A conjecture on the relationship between critical residual entropy and finite temperature pseudo-transitions of one-dimensional models
Extracted source text · arXiv 1810.07817
%% LyX 2.2.3 created this file. For more info, see http://www.lyx.org/.
%% Do not edit unless you really know what you are doing.
\documentclass[10pt,english]{elsarticle}
\usepackage[T1]{fontenc}
\usepackage[latin9]{inputenc}
\usepackage{geometry}
\geometry{verbose,tmargin=3cm,bmargin=3cm,lmargin=3cm,rmargin=3cm}
\setcounter{secnumdepth}{4}
\setcounter{tocdepth}{4}
\synctex=-1
\usepackage{color}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{graphicx}
\makeatletter
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% User specified LaTeX commands.
%\usepackage{scalefnt}
%\journal{}
%\include{shadowtext}
\makeatother
\usepackage{babel}
\begin{document}
\title{Phase boundary residual entropy and finite temperature pseudo-transition
for one-dimensional models}
\author{Onofre Rojas}
\address{Departamento de Física, Universidade Federal de Lavras, 37200-000,
Lavras - MG, Brazil}
\address{The Abdus Salam International Centre for Theoretical Physics (ICTP),
34151 Trieste, Italy}
\begin{abstract}
Recently was observed clues of pseudo-critical temperature in one-dimensional
spin models, such as the Ising-Heisenberg spin models, among others,
exhibiting the pseudo-transitions. Here we report an intrinsic relationship
between the zero temperature phase boundary residual entropy and pseudo-
transition. Usually, the residual entropy increase at the phase boundary,
which means the system becomes with more accessible states in the
phase boundary compared to its adjacent states. However, this is not
always the case; there are some phase boundaries where the entropy
remains equal to the largest residual entropy of the adjacent states.
Therefore, we propose the following statement at zero temperature.
If the phase boundary residual entropy is continuous at least from
the one-sided limit, then the analytic free energy exhibits a pseudo-transition
at finite temperature. This condition would be essential to study more
realistic models. Just by analyzing at zero temperature behavior of
the residual entropy, we can know whether the system will exhibit
pseudo-transition. To illustrate our argument, we use a couple of
examples of Ising-Heisenberg models to show the pseudo-transitions
behaviors due to the phase boundary residual entropy continuity. These
are a frustrated coupled double tetrahedral chain and an unfrustrated
diamond chain.
\end{abstract}
\begin{keyword}
Residual entropy; Quasi-phases; Pseudo-transitions; Ising-Heisenberg
\end{keyword}
\maketitle
\section{Introduction}
In 1950 van Hove\cite{Hove}, verified the absence of phase transition
with a short-range order for the uniform one-dimensional system. Writing
the partition function by using the transfer matrix technique and
reducing the problem to the largest eigenvalue, which implies that
the free energy is an analytic function. Thus established a theorem
to prove that one-dimensional models with short-range coupling do
not exhibit any phase transitions. Recently, Cuesta and Sanchez\cite{cuesta}
argued that van Hove's theorem should satisfy the following conditions:
(i) Homogeneity, excluding automatically inhomogeneous system, i.e.,
disordered or periodic. (ii) The Hamiltonian does not include particles
position terms, e.g., external fields. (iii) Hard-core particles,
this means the theorem cannot be applied to point-like or soft particles.
Hence, Cuesta and Sanchez\cite{cuesta} proposed a more general non-existence
theorem for phase transition at finite temperature. Mainly they included
an external field and considered point-like particles. Which broadens
the non-existence theorem, but it is not yet a fully general theorem,
e.g., were not included mixed particle chains and more general external
fields.
Unlike, there are some one-dimensional models with a short-range coupling
that exhibit a first-order phase transition at finite temperature.
The Kittel model (also known as a zipper model)\cite{kittel}, is
a typical simple model with a finite transfer matrix. Whose constraint
on zipper corresponds to an infinite potential, and this condition
leads to the non-analytic free energy. Consequently, the system exhibits
a first-order phase transition. Another model is that considered by
Chui-Weeks model\cite{chui}, with a typical set of models called
solid-on-solid for surface growth. Whose transfer matrix dimension
is infinite, but can be solved exactly. Imposing the impenetrable
condition to subtract, the model shows the existence of phase-transition.
Dauxois-Peyrard model\cite{dauxois}, is another model with infinite
transfer matrix dimension, which can only be solved numerically. Recently
Sarkanych et al.\cite{sarkanych} proposed a one-dimensional Potts
model with invisible states and short-range coupling. The term invisible
essentially refers to an additional energy degeneracy, which contributes
to the entropy, but not the interaction energy. These invisible states
are the responsible for generating a first order phase transition.
This one is similar to the infinite energy potential in Kittel model\cite{kittel}.
In a nutshell, all these models break the Perron-Frobenius Theorem,
because free energy becomes non-analytical at the phase transition
temperature, or equivalently some elements of the transfer matrix
become null (corresponding to infinite energy).
On the other hand, the term \textquotedbl{}pseudo-transition\textquotedbl{}
was introduced by Timonin\cite{Timonin} in 2011 when studied the
spin ice in a field, to refer to a sudden change in first derivative
of the free energy, while a strong vigorous peak appears in the second
derivative of free energy, although there are no discontinuity or
divergence, respectively. Later this definition was adopted for our
group\cite{Isaac,Isaac 2,pseudo} because we found the same kind of
property. The pseudo-transition does not violate the Perron-Frobenius
theorem, because the free energy is always analytic. Equivalently,
some elements (Boltzmann factor) of the transfer matrix become only
a tiny amount compared to other elements, or the corresponding energy
becomes large but finite compared to ground state energy. So, we can
ignore the small Boltzmann factor without changing its exact result
significantly. To illustrate better this behavior, let us consider
a decorated model that can be mapped into a simple spin-1/2 Ising-like
model with Hamiltonian
\begin{equation}
H=-\sum_{i=1}^{N}\left[K_{0}+Ks_{i}s_{i+1}+\tfrac{1}{2}B(s_{i}+s_{i+1})\right],
\end{equation}
where $K_{0}$, $K$ and $B$ are effective parameters which in general
could depend on temperature, and with $N$ unit cells. Therefore,
the transfer matrix can be expressed as $\mathbf{V}=\left[\begin{array}{cc}
w_{1} & w_{0}\\
w_{0} & w_{-1}
\end{array}\right]$, like discussed in reference \cite{pseudo}. Each elements of transfer
matrix $w_{n}$ or Boltzmann factors with $n=\{-1,0,1\}$, (we will
call as sectors), can be written as
\begin{equation}
w_{n}=\sum_{k=0}g_{n,k}{\rm e}^{-\beta\varepsilon_{n,k}}.\label{eq:G-w}
\end{equation}
Here $\varepsilon_{n,k}$ represent the energy levels $k=\{0,1,\ldots\}$
and $g_{n,k}$ denotes the degeneracy for each energy level and assuming
$g_{n,k}=\{1,2,3,\dots\}$. With $\beta=1/k_{B}T$, $k_{B}$ Boltzmann
constant and $T$ is the absolute temperature.
With the corresponding transfer matrix eigenvalues provided by
\begin{equation}
\lambda_{\pm}=\tfrac{1}{2}\Bigl(w_{1}+w_{-1}\pm\sqrt{(w_{1}-w_{-1})^{2}+4w_{0}^{2}}\Bigr).\label{eq:L12}
\end{equation}
Assuming the chain with a periodic boundary condition, the partition
function becomes $\mathcal{Z}_{N}=\lambda_{+}^{N}+\lambda_{-}^{N}$.
Consequently, the free energy can be obtained in the thermodynamic
limit ($N\rightarrow\infty$) as
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left[\tfrac{1}{2}\Bigl(w_{1}+w_{-1}+\sqrt{(w_{1}-w_{-1})^{2}+4w_{0}^{2}}\Bigr)\right].\label{eq:free-energ}
\end{equation}
As discussed in the previous literature\cite{pseudo}, if we consider
$\bar{w}_{0}\equiv\frac{w_{0}}{|w_{1}-w_{-1}|}\rightarrow0$, or in
terms of $\min(\varepsilon_{0,0}-\varepsilon_{1,0},\varepsilon_{0,0}-\varepsilon_{-1,0})\rightarrow+\infty$.
Ignoring the higher order terms correction ${\cal O}(\bar{w}_{0}^{2})$,
the free energy reduces to
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left[\max\left(w_{1},w_{-1}\right)\right].\label{eq:free-dsc}
\end{equation}
This result could mean, the presence of a genuine phase transition
at finite temperature, because \eqref{eq:free-dsc} becomes a non-analytic
function when $w_{1}=w_{-1}$. Of course, this cannot happen, unless
the lowest energy in sector $n=0$ satisfies.
From now on, we will focus on the thermodynamic properties when $w_{1}$
and $w_{-1}$ are competing term or equivalently means around the
``quasi-phase'' transition between sectors $n=1$ and $n=-1$.
Recent investigations revealed a number of decorated one-dimensional
models, particularly the Ising and Heisenberg models with a variety
of structures. Such as the Ising-Heisenberg in the diamond chain\cite{torrico,torrico2}.
One-dimensional double-tetrahedral model, where the nodal site is
assembled by localized Ising spin, and alternating with a pair of
mobile electrons delocalized within a triangular plaquette\cite{Galisova}.
Ladder model with alternating Ising-Heisenberg coupling\cite{on-strk}.
As well as the triangular tube model with Ising-Heisenberg coupling
\cite{strk-cav}. In all these models pseudo-transition clues were
observed. The first derivative of the free energy, such as entropy,
internal energy or magnetization show an abrupt jump as varying the
temperature. Which is similar to the first order phase transition,
but the function is continuous. While a second order derivative of
free energy, like the specific heat and magnetic susceptibility, resembles
a typical behavior of second-order phase transition at finite temperature.
Therefore, this peculiar behavior drew attention to a more careful
study, as considered in reference \cite{pseudo}. Lately, in reference\cite{Isaac}
has been made an additional discussion on this property and detailed
study of the correlation function for arbitrarily distant spins around
the pseudo-transition.
The rest of the article is organized as follows: In sec.2 we present
the phase boundary residual entropy and its connection to the pseudo
transitions. In Sec.3 the frustrated Ising-Heisenberg tetrahedral
chain was investigated, and the pseudo-transition using the conditions
proposed in Sec. 2 is taken into account. Analogously, in Sec. 4 is
considered an unfrustrated Ising-Heisenberg diamond chain, and we
investigate phase boundary residual entropy according to Sec.2. Finally,
in sec.5 our conclusions and perspectives are provided.
\section{Phase boundary residual entropy}
Now we analyze the residual entropy in the phase boundary of states
in sectors $n=1$ and $n=-1$. For this purpose, we will use the free
energy given in \eqref{eq:free-energ} when $T\rightarrow0$. Hence,
we can get the residual entropy at zero temperature. Let us assume
that the energies $\varepsilon_{_{1,0}}$ and $\varepsilon_{_{-1,0}}$
depend on some arbitrary parameter $x$ (as illustrated in fig.\ref{fig:Spectral-energy}),
e.g., magnetic field. Changing the parameter $x$, we can tune for
a particular $x_{c}$ where adjacent states coexist, with the energy
$\varepsilon_{c}$ of the phase boundary and corresponding degeneracy
$G_{c}$.
\begin{figure}[h]
\centering{}\includegraphics[scale=0.7]{grnd-st-enr}\caption{\label{fig:Spectral-energy}Ground state-energy phase transition,
the phase boundary occurs at $x_{c}$. Blue lines correspond to sector
$n=-1$ energy levels, red lines correspond to sector $n=1$ energy
levels, and black line denoting the sector $n=0$ energy levels. Tick
lines correspond to lowest energies in each sector, and thin lines
mean excited energies. In (a) boundary energy is composed only by
lowest energies of sectors $n=1$ and $n=-1$. In (b) Some excited
energies could contribute in the phase boundary. }
\end{figure}
Thus the energies $\varepsilon_{_{1,0}}(x)$ and $\varepsilon_{_{-1,0}}(x)$,
in the phase boundary becomes $\varepsilon_{_{1,0}}(x_{c})=\varepsilon_{_{-1,0}}(x_{c})=\varepsilon_{c}$,
with its corresponding degeneracies at phase boundary $g_{1,0}^{c}$
and $g_{-1,0}^{c}$. The degeneracy in phase boundary could become
greater than in adjacent states degeneracies, i.e. $g_{1,0}^{c}\geqslant g_{1,0}$
and $g_{-1,0}^{c}\geqslant g_{-1,0}$ (see fig.\ref{fig:Spectral-energy}b).
The lowest energy, in sector $n=0$ satisfy $\varepsilon_{_{0,0}}(x_{c})>\varepsilon_{c}$,
then we have $\bar{w}_{0}\rightarrow0$. Thus, the free energy around
the boundary state becomes
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left[\tfrac{1}{2}\Bigl(g_{1,0}^{c}+g_{-1,0}^{c}+|g_{1,0}^{c}-g_{-1,0}^{c}|\Bigr){\rm e}^{-\beta\varepsilon_{c}}\right]=\varepsilon_{c}-\tfrac{1}{\beta}\ln\left[\max\left(g_{1,0}^{c},g_{-1,0}^{c}\right)\right].
\end{equation}
Afterward, we can obtain the corresponding phase boundary residual
entropy,
\begin{equation}
\mathcal{S}_{c}=\ln\left[\max\left(g_{1,0}^{c},g_{-1,0}^{c}\right)\right],\label{eq:Sg-psd}
\end{equation}
through the article we will consider the entropy in units of $k_{B}$.
The phase boundary degeneracy per unit cell, results in $G_{c}=\max\left(g_{1,0}^{c},g_{-1,0}^{c}\right)$.
It is worth mentioning, according to the third law of thermodynamics
or often referred to as Nernst's postulate. At zero temperature, the
entropy leads to a constant and must be independent of any parameter
(such as $x$), so the residual entropy is determined only by its
degeneracy of the ground state energy.
\begin{figure}[h]
\centering{}\includegraphics[scale=0.8]{zero-T-entropy}\caption{\label{fig:entropy-T0}(a) Typical zero temperature continuous entropy
as a function of parameter $x$, assuming the system has two phases.
(b) Continuous entropy from the one-sided limit at $x_{c}$, at zero
temperature. (c-d) Typical discontinuous entropy from the non-sided
limit at $x_{c}$, phase boundary residual entropy is strictly more
significant than its adjacent states.}
\end{figure}
\begin{itemize}
\item First, let us assume there is no additional degeneracy in phase boundary,
this means $g_{1,0}^{c}=g_{1,0}$ and $g_{-1,0}^{c}=g_{-1,0}$ (see
fig.\ref{fig:Spectral-energy}a) . Then the phase boundary residual
entropy becomes
\begin{equation}
\mathcal{S}_{c}=\ln\left[\max\left(g_{1,0},g_{-1,0}\right)\right],\label{eq:Smin}
\end{equation}
here the phase boundary residual entropy is never larger than the
largest residual entropy of adjacent states, $G_{c}=g_{1,0}$ or $G_{c}=g_{-1,0}$.
When $g_{1,0}=g_{-1,0}$, the residual entropy is illustrated schematically
in fig.\ref{fig:entropy-T0}a. Therefore, the entropy as a function
of $x$ at zero the temperature, has left and right limits,
\begin{equation}
\lim_{x\rightarrow x_{c}^{-}}\mathcal{S}(x)=\mathcal{S}(x_{c})=\mathcal{S}_{c}\quad\text{and}\quad\lim_{x\rightarrow x_{c}^{+}}\mathcal{S}(x)=\mathcal{S}(x_{c})=\mathcal{S}_{c},
\end{equation}
and both limits are identical, then we say the entropy is continuous
at $x_{c}$.
In principle, this would mean the absence of phase transition. But
there are types of phase with identical entropy, which satisfy this
condition. We will see later this case when we apply to a specific
unfrustrated model.
When $g_{1,0}\ne g_{-1,0}$, the residual entropy is illustrated schematically
in fig.\ref{fig:entropy-T0}b. Assuming the phase adjacent residual
entropies satisfy $\mathcal{S}_{1}<\mathcal{S}_{-1}$, then the left
and right limit of entropy at $x_{c}$ becomes,
\begin{equation}
\lim_{x\rightarrow x_{c}^{-}}\mathcal{S}(x)<\mathcal{S}(x_{c})=\mathcal{S}_{c}\quad\text{and}\quad\lim_{x\rightarrow x_{c}^{+}}\mathcal{S}(x)=\mathcal{S}(x_{c})=\mathcal{S}_{c}.
\end{equation}
In this case, the entropy is continuous from the right-sided limit
but discontinuous from the left-sided limit. Therefore we can say
the entropy is continuous from the one-sided limit at $x_{c}$.
\item Second, when $g_{1,0}^{c}>g_{1,0}$ and/or $g_{-1,0}^{c}>g_{-1,0}$,
the energy levels are illustrated in fig.\ref{fig:Spectral-energy}b.
In this case, the phase boundary residual entropy satisfy $G_{c}>g_{1,0}$
and $G_{c}>g_{-1,0}$. The entropy is discontinuous at $x_{c}$, and
we have two possibilities:
The case when $g_{1,0}^{c}\ne g_{-1,0}^{c}$, is reported in fig.\ref{fig:entropy-T0}c
\begin{equation}
\lim_{x\rightarrow x_{c}^{-}}\mathcal{S}(x)\ne\lim_{x\rightarrow x_{c}^{+}}\mathcal{S}(x)<\mathcal{S}(x_{c})=\mathcal{S}_{c}.
\end{equation}
So the entropy in $x_{c}$ has non-sided limit, with a jump-point
discontinuity at $x_{c}$.
Similarly, for the case $g_{1,0}^{c}=g_{-1,0}^{c}$, as depicted in
fig.\ref{fig:entropy-T0}d,
\begin{equation}
\lim_{x\rightarrow x_{c}^{-}}\mathcal{S}(x)=\lim_{x\rightarrow x_{c}^{+}}\mathcal{S}(x)<\mathcal{S}(x_{c})=\mathcal{S}_{c},
\end{equation}
the entropy is discontinuous at $x_{c}$. Because there is no left
or right sided limit, so it has a point discontinuity at $x_{c}$.
\end{itemize}
Further phase boundary residual entropies for other combinations among
sectors are given in \ref{Appdx-A}. Where we show a number of combinations
among other sectors, and always the phase boundary degeneracy is strictly
larger than their adjacent phase degeneracies.
An alternative way to obtain phase boundary residual entropy could
be by using the combinatorial technique. Similar to that discussed
in the reference\cite{Torr-prb-16,Torr-jmmm-18} to find residual
entropy.
In summary, if the residual entropy is continuous at least from the
one-sided limit, then the analytic free energy exhibits a pseudo-transition
at finite temperature. For our case, the entropy provided by \eqref{eq:Smin}
automatically satisfies the condition of continuity in at least from
the one-sided limit.
\subsection{Low temperature free energy and pseudo transition}
Using the relation \eqref{eq:free-dsc}, we can write the free energy
explicitly as a function of temperature and parameter $x$,
\begin{equation}
f(x,T)=-T\ln\left[\max\left(w_{1}(x,T),w_{-1}(x,T)\right)\right].
\end{equation}
First, let us consider the Boltzmann factor in the low-temperature
region, which we express with good approximation just including ground
state and lowest excited state energy in each sector. Thus we have
\begin{alignat}{1}
w_{1}(x,T)= & g_{1,0}{\rm e}^{-\beta\varepsilon_{1,0}(x)}+g_{1,1}{\rm e}^{-\beta\varepsilon_{1,1}}={\rm e}^{-\beta\bar{\varepsilon}(x)}\,g_{1,0}\;{\rm e}^{-\beta\epsilon(x)/2}\left[1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right],\\
w_{-1}(x,T)= & g_{-1,0}{\rm e}^{-\beta\varepsilon_{-1,0}(x)}+g_{-1,1}{\rm e}^{-\beta\varepsilon_{-1,1}}={\rm e}^{-\beta\bar{\varepsilon}(x)}\,g_{-1,0}\;{\rm e}^{\beta\epsilon(x)/2}\left[1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right],
\end{alignat}
where $\bar{\varepsilon}(x)=\frac{\varepsilon_{1,0}(x)+\varepsilon_{-1,0}(x)}{2}$
is the average energy, $\epsilon(x)=\varepsilon_{1,0}(x)-\varepsilon_{-1,0}(x)$
is energy gap between sectors, and $\delta_{1}(x)=\varepsilon_{1,1}(x)-\varepsilon_{1,0}(x)$
and $\delta_{-1}(x)=\varepsilon_{-1,1}(x)-\varepsilon_{-1,0}(x)$
are energy gap within each sectors.
Next, the free energy becomes
\begin{equation}
f(x,T)=\bar{\varepsilon}(x)-T\ln\left\{ \max\left[g_{1,0}{\rm e}^{-\beta\epsilon(x)/2}\left(1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right),g_{-1,0}{\rm e}^{\beta\epsilon(x)/2}\left(1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right)\right]\right\} .
\end{equation}
For the particular case when $x\rightarrow x_{c}$, and assuming the
energy gap temperature leads to $\beta\delta_{1}(x_{c})\rightarrow+\infty$
and $\beta\delta_{-1}(x_{c})\rightarrow+\infty$ , which implies $\epsilon(x_{c})\rightarrow0$
and $\bar{\varepsilon}(x)\rightarrow\varepsilon_{c}$. So the free
energy can be expressed as
\begin{equation}
f(x_{c},T)=\varepsilon_{c}-T\ln\left[\max\left(g_{1,0},g_{-1,0}\right)\right],
\end{equation}
recovering the previous result in \eqref{eq:Sg-psd}.
When the degeneracies satisfy $g_{1,0}=g_{-1,0}=1$, the free energy
becomes independent of temperature. Hence, we need to include the
first excited energy to express the free energy in the low-temperature
limit,
\begin{equation}
f(x,T)=\bar{\varepsilon}(x)-T\ln\left\{ g_{c}\max\left[\left(1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right),\left(1+\tfrac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta\delta_{-1}(x)}\right)\right]\right\} .
\end{equation}
\subsection{Pseudo-critical temperature}
As already discussed above, a condition to emerge the pseudo-critical
temperature is because we can write the phase boundary residual entropy
as ${\cal S}_{c}=\ln\left[\max\left(g_{1,0},g_{-1,0}\right)\right]$.
Thus the pseudo-critical temperature\cite{Isaac,pseudo} can be found
using the following relation
\begin{equation}
w_{1}(x_{p},T_{p})=w_{-1}(x_{p},T_{p}).\label{eq:w1-w2}
\end{equation}
where $x_{p}$ and $T_{p}$ correspond the parameters where occurs
the pseudo-transition.
\begin{figure}[h]
\centering{}\includegraphics[scale=0.5]{pseud-pDGM}\caption{\label{fig:QP-dmg}Quasi-phase\cite{Timonin} diagram $x$ against
$T$. Only in zero temperature there is real phase transition at $(x_{c},0)$.
For finite temperature arises a pseudo transition at $(x_{p},T_{p})$
for $T>0$.}
\end{figure}
As a first approximation, we can consider just the ground state energy
of the Boltzmann factor. For each of the sectors we have $w_{1}=g_{1,0}{\rm e}^{-\beta\varepsilon_{1,0}}$
and $w_{-1}=g_{-1,0}{\rm e}^{-\beta\varepsilon_{-1,0}}$.
Assuming $\epsilon_{p}=\varepsilon_{1,0}(x_{p})-\varepsilon_{-1,0}(x_{p})$
in \eqref{eq:w1-w2}, we find the following relation,
\begin{equation}
{\rm e}^{-\beta_{p}\epsilon_{p}}=\frac{g_{-1,0}}{g_{1,0}},\quad\Rightarrow\quad T_{p}=\frac{\epsilon_{p}}{\ln\left(\frac{g_{1,0}}{g_{-1,0}}\right)}=\frac{\varepsilon_{1,0}(x_{p})-\varepsilon_{-1,0}(x_{p})}{\ln\left(\frac{g_{1,0}}{g_{-1,0}}\right)}.\label{eq:gen-eq-ws-1}
\end{equation}
The result \eqref{eq:gen-eq-ws-1} was already found in reference
\cite{Galisova,on-strk,strk-cav} , and in this limiting case, we
can obtain an analytical expression for pseudo-critical temperature.
It is also worth mentioning that the critical temperature for the
Kittel model\cite{Galisova,on-strk,strk-cav} has a quite similar
expression.
However, eq.\eqref{eq:gen-eq-ws-1} cannot be useful, for the case
$g_{1,0}=g_{-1,0}$. Then it is necessary to include the lowest excited
energies in at least one sector. Hence, we assume the ground state
energy and the lowest excited energy levels in each of the sectors
as $w_{1}=g_{1,0}{\rm e}^{-\beta\varepsilon_{1,0}}+g_{1,1}{\rm e}^{-\beta\varepsilon_{1,1}}$
and $w_{-1}=g_{-1,0}{\rm e}^{-\beta\varepsilon_{-1,0}}+g_{-1,1}{\rm e}^{-\beta\varepsilon_{-1,1}}$.
Therein we can write a transcendental equation as
\begin{equation}
{\rm e}^{-\beta_{p}\epsilon_{p}}=\frac{g_{-1,0}+g_{-1,1}{\rm e}^{-\beta_{p}\delta_{-1,p}}}{g_{1,0}+g_{1,1}{\rm e}^{-\beta_{p}\delta_{1,p}}}=\left(\frac{g_{-1,0}}{g_{1,0}}\right)\frac{1+\frac{g_{-1,1}}{g_{-1,0}}{\rm e}^{-\beta_{p}\delta_{-1,p}}}{1+\frac{g_{1,1}}{g_{1,0}}{\rm e}^{-\beta_{p}\delta_{1,p}}},\label{eq:gen-eq-ws}
\end{equation}
where $\epsilon_{p}=\varepsilon_{1,0}(x_{p})-\varepsilon_{-1,0}(x_{p})$
is the energy gap between sectors at $x_{p}$, $\delta_{1,p}=\varepsilon_{1,1}(x_{p})-\varepsilon_{1,0}(x_{p})$
and $\delta_{-1,p}=\varepsilon_{-1,1}(x_{p})-\varepsilon_{-1,0}(x_{p})$
are the energy gap for each of the sectors. As schematically reported
in fig.\ref{fig:Spectral-energy}a.
Usually, when ground state degeneracy is unequal, it is enough to
use \eqref{eq:gen-eq-ws-1}. Nevertheless, when the ground state degeneracy
is identical, we need to use eq.\eqref{eq:gen-eq-ws}. If higher energies
demand to find pseudo-critical temperature, is better to consider
the full expression \eqref{eq:w1-w2}.
In fig.\ref{fig:QP-dmg} we illustrate schematically a typical pseudo-transition
curve given by \eqref{eq:gen-eq-ws-1} and \eqref{eq:gen-eq-ws}.
Here we remark that only at zero temperature occurs a real transition
at $x_{c}$.
In ref.\cite{karlova} a similar approach was considered, when analyzing
the maximum of the peak for the specific heat, and to relate the height
of the peak with the degeneracy of the ground state.
\section{Coupled tetrahedral Ising-Heisenberg chain}
\begin{figure}[h]
\centering{}\includegraphics[scale=0.9]{spin-chains-dthd}\caption{\label{fig:Sch-trd-ch}Schematic representation of coupled tetrahedral
Ising-Heisenberg chain. Small balls ($\sigma_{i}$) corresponds to
Ising spins, and large balls ($\boldsymbol{S}_{a(b),i}$) correspond
to Heisenberg spins.}
\end{figure}
Previously, in reference \cite{mambrini,roj-alc} the Heisenberg version
of the coupled tetrahedral Heisenberg chain was considered. Whereas,
Galisova and Strecka\cite{Galisova,galisova17} considered the Ising
and delocalized electrons in the tetrahedral chain. Later, in the
reference \cite{vadim-1,Vadim-2} the Ising-Heisenberg version of
the model was introduced (see eq. \ref{fig:Sch-trd-ch}). Although
this model has already been discussed in reference \cite{vadim-1,Vadim-2},
here we focus on the pseudo-transition property that has not yet been
explored. Thus, we present the Hamiltonian of the model as
\begin{alignat}{1}
H=-\sum_{i=1}^{N} & \left\{ J(\boldsymbol{S}_{a,i},\boldsymbol{S}_{b,i})_{z}+J(\boldsymbol{S}_{b,i},\boldsymbol{S}_{c,i})_{z}+J(\boldsymbol{S}_{c,i},\boldsymbol{S}_{a,i})_{z}+\tfrac{h}{2}\left(\sigma_{i}+\sigma_{i+1}\right)\right.\nonumber \\
& \left.+\left(S_{a,i}^{z}+S_{b,i}^{z}+S_{c,i}^{z}\right)\left[h_{z}+J_{0}(\sigma_{i}+\sigma_{i+1})\right]\right\} ,\label{eq:H-tetra}
\end{alignat}
where $J(\boldsymbol{S}_{a,i},\boldsymbol{S}_{b,i})_{z}=JS_{a,i}^{x}S_{b,i}^{x}+JS_{a,i}^{y}S_{b,i}^{y}+J_{z}S_{a,i}^{z}S_{b,i}^{z}$,
with $S_{a,i}^{\alpha}$ denoting the Heisenberg spin-1/2, and $\alpha=\{x,y,z\}$,
while $\sigma_{i}$ denotes the Ising spin ($\sigma_{i}=\pm\frac{1}{2}$).
In a similar way are defined for sites $b$ and $c$ in \eqref{eq:H-tetra}.
The triangle structure with Heisenberg coupling is an operator of
dimension $8\times8$. Thus we can express as block matrices, one
quadruplet and two doublet states, which can be diagonalized readily.
Below we present the triplet eigenvalues and eigenstates
\begin{alignat*}{1}
e_{_{\frac{3}{2},\frac{3}{2}}}= & -\frac{3J_{z}}{4},
\end{alignat*}
with corresponding eigenvectors
\begin{equation}
\left|\tfrac{3}{2},+\tfrac{3}{2}\right\rangle =\left|\substack{+\\
+\\
+
}
\right\rangle ,\;\text{and}\;\left|\tfrac{3}{2},-\tfrac{3}{2}\right\rangle =\left|\substack{-\\
-\\
-
}
\right\rangle .
\end{equation}
The other eigenvalue is
\begin{alignat*}{1}
e_{_{\frac{3}{2},\frac{1}{2}}}= & -J+\frac{J_{z}}{4},
\end{alignat*}
whose eigenvectors are given by
\begin{alignat}{2}
\left|\tfrac{3}{2},+\tfrac{1}{2}\right\rangle = & \tfrac{1}{\sqrt{3}}\left(\left|\substack{+\\
+\\
-
}
\right\rangle +\left|\substack{+\\
-\\
+
}
\right\rangle +\left|\substack{-\\
+\\
+
}
\right\rangle \right),\quad\text{and}\quad & \left|\tfrac{3}{2},-\tfrac{1}{2}\right\rangle = & \tfrac{1}{\sqrt{3}}\left(\left|\substack{-\\
-\\
+
}
\right\rangle +\left|\substack{-\\
+\\
-
}
\right\rangle +\left|\substack{+\\
-\\
-
}
\right\rangle \right).
\end{alignat}
While the two doublet states are degenerate. Whose eigenvalue is
\begin{alignat*}{1}
e_{_{\frac{1}{2},\frac{1}{2}}}= & \frac{J}{2}+\frac{J_{z}}{4},
\end{alignat*}
and the corresponding four-fold eigenstates are
\begin{alignat}{2}
\left|\tfrac{1}{2},+\tfrac{1}{2}\right\rangle = & \frac{1}{\sqrt{6}}\left(\left|\substack{+\\
+\\
-
}
\right\rangle -2\left|\substack{+\\
-\\
+
}
\right\rangle +\left|\substack{-\\
+\\
+
}
\right\rangle \right),\quad & \left|\tfrac{1}{2},+\tfrac{1}{2}\right\rangle = & \frac{1}{\sqrt{2}}\left(\left|\substack{-\\
+\\
+
}
\right\rangle -\left|\substack{+\\
+\\
-
}
\right\rangle \right),
\end{alignat}
\begin{alignat}{2}
\left|\tfrac{1}{2},-\tfrac{1}{2}\right\rangle = & \frac{1}{\sqrt{6}}\left(\left|\substack{-\\
-\\
+
}
\right\rangle -2\left|\substack{-\\
+\\
-
}
\right\rangle +\left|\substack{+\\
-\\
-
}
\right\rangle \right),\quad\text{and}\quad & \left|\tfrac{1}{2},-\tfrac{1}{2}\right\rangle = & \frac{1}{\sqrt{2}}\left(\left|\substack{-\\
-\\
+
}
\right\rangle -\left|\substack{+\\
-\\
-
}
\right\rangle \right).
\end{alignat}
\subsection{Zero temperature phase diagram}
Now using the eigenvalues found above. We can express the energy levels
and corresponding degeneracy for $n=\{-1,0,1\}$,
\begin{alignat}{2}
\varepsilon_{n,0}= & \frac{1}{2}\left(J_{0}-h\right)n-J+\frac{J_{z}}{4}+\frac{h_{z}}{2}, & g_{n,0}= & 1,\label{eq:e-Trd-0}\\
\varepsilon_{n,1}= & \frac{1}{2}\left(3J_{0}-h\right)n-\frac{3J_{z}}{4}+\frac{3h_{z}}{2}, & g_{n,1}= & 1,\\
\varepsilon_{n,2}= & {\normalcolor {\normalcolor {\color{red}{\normalcolor \frac{1}{2}\left(J_{0}-h\right)n+\frac{J}{2}+\frac{J_{z}}{4}+\frac{h_{z}}{2}}},}} & g_{n,2}= & 2,\\
\varepsilon_{n,3}= & {\color{blue}{\normalcolor -\frac{1}{2}\left(J_{0}+h\right)n+\frac{J}{2}+\frac{J_{z}}{4}-\frac{h_{z}}{2}}},\quad & g_{n,3}= & 2,\\
\varepsilon_{n,4}= & {\color{green}{\normalcolor -\frac{1}{2}\left(3J_{0}+h\right)n-\frac{3J_{z}}{4}-\frac{3h_{z}}{2}}},\quad & g_{n,4}= & 1,\\
\varepsilon_{n,5}= & -\frac{1}{2}\left(J_{0}+h\right)n-J+\frac{J_{z}}{4}-\frac{h_{z}}{2},\quad & g_{n,5}= & 1.\label{eq:e-Trd-5}
\end{alignat}
To analyze the phase diagram at zero temperature, we express some
relevant states below.
\begin{figure}[h]
\begin{centering}
\includegraphics[scale=0.4]{diagm-h-Jz-F}
\par\end{centering}
\caption{\label{fig:(a)-Zero-temperature}(a) Zero temperature phase diagram
in the plane of $J_{z}-h$, assuming fixed parameters $J=-10$, $J_{0}=-10$
and $h_{z}=h$. (b) Entropy density plot for temperature $T=0.6$,
assuming the same set of parameters considered in (a). Here the prefix
letter \textquotedbl{}q\textquotedbl{} is to assign the quasi-phases. }
\end{figure}
Therefore, first, we report the ground state energy of the saturated
phase (SA) which read as,
\begin{alignat*}{1}
E_{SA}=\varepsilon_{1,4}= & -\frac{1}{2}\left(3J_{0}+h\right)-\frac{3J_{z}}{4}-\frac{3h_{z}}{2}.
\end{alignat*}
Whose ground states, Ising spin magnetization, Heisenberg spin magnetization,
and total magnetization, respectively are
\begin{equation}
|SA\rangle=\prod_{i=1}^{N}\left|\tfrac{3}{2},+\tfrac{3}{2}\right\rangle _{i}|+\rangle_{i},\quad\text{with}\quad m_{I}=\frac{1}{2},\quad m_{H}=\frac{1}{2},\quad\text{and}\quad m_{t}=2.\label{Trt-SA}
\end{equation}
Similarly, the ground state energy for ferrimagnetic (FI) phase can
be expressed as
\begin{alignat*}{1}
E_{FI}=\varepsilon_{-1,4}= & \frac{1}{2}\left(3J_{0}+h\right)-\frac{3J_{z}}{4}-\frac{3h_{z}}{2},
\end{alignat*}
and its ground state and magnetizations become,
\begin{equation}
|FI\rangle=\prod_{i=1}^{N}\left|\tfrac{3}{2},+\tfrac{3}{2}\right\rangle _{i}|-\rangle_{i},\quad\text{with}\quad m_{I}=-\frac{1}{2},\quad m_{H}=\frac{1}{2},\quad\text{and}\quad m_{t}=1.\label{Trt-FI}
\end{equation}
The next phase we consider is a frustrated phase, given by
\begin{alignat*}{1}
E_{_{FR_{1}}}=\varepsilon_{1,2}= & \frac{1}{2}\left(J_{0}-h\right)+\frac{J}{2}+\frac{J_{z}}{4}+\frac{h_{z}}{2},
\end{alignat*}
with corresponding ground state and magnetizations, are written as
\begin{equation}
|FR_{1}\rangle=\prod_{i=1}^{N}\left|\tfrac{1}{2},-\tfrac{1}{2}\right\rangle _{i}|+\rangle_{i},\quad\text{with}\quad m_{I}=\frac{1}{2},\quad m_{H}=-\frac{1}{6},\quad\text{and}\quad m_{t}=0.\label{Trt-FR1}
\end{equation}
The other frustrated ground state energy is
\begin{alignat*}{1}
E_{_{FR_{2}}}=\varepsilon_{1,3}= & -\frac{1}{2}\left(J_{0}+h\right)+\frac{J}{2}+\frac{J_{z}}{4}-\frac{h_{z}}{2},
\end{alignat*}
and its respective ground state and magnetizations read as
\begin{equation}
|FR_{2}\rangle=\prod_{i=1}^{N}\left|\tfrac{1}{2},+\tfrac{1}{2}\right\rangle _{i}|+\rangle_{i},\quad\text{with}\quad m_{I}=\frac{1}{2},\quad m_{H}=\frac{1}{6},\quad\text{and}\quad m_{t}=1.\label{Trt-FR2}
\end{equation}
In fig\ref{fig:(a)-Zero-temperature}a, the phase diagram is shown
at zero temperature, where the ground states are given for each region.
The phase boundary between $FR_{1}$ and $FR_{2}$ is given by $h=10$,
the phase boundary degeneracy is composed by $g_{1,0}=2$, $g_{1,1}=2$,
$g_{-1,0}=2$ and $g_{0,0}=2$, then the residual entropy can be obtained
using the eq.\eqref{eq:Sg3}, which becomes ${\cal S}=\ln(3+\sqrt{5})$.
The straight line $h=J_{z}$ describes the interface between $FR_{1}$
and $FI$. Hence, the phase boundary residual entropy can be get using
the eq.\eqref{eq:S_01}, thus we have $\mathcal{S}=\ln(3)$, because
$g_{1,0}=2$ and $g_{-1,1}=1$. In the same way, the boundary between
$FI$ and $SA$ is given by $h=30$. So we can obtain the residual
entropy using the eq.\eqref{eq:Sg3}, which becomes $\mathcal{S}=\ln(2)$
since $g_{1,0}=1$ and $g_{-1,0}=1$. Another case is the boundary
between $SA$ and $FR_{2}$ given by $h=J_{z}$. The phase boundary
entropy at zero temperature can obtained using the eq.\eqref{eq:S-single-sct},
where the residual entropy becomes $\mathcal{S}=\ln(3)$, because
$g_{1,0}=2$ and $g_{1,1}=1$. All the above phase boundaries are
entirely discontinuous, indicating the absence of the pseudo-transition.
At last, we obtain the phase boundary in the interface of $FI$ and
$FR_{2}$ described by $J_{z}=-15$ (red dashed line). The residual
entropy is given by eq.\eqref{eq:Smin} with degeneracies $g_{1,0}=2$
and $g_{-1,0}=1$. Hence, the phase boundary residual entropy becomes
${\cal S}=\ln(2)$. It is worth to remark the phase boundary residual
entropy is equal to the largest adjacent residual entropy. Therefore,
we can affirm this boundary should lead to a pseudo-transition, as
we must confirm further below.
\subsection{Thermodynamics }
Now we can obtain the free energy \eqref{eq:free-energ} as a function
of the Boltzmann factors. For the present model, the Boltzmann factor
is taken using the energy levels given by (\ref{eq:e-Trd-0}-\ref{eq:e-Trd-5}),
\begin{equation}
w_{n}=\sum_{k=0}^{5}g_{n,k}{\rm e}^{-\beta\varepsilon_{n,k}}.
\end{equation}
Hence, the Boltzmann factor can be expressed as follows,
\begin{equation}
w_{n}=2{\rm e}^{\beta\left(\frac{nh}{2}-\frac{J_{z}}{4}\right)}\left\{ \left({\rm e}^{\beta J}+2{\rm e}^{-\beta J/2}\right)\cosh\left(\tfrac{J_{0}+h_{z}}{2}\right)+{\rm e}^{\beta J_{z}}\cosh\left(\tfrac{3}{2}(J_{0}+h_{z})\right)\right\} ,
\end{equation}
where $n=\{-1,0,1\}$.
Using the free energy \eqref{eq:free-energ}, we can find the entropy
of the system at finite temperature. In fig.\ref{fig:(a)-Zero-temperature}b
we illustrate the density plot of the entropy as a function of $J_{z}$
and $h$, for fixed $T=0.6$, and using the same scale of fig.\ref{fig:(a)-Zero-temperature}a.
Here we can observe that the entropy follows the vestige of zero temperature
phase diagram. Definitely, the thermal excitation influences the phase
boundaries. All except one, display an increase of the entropy around
the phase boundaries. The entropy in the boundary $qFI$ and $qFR_{2}$
(the prefix \textquotedbl{}$q$\textquotedbl{} is to assign the quasi-phases
defined by Timonin\cite{Timonin}) does not increase. Because in zero
temperature the entropy is continuous from the one-sided limit at
the phase boundary.
\begin{figure}[h]
\begin{centering}
\includegraphics[scale=0.81]{quasi-phasesSM}\caption{\label{fig:phs-dgm-T}The first column corresponds to density plot,
in the plane $J_{z}-T$ (top) for a fixed $J=-10$, $J_{0}=-10$ and
$h=20$, and in the plane $h-T$ (bottom) for $J=-10$, $J_{0}=-10$
and $J_{z}=-14.6$. Middle column reports the density plot of Ising
spin magnetization. The last column illustrates the density plot of
Heisenberg spin magnetization. }
\par\end{centering}
\end{figure}
In fig.\ref{fig:phs-dgm-T}(left column) reports density plot of entropy
in the plane $T-J_{z}$ (top) and $T-h$ (bottom), for the parameters
considered in the caption. The phase boundary between quasi-phases
$qFR_{2}$ and $qFI$ are easily identified. In fig.\ref{fig:phs-dgm-T}(middle
column) illustrates the Ising spin magnetization $m_{I}$ in the plane
$T-J_{z}$ (b), and in the plane $T-h$ (e). While the right column
reports for Heisenberg spin magnetization ($3m_{H}$) in the plane
$T-J_{z}$ (c), and in the plane $T-h$ (f).
\begin{figure}[h]
\includegraphics[scale=0.76]{Sr-Jz-h-0T}\caption{\label{fig:Entrp-tetra}(a-c) Entropy as a function of $J_{z}$ assuming
fixed $J=-10$, $J_{0}=-10$, for a range of temperature, $T=\{0.2,0.5,1.0,1.5,2.0\}$:
(a) for $h=35$; (b) for $h=20$; (c) for $h=10$. (d) Entropy as
a dependence of $h$ for fixed $J=-10$, $J_{0}=-10$, $J_{z}=-20$
and the same set of temperature in (a-c).}
\end{figure}
In fig.\ref{fig:Entrp-tetra}a is depicted the entropy as a function
of $J_{z}$ in the low-temperature region. We can observe the track
of frustrated ($FR_{2}$) phase which is a state macroscopically degenerate,
with residual entropy $\mathcal{S}=\ln(2)$. The peak corresponds
to the phase boundary between $FR_{2}$ and $SA$ with corresponding
phase boundary residual entropy $\mathcal{S}=\ln(3)$. Here we can
see how a discontinuous residual entropy spreads due to thermal excitation.
In fig.\ref{fig:Entrp-tetra}b is illustrated the entropy as a dependence
of $J_{z}$, where we can observe the phase boundary between $FR_{2}$
with $\mathcal{S}=\ln(2)$ and $FI$ with $\mathcal{S}=0$. Whose
phase boundary entropy is shown and clearly remains at $\mathcal{S}=\ln(2)$
for $T\lesssim1$. This is because the entropy is continuous from
the one-sided limit at phase boundary. In fig.\ref{fig:Entrp-tetra}c
is reported the entropy as a function of temperature, in the interface
between $qFR_{1}$ ($\mathcal{S}=\ln(2)$ ) and $qFR_{2}$ ($\mathcal{S}=\ln(2)$
), whose phase boundary residual entropy is $\mathcal{S}=\ln(3+\sqrt{5})$
for $J_{z}\lesssim-20$. Whereas for $J_{z}=-15$, the phase boundary
joins three phases $qFR_{1}$, $qFR_{2}$ and $qFI$, at first glance
this is similar to fig.\ref{fig:Entrp-tetra}b. But the phase boundary
residual entropy is larger than its adjacent phases, which can be
obtained using the eq.\eqref{eq:Sg3}. Where the degeneracy for each
sector are $g_{1,0}=4$, $g_{-1,0}=3$ and $g_{0,0}=2$, under these
circumstances the entropy becomes $\mathcal{S}=\ln(\frac{7+\sqrt{17}}{2})$.
We can observe better this point in a magnified plot in the inner
part of the fig.\ref{fig:Entrp-tetra}c. Due to this small peak, the
right side curve spreads destroying any evidence of pseudo-transition.
In fig.\ref{fig:Entrp-tetra}d is plotted the entropy as a function
of magnetic field $h$. And we observe a residual entropy between
phase boundaries, which are in agreement with previous plots.
\begin{figure}[h]
\includegraphics[scale=0.81]{qph-trans-Tt}\caption{\label{fig:Phs-qnty-T}Considering fixed parameters $J=-10$, $J_{0}=-10$,
$h=20$ and $J_{z}=\{-13,-14.1,-14.3,-14.5,-14.6\}$ . (a) Entropy
${\cal S}$ as a function of temperature, at pseudo-critical temperature,
is observed a strong change of entropy. (b) Correlation length $\xi$
as dependence of temperature. (c) Specific heat $C$ against temperature.
(d) Magnetic susceptibility $\chi$ as a function of temperature.}
\end{figure}
In fig.\ref{fig:Phs-qnty-T}a is depicted the entropy as a function
of temperature assuming fixed parameters $J=-10$, $J_{0}=-10$, $h=20$
and $J_{z}=\{-13,-14.1,-14.3,-14.5,-14.6\}$. A strong change in the
entropy curvature at the pseudo-critical temperature is evident. So,
when the temperature increases, the strong jump becomes smoother and
then gradually vanishing. In fig.\ref{fig:Phs-qnty-T}b is illustrated
the correlation length as a dependence of temperature. Hence, it is
shown a sharp and robust peak at pseudo-critical temperature. In fig.\ref{fig:Phs-qnty-T}c
is reported the specific heat, and we observe once again a similar
behavior at pseudo-critical temperature. Whereas in fig.\ref{fig:Phs-qnty-T}d
is plotted the magnetic susceptibility as a function of temperature.
The magnetic susceptibility exhibits small sharp peaks, because the
magnetization at zero temperature is identical for $Fr_{2}$ and $FI$
phases, according to \eqref{Trt-FI} and \eqref{Trt-FR2} respectively.
But thermal excitation in Ising and Heisenberg spins provokes different
responses showing small sharp peaks around pseudo-critical temperature.
\section{Ising-XYZ diamond chain}
\begin{figure}[h]
\centering{}\includegraphics[scale=0.8]{spin-chains-dmd}\caption{\label{schm-dmd}Schematic representation of Ising-Heisenberg diamond
chain. Small balls ($\sigma_{i}$) correspond to Ising spins and large
balls ($\boldsymbol{S}_{a(b),i}$) correspond to Heisenberg spins.}
\end{figure}
Another model we consider here is, the Ising-XYZ diamond chain structure
as illustrated in fig.\ref{schm-dmd}, which was previously discussed
in reference \cite{torrico,torrico2}. Where $\sigma_{i}$ (small
balls) represents the Ising spin-1/2, and $S_{a(b),i}^{\alpha}$ (large
balls) denotes the Heisenberg spin-1/2, with $\alpha=\{x,y,z\}$.
However, we will direct our attention to the non-frustrated property.
Thus, here we just give a revisiting of the model\cite{torrico,torrico2},
whose Ising-XYZ Hamiltonian is expressed as
\begin{alignat}{1}
H=-\sum_{i=1}^{N} & \left[J(1+\gamma)S_{a,i}^{x}S_{b,i}^{x}+J(1-\gamma)S_{a,i}^{y}S_{b,i}^{y}+J_{z}S_{a,i}^{z}S_{b,i}^{z}+J_{0}(S_{a,i}^{z}+S_{b,i}^{z})(\sigma_{i}+\sigma_{i+1})\right.\nonumber \\
& \left.+h_{z}(S_{a,i}^{z}+S_{b,i}^{z})+\tfrac{h}{2}(\sigma_{i}+\sigma_{i+1})\right],\label{eq:Hamt}
\end{alignat}
here $J$ corresponds to $xy$-axes exchanges and $\gamma$ being
the XY-anisotropy, $J_{z}$ corresponds to $z$-axis of Heisenberg
spins exchange. While $J_{0}$ denotes Ising-Heisenberg spins exchange,
and $h_{z}$($h$) corresponds to the external magnetic field acting
in Heisenberg spins (Ising spins) respectively, along the $z$-axis.
Some of the relevant ground state energy found in reference \cite{torrico,torrico2}
are summarized below, assuming $n=\sigma_{i}+\sigma_{i+1}$ :
(i) For sector $n=1$ ($\uparrow\uparrow$ ) the first ground state
energy is
\begin{equation}
\varepsilon_{1,0}=E_{_{MF_{2}}}=-\tfrac{J_{z}}{4}-\tfrac{h}{2}-\sqrt{(h_{z}+J_{0})^{2}+\tfrac{1}{4}J^{2}\gamma^{2}},
\end{equation}
Named as modulated ferromagnetic Heisenberg spin ($MF_{2}$) phase.
With corresponding ground state
\begin{alignat}{1}
\left|MF_{2}\right\rangle = & \overset{N}{\underset{i=1}{\prod}}\left(\cos\theta_{1}|\begin{smallmatrix}+\\
+
\end{smallmatrix}\rangle_{i}+\sin\theta_{1}|\begin{smallmatrix}-\\
-
\end{smallmatrix}\rangle_{i}\right)\otimes|\uparrow\rangle_{i},\label{eq:state3}
\end{alignat}
where $\theta_{n}=\frac{1}{2}\tan^{-1}\frac{J\gamma}{2(h_{z}+J_{0}n)}$
defined in $0<\theta_{n}<\pi$.
Whereas the ferrimagnetic (FI) phase, is given by
\begin{equation}
\varepsilon_{1,1}=E_{_{FI}}=-\tfrac{J+h}{2}+\tfrac{J_{z}}{4},
\end{equation}
and the corresponding ground states is expressed as
\begin{alignat}{1}
\left|FI\right\rangle = & \overset{N}{\underset{i=1}{\prod}}\frac{1}{\sqrt{2}}\left(|\begin{smallmatrix}-\\
+
\end{smallmatrix}\rangle_{i}+|\begin{smallmatrix}+\\
-
\end{smallmatrix}\rangle_{i}\right)\otimes|\uparrow\rangle_{i}.\label{eq:state4}
\end{alignat}
(ii) For sector $n=-1$ ($\downarrow\downarrow$), the ground state
energy, becomes
\begin{equation}
\varepsilon_{-1,0}=E_{_{MF_{0}}}=-\tfrac{J_{z}}{4}+\tfrac{h}{2}-\sqrt{(h_{z}-J_{0})^{2}+\tfrac{1}{4}J^{2}\gamma^{2}},
\end{equation}
with respective modulated ferromagnetic ($MF_{0}$) state, given by
\begin{alignat}{1}
\left|MF_{0}\right\rangle = & \overset{N}{\underset{i=1}{\prod}}\left(\cos\theta_{-1}|\begin{smallmatrix}+\\
+
\end{smallmatrix}\rangle_{i}+\sin\theta_{-1}|\begin{smallmatrix}-\\
-
\end{smallmatrix}\rangle_{i}\right)\otimes|\downarrow\rangle_{i}.\label{eq:state2}
\end{alignat}
The free energy for this model was obtained in reference \cite{torrico,torrico2}.
So from \eqref{eq:free-energ} we express using the Boltzmann factor
for Ising-XYZ diamond chain\cite{pseudo},
\begin{equation}
w_{n}=2{\rm e}^{\frac{\beta nh}{2}}\left[{\rm e}^{-\frac{\beta J_{z}}{4}}{\rm ch}\left(\tfrac{\beta J}{2}\right)+{\rm e}^{\frac{\beta J_{z}}{4}}{\rm ch}\left(\beta\Delta_{n}\right)\right],
\end{equation}
where $\Delta_{n}=\sqrt{(h_{z}+J_{0}n)^{2}+\frac{1}{4}J^{2}\gamma^{2}}$.
Performing the derivative of \eqref{eq:free-energ} one can find the
Heisenberg spin magnetization $m_{H}=-\frac{\partial f}{\partial h_{z}}$
and Ising spin magnetization $m_{I}=-\frac{\partial f}{\partial h}$.
At zero temperature, the interface between $MF_{0}$ and $MF_{2}$
(see fig.\ref{fig:Phase-diagram}) assuming $h_{z}=h$, occurs at
a critical magnetic field
\begin{equation}
h_{c}=\frac{(\gamma^{2}-1)+2J_{z}J+4J_{0}^{2}-J_{z}^{2}}{4J+8J_{0}-4J_{z}}.
\end{equation}
For $h<h_{c}$ the system is in $MF_{0}$ state, and for $h>h_{c}$
the system becomes in another state $MF_{2}$.
\begin{figure}[h]
\centering{}\includegraphics[scale=0.8]{phs-dg-IHDm-ch}\caption{\label{fig:Phase-diagram}Density plot for fixed parameters $J=100$,
$\gamma=0.8$, $J_{z}=24$ and $J_{0}=-24$. (a) Entropy (b) Heisenberg
spin magnetization. (c) Ising spin Magnetization. }
\end{figure}
In fig.\ref{fig:Phase-diagram}a we illustrate the density plot entropy
as a function of temperature for fixed parameters $J=100$, $\gamma=0.8$,
$J_{z}=24$ and $J_{0}=-24$. The boundary between quasi-phases $qMF_{0}$
and $qMF_{2}$ is indistinguishable. Because, in zero temperature
both phases $MF_{0}$ and $MF_{2}$ are non-degenerate with entropy
${\cal S}=0$. Thus, the phase boundary residual entropy also becomes
null according to relation \eqref{eq:Smin}, which is easily observed
in the density plot entropy. However, in fig.\ref{fig:Phase-diagram}b
is displayed the density plot of Heisenberg spin magnetization ($m_{H}$),
the boundary between quasi-phase $qMF_{2}$ and quasi-phase $qFM_{0}$
is undoubtedly distinguishable. At $T=0$ and $h<h_{c}=13.063045$,
the Heisenberg spins are parallel ordered with greater probability
pointing up. The maximum magnetization per spin is $m_{H}\sim0.3$,
which is pictorially denoted as effective canting spin (yellow region).
Similarly, for $h>h_{c}=13.063945$ the Heisenberg spin magnetization
becomes negative $m_{H}\sim-0.1$, pictorially illustrated by effective
canting spin down (cyan region). For more detailed information concerning
Heisenberg spin magnetization, we refer the reader to ref. \cite{Isaac}.
In fig.\ref{fig:Phase-diagram}c is displayed the density plot of
Ising spin magnetization, for $h<h_{c}$ the magnetization is nearly
$m_{I}=-0.5$ what means most of the Ising spins are aligned with
spin down, whereas for $h>h_{c}$ most of the spins are pointing up
aligning with the external magnetic field. Consequently, this behavior
could be easily misinterpreted as a real phase transition. Nevertheless,
we do not expect a true phase transition at finite temperature, because
all derivative of the free energy is analytic. At finite temperature,
there is no critical magnetic field, but only a pseudo-critical magnetic
field $h_{p}\lesssim h_{c}$, which vanishes roughly at $T\sim1.0$,
for temperature $T\gtrsim1.0$ the system becomes a standard disordered
system predominantly.
\begin{figure}[h]
\includegraphics[scale=0.79]{Sr-Jz-h-0T-1}\caption{\label{fig:Ent-h-g}Entropy for fixed parameters $J=100$, $J_{z}=24$,
$J_{0}=-24$ and temperature $T=\{0.3,0.5,1.0,1.5,2.0\}$. (a) Entropy
as a function of $h$ considering $\gamma=0.8$. (b) Entropy as a
function of $h$ considering $\gamma=0.7$. (c) Entropy as a function
of $\gamma$ considering $h=18$. (d) Entropy as a function of $\gamma$
considering $h=12$. }
.
\end{figure}
Fig.\ref{fig:Ent-h-g}a illustrates the entropy as a function of $h$,
for a range of temperature $T=\{0.3,0.5,1.0,1.5,2.0\}$. There is
a sudden change for $T\lesssim1.0$, this corresponds to the pseudo-transition
at $h=13.063945$. Therefore, for any region or phase, the entropy
vanishes when $T\rightarrow0$. Similarly, in fig.\ref{fig:Ent-h-g}b
is plotted entropy as a dependence of $h$. We observe a sudden change
for $h=12.8$, again as soon as temperature decreases the entropy
vanishes. However, around $h\approx39$ corresponds to the phase boundary
between $FI$ and $MF_{2}$\cite{pseudo,torrico,torrico2} with a
residual entropy $\mathcal{S}=\ln(2)$, and obviously in this boundary
there is no pseudo-transition. In fig.\ref{fig:Ent-h-g}c-d is plotted
the entropy against $\gamma$, for $h=18$ and $h=12$. For temperature
$T\lesssim1$ arises a sudden change showing the pseudo-transition.
Here, also the entropy vanishes when $T\rightarrow0$ according to
the phase boundary residual entropy. In reference \cite{Isaac,pseudo,torrico,torrico2},
we can find other detailed discussions concerning this model.
\section{Conclusions.}
Although few one-dimensional models exhibit the phase transition,
recently were investigated a pseudo-critical temperature in one-dimensional
spin models\cite{Isaac,pseudo}. There is a number of models exhibiting
pseudo-transition, like the Ising-Heisenberg spin models with a variety
of structures\cite{torrico,torrico2,Galisova,on-strk,strk-cav}. Here
we report an intrinsic relation between zero temperature phase boundary
residual entropy and pseudo-transition. In general, the residual entropy
increases in the interface where occurs the phase transition. What
means the system increased its accessible states in the interface
compared to adjacent states. However, there exist some cases where
the phase boundary entropy remains equal to the largest residual entropy
of neighboring states, and the entropy is given by $\mathcal{S}=\ln\left[\max\left(g_{1,0},g_{-1,0}\right)\right]$.
Our main result dwells in a simple condition to recognize pseudo-transition.
If the phase boundary residual entropy is continuous at least from
the one-sided limit, then analytical free energy exhibits a pseudo-transition.
To show this property, we have considered two Ising-Heisenberg spins
models, one frustrated model in a coupled double tetrahedral chain
and another unfrustrated diamond chain.
Finding pseudo-transition in more realistic systems would be a fascinating
investigation. However, this would be a tough numerical task, even
seeing vestiges of pseudo-transition, one could confuse with a true
phase transition. In this sense, the condition of the phase boundary
entropy is essential. Because we can apply this condition at zero
temperature, and searching for the continuity of entropy would be
a more easy task, compared to the study full thermodynamic quantities.
Once found the entropy is continuous in at least from the one-sided
limit. Thus, for analytic free energy we can observe a pseudo-transitions
at a finite temperature around a specific region using numerical techniques.
\section*{Acknowledgment}
This work was partially supported by ICTP, Brazilian agencies CNPq
and FAPEMIG.
\appendix
\section{\label{Appdx-A}Other phase boundaries}
\subsection{Phase boundary between states of sector $n=0$ and $n=\pm1$ }
First let us consider the sector $n=1$ and $n=0$. Assuming the ground
state energies are given by $\varepsilon_{_{1,0}}(x)$ and $\varepsilon_{_{0,0}}(x)$,
the boundary state occurs when $\varepsilon_{_{1,0}}(x_{c})=\varepsilon_{_{0,0}}(x_{c})=\varepsilon_{c}$
with corresponding degeneracy $g_{1,0}^{c}\geqslant g_{1,0}$ and
$g_{0,0}^{c}\geqslant g_{0,0}$. The lowest energy $\varepsilon_{-1,0}(x)$
in sector $n=-1$, will be strictly higher than $\varepsilon_{c}$
($\varepsilon_{-1,0}>\varepsilon_{c}$), what means $w_{-1}/w_{0}\rightarrow0$
and $w_{-1}/w_{1}\rightarrow0$ when $T\rightarrow0$. So the free
energy in \eqref{eq:free-energ} at sufficiently low temperature reduces
to
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left[\tfrac{1}{2}\Bigl(g_{1,0}^{c}+\sqrt{(g_{1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr){\rm e}^{-\beta\varepsilon_{c}}\right].
\end{equation}
Consequently, the phase boundary residual entropy is
\begin{equation}
\mathcal{S}_{c}=\ln\left[\tfrac{1}{2}\Bigl(g_{1,0}^{c}+\sqrt{(g_{1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr)\right],\label{eq:S_01}
\end{equation}
where the degeneracy in phase boundary results in $G_{c}=\tfrac{1}{2}\Bigl(g_{1,0}^{c}+\sqrt{(g_{1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr)$.
We can observe the boundary degeneracy is strictly larger than its
adjacent degeneracies: $G_{c}>g_{1,0}$ and $G_{c}>g_{0,0}$. The
residual entropy is reported schematically in fig.\ref{fig:entropy-T0}c.
Entropy ${\cal S}_{a}$ and ${\cal S}_{b}$ denote in general a residual
entropy between adjacent states.
For sector $n=-1$ and $n=0$, the result of free energy will be equivalent
to the previous case. We can obtain by merely exchanging $\varepsilon_{_{1,0}}(x_{c})\rightarrow\varepsilon_{_{-1,0}}(x_{c})$,
in expression \eqref{eq:S_01}.
\subsection{Phase boundary lies in a single sector}
Considering $\varepsilon_{1,0}(x)$ and $\varepsilon_{1,1}(x)$ are
ground states energies describing each adjacent states. Thus assuming
that occurs a phase transition between these states, then the boundary
energy is given by $\varepsilon_{1,0}(x_{c})=\varepsilon_{1,1}(x_{c})=\varepsilon_{c}$
at $T=0$. In general, it is possible that some additional states
can coincide in the phase boundary, then the degeneracies can be expressed
as $g_{1,0}^{c}\geqslant g_{1,0}$ and $g_{1,1}^{c}\geqslant g_{1,1}$.
Therefore, all other energy levels must be higher than $\varepsilon_{c}$,
so when $T\rightarrow0$ the spectral energy in other sectors can
be neglected ($w_{0}/w_{1}\rightarrow0$ and $w_{-1}/w_{1}\rightarrow0$).
Hence, the free energy in the low-temperature limit is expressed as
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left(w_{1}\right)=-\tfrac{1}{\beta}\ln\left(g_{1,0}^{c}{\rm e}^{-\beta\varepsilon_{c}}+g_{1,1}^{c}{\rm e}^{-\beta\varepsilon_{c}}\right)=-\tfrac{1}{\beta}\ln\left[\left(g_{1,0}^{c}+g_{1,1}^{c}\right){\rm e}^{-\beta\varepsilon_{c}}\right].
\end{equation}
Whereas, the corresponding boundary residual entropy, reduce to
\begin{equation}
\mathcal{S}_{c}=\ln\left(g_{1,0}^{c}+g_{1,1}^{c}\right).\label{eq:S-single-sct}
\end{equation}
Thereby, the degeneracy in phase boundary is given by $G_{c}=(g_{1,0}^{c}+g_{1,0}^{c})$.
Once again, the phase boundary residual entropy is strictly higher
than any residual entropy of adjacent states, because $G_{c}>g_{1,0}^{c}$
and $G_{c}>g_{1,1}^{c}$.
A schematic representation of the phase boundary residual entropy
is illustrated in fig.\ref{fig:entropy-T0}c.
\subsection{Phase boundary lies in three sectors}
The three sectors can constitute phase boundary, this means the states
with energies $\varepsilon_{_{1,0}}(x)$, $\varepsilon_{_{0,0}}(x)$
and $\varepsilon_{_{-1,0}}(x)$, can coexist for particular $x_{c}$.
So we assume $\varepsilon_{_{1,0}}(x_{c})=\varepsilon_{_{0,0}}(x_{c})=\varepsilon_{_{-1,0}}(x_{c})=\varepsilon_{c}$,
and the respective degeneracies are $g_{1,0}^{c}$, $g_{0,0}^{c}$
and $g_{-1,0}^{c}$. In this case, the free energy becomes
\begin{equation}
f=-\tfrac{1}{\beta}\ln\left[\tfrac{1}{2}\Bigl(g_{1,0}^{c}+g_{-1,0}^{c}+\sqrt{(g_{1,0}^{c}-g_{-1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr){\rm e}^{-\beta\varepsilon_{c}}\right].
\end{equation}
Finally, the phase boundary residual entropy reads
\begin{equation}
\mathcal{S}_{c}=\ln\left[\tfrac{1}{2}\Bigl(g_{1,0}^{c}+g_{-1,0}^{c}+\sqrt{(g_{1,0}^{c}-g_{-1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr)\right].\label{eq:Sg3}
\end{equation}
Then the energy boundary degeneracy is $G_{c}=\tfrac{1}{2}\Bigl(g_{1,0}^{c}+g_{-1,0}^{c}+\sqrt{(g_{1,0}^{c}-g_{-1,0}^{c})^{2}+4(g_{0,0}^{c})^{2}}\Bigr)$.
Similar to the previous cases, the phase boundary residual entropy
is strictly larger than its adjacent states residual entropies, because
$G_{c}>g_{1,0}$, $G_{c}>g_{0,0}$ and $G_{c}>g_{-1,0}$.
In all cases, the entropy inevitably exhibits a jump-point discontinuity
or a point discontinuity at $x_{c}$. Because the phase boundary
residual entropy is strictly larger than its adjacent states. Even
for the case with an energy gap, the system still has entropy discontinuity
(non-sided limit at $x_{c}$).
\begin{thebibliography}{10}
\bibitem{Hove}L. van Hove, Physica \textbf{16},137 (1950).
\bibitem{cuesta}J. A. Cuesta and A. Sanchez, J. Stat. Phys. \textbf{115},
869 (2003).
\bibitem{kittel}C. Kittel, Am. J. Phys. \textbf{37}, 917(1969).
\bibitem{chui}S. T. Chui and John D. Weeks, Phys. Rev. B \textbf{23},
2438 (1981)
\bibitem{dauxois}T. Dauxois and M. Peyrard, Phys. Rev. E \textbf{51},
4027 (1995).
\bibitem{sarkanych}P. Sarkanych, Y. Holovatch and R. Kenna, Phys.
Lett. A \textbf{381}, 3589 (2017).
\bibitem{Timonin}P. N. Timonin, J. Exp. Theor. Phys. 113, 251 (2011).
\bibitem{Isaac}I. M. Carvalho, J. Torrico, S. M. de Souza, O. Rojas,
Oleg Derzhko, arXiv:1808.04158.
\bibitem{Isaac 2}I. M. Carvalho, J. Torrico, S. M. de Souza, M. Rojas,
O. Rojas, J. Magn. Magn. Mats. \textbf{465}, 323 (2018).
\bibitem{pseudo}S. M. de Souza and O. Rojas, Sol. Stat. Comm. \textbf{269},
131 (2017)
\bibitem{torrico}J. Torrico, M. Rojas, S. M. de Souza, O. Rojas and
N. S. Ananikyan, Eur. Phys. Lett. \textbf{108}, 50007 (2014).
\bibitem{torrico2}J. Torrico, M. Rojas, S. M. de Souza and O. Rojas,
Phys. Lett. A \textbf{380}, 3655 (2016).
\bibitem{Galisova}L. Galisova and J. Strecka, Phys. Rev. E \textbf{91},
0222134 (2015)
\bibitem{on-strk}O. Rojas, J. Stre\v{c}ka and S.M. de Souza, Sol.
Stat. Comm. \textbf{246}, 68 (2016).
\bibitem{strk-cav}J. Strecka, R. C. Alecio, M. Lyra and O. Rojas,
J. Magn. Magn. Mats. \textbf{409}, 124 (2016).
\bibitem{Torr-prb-16}J. Torrico, M. Rojas, M. S. S. Pereira, J. Strecka,
M. L. Lyra, Phys. Rev. B \textbf{93}, 014428 (2016).
\bibitem{Torr-jmmm-18}J. Torrico, J. Strecka, M. Hagiwara, O. Rojas,
S. M. de Souza, Y. Han, Z. Honda, M. L. Lyra, J. Magn. Magn. Mats.
\textbf{460}, 368 (2018).
\bibitem{karlova}K. Karlova, J. Strecka, T. Madaras, Physica B \textbf{488},
49 (2016)
\bibitem{mambrini}M. Mambrini, J. Trébosc and F. Mila, Phys. Rev.
B \textbf{59} 13806 (1999).
\bibitem{roj-alc}O. Rojas, F. C. Alcaraz, Phys. Rev. B \textbf{67},
174401 (2003)
\bibitem{galisova17}L. Galisova, Phys. Rev. E \textbf{96}, 052110
(2017).
\bibitem{vadim-1}V. Ohanyan, Cond. Matt. Phys., \textbf{12}, 343
(2009)
\bibitem{Vadim-2}D. Antonosyan, S. Bellucci, V. Ohanyan, Phys. Rev.
B \textbf{79}, 014432 (2009)
\end{thebibliography}
\end{document}