1:
2: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,superscriptaddress]{revtex4}
3: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
4:
5: % Some other (several out of many) possibilities
6: %\documentclass[preprint,aps]{revtex4}
7: %\documentclass[preprint,aps,draft]{revtex4}
8: %\documentclass[prb]{revtex4}% Physical Review B
9:
10: \usepackage{graphicx}% Include figure files
11: \usepackage{dcolumn}% Align table columns on decimal point
12: \usepackage{bm}% bold math
13:
14: %\nofiles
15:
16: \begin{document}
17:
18: %\preprint{APS/123-QED}
19:
20: \title{Temperature and Magnetic Field Dependencies\\in a Model of Diamagnetic Hysteresis in Beryllium}% Force line breaks with \\
21:
22: \author{Nathan Logoboy}
23:
24: \email{logoboy@phys.huji.ac.il}
25:
26:
27: \affiliation{Grenoble High Magnetic Field Laboratory, MPI-FKF and
28: CNRS P.O. 166X, F-38042 Grenoble Cedex 9, France}
29:
30: \affiliation {The Racah Institute of Physics, The Hebrew University
31: of Jerusalem, 91904 Jerusalem, Israel}
32:
33: \author{Walter Joss}
34: \affiliation{Grenoble High Magnetic Field Laboratory, MPI-FKF and
35: CNRS P.O. 166X, F-38042 Grenoble Cedex 9, France}
36:
37: \affiliation {Universit$\acute{e}$ Joseph Fourier, B.P. 53, F-38041
38: Grenoble Cedex 9, France}
39:
40: \date{\today}% It is always \today, today,
41: % but any date may be explicitly specified
42:
43: \begin{abstract}
44: The model of diamagnetic hysteresis loop of strongly correlated
45: electron gas at the conditions of dHvA effect is developed. It is
46: shown, that in the framework of Rayleigh theory for magnetization
47: loop, the coercive force and remnant magnetization in every period
48: of dHvA oscillations are characterized by strong dependencies on
49: temperature, magnetic field and Dingle temperature.
50: \end{abstract}
51:
52: \pacs{75.20.En, 75.60.Ch, 71.10.Ca, 71.70.Di, 71.25.-s; 71.25. Hc; 75.40-s; 75.40.Cx.}% PACS, the Physics and Astronomy
53: % Classification Scheme.
54: %\keywords{Suggested keywords}%Use showkeys class option if keyword
55: %display desired
56: \maketitle
57:
58: \section{\label{sec:Introduction}Introduction}
59:
60: At high magnetic field and low temperature the strongly correlated
61: electron gas is characterized by instability due to magnetic
62: interaction between electrons \cite{Shoenberg}, which results in
63: phase transition from homogeneous state into inhomogeneous one and
64: stratification of the sample into diamagnetic domains, e.g. Condon
65: domains (CDs), with alternating from one domain to another
66: diamagnetic moment. This instability of an electron gas is called
67: diamagnetic phase transition and extensively studied, both
68: theoretically and experimentally \cite{Condon}-\cite{Solt2}. The
69: diamagnetic domains were observed in Be by magnetization
70: measurements \cite{Condon}; in Ag by nuclear magnetic resonance
71: (NMR) \cite{Condon_Walstedt} and recently by Hall probe spectroscopy
72: \cite{Kramer1}. They have been observed in Be, Sn, Al and Pb by muon
73: spin-rotation ($\mu$SR) spectroscopy \cite{Solt1}. Usually, for
74: plate-like samples, this non-uniform phase is realized as periodic
75: domain structure \cite{Shoenberg}-\cite{Kramer1} with alternating
76: from one domain to another diamagnetic moment. Magnetic interaction
77: of strongly correlated electrons gas results in non-linear
78: relationship between local diamagnetic moments and external magnetic
79: field. It was shown \cite{Logoboy1} that in CD phase the local
80: diamagnetic moments, which contribute to the process of
81: magnetization, are characterized by strong temperature and magnetic
82: field dependencies.
83:
84: The existence of metastable states for local diamagnetic moments in
85: each period of dHvA oscillations can results in hysteresis, e. g.
86: irreversible changes in magnetization due to change in external
87: magnetic field, similarly to magnetic materials of spin origin. The
88: microscopic mechanism of magnetization loop, relates to defects of
89: crystal lattice, intrinsic stress and elastic deformations. The
90: impurities, dislocations and other defects of crystal lattice cause
91: an internal friction which opposes to the wall displacement. The
92: domain walls are pinned by such defects. Unpinning is realized by
93: small dissipative jumps through the increasing of the value of
94: applied magnetic field.
95:
96: Although, the possibilities of irreversible behavior of
97: magnetization curve in diamagnetic materials have been discussed
98: earlier \cite{Shoenberg}, only recent experimental investigation of
99: CDs in Be \cite{Kramer2} reveals hysteresis loop in the
100: magnetization process. The hysteresis magnetization curve of Be
101: \cite{Kramer2} was measured directly by Hall probes in dc field and
102: standard ac method with different modulation levels, frequencies and
103: magnetic field ramp rates, and the shape of magnetization curve was
104: reconstructed by several higher harmonics of the ac pickup voltage.
105: Detection of the hysteresis in Be \cite{Kramer2} substantiates a far
106: reaching analogy between an electron gas, which has unstable,
107: metastable and stable states within one dHvA period, and the
108: liquid-gas phase coexistence and transformation. Recent progress
109: \cite{Kramer1}, \cite{Kramer2} in experiment on observation of
110: Condon domain structure in Ag and Be, in particular, the detection
111: of diamagnetic hysteresis, provides a natural stimulus towards a
112: more detailed understanding of the properties of strongly correlated
113: electron gas in the conditions of dHvA effect. CDs in a $3d$
114: electron gas are the only type of magnetic structures, for which the
115: magnetization process has not been considered so far.
116:
117: In present paper we develop the model of diamagnetic hysteresis,
118: which fits the recent data on observation of diamagnetic hysteresis
119: in Be \cite{Kramer2}. The hysteresis loop appears periodically in
120: every period of dHvA oscillations, when $\mu_{0}\chi
121: >1$ ( $\chi$ is differential magnetic susceptibility, $M$ is the oscillatory
122: part of the magnetization, and $B$ is magnetic induction
123: \cite{Shoenberg}). There are three main contributions into the
124: process of magnetization of normal metals in the condition of strong
125: dHvA effect which have to be distinguished, the dependence of local
126: diamagnetic moments on external magnetic field through the reduced
127: amplitude of dHvA oscillations \cite{Logoboy1}, the domain wall
128: displacement and nucleation of the domains. The rotation processes,
129: that are intrinsic property of magnetization loop in magnetic
130: materials of spin origin, can be neglected in rearrangement of CD
131: structure at least for metals with relatively simple Fermi surfaces
132: \cite{Shoenberg} due to one-dimensional character of the diamagnetic
133: moments. We show that the remnant magnetization and coercivity, as
134: well as the hysteresis loop square, which are the fundamental
135: properties of the magnetization curve, have strong dependence on
136: temperature, magnetic field and Dingle temperature.
137:
138: The paper is organized as follows. Section \ref{sec:Introduction} is
139: an introduction. In Section \ref{sec:Model} we consider the model of
140: diamagnetic hysteresis, based on Rayleigh model in the framework of
141: Lifshitz-Kosevich-Shoenberg approximation. In Section
142: \ref{sec:Results and Discussions} we calculate the temperature and
143: magnetic field dependence of remnant magnetization and coercivity.
144: Section IV is conclusions. The last section contains
145: acknowledgements.
146:
147: \section{\label{sec:Model}Model}
148:
149: The oscillator part of the thermodynamic potential density in the
150: Lifschitz-Kosevich-Shoenberg formalism \cite{Shoenberg} can be
151: written in reduced form with taking into account the shape sample
152: effects:
153:
154: \begin{equation} \label{eq:free energy density}
155: \Omega =\frac{1} {4 \pi k^{2}} \left [ a \cos {b}+\frac{1}{2}a^{2}(1-n) \sin^{2} {b} \right ], \\
156: \end{equation}
157: where $b=k(B-\mu_{0}H)=k[\mu_{0}h_{ex}+(1-n)M]=x+(1-n)y$, $\mu_{0}H$
158: is the magnetic field inside the material at the center of dHvA
159: period, $k=2 \pi F/(\mu_{0}H)^{2}$, $F$ is the fundamental
160: oscillation frequency), $H_{ex}$ is external magnetic field,
161: $h_{ex}=H_{ex}-H$ is the small increment to the magnetic field $H$,
162: $n$ is the demagnetization factor. All the components of vectors are
163: taken along the direction of the magnetic induction. The quality
164: $a=4 \pi k A=\mu _{0} \partial M/\partial B$ \cite{Shoenberg} is the
165: reduced amplitude of the dHvA oscillations (the ratio between the
166: amplitude of oscillations $A$ and their period $\Delta H$).
167:
168: In the first harmonic approximation the magnetization is found from
169: the implicit equation of state \cite{Shoenberg}:
170:
171: \begin{equation} \label{eq:magnetization}
172: y=a \sin {[x+(1-n)y]}, \\
173: \end{equation}
174: The validity of thermodynamic potential density $\Omega$ (\ref{eq:free energy density}), and the expression for the reduced
175: magnetization $y$ (\ref{eq:magnetization}), derived from
176: Eq.~(\ref{eq:free energy density}), is restricted by applicability
177: to homogeneous phase only, where the existence of demagnetization
178: coefficient $n$ is justified. In the conditions of strong magnetic
179: interaction between electrons, when
180:
181: \begin{equation} \label{eq:CDCondition}
182: a \left ( \mu_{0}H, T, T_{D}\right )\ge 1, \\
183: \end{equation}
184: a state of lower thermodynamic potential is achieved over part of
185: dHvA oscillation cycle by the formation of CD structure, for which
186: the local value of magnetization alternates in sign from one domain
187: to the next, resulting in essential reduce of magneto-static energy.
188: The usually observed diamagnetic domain structure is of
189: stripe-domain type \cite{Shoenberg},\cite{Kramer1}, and the
190: contribution of magneto-static energy into the free energy density
191: is negligibly small in comparing to the magneto-static energy of
192: homogeneous state. In the domain state, when the reduced amplitude
193: of dHvA oscillations satisfies to the condition
194: (\ref{eq:CDCondition}), the demagnetization factor $n$, which is
195: characteristic of the uniformly magnetized sample, is replaced by
196: the coefficient $\alpha$ \cite{Logoboy1}. This coefficient depends
197: on magnetic field, temperature, impurities and geometry of the
198: sample and takes into account the magneto-static energy for given
199: type of domain structure. Typically for plate-like sample,
200: $\alpha<<1$, independently on the types of domain structures
201: \cite{Kittel} and important mainly in calculation of the
202: steady-state period $2D$ of the domain structure by minimization of
203: the free energy density, containing two terms, magneto-static energy
204: and the surface energy of the domain walls
205: \cite{Kittel}-\cite{Chikazumi}. For observed CDs structures
206: \cite{Condon_Walstedt},\cite{Kramer1},\cite{Kramer2} the inequality
207: $D/L<<1$, where $L$ is the width of the sample, is fulfilled and
208: $\alpha \sim D/L<<1$.
209:
210: The equation
211: \begin{equation} \label{eq:DPT}
212: a ( \mu_{0}H, T, T_{D} )= 1, \\
213: \end{equation}
214: defines the critical surface in three dimensions $\mu_{0}H-T-T_{D}$
215: . Above this surface the uniform diamagnetic phase exists, but below
216: it, the CD phase can appear in the part of every period of the dHvA
217: oscillations.
218:
219: For proper calculation of the reduced amplitude of the dHvA
220: oscillations a and constructing the hysteresis loop, the correct
221: topology of Fermi surface has to be taken into account. In case of
222: Be the standard procedure of calculation of the amplitude of dHvA
223: oscillations \cite{Shoenberg}, based on series expansions near the
224: extreme cross-sections, requires considerable modification due to
225: negligible small curvature of extreme cross-sections of the
226: 'cigar'-like part of the Fermi surface, contributed to the dHvA
227: oscillations \cite{Logoboy3}. The Fermi surface of Be consists of
228: the second zone monster ('coronet') and the third zone 'cigar'. It
229: is well-established that the dHvA oscillations originates from the
230: three maximum cross-sections of 'cigar' ('waist' and 'hips') which
231: are characterized by a very small curvature. In particular, the
232: small curvature of the cylinder like Fermi surface explains
233: relatively high amplitude of dHvA oscillations in Be at magnetic
234: field $\boldmath H$ applied parallel to the hexagonal axis, e.g.
235: $\boldmath H \parallel [0001]$, in comparing to Ag
236: \cite{Kramer1},\cite{Kramer3}, where the small amplitude of dHvA
237: oscillations is explained in the framework of spherical Fermi
238: surface \cite{Shoenberg}. In the framework of the model of slightly
239: corrugated cylinder like Fermi surface of Be \cite{Logoboy3}, the
240: reduced amplitude of the dHvA oscillations takes a form
241: \begin{equation} \label{eq:Reduced Amplitude}
242: a=\frac{4\kappa (\mathcal{A}_{0}\hbar)^{2}}{\pi^{3}m_{c}(\mu_{0}H)^{2}} |Q(\mu_{0}H)| R(\mu_{0}H,T,T_{D}), \\
243: \end{equation}
244: where $R(\mu_{0}H,T,T_{D})$ is a reduction factor \cite{Shoenberg},
245: \cite{Logoboy3}, which takes into account the influence of the
246: temperature $T$ and impurities on the amplitude of the dHvA
247: oscillations.
248:
249: The complex function
250: \begin{eqnarray} \label{eq:Q}
251: Q=Q_{1}+jQ_{2}=\mid Q\mid \exp{(j\psi)}, \qquad \qquad \nonumber \\
252: Q_{1}=3\beta \cos{\lambda}+2(1-2\beta) J_{0}(\lambda), \quad
253: Q_{2}=\beta \sin{\lambda}
254: \end{eqnarray}
255: where $\lambda=l^{2}\mathcal{A}_{0}\xi$, depends on applied magnetic
256: field through parameter $l^{2}=c\hbar/e\mu_{0}H$ and describes the
257: beatings in the amplitude of dHvA oscillations resulting from two
258: different extreme cross-section of Fermi surface of Be
259: \cite{Logoboy3}. Here, $\kappa=0.217 A^{o~-1}$ \cite{Shoenberg} is
260: the distance in reciprocal space between two extreme cross sections
261: of 'cigar', $A_{0}$ is average cross section area, $J_{0}(\lambda)$
262: is Bessel function of the first order and $\beta$ is the half-width
263: of the cylinder like extreme cross sections.
264:
265: The Eqs.~(\ref{eq:DPT}),(\ref{eq:Reduced Amplitude}) define in
266: explicit form the critical temperature for diamagnetic phase
267: transition as a function of magnetic field $\mu_{0}H$ and Dingle
268: temperature $T_{D}$. It allows us to construct the phase diagrams
269: for Be and investigate the properties of the diamagnetic hysteresis.
270:
271: \section{\label{sec:Results and Discussions}Results and Discussions}
272:
273: At the condition $a ( \mu_{0}H, T, T_{D} )>1$ the sample is divided
274: into domains with up and down magnetization. In the center of the
275: dHvA period $h=0$ the domains are characterized by the same width.
276: An external magnetic field removes the equivalence of the states,
277: the energy balance is altered and rearrangements of the domain
278: structure take place. For CD structure this process is realized not
279: only through the motion of domain walls, which results in
280: corresponding changes of domain volumes, and nucleation of the
281: domains, but also through magnetic field dependence of the local
282: magnetization in every adjacent domains \cite{Logoboy1}. To
283: understand the magnetization process in CD phase it is convenient to
284: define the approximate anhysteretic, or pinning-free, magnetization
285: \begin{equation} \label{eq:Anhysteretic Magnetization}
286: y^{(anhys)}=\frac {1}{2}(1-\frac{x}{\Delta})y_{-}+\frac {1}{2}(1+\frac{x}{\Delta})y_{+}, \\
287: \end{equation}
288: where the function $y_{\pm}=y_{\pm}(x)=-y_{\mp}(-x)$ is the
289: solution of the equation $y_{\pm}=a \sin {(x+y_{\pm})}$ at the range
290: $|x|\le \Delta \le k \cdot \Delta H/2$ in every period $\Delta H$ of
291: the dHvA oscillations and describes magnetic field dependence of the
292: local diamagnetic moments of the up ($+$) and down ($-$) domains
293: (Fig.~\ref{Hysteresis}). In Eq.~(\ref{eq:Anhysteretic
294: Magnetization}) we assumed that the volume fractions of the up and
295: down domains are linear functions of the increment of the reduced
296: magnetic field in every dHvA period. This assumption was justified
297: by NMR experiments in Ag \cite{Condon_Walstedt} and $\mu$SR
298: spectroscopy in Be \cite{Solt1}- \cite{Solt2}. Both experiments
299: reveal the domain structure by the appearance of a doublet
300: corresponding to two domain magnetic inductions with line
301: intensities, that show the linear dependence on the external
302: magnetic field of the volume fractions occupied by the neighboring
303: domains.
304:
305: The range of existence of the CDs $2 \Delta$ is defined by
306: \cite{Logoboy1}
307: \begin{equation} \label{eq:Delta}
308: \Delta =\sqrt {a^{2}-1}- \cos^{-1}{ \frac {1}{a}} \ge 0, \quad a \ge 1. \\
309: \end{equation}
310: It follows from Eq.~(\ref{eq:Anhysteretic Magnetization}), that the
311: anhysteretic magnetization in every period of the dHvA oscillations
312: is a function of magnetic field $H$, temperature $T$ and Dingle
313: temperature $T_{D}$. The expression for $y^{(anhys)}$
314: Eq.~(\ref{eq:Anhysteretic Magnetization}) can be simplified for
315: relatively high reduced amplitude of the dHvA oscillations, e.g.
316: $a\gg 1$. In this case $y_{+} \approx y_{0}-(x/ \Delta)y_{0}$,
317: $y_{-} \approx y_{+}-2y_{0}$, where $y_{0}$ is the half-splitting in
318: the center of the period, defined explicitly form by the equation
319: $y_{0}/ \sin{y_{0}=a}$ \cite{Shoenberg}, and the
320: Eq.~(\ref{eq:Anhysteretic Magnetization}) is replaced by the
321: following one
322: \begin{equation} \label{eq:Approximation}
323: y^{(anhys)} \approx y_{0}(\frac {1}{\Delta}-\frac {1}{\pi})x, \quad |x| \le \Delta \le \pi. \\
324: \end{equation}
325: Although, the local diamagnetic moments $y_{\pm}=y_{\pm}(x_{ex})$
326: are characterized by average {\it negative} slope as the functions
327: of magnetic field increment $x_{ex}$ \cite{Logoboy1}, it follows
328: from Eq.~(\ref{eq:Approximation}), that the anhysteretic curve has a
329: {\it positive} slope in every period of the dHvA oscillations
330: (Fig.~\ref{Hysteresis}). We will see that the diamagnetic hysteresis
331: is also characterized by the positive slope: in the process of the
332: diamagnetic magnetization the volume fraction of the up-domains
333: increases on expense of the volume fraction of the down-domains with
334: increase of magnetic field increment $x_{ex}$ \cite{Shoenberg}.
335:
336: \begin{figure}
337: \includegraphics[width=0.4\textwidth]{Fig_1.eps}
338: \caption{ Diamagnetic magnetization loop $y=y(x;a)$ versus magnetic
339: field increment $x$ is plotted in one period of dHvA oscillations.
340: The local diamagnetic moments $y_{\pm}=y_{\pm}(x;a)$ reproduced from
341: \cite{Logoboy1} contribute to the hysteresis curve
342: $y_{a,d}=y_{a,d}(x;a)$. Three fundamental parameters of the
343: hysteresis loop, e.g. coercivity $x_{c}$, remnant magnetization
344: $y_{r}$ and saturation magnetization $y_{s}$ are shown. The dot
345: lines correspond to the range of existence of the hysteresis
346: ($2\Delta$). The existence of the hysteresis is restricted by
347: $2\Delta \le 2\pi$. The dash lines show the states which are not
348: realized. In numerical calculations the value of $a=3$ and $n=0.52$
349: are used.} \label{Hysteresis}
350: \end{figure}
351:
352: The real diamagnetic magnetization process in CD phase is
353: characterized by hysteresis, or magnetization loop, which appears
354: periodically in applied magnetic field in every period of dHvA
355: oscillations, when the condition (\ref{eq:CDCondition}) is
356: fulfilled. According to experimental data [4] the diamagnetic
357: hysteresis is usually small, related to the domain wall displacement
358: and domain nucleation. Thus, it can be described in the framework of
359: the Lord Rayleigh model with two parameters: initial susceptibility
360: $\chi$ and Rayleigh constant $\eta$ (see, e.g. \cite{Chikazumi}):
361: \begin{equation} \label{eq:Rayleigh Model}
362: y_{a,d}= \pm y_{s}+\chi(x \mp \Delta) \pm \frac{1}{2} \eta (x \mp \Delta)^{2},\\
363: \end{equation}
364: where the upper (lower) sign corresponds to the ascending
365: (descending) magnetization in increasing (decreasing) magnetic field
366: (Fig.~\ref{Hysteresis}). The three fundamental parameters of the
367: hysteresis loop: the remnant magnetization $y_{r}=y_{r}(a)$, the
368: saturation magnetization $y_{s}$ and the coercivity $x_{c}=x_{a}$,
369: which is the reverse field to bring the magnetization to zero from
370: initial saturation, are characterized by strong dependence on the
371: applied magnetic field, temperature and Dingle temperature through
372: the reduced amplitude of the dHvA oscillations
373: $a=a(\mu_{0}H,T,T_{D})$. The saturation magnetization $y_{s}=y_{a}$
374: can be calculated numerically from the explicit equation
375: \begin{equation} \label{eq:Saturation Magnetization}
376: y_{s}=a \sin{[\Delta + (1-n)y_{s}]}.\\
377: \end{equation}
378: The Eq.~(\ref{eq:Rayleigh Model}) gives the following expressions
379: for the coercivity factor $x_{c}$ and remnant magnetization $y_{s}$:
380: \begin{eqnarray}
381: x_{c}=\frac{\delta\Delta^{2}}{1-\delta\Delta+\sqrt{(1-\delta\Delta)^{2}+(\delta\Delta)^{2}}}
382: \approx \delta\Delta^{2},\label{eq:Coercivity}\\
383: y_{r}=\frac{1}{2}\eta \Delta^{2}, \qquad \qquad
384: \label{eq:Remnant}
385: \end{eqnarray}
386: where $\delta=\eta/\chi$. In calculation of
387: Eqs.~(\ref{eq:Coercivity}), (\ref{eq:Remnant}) we used $\delta
388: \Delta \ll 1$, which is justified by the observed narrow hysteresis
389: loop in Be \cite{Kramer2}. In this limit, coercivity factor $x_{c}$
390: and remnant magnetization $y_{r}$, being quadratic functions of the
391: range of CD existence $\Delta$ (\ref{eq:Delta}), have the same
392: temperature and magnetic field dependencies. The initial
393: susceptibility $\chi$ can be calculated from measured remnant
394: magnetization $y_{r}$ and coercivity factor $x_{c}$ according to the
395: equation $\chi=2y_{r}/x_{c}$. The hysteresis loop square which
396: characterizes the losses in the process of diamagnetic
397: magnetization, is defined as follows
398: \begin{equation} \label{eq:Losses}
399: S=\frac{4}{3}\eta \Delta^{3}.\\
400: \end{equation}
401: and also temperature and magnetic-field dependent. The loop square
402: $S$ Eq.~(\ref{eq:Losses}) reaches its maximum
403: $S_{max}=(4/3)\pi^{3}\eta$ at $a \gg 1$, inasmuch as the range of
404: the existence of the CDs $2\Delta$ in every period of the dHvA
405: oscillations is restricted by the period of the oscillations (
406: $2\pi$ in relative units). Analysis of the
407: Eqs.~(\ref{eq:Coercivity})-(\ref{eq:Losses}) with taking into
408: consideration Eq.~(\ref{eq:Delta}) shows that the parameters of the
409: hysteresis loop $x_{c}, y_{r}, S \to 0$ at the point of diamagnetic
410: phase transition, when $a \to 1$.
411:
412: \begin{figure}
413: \includegraphics[width=0.4\textwidth]{Fig_2.eps}
414: \caption{\ Magnetic field dependence of the reduced coercivity
415: $x_{c}=x_{c}(\mu_{0}H)$ as a function of applied magnetic field at
416: $T=0.5K$ and $T_{D}=2K$ in the whole interval of CD existence in Be
417: (a) and at three different temperatures $T=0.5K$ (solid line),
418: $T=1K$ (broken line) and $T=1.5 K$ (dot line) (b) in the part of the
419: magnetic field interval. For calculations the value of $\delta=0.1$
420: is used.} \label{Magnetic Field Dependence}
421: \end{figure}
422:
423: The magnetic field dependence of the coercivity
424: $x_{c}=x_{c}(\mu_{0}H)$ (\ref{eq:Coercivity}) at the Dingle
425: temperature $T_{D}=2K$ of the experiment \cite{Kramer2} and three
426: different temperatures $T$ from $0.5K$ to $1.5K$ is represented in
427: Fig.~(\ref{Magnetic Field Dependence}). To construct the plots we
428: used $\delta=0.1$, which is justified by comparison with the data
429: \cite{Kramer3}. The function $x_{c}=x_{c}(\mu_{0}H)$ is plotted at
430: the whole range of applied magnetic field $H$ from $1T$ till $6T$
431: (Fig.~\ref{Magnetic Field Dependence}(a)), where the non-uniform
432: diamagnetic phase exists at given Dingle temperature $T_{D}=2K$
433: \cite{Kramer3}. The function $x_{c}=x_{c}(\mu_{0}H)$ has asymmetric
434: multi-bell like shape, typical for the phase diagrams of Be
435: $\mu_{0}H-T-T_{D}$ \cite{Logoboy3}, \cite{Kramer3}. The maximums of
436: the function $x_{c}=x_{c}(\mu_{0}H)$ appear periodically on the
437: scale of inverse magnetic field with the period inversely
438: proportional to the discrepancy $\Delta F=F_{h}-F_{w}$ of the two
439: fundamental frequencies, $F_{h}$ and $F_{w}$, corresponding to two
440: extreme cross sections of cigar-like Fermi surface of Be:
441: \begin{equation} \label{eq:Period}
442: \Delta(\frac{1}{\mu_{0}H})=\frac{1}{\Delta F}.\\
443: \end{equation}
444: For $F_{h}=970.9T$ and $F_{w}=942.2T$
445: \cite{Shoenberg},\cite{Kramer2} we obtain $\Delta F=28.7T$. The
446: increase of the temperature $T$ results in decrease of the maximums
447: of the function $x_{c}=x_{c}(\mu_{0}H)$ till its disappearing, when
448: the point of phase transition is reached ($a=1$). The position of
449: the maximums of the function $x_{c}=x_{c}(\mu_{0}H)$ does not depend
450: on temperature in accordance with the phase diagrams for Be
451: \cite{Logoboy3}. The periodicity of the coercivity on reciprocal
452: magnetic field is illustrated in Fig.~\ref{Inverse Magnetic
453: Field}(a), where the coercive force $X_{c}=x_{c}/k=X_{c}(\mu_{0}H)$
454: measured in Gauss is plotted as a function of $(\mu_{0}H)^{-1}$ at
455: $T=1.3K$ and $T_{D}=1.8K$. It follows from Fig.~\ref{Inverse
456: Magnetic Field}(a), that the preferable interval of applied magnetic
457: field for measuring the coercive force is $2-4T$ , where the
458: function $X_{c}$ can reach maximum values from $1G$ to $2G$ at given
459: values of $T=1.3K$ and $T_{D}=1.8K$.
460:
461: \begin{figure}[b]
462: \includegraphics[width=0.4\textwidth]{Fig_3.eps}
463: \caption{(a) Coercive force $X_{c}=x_{c}/k=X_{c}(\mu_{0}H)$ as a
464: function of inverse magnetic field at $T=1.3K$ and $T_{D}=1.8K$, and
465: (b) reduced coercivity $x_{c}=x_{c}(\mu_{0}H)$ at $T=1.3K$ and three
466: different Dingle temperatures: $T_{D}=1.8$ (solid line), $T_{D}=2K$
467: (broken line) and $T_{D}=2.2K$ (dot line).} \label{Inverse Magnetic
468: Field}
469: \end{figure}
470:
471: The increase of the Dingle temperature $T_{D}=\hbar /2\pi k_{B}\tau$
472: which is inversely proportional to the scattering lifetime $\tau$ of
473: the conduction electrons, leads to the reduction of the amplitude of
474: the dHvA oscillations, resulting in the phase transition into
475: uniform diamagnetic phase. Thus, the range of the existence of the
476: CD phase $\Delta$ Eq.~(\ref{eq:Delta}) decreases ($\Delta \to 0$,
477: when $a \to 1$). Assuming that this contribution to the coercive
478: force $x_{c}$ Eq.~(\ref{eq:Coercivity}) is dominant and neglecting
479: an unessential change of the parameter $\delta=\chi/\eta$, one can
480: come to the bells-like magnetic field dependencies of $x_{c}$, as
481: illustrated in Fig.~\ref{Inverse Magnetic Field}.
482:
483: The temperature dependencies of the reduced coercivity factor
484: $x_{c}=x_{c}(T)$ and measured coercive force $X_{c}=X_{c}(T)$ at
485: $\mu_{0}H=3.6T$ and different Dingle temperatures are shown in
486: Fig.~\ref{Temperature Dependence}. The circles in
487: Fig.~\ref{Temperature Dependence}(b) represent the results of
488: measurement of the coercive force in Be \cite{Kramer3}. The
489: theoretical curve $X_{c}=X_{c}(T)$ plotted at$\mu_{0}H=3.6T$ and
490: $T_{D}\approx 1.8K$ fits the experimental points. The reported
491: Dingle temperature in \cite{Kramer3} is $T_{D}=2K$. There is a good
492: agreement between predicted temperature dependence of the coercive
493: force and data \cite{Kramer3}.
494:
495: The observation of the hysteresis loop in Be \cite{Kramer1} provides
496: the possibility of the verification of the developed theory of the
497: diamagnetic hysteresis and evaluate parameters of the loop. The
498: periodical irreversible behavior of the magnetization process in the
499: rod-like sample $8\times2\times1mm^{3}$ of Be is demonstrated at
500: $T=1.3K$ in the whole range of applied magnetic field $\mu_{0}H
501: \approx 2-6T$, where the CD phase exists. The Dingle temperature is
502: $T_{D}=2K$. The diamagnetic phase diagrams estimated from
503: Eq.~(\ref{eq:DPT}) with taking into account the Eq.~(\ref{eq:Reduced
504: Amplitude}) for the reduced amplitude of oscillations $a$, obtained
505: in the framework of the slightly corrugated Fermi surface of Be
506: \cite{Logoboy3}, are consistent with the data \cite{Kramer2},
507: \cite{Kramer3}. At $\mu_{0}H=3.6T$ the calculated period of the dHvA
508: oscillations $\Delta H=2\pi/k=(\mu_{0}H)^{2}/F_{0} \approx 138G$
509: ($F_{0}=955T$ is average area of the extreme cross-sections of the
510: Fermi surface sheet, relevant to dHvA oscillations \cite{Shoenberg})
511: is very close to the observed period of the dHvA oscillations
512: $\Delta H \approx 134-138G$ \cite{Kramer2}. According to the data
513: \cite{Kramer2}, the coercive force $X_{c}\approx 1.2G$, the remnant
514: magnetization $Y_{r}=y_{r}/k\approx 1.2G$, the saturation
515: magnetization $Y_{s}$ and the range of existence of the non-uniform
516: phase $2X_{\Delta}$ can be evaluated as $Y_{s}\approx 30G$ and
517: $2X_{\Delta}\approx 34G$ consequently. Thus, the experimental value
518: of the reduced parameter $\Delta^{(exp)}=X_{\Delta}k\approx 0.8$ can
519: be compared with the corresponding theoretical value of $\Delta$
520: Eq.~(\ref{eq:Delta}) in the following way. At $T=1.3K, T_{D}=2K$ and
521: $\mu_{0}H=3.6T$, the reduced amplitude of the dHvA oscillations
522: according to the Eq.~(\ref{eq:Reduced Amplitude}) is $a\approx 2$,
523: giving the value of $\Delta\approx0.7$ in accordance with the
524: experimental value $\Delta^{(exp)}=X_{\Delta}k\approx 0.8$
525: \cite{Kramer2}. According to this calculation, at the conditions of
526: the experiment \cite{Kramer2}, CD phase occupies $\approx 0.25$ part
527: of the period. The data \cite{Kramer2} allow us to evaluate the
528: parameters of the Rayleigh hysteresis loop, $\chi$ and $\eta$.
529: First, we calculate $y_{r}=kY_{r}\approx0.053$, that gives
530: $\eta=2y_{r}/\Delta^{2}\approx0.18$, $\chi=2y_{r}/x_{c}\approx2$ and
531: $\delta=0.09\approx0.1$. Then, we can compare the experimental and
532: theoretical values for saturation magnetization. According to the
533: data \cite{Kramer2} $y^{(exp)}_{s}=kY_{s}\approx1.4$. The
534: theoretical value of the saturation magnetization $y_{s}$ can be
535: calculated from Eq.~(\ref{eq:Saturation Magnetization}). Using the
536: value of the demagnetizing coefficient $n\approx0.05$, calculated
537: for prolate ellipsoid \cite{Osborn}, $a\approx2$ and
538: $\Delta\approx0.7$ one can obtain $y_{s}\approx1.6$ in accordance
539: with the experimental value of $y^{exp}_{s}=1.4$.
540:
541: \begin{figure}
542: \includegraphics[width=0.4\textwidth]{Fig_4.eps}
543: \caption{(a) Temperature dependence of the reduced coercivity factor
544: $x_{c}=x_{c}(T)$ and (b) coercive force $X_{c}=x_{c}/k=X_{c}(T)$ at
545: $\mu_{0}H=3.6T$ and six different Dingle temperatures $T_{D}:
546: 1~-~1.8K, 2~-~1.9K, 3~-~2K, 4~-~2.1K, 5~-~2.2K, 6~-~2.3K$. Close
547: circles are referred from R.Kramer {\it et al
548: }\cite{Kramer3}.}\label{Temperature Dependence}
549: \end{figure}
550:
551: \section{Conclusions}
552:
553: Magnetic interaction of strongly correlated electron gas results in
554: non-linear relationship between magnetization and magnetic field,
555: which can give rise an effect of hysteresis in magnetization curve
556: due to existence of metastable states for local diamagnetic moments
557: in each period of dHvA oscillations. There are three main
558: contributions into the process of magnetization of normal metals in
559: the condition of strong dHvA effect: the domain wall displacement,
560: the nucleation of domains and the dependence of local diamagnetic
561: moments on external magnetic field \cite{Logoboy1}.
562:
563: We developed the model of diamagnetic hysteresis in CD phase in the
564: framework of Rayleigh hysteresis loop. Rayleigh law for the
565: magnetization process of spin origin is generally believed to be
566: caused only by the wall displacement and in relatively small
567: magnetic field. There are two circumstances that can justify the
568: applicability of the Rayleigh model for magnetization process of
569: strongly correlated electron gas. The first, although for
570: observation of the dHvA effect the high magnetic field in the scale
571: of $1-10T$ is need, the interplay between uniform and non-uniform
572: phases realized at the fine magnetic field scale of the magnetic
573: field increment $|h|\le\Delta H=(\mu_{0}H)^{2}/F$, e.g. $\sim mT$,
574: in every period of the dHvA oscillations. And, the second, the
575: magnetization process for electron gas in the conditions of dHvA
576: effect is dominated by domain wall motion which is similar to the
577: magnetization process of soft thin magnetic films in the physics of
578: magnetism of spin origin. There is a good agreement between the
579: theoretical results obtained in the framework of the developed model
580: of diamagnetic hysteresis and recent experimental results on
581: observation of the hysteresis in Be \cite{Kramer2}. The diamagnetic
582: hysteresis is usually small, appears periodically in every period of
583: dHvA oscillations with changing the value of applied magnetic field,
584: and is described in the framework of the Rayleigh model with two
585: fundamental parameters: initial susceptibility $\chi$ and Rayleigh
586: constant $\eta$. We calculate the parameters of the diamagnetic
587: hysteresis from the data \cite{Kramer2}. The zero field saturation
588: (remnant) magnetization $y_{c}$, coercivity $x_{c}$ and hysteresis
589: loop square $S$ are characterized by strong dependence on
590: temperature, magnetic field and impurity of the sample through the
591: reduced amplitude of the dHvA oscillations $a(\mu_{0}H,T,T_{D})$.
592: The irreversible changes in magnetization are the result of
593: existence of metastable states due to defects of crystal lattice,
594: intrinsic stress and deformation. Generally, the diamagnetic
595: hysteresis occurs because the domain boundaries do not return to
596: their original positions when the external magnetic field approaches
597: to its value corresponding to the centre of period of dHvA
598: oscillations. At the initial permeability range the domain walls are
599: displaced reversibly from their stable positions which essentially
600: determined by the homogeneity of materials. Beyond the initial
601: permeability range, the intensity of magnetization increases and the
602: process is irreversible. The irreversible magnetization range is
603: attained by irreversible displacement of domain walls. Although the
604: rotation processes, that are intrinsic property of magnetization
605: loop in magnetic materials of spin origin, can be neglected in
606: rearrangement of CD structure, the microscopic theory of diamagnetic
607: loop has to include into consideration the possible bending of the
608: domain wall due to magnetization current densities localized in the
609: domain wall close to the sample surface \cite{Logoboy2}. The
610: irreversibility of the magnetization process is believed to be the
611: intrinsic property of the CD structure for all normal metals. The
612: present model of diamagnetic hysteresis can be applied for
613: investigation of temperature and magnetic field dependencies of
614: diamagnetic magnetization not only in Be, but in other systems,
615: where the CDs were observed. We hope that our theoretical results
616: will motivate the experimentalists to carry out further experiments
617: in investigation such an exotic phenomenon as diamagnetic
618: hysteresis.
619:
620: \begin{acknowledgments}
621: We are indebted to V. Egorov and I. Sheikin for illuminating
622: discussions. We are also grateful to R. Kramer for providing us with
623: his experimental results before publication.
624: \end{acknowledgments}
625:
626: \begin{thebibliography}{99}
627:
628: \bibitem{Shoenberg} D.~Shoenberg, {\it Magnetic Oscillations in Metals} (Cambridge University Press, Cambridge, England, 1984).
629: \bibitem{Condon} J.~H.~Condon, Phys. Rev. {\bf 45}, 526 (1966).
630: \bibitem{Condon_Walstedt} J.~H.~Condon and R.~E.~Walstedt, Phys. Rev. Lett. {\bf 21}, 612 (1968).
631: \bibitem{Kramer1} R.~G. Kramer, V.~S.~Egorov,~V.~A.~Gasparov, A.~G.~M.~Jansen, and W.~Joss, Phys. Rev. Lett. {\bf 95}, 267209 (2005).
632: \bibitem{Solt1} G.~Solt and V.~S.~Egorov, Physica B {\bf 318}, 231 (2002).
633: \bibitem{Solt2} G.~Solt, V.~S.~Egorov, C.~Baines, D.~Herlach, and U.~Zimmermann, Physica B {\bf 326}, 536 (2003).
634: \bibitem{Logoboy1} N.~Logoboy, A.~Gordon, I.~D.~Vagner, and W.~Joss, Solid State Comm. {\bf 134}, 497 (2005).
635: \bibitem{Kramer2} R.~B.~G.~Kramer, V.~S.~Egorov, A.~G.~M.~Jansen, and W.~Joss, Phys. Rev. Lett. {\bf 95}, 187204 (2005).
636: \bibitem{Logoboy2} N.~Logoboy, V.~S.~Egorov, and W.~Joss, Solid State Comm. {\bf 137}, 570 (2006).
637: \bibitem{Kittel} C.~Kittel, Rev. Mod. Phys. {\bf 21}541 (1949) .
638: \bibitem{Chikazumi} S. ~Chikazumi, {\it Physics of Ferromagnetism } (Clarendon Press, Oxford,
639: England, (1997).
640: \bibitem{Logoboy3} N.~Logoboy and W.~Joss, Solid State Comm. {\bf 139}, 191 (2006).
641: \bibitem{Kramer3} R.~B.~G.~Kramer, (2006) ({\it private communiation}).
642: \bibitem{Osborn} J.~A.~Osborn, Phys. Rev. {\bf 67} 351 (1945)
643:
644: \end{thebibliography}
645:
646: \end{document}
647: