: which has the least GMR, the fit is not so good ($R^2 \sim
351: 0.90$). Figures \ref{fig:S2ln}, \ref{fig:x4alllog}, and \ref{fig:x6alllog}
352: show the raw M vs. ln t data and the fits for samples 2, 3, and 4,
353: respectively at different temperatures. For each sample,
354: $M_0$ decreases with increasing temperature which can
355: be understood as the decrease in magnetization due to thermal 
356: fluctuations. The coefficient S, which is termed as the magnetic viscosity, 
357: initially increases linearly with temperature; then it decreases at higher
358: temperatures. Larger S implies larger change in the magnetization during the
359: observational time period. S shows a peak around $T_g$ of the ZFC 
360: magnetization curves (samples 2 - 4 show $T_g$ at $\sim$ 62, 123, and 83 K,
361: respectively taken at 100 Oe \cite{Patel:2005}).
362: This can be understood as follows: At temperatures much lower than 
363: the freezing temperature, due to the low temperature (less thermal 
364: energy), the magnetic moments can not relax much (as they are nearly frozen). So 
365: lower the temperature lower is the value of S. At temperatures
366: much higher than the freezing temperature, due to higher thermal energy 
367: the system behaves more and more like a paramagnetic/superparamagnetic 
368: system. In the paramagnetic region the entire magnetization relaxes very
369: rapidly ($<$10 s)\cite{Chamberlin:1984}. So in our measurement time window
370: ($10^2 - 10^4$ s) we hardly observe any relaxation and hence S looks small.
371: At around $T_g$ in the ZFC magnetization curve, the
372: thermal energy provides the moments freedom to move and they can also 
373: interact with one another. In this region, the system is like a spin glass
374: which is about to unfreeze and so the
375: magnetization responds much faster. That is why we find maxima in 
376: the magnetic viscosity vs. temperature curves. This is shown in 
377: Fig. \ref{fig:pwrp}. To make the comparison more clear we have shown the percentage
378: change in magnetization with time at different temperatures for sample 2 in Fig. \ref{fig:f83d}.
379: Similar variation of S with temperature was found by Guy \cite{Guy:1978} in a
380: Au-Fe spin-glass alloy containing 2 at. \% Au. However, Lottis et al. \cite {Lottis:1988} found 
381: broad maxima in the temperature dependence of S in Co-Cr films. 
382: 
383: Figure  \ref{fig:S1ln} shows the raw M vs. ln t data and the fit for sample 1. Sample 1
384: showed the least GMR. The interpretation for the least GMR is that in this
385: sample with Cr thickness of 6 \AA \ most Fe layers are coupled ferromagnetically to the neighbouring Fe
386: layers. So the whole sample is like a FM material. The ZFC magnetization measurements
387: of this sample show that the magnetization decreases monotonically with 
388: increasing temperature. So measurement temperatures of 50,
389: 100, and 150 K are much above the $T_B$ of this sample. The magnetization,
390: plotted on a logarithmic scale, has clearly an `S'- shape as shown in Fig. \ref{fig:S1ln}.
391: 
392: 
393: Next we tried to fit the data to the power law in the form of Eq. 
394: (~\ref{eq:power}). We find that the power law
395: gives equally good fits for samples 2 - 4, with $R^2 >$ 0.99 along with
396: $\chi^2$ consistent with the experimental resolution. 
397: The numerical value of the parameter $\beta$
398: is the same as of the parameter `S' divided by $M_0$ of Eq. (~\ref{eq:ln}). We can rewrite
399: Eq. (~\ref{eq:power}) as
400: 
401: \begin{equation}
402: \label{eq:power1}
403: M(t) = M_0 \ t^{-\beta} = M_0(e^{-\beta \ ln(t)}) \\
404: \approx M_0 - (M_0\beta) \ ln(t),
405: \end{equation}
406: since $\beta$ is very small and so we can neglect higher order terms. This has
407: a form similar to that of Eq. (~\ref{eq:ln}). For larger time scale $t > 10^5$ s, $t^{-\beta}$
408: decreases faster than $ln(t)$. We have tried to show this feature in Fig. 
409: \ref{fig:S3Ex}. In other words, from our measurement time window we shall not
410: be able to distinguish log and power fits. However, a careful observation from Tables I-IV 
411: indicates a small difference between the two fits. Eq.
412: (~\ref{eq:ln}) gives better fit than Eq. (~\ref{eq:power}) at low temperatures
413: but at higher temperatures (much above $T_g$) Eq. (~\ref{eq:power}) gives better
414: fits. Sample 1 does not have $T_g$ or $T_B$ in the measured temperature
415: range and so the power law fit is better for all the temperatures. A stronger proof would 
416: have been possible if we could take data till $10^5 - 10^6$ s.
417: 
418: 
419: 
420: Next we tried the stretched exponential fit in the form of Eq. (~\ref{eq:strexp}).
421: The low-field magnetization measurements on samples 2-4 had shown
422: spin-glass-like history dependent behaviour. But this fitting gave unrealistic
423: large time constants $(\tau)$ with equally large errors.
424: 
425: 
426: To conclude, in these sputtered GMR multilayers which are structural combinations
427: of FM Fe films, AF Cr Films, and Fe-Cr mixed interfaces, the magnetization decays
428: logarithmically at low temperatures and above $T_g$ the magnetization decays
429: algebraically. In the multilayers one expects less distribution of energy barriers
430: but we could not find any exponential decay. For `AFM' (sample 2, 3, 
431: and 4) and `FM' (sample 1) multilayers we found different decay mechanisms. The power law decay
432: is better at higher temperatures. It is difficult to distinguish between `good
433: AFM' (sample 2 with GMR of 33 \%) and `AFM' (sample 3 with GMR of 21 \%
434: and sample 4 with GMR of 20 \%) samples from their decay behavior.
435: 
436: 
437: \begin{acknowledgments}
438: We sincerely thank Drs. D. Temple and C. Pace of MCNC, Electronic Technologies
439: Division, Research Triangle Park, North Carolina for providing us the samples.
440: One of us (R.S.P.) acknowledges CSIR, Government of India, for financial support.
441: \end{acknowledgments}
442: 
443: \begin{thebibliography}{6}
444: \bibitem{Street:1949} R. Street and J. C. Woolley, Proc. Phys. Soc.
445: London Sec. A {\bf 62}, 562 (1949).
446: 
447: \bibitem{Folks:1994}  L. Folks and R. Street, J. Appl. Phys. {\bf 76}, 
448: 6391 (1994).
449: 
450: \bibitem{Dahlberg:1994} E. Dan Dahlberg, D. K. Lottis, R. M. White, M. M. 
451: Matson, and E. Engle, J. Appl. Phys. {\bf 76}, 6401 (1994).
452: 
453: \bibitem{Sinha:1996} G. Sinha, R. Chatterjee, M. Uehara, and A. K. Majumdar, J. 
454: Magn. Magn. Mater. {\bf 164}, 345 (1996).
455: 
456: \bibitem{Chamberlin:1991} R. V. Chamberlin and F. Holtzberg, Phys. Rev. Lett.
457: {\bf 67}, 1606 (1991).
458: 
459: \bibitem{Ulrich:2003} M. Ulrich, J. Garcia-Otero, J. Rivas, and A. Bunde, 
460: Phys. Rev. B {\bf 67}, 024416 (2003).
461: 
462: \bibitem{Patel:2002} R. S. Patel, D. Kumar, and A. K. Majumdar,
463: Phys. Rev. B {\bf B 66}, 054408 (2002).
464: 
465: \bibitem{Leighton:2001} C. Leighton and Ivan K. Schuller, Phys. Rev. B {\bf 
466: 63}, 174419 (2001).
467: 
468: \bibitem{Fullerton:1996} E. E. Fullerton and S. D. Bader, Phys. Rev. B 
469: {\bf 53}, 5112 (1996).
470: 
471: \bibitem{Chamberlin:1984} R. V. Chamberlin, G. Mozurkewich, and R. Orbach, 
472: Phys. Rev. Lett. {\bf 52}, 867 (1984).
473: 
474: \bibitem{Chamberlin:1984b} R. V. Chamberlin, Phys. Rev. B {\bf 30}, 
475: R5393 (1984).
476: 
477: \bibitem{Chubykalo:1999} O. A. Chubykalo, J. M. Gonzez, J. Magn. Magn. 
478: Mater. {\bf 196-197}, 810 (1999).
479: 
480: \bibitem{Panagioto:2002} I. Panagiotopoulos, N. Moutis, and C. Christides, 
481: Phys. Rev. B {\bf 65}, 132407 (2002).
482: 
483: \bibitem{Chen:2003} X. Chen, W. Kleemann, O. Petracic, O. Sichelschmidt, S. 
484: Cardoso, and P. P. Freitas, Phys. Rev. B {\bf 68}, 054433 (2003).
485: 
486: \bibitem{Street:1994}  R. Street and S. D. Brown, J. Appl. Phys. {\bf 76}, 
487: 6386 (1994).
488: 
489: \bibitem{Chantrell:1994}  R. W. Chantrell, A. Lyberatos, M. El-Hilo, and K. 
490: O'Grady, J. Appl. Phys. {\bf 76}, 6407 (1994).
491: 
492: \bibitem{Sharrock:1994} M. P. Sharrock, J. Appl. Phys. {\bf 76}, 6413 (1994).
493: 
494: \bibitem{Goodman:2000} A. M. Goodman, H. Laider, K. O'Grady, N. W. Owen, and 
495: A. K. Petford-Long, J. Appl. Phys. {\bf 87}, 6409 (2000).
496: 
497: \bibitem{Lannon:2002} J. M. Lannon Jr., C. C. Pace, D. Temple G. E. McGuire,
498: A. F. Hebard, and M. Ray, Materials Research Society Symposium - Proceedings, 
499: {\bf 690}, 157 (2002).
500: 
501: \bibitem{Shi:1997} Z. P. Shi and R. S. Fishman, Phys. Rev. Lett. {\bf 78},
502: 1351 (1997).
503: 
504: \bibitem{Berger:1997} A. Berger and E. E. Fullerton, J. Magn. Magn. Mater.
505: {\bf 121}, 471 (1997).
506: 
507: \bibitem{Chamberlin:1994} R. V. Chamberlin, J. Appl. Phys. {\bf 76}, 
508: 6401 (1994).
509: 
510: \bibitem{Patel:2005} R. S. Patel, A. K. Majumdar, A. K. Nigam, D. Temple,
511: and C. Pace, arXiv:cond-mat/0504275.
512: 
513: \bibitem{Guy:1978} C. N. Guy, J. Phys. F: Metal Phys., {\bf 8}, 1309 (1978).
514: 
515: \bibitem{Lottis:1988} Daniel K. Lottis, E. Dan Dahlberg, J. A. Christner, J. I. Lee. 
516: R. L. Peterson, and R. M. White, J. Appl. Phys., {\bf 63}, 2920 (1988).
517: 
518: 
519: \end{thebibliography}
520: \vspace{15cm}
521: 
522: \begin{figure}
523: \includegraphics[bb=1pt 13pt 393pt 195pt, width=8.5cm]{Fig-1}
524: %\includegraphics[bb=2.5cm 11.4cm 16.3cm 17.8cm,width=8.5cm]{sam}
525: \caption{\label{fig:sample} Schematic diagram of the sputtered GMR samples. Black 
526: and grey spheres represent Cr and Fe atoms, respectively. Typical 
527: estimates of the relative strengths of the Fe-Fe, Fe-Cr, and Cr-Cr 
528: couplings are 1 : -0.3 : -0.18 \cite{Shi:1997} and 1 : -0.55 : 
529: -0.3 \cite{Berger:1997}. Directions of the arrow show the spin alignment
530: direction. `?' shows the uncertainty(frustration) in the spin direction at that site.}
531: \end{figure}
532: 
533: \begin{figure}
534: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-2}
535: \caption{\label{fig:S2ln} (Color online) M (raw data) is 
536: plotted against {\it log} (t in s) for
537: sample 2. The solid lines are the logarithmic fits to Eq. (~\ref{eq:ln}).
538: The data and the fits are multiplied by 1.285, 1.66, 1.73, 2.27, 2.85, and 3.92
539: for 30, 45, 60, 80, 100, and 140 K, respectively to show all the
540: curves on the same figure.}
541: \end{figure}
542: 
543: \begin{figure}
544: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-3}
545: \caption{\label{fig:x4alllog} (Color online) M (raw data) is 
546: plotted against {\it log} (t in s) for
547: sample 3. The solid lines are the logarithmic fits to Eq. (~\ref{eq:ln}).
548: The data and the fits are multiplied by 1.3, 2.3, 3.2, 4.42, and 8.4
549: for 80, 120, 140, 160, and 200 K, respectively to show all the
550: curves on the same figure.}
551: \end{figure}
552: 
553: \begin{figure}
554: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-4}
555: \caption{\label{fig:x6alllog} (Color online) M (raw data) is 
556: plotted against {\it log} (t in s) for
557: sample 4. The solid lines are the logarithmic fits to Eq. (~\ref{eq:ln}).
558: The data and the fits are multiplied by 1.19, 1.45, 1.9, 2.5, 3.45, 4, and 8.2
559: for 40, 60, 80, 100, 120, 140, and 200 K, respectively 
560: to show all the curves on the same figure.}
561: \end{figure}
562: 
563: \begin{figure}
564: %\includegraphics[bb=0.4cm 0.5cm 9.5cm 7.2cm,width=8.5cm]{PwrN}
565: \includegraphics[bb=14pt 16pt 263pt 222pt,width=8.5cm]{Fig-5}
566: \caption{\label{fig:pwrp} (Color online) Variation of the parameter S of  Eq. 
567: (~\ref{eq:ln}) with temperature. The solid lines are just guides 
568: to the eye.}
569: \end{figure}
570: 
571: \begin{figure}
572: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-6}
573: \caption{\label{fig:f83d} (Color online) Percentage change in 
574: M is plotted against t (s) at various temperatures for
575: sample 2.}
576: \end{figure}
577: 
578: \begin{figure}
579: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-7}
580: \caption{\label{fig:S1ln} (Color online) M(raw data) is 
581: plotted against {\it log} (t in s) for
582: sample 1. The solid lines are the logarithmic fits to Eq. (~\ref{eq:ln}). 
583: The data and the fits for 50 and 150 K are shifted along y-axis for clarity.}
584: \end{figure}
585: 
586: \begin{figure}
587: \includegraphics[bb=0.4cm 0.5cm 9.0cm 7.7cm,width=8.5cm]{Fig-8}
588: \caption{\label{fig:S3Ex} (Color online) M/M$_0$ is plotted against {\it log} (t in 
589: s) for sample 3 at 160 K. The solid line is the logarithmic 
590: fit to Eq. (~\ref{eq:ln}) and the dashed line is the power fit to Eq. (~\ref{eq:power}).
591: Both the fits are extended till $10^6$ s to show that in larger time window 
592: the power fit decreases slower than the logarithmic fit. However, in our
593: measurement time span there is only a small difference between the two fits.}
594: \end{figure}
595: 
596: 
597: 
598: \begin{table}
599: \caption{Values of $\chi^2$, correlation coefficient 
600: R$^2$, the parameters $M_0$, $S$ of Eq. (~\ref{eq:ln}), $M_0$, and
601: $\beta$ of Eq. (~\ref{eq:power}) for sample 1.}
602: \begin{tabular}{|c|c|c|c|c|} \hline
603: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$  & S(10$^{-7} emu)$ \\ \hline
604: 50 &  1.407 & 0.8778 & 197.24 $\pm $ 0.02 & 5.1 $\pm $ 0.3 \\ \hline
605: 100 & 2.110 & 0.8707 & 193.48 $\pm $ 0.03 & 5.9 $\pm $ 0.3 \\ \hline
606: 150 & 1.713 & 0.9497 & 189.30 $\pm $ 0.02 & 8.7 $\pm $ 0.3 \\ \hline \hline
607: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & $\beta$(10$^{-4})$ \\ \hline
608: 50 & 1.405 & 0.8779 & 197.00 $\pm $ 0.02 & 2.6 $\pm $ 0.2 \\ \hline
609: 100 & 2.108 & 0.8708 & 193.00 $\pm $ 0.03 & 3.1 $\pm $ 0.2 \\ \hline
610: 150 & 1.709 & 0.9498 & 189.00 $\pm $ 0.02 & 4.6 $\pm $ 0.2 \\ \hline
611: \end{tabular}
612: \label{tab:F06}
613: \end{table}
614: 
615: 
616: 
617: \begin{table}
618: \caption{Values of $\chi^2$, correlation coefficient R$^2$,
619: and the parameters $M_0$, $S$ of Eq. (~\ref{eq:ln}), $M_0$, and
620: $\beta$ of Eq. (~\ref{eq:power}) for sample 2.}
621: \begin{tabular}{|c|c|c|c|c|}\hline
622: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & S(10$^{-7} emu)$ \\ \hline
623: 20 & 5.1674 & 0.9965 & 25.721 $\pm $0.006 & 7.96 $\pm $ 0.08 \\ \hline
624: 30 & 11.621 & 0.9988 & 20.274 $\pm $ 0.004 & 13.03 $\pm $ 0.06 \\ \hline
625: 45 & 29.432 & 0.9992 & 16.066 $\pm $ 0.006 & 18.64 $\pm $ 0.07 \\ \hline
626: 60 & 25.028 & 0.9986 & 15.140 $\pm $ 0.008 & 15.5 $\pm $ 0.1 \\ \hline
627: 80 & 20.487 & 0.9895 & 11.54 $\pm $ 0.01 & 12.2 $\pm $ 0.2 \\ \hline
628: 100 & 24.683 & 0.9983 & 9.037 $\pm $ 0.004 & 8.30 $\pm $ 0.05 \\ \hline
629: 140 & 89.531 & 0.9833 & 6.320 $\pm $ 0.006 & 3.56 $\pm $ 0.08 \\ \hline
630: 200 & 1119.8 & 0.9969 & 1.698 $\pm $ 0.004 & 6.21 $\pm $ 0.06 \\ \hline \hline
631: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & $\beta$(10$^{-3})$ \\ \hline
632: 20 & 5.4684 & 0.9963 & 25.727 $\pm $0.006 & 3.17 $\pm $ 0.03 \\ \hline
633: 30 & 12.7867 & 0.9987 & 20.296 $\pm $ 0.005 & 6.74 $\pm $ 0.03 \\ \hline
634: 45 & 36.7689 & 0.999 & 16.126 $\pm $ 0.007 & 12.65 $\pm $ 0.05 \\ \hline
635: 60 & 29.2681 & 0.9983 & 15.187 $\pm $ 0.009 & 11.06 $\pm $ 0.07 \\ \hline
636: 80 & 22.0195 & 0.9988 & 11.570 $\pm $ 0.006 & 11.41 $\pm $ 0.06 \\ \hline
637: 100 & 19.4450 & 0.9987 & 9.059 $\pm $ 0.004 & 9.86 $\pm $ 0.06 \\ \hline
638: 140 & 88.8819 & 0.9834 & 6.328 $\pm $ 0.006 & 5.9 $\pm $ 0.1 \\ \hline
639: 200 & 1807.83 & 0.995 & 1.7690 $\pm $ 0.007 & 48.3 $\pm $ 0.6 \\ \hline
640: \end{tabular}
641: \label{tab:F08}
642: \end{table}
643: 
644: 
645: 
646: 
647: 
648: \begin{table}[htbp]
649: \caption{Values of $\chi^2$, correlation coefficient R$^2$,
650: the parameters $M_0$, $S$ of Eq. (~\ref{eq:ln}),  $M_0$, and $\beta$
651: Eq. (~\ref{eq:power}) for sample 3.}
652: \begin{tabular} {|c|c|c|c|c|} \hline
653: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & S(10$^{-7} emu)$ \\ \hline
654: 10  & 15.52408 & 0.9872 & 26.223 $\pm $ 0.002 & 1.04 $\pm $ 0.02 \\ \hline
655: 15  & 17.2138 & 0.9944 & 25.8571 $\pm $ 0.002 & 1.64  $\pm $ 0.02 \\ \hline
656: 20 & 63.2449 & 0.9885 & 25.5907 $\pm $ 0.003 & 2.15 $\pm $ 0.04 \\ \hline
657: 25 & 44.2073 & 0.9930 & 26.2994 $\pm $ 0.003 & 2.39 $\pm $ 0.03 \\ \hline
658: 35 & 87.8776 & 0.9954 & 34.2492 $\pm $ 0.004 & 3.97 $\pm $ 0.04 \\ \hline
659: 45  & 12.3645 & 0.9872 & 27.525 $\pm $ 0.007 & 5.60 $\pm $ 0.09 \\ \hline
660: 55  & 54.5099 & 0.9993 & 22.8914 $\pm $ 0.002 & 7.47 $\pm $ 0.03 \\ \hline
661: 80  & 12.2490 & 0.9989 & 21.588 $\pm $ 0.005 & 14.3  $\pm $ 0.07 \\ \hline
662: 120 & 16.2190 & 0.9997 & 12.584 $\pm $ 0.003 & 17.29 $\pm $ 0.04 \\ \hline
663: 140 & 36.9160 & 0.9994 & 9.056 $\pm $ 0.003 & 13.83 $\pm $ 0.05 \\ \hline
664: 160 & 75.9658 & 0.9988 & 6.464 $\pm $ 0.003 & 9.48  $\pm $ 0.05 \\ \hline
665: 200 & 84.3582 & 0.9988 & 3.409 $\pm $ 0.002 & 5.37 $\pm $ 0.03 \\ \hline \hline
666: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & $\beta$(10$^{-3})$ \\ \hline
667: 10 & 0.3085 & 0.9872 & 29.814 $\pm $ 0.002 & 0.397 $\pm $ 0.004 \\ \hline
668: 15 & 0.3483 & 0.9944 & 29.398 $\pm $ 0.002 & 0.635  $\pm $ 0.008 \\ \hline
669: 20 & 1.2901 & 0.9884 & 29.096 $\pm $ 0.003 & 0.85 $\pm $ 0.02 \\ \hline
670: 25 & 0.8924 & 0.9930 & 29.901 $\pm $ 0.003 & 0.92 $\pm $ 0.02 \\ \hline
671: 35 & 1.8190 & 0.9953 & 28.847 $\pm $ 0.004 & 1.58  $\pm $ 0.02 \\ \hline
672: 45 & 12.3258 & 0.9872 & 27.527 $\pm $ 0.007 & 2.06 $\pm $ 0.03 \\ \hline
673: 55 & 1.1379 & 0.9993 & 26.034 $\pm $ 0.003 & 3.35 $\pm $ 0.02 \\ \hline
674: 80 & 13.4735 & 0.9988 & 21.613 $\pm $ 0.005 & 6.95  $\pm $ 0.03 \\ \hline
675: 120 & 22.8617 & 0.9996 & 12.652 $\pm $ 0.004 & 15.24 $\pm $ 0.04 \\ \hline
676: 140 & 26.3087 & 0.9996 & 9.116 $\pm $ 0.003 & 17.15 $\pm $ 0.05 \\ \hline
677: 160 & 56.4284 & 0.9991 & 6.504 $\pm $ 0.003 & 16.39  $\pm $ 0.07 \\ \hline
678: 200 & 5296.38 & 0.9456 & 3.364 $\pm $ 0.001 & 15.0 $\pm $ 0.4 \\ \hline
679: \end{tabular}
680: \label{tab:X4}
681: \end{table}
682: 
683: 
684: \begin{table}[htbp]
685: \caption{Values of $\chi^2$, correlation coefficient R$^2$,
686: and the parameters $M_0$, S of Eq. (~\ref{eq:ln}), 
687: $M_0$, and $\beta$ of Eq. (~\ref{eq:ln}) for sample 4.}
688: \begin{tabular} {|c|c|c|c|c|} \hline \hline
689: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & S(10$^{-7} emu)$ \\  \hline
690: 20 & 1.8586 & 0.9969 & 29.175 $\pm $ 0.004 & 5.64 $\pm $ 0.05 \\ \hline
691: 40 & 20.7856 & 0.9969 & 24.832 $\pm $ 0.007 & 12.08 $\pm $ 0.09 \\ \hline
692: 60 & 27.4387 & 0.9980 & 20.73 $\pm $ 0.01& 18.5 $\pm $ 0.1 \\ \hline
693: 80 & 28.9706 & 0.9989 & 16.007 $\pm $ 0.008 & 19.3 $\pm $ 0.1 \\ \hline
694: 100 & 49.0388 & 0.9985 & 12.212 $\pm $ 0.008 & 16.07 $\pm $ 0.1 \\ \hline
695: 120 & 103.588 & 0.9981 & 8.749 $\pm $ 0.005 & 11.25 $\pm $ 0.05 \\ \hline
696: 140 & 248.365 & 0.9925 & 7.46 $\pm $ 0.01 & 9.8 $\pm $ 0.1 \\ \hline
697: 200 & 346.956 & 0.9883 & 3.481 $\pm $ 0.004 & 3.50 $\pm $ 0.05 \\ \hline \hline
698: T(K) & $\chi^{2}(10^{-8})$ & R$^{2}$ & M$_0(10^{-5} emu)$ & $\beta$(10$^{-3})$ \\ \hline
699: 20& 26.4418 & 0.9955& 2.9180 $\pm $ 0.0005& 1.97 $\pm $ 0.02 \\ \hline
700: 40& 20.4407 & 0.9969& 2.4848 $\pm $ 0.0007& 5.05 $\pm $ 0.04 \\ \hline
701: 60& 33.6017 & 0.9976& 2.078 $\pm $ 0.001& 9.58 $\pm $ 0.07 \\ \hline
702: 80& 33.0043 & 0.9988& 1.608 $\pm $ 0.001& 13.28 $\pm $ 0.07 \\ \hline
703: 100& 32.6814 & 0.9990& 1.2280 $\pm $ 0.0007& 14.62 $\pm $ 0.07 \\ \hline
704: 120& 79.7989 & 0.9985& 0.8793 $\pm $ 0.0005& 14.22 $\pm $ 0.07 \\ \hline
705: 140& 194.536 & 0.9942& 0.750 $\pm $ 0.001& 14.7 $\pm $ 0.2 \\ \hline
706: 200& 333.72 & 0.9887& 0.3492 $\pm $ 0.0004& 10.9 $\pm $ 0.1 \\ \hline
707: \end{tabular}
708: \label{tab6:X9004pwr1}
709: \end{table}
710: 
711: 
712: \end{document}
713: 
714: 
715: