This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Orbital magnetic susceptibility of zigzag carbon nanobelts: a tight-binding study

Norio Inui Graduate School of Engineering, University of Hyogo, Himeji, Hyogo 671-2280, Japan
Abstract

The magnetic properties of a circular graphene nanoribbon (carbon belt) in a magnetic field parallel to its central axis is studied using a tight-binding model. Orbital magnetic susceptibility is calculated using an analytical expression of the energy eigenvalues as a function of the magnetic flux density for any size, and its temperature dependence is considered. In the absence of electron hopping parallel to the magnetic field, the orbital magnetic susceptibility diverges at absolute zero if the chemical potential is zero and the number of atoms is a multiple of four. As the temperature increases, the magnitude of susceptibility decreases according to the power law, whose exponent depends on the size. In the presence of electron hopping parallel to the magnetic field, the divergence of the susceptibility near absolute zero disappears, and the sign changes with the transfer integral parallel to the magnetic field and the temperature.

I Introduction

Low-dimensional materials with honeycomb structures such as C60 Elser1987 ; Coffey1992 , carbon nanotube Ajiki1993 ; Lu1995 ; Chen2005 , and graphene Sepioni2010 ; Ortega2013 , have attracted attention. Recently, a circular graphene nanoribbon, called a carbon nanobelt Povie2017 , has been added to allotropes of carbon. As defined by Itami et al. Segawa2016 ; Itami2023 , a carbon nanobelt is a ring-shaped segment of a carbon nanotube, and the cleavage of at least two C-C bonds is necessary to open it. The first carbon nanobelt is a segment of (6,6) carbon nanotubes and consists of 12 benzene rings. Longer and complex carbon nanobelts can be achieved using current synthesis techniques Li2021 ; Wang2023 ; Wang2024 . Although many physical properties of carbon nanobelts remain unknown, researchers have reported high electric conductance between the tip of a scanning tunneling microscope and carbon nanobelts; this excellent property is expected to benefit electric applications Li2022 .

A carbon nanotube can be regarded as a graphene sheet rolled into a tube Li2013 . Similarly, a carbon nanobelt can be regarded as a graphene nanoribbon that can be seamlessly bent into cylindrical shapes. Graphene exhibits unique magnetic properties not observed in other materials Goerbig2011 . For example, divergent orbital diamagnetism at the Dirac point has been theoretically predicted and was recently detected Bustamante2021 . The magnetic properties of graphene ribbons are primarily determined by two factors: orbital magnetization and spin angular momentum Sepioni2010 . This diamagnetic property originates from a previous contribution. Unlike the spin, the orbital magnetization of a graphene sheet is generated by a global electric current, which depends on the shape of the sheet. In this study, we focus on the magnetization of carbon nanobelts caused by the orbital current circulating around them.

Extensive theoretical studies on the orbital magnetization of graphene disks have revealed that it strongly depends on their size and type of edges, temperature, and defects Ezawa2007 ; Liu2008 ; Ominato2013 ; Deyo2021 . However, little is known about the magnetic properties of carbon nanobelts. Thus, we introduce a tight-binding model for carbon nanobelts and investigate the dependence of the magnetic susceptibility on their size and temperature. The advantage of using a tight-binding model over other theoretical methods, such as density functional theory Shi2020 ; Wu2013 ; Negrin2023 , is that the energy eigenvalue can often be expressed in analytical forms for arbitrary sizes. The magnetic susceptibility is obtained by calculating the second derivative of the free energy Ominato2013 . Thus, the dependence of the magnetic susceptibility on the size can be precisely discussed.

The remainder of this paper is organized as follows: In Section 2, we introduce a tight-binding model for a carbon nanobelt based on cyclacene, which is the first proposed carbon nanobelt Turker2004 . Although cyclacene has not yet been synthesized, its structure is the simplest among carbon nanobelts Gleiter2009 . Cyclacene can be considered as two interacting one-dimensional atomic arrays, hereafter referred to as rings. Therefore, we address the energy eigenvalue problem of a ring and explain its size dependence. In Section 3, the dependence of the magnetic susceptibility of the ring on its size and temperature is discussed. In Section 4, the magnetic properties of the carbon nanobelts, hereafter referred to as belts, are investigated. In particular, the difference in the magnetic susceptibility between the rings and the belts is discussed. In Section 5, the relations between the change in the energy level by applying a magnetic field and the magnetic properties is summarized.

II Tight-binding model of a ring

The structure of the belt considered in this study consists of upper and lower rings, as shown in Fig. 1(a), where the solid circles indicate the atoms. A uniform magnetic field is applied parallel to the central axis (zz-axis) of the belts. Figure 1(b) shows the position of atoms in the upper ring projected onto the xyxy-plane. We assume that atoms in a ring exist on a circle. The electrons can hop along the circumference and is characterized by the transfer integral γ\gamma. Additionally, electrons can hop between rings, whose transfer integral γ\gamma^{\prime} may be different from γ\gamma. Because the magnetic response to the these parameter dependencies cannot be easily considered simultaneously, we prohibited hopping between rings in this section and considered the magnetic response of a single ring using a tight-binding model.

Refer to caption
Figure 1: (a) Configuration of a belt fused 12 benzene rings. Magnetic field is applied parallel to zz-axis. (b) Labeling and positions of upper sites indicated by blue circles, where θ\theta = 2π/N2\pi/N.

Let the number of sites (atoms) in the ring be NN. The radius of a ring, whose nearest distance between sites in the xyxy-plane is aa is expressed as

RN\displaystyle R_{N} =\displaystyle= a2sin(πN).\displaystyle\frac{a}{2\sin(\frac{\pi}{N})}. (1)

Each site in the ring is labeled with an integer jj \in {1,2,,N}\{1,2,\ldots,N\} as shown in Fig. 1(b), and the position of the jj-th site on the xyxy-plane rj\vec{r}_{j} = (xj,yj)(x_{j},y_{j}) is defined by

{xj = RNcos(2π(j1)N),yj = RNsin(2π(j1)N).\displaystyle\left\{\begin{tabular}[]{l}$x_{j}$ = $R_{N}\cos\left(\frac{2\pi(j-1)}{N}\right)$,\\ $y_{j}$ = $R_{N}\sin\left(\frac{2\pi(j-1)}{N}\right)$.\end{tabular}\right. (4)

The Hamiltonian of the tight-binding model Ezawa2007 , in which the contribution of spin is neglected, is defined by

HRing\displaystyle H_{\mbox{\tiny Ring}} =\displaystyle= γj=1N(eiϕjcj+1cj+eiϕjcjcj+1).\displaystyle-\gamma\sum_{j=1}^{N}(e^{i\phi_{j}}c_{j+1}^{{\dagger}}c_{j}+e^{-i\phi_{j}}c_{j}^{{\dagger}}c_{j+1}). (5)

where the operator cjc_{j}^{{\dagger}} creates an electron at the jj-th site, and cjc_{j} annihilates an electron at the jj-th site. The periodic condition requires cN+1c_{N+1}^{{\dagger}} = c1c_{1}^{{\dagger}} and cN+1c_{N+1} = c1c_{1}. The effect of the magnetic field in the Hamiltonian is expressed through the Peierls phase ϕj\phi_{j} between the jj-th site and the (j+1)(j+1)-th site defined by

ϕj\displaystyle\phi_{j} =\displaystyle= erjrj+1𝑑rA,\displaystyle-\frac{e}{\hbar}\int_{\vec{r}_{j}}^{\vec{r}_{j+1}}d\vec{r}\cdot\vec{A}, (6)

where A\vec{A} is the vector potential and is related to the applied magnetic flux density B\vec{B} in the form of B\vec{B} = ×A\nabla\times\vec{A}. Using the Landau gauge, the vector potential can be expressed as

A\displaystyle\vec{A} =\displaystyle= (0,Bx,0).\displaystyle(0,Bx,0). (7)

In the continuous limit of aa \rightarrow 0, the Hamiltonian (58) can be approximately regarded as (p^+eA)2/2m(\hat{p}+e\vec{A})^{2}/2m, where p^\hat{p} and mm are the momentum operator and effective mass related to γ\gamma, respectively Boykin2001 ; Inui2023 . Combining Eq. (7) with Eq. (6) leads to

ϕj\displaystyle\phi_{j} =\displaystyle= eB2(xj+1+xj)(yjyj+1),\displaystyle\frac{eB}{2\hbar}(x_{j+1}+x_{j})(y_{j}-y_{j+1}), (8)
=\displaystyle= αsin(2πN)cos2((2j1)πN),\displaystyle-\alpha\sin\left(\frac{2\pi}{N}\right)\cos^{2}\left(\frac{(2j-1)\pi}{N}\right), (9)

where α\alpha is a dimensionless parameter defined by

α\displaystyle\alpha =\displaystyle= eRN2B.\displaystyle\frac{eR_{N}^{2}B}{\hbar}. (10)

If the magnetic flux density through the ring is defined by ϕ\phi \equiv πRN2B\pi R_{N}^{2}B, then the parameter α\alpha is expressed as ϕ/ϕ0\phi/\phi_{0} where ϕ0\phi_{0} \equiv hh/(2ee) is magnetic flux quantum.

The energy eigenvalue ϵn,N(α)\epsilon_{n,N}(\alpha) of the Hamiltonian HRingH_{\mbox{\tiny Ring}} with quantum number nn = 1, 2, \ldots NN is given by

ϵn,N(α)\displaystyle\epsilon_{n,N}(\alpha) =\displaystyle= 2γcos[2nπN+α2sin(2πN)].\displaystyle-2\gamma\cos\left[\frac{2n\pi}{N}+\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\right]. (11)

The jj-th component of the eigenstate ψn\psi_{n} corresponding to ϵn,N(α)\epsilon_{n,N}(\alpha) is given by

ψn,j\displaystyle\psi_{n,j} =\displaystyle= 1Neiθn,j,\displaystyle\frac{1}{\sqrt{N}}e^{i\theta_{n,j}}, (12)

where

θn,j\displaystyle\theta_{n,j} =\displaystyle= 2njπNα4[sin(4πN)+sin(4(j1)πN)].\displaystyle\frac{2nj\pi}{N}-\frac{\alpha}{4}\left[\sin\left(\frac{4\pi}{N}\right)+\sin\left(\frac{4(j-1)\pi}{N}\right)\right]. (13)

The combination of the eigenvalue and eigenstate shown above satisfies the Schrödinger equation HRingψnH_{\mbox{\tiny Ring}}\psi_{n} = ϵn,Nψn\epsilon_{n,N}\psi_{n}. Applying the Hamiltonian on the eigenstate, we have

[HRingψn]j\displaystyle[H_{\mbox{\tiny Ring}}\psi_{n}]_{j} =\displaystyle= γ(eiϕj1ψn,j1+eiϕjψn,j+1),\displaystyle-\gamma(e^{i\phi_{j-1}}\psi_{n,j-1}+e^{-i\phi_{j}}\psi_{n,j+1}), (14)

where [.]j[.]_{j} denotes the jj-th component of HRingψnH_{\mbox{\tiny Ring}}\psi_{n}. The eigenstates at the sites j+1j+1 and j1j-1 can be expressed as

ψn,j+1\displaystyle\psi_{n,j+1} =\displaystyle= eiδn,jψn,j,\displaystyle e^{i\delta_{n,j}}\psi_{n,j}, (15)
ψn,j1\displaystyle\psi_{n,j-1} =\displaystyle= eiδn,j1ψn,j,\displaystyle e^{-i\delta_{n,j-1}}\psi_{n,j}, (16)

where

δn,j=2nπNα2sin(2πN)cos(2(2j1)πN).\displaystyle\delta_{n,j}=\frac{2n\pi}{N}-\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\cos\left(\frac{2(2j-1)\pi}{N}\right). (17)

Combining Eqs. (14), (15), and (16), we obtain

[HRingψn]j\displaystyle[H_{\mbox{\tiny Ring}}\psi_{n}]_{j} =\displaystyle= γ(ei(2nπN+α2sin(2πN))ψn,j+ei(2nπN+α2sin(2πN))ψn,j),\displaystyle-\gamma\left(e^{-i\left(\frac{2n\pi}{N}+\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\right)}\psi_{n,j}+e^{i\left(\frac{2n\pi}{N}+\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\right)}\psi_{n,j}\right), (18)
=\displaystyle= 2γcos[2nπN+α2sin(2πN)]ψn,j,\displaystyle-2\gamma\cos\left[\frac{2n\pi}{N}+\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\right]\psi_{n,j}, (19)
=\displaystyle= ϵn,N(α)[ψn]j.\displaystyle\epsilon_{n,N}(\alpha)[\psi_{n}]_{j}. (20)

We define the energy eigenvalue normalized by γ\gamma as λn,N(α)\lambda_{n,N}(\alpha) (\equiv ϵn,N(α)/γ\epsilon_{n,N}(\alpha)/\gamma). Figure 2 shows λn,N(α)\lambda_{n,N}(\alpha) for NN = 4, 5, and 6 as a function of α\alpha. In the absence of the magnetic field, i.e., α\alpha = 0, each energy level has a degeneracy of two except for nn = 1 (grand state) and nn = NN. Note that λ2,4(α)\lambda_{2,4}(\alpha) and λ3,4(α)\lambda_{3,4}(\alpha) are zero at α\alpha = 0. This holds for any NN if NN mod 4 = 0, i.e., if NN is a multiple of 4.

Refer to caption
Figure 2: Energy eigenvalues of rings normalized by the transfer integral γ\gamma for NN = 4, 5, and 6.

The number of six-membered rings in carbon nanobelts reported in Ref. Povie2017 is 12, which is greater than that in the site considered above. Although we showed the eigenvalues of the rings with a small NN for visibility, the exact eigenvalues can be obtained for an arbitrary NN and magnetic flux BB using Eq. (11).

III Magnetic susceptibility of a ring

III.1 Relation between magnetic susceptibility and energy eigenvalues

Magnetic susceptibility is an important parameter that determines the magnetic properties of a ring and depends on the temperature TT. The temperature dependence is calculated from free energy as follows Ominato2013 :

FN(α,T)\displaystyle F_{N}(\alpha,T) =\displaystyle= 2kBTnln[1+eμϵn,N(α)kBT],\displaystyle-2k_{\mbox{\tiny B}}T\sum_{n}\ln[1+e^{\frac{\mu-\epsilon_{n,N}(\alpha)}{k_{\mbox{\tiny B}}T}}], (21)

where the factor “2” originates from the degree of spin polarization, and kBk_{\mbox{\tiny B}} and μ\mu are the Boltzmann constant and chemical potential, respectively Deyo2021 ; Maebuchi2023 . In the following, we consider only the case of μ\mu = 0. The magnetic susceptibility per unit length Wakabayashi1999 is given by

χ\displaystyle\chi =\displaystyle= 12πRN(2FB2)T|B=0.\displaystyle\left.-\frac{1}{2\pi R_{N}}\left(\frac{\partial^{2}F}{\partial B^{2}}\right)_{T}\right|_{B=0}. (22)

Figure 3 shows the magnetic susceptibility normalized by χ0\chi_{0} \equiv γe2a3/8π2\gamma e^{2}a^{3}/8\pi\hbar^{2} as a function of the normalized temperature defined by kBT/γk_{\mbox{\tiny B}}T/\gamma. The magnetic susceptibility for NN = 4 diverges, as TT approaches zero.

Refer to caption
Figure 3: Temperature dependence of the magnetic susceptibilities of rings normalized with χ0\chi_{0} for NN = 4, 5, and 6.

Let the inverse temperature normalized with kB/γk_{\mbox{\tiny B}}/\gamma be β\beta \equiv γ/kBT\gamma/k_{\mbox{\tiny B}}T. Substituting Eq. (21) into Eq. (22), the susceptibility is expressed as

χ\displaystyle\chi =\displaystyle= χ04sin3(πN)n=1N[4λn,N′′(0)G1(βλn,N(0))\displaystyle-\frac{\chi_{0}}{4\sin^{3}\left(\frac{\pi}{N}\right)}\sum_{n=1}^{N}\left[4\lambda_{n,N}^{\prime\prime}(0)G_{1}\left(-\beta\lambda_{n,N}(0)\right)\right. (23)
β(λn,N(0))2G2(β2λn,N(0))],\displaystyle\left.-\beta(\lambda_{n,N}^{\prime}(0))^{2}G_{2}\left(-\frac{\beta}{2}\lambda_{n,N}(0)\right)\right],

where

G1(x)\displaystyle G_{1}(x) =\displaystyle= 11+ex,\displaystyle\frac{1}{1+e^{-x}}, (24)
G2(x)\displaystyle G_{2}(x) =\displaystyle= sech2(x)4(ex+ex)2.\displaystyle\mbox{sech}^{2}(x)\equiv\frac{4}{(e^{x}+e^{-x})^{2}}. (25)

Equation (23) indicates that the susceptibility is determined from the energy eigenvalues and their derivative up to the second order, and this is necessary for considering the relation between the susceptibility and the energy eigenvalues. Let ξn\xi_{n} be 2πn/N2\pi n/N. Using the expression for eigenvalues Eq. (11), we have

χ\displaystyle\chi =\displaystyle= χ0cos2(πN)sin(πN)n=1N(ζn(1)(β)+ζn(2)(β)),\displaystyle-\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)}\sum_{n=1}^{N}(\zeta_{n}^{(1)}(\beta)+\zeta_{n}^{(2)}(\beta)), (26)

where

ζn(1)(β)\displaystyle\zeta_{n}^{(1)}(\beta) =\displaystyle= 2cos(ξn)G1[2βcos(ξn)],\displaystyle 2\cos\left(\xi_{n}\right)G_{1}\left[2\beta\cos\left(\xi_{n}\right)\right], (27)
ζn(2)(β)\displaystyle\zeta_{n}^{(2)}(\beta) =\displaystyle= βsin2(ξn)G2[βcos(ξn)].\displaystyle-\beta\sin^{2}\left(\xi_{n}\right)G_{2}\left[\beta\cos\left(\xi_{n}\right)\right]. (28)

III.2 Magnetic susceptibility at absolute zero

To consider the magnetic susceptibility at low temperatures, two contributions ζn(1)\zeta_{n}^{(1)} and ζn(2)\zeta_{n}^{(2)} to susceptibility should be examined separately. The former is related to the second derivative at α\alpha = 0, and the latter is related to the first derivative at α\alpha = 0. In the limit of TT \rightarrow 0 (i.e. β\beta \rightarrow \infty), ζn(1)(α,T)\zeta_{n}^{(1)}(\alpha,T) converges as follows:

limβζn(1)\displaystyle\lim_{\beta\rightarrow\infty}\zeta_{n}^{(1)} =\displaystyle= {2cos(ξn),cos(ξn)  0,0,cos(ξn) < 0.\displaystyle\left\{\begin{tabular}[]{ll}$2\cos\left(\xi_{n}\right)$,&$\cos(\xi_{n})$ $\geq$ 0,\\ 0,&$\cos(\xi_{n})$ $<$ 0.\end{tabular}\right. (31)

Accordingly, the sign of cos(ξn)\cos(\xi_{n}) is important. The conditions of cos(ξn)\cos(\xi_{n}) = 0 are that NN mod 4 = 0 and nn = N/4N/4, 3N/43N/4. This is the same as the condition that the energy eigenvalue is zero in the absence of the magnetic field. The range where cos(ξn)\cos(\xi_{n}) is positive is given by

J+(N)\displaystyle J_{+}(N) =\displaystyle= {1nN(Nmod 4)4}{3N+(Nmod 4)4nN}.\displaystyle\left\{1\leq n\leq\frac{N-(N\,\mbox{mod}\,4)}{4}\right\}\bigcap\left\{\frac{3N+(N\,\mbox{mod}\,4)}{4}\leq n\leq N\right\}.

The summation of cos(ξn)\cos(\xi_{n}) over nn satisfying cos(ξn)\cos(\xi_{n}) >> 0 is written as

nJ+(N)cos(ξn)\displaystyle\sum_{n\in J_{+}(N)}\cos\left(\xi_{n}\right) =\displaystyle= {1tan(πN),N mod 4 = 0,12sin(π2N),N mod 4 = 1, 3,1sin(πN),N mod 4 = 2.\displaystyle\left\{\begin{tabular}[]{ll}$\frac{1}{\tan\left(\frac{\pi}{N}\right)}$,&$N$ mod 4 = 0,\\ $\frac{1}{2\sin\left(\frac{\pi}{2N}\right)}$,&$N$ mod 4 = 1, 3,\\ $\frac{1}{\sin\left(\frac{\pi}{N}\right)}$,&$N$ mod 4 = 2.\\ \end{tabular}\right. (36)

We now consider the second contribution. If cos(ξn)\cos(\xi_{n}) is not zero, βcos(ξn)\beta\cos(\xi_{n}) diverges in the limit of β\beta\rightarrow\infty. Thus, G2(βcos(ξn))G_{2}(\beta\cos(\xi_{n})) converges to zero at absolute zero. Namely, the second contribution is neglectable near TT = 0 if cos(ξn)\cos(\xi_{n}) is not zero. If cos(ξn)\cos(\xi_{n}) is zero, ζn(2)\zeta_{n}^{(2)} is expressed as β-\beta, as β\beta approaches \infty. Therefore, if NN is a multiple of 4, then the contribution of ζn(2)\zeta_{n}^{(2)} dominates the temperature dependence near absolute zero. Because the values of cos(ξn)\cos(\xi_{n}) is zero at nn = N/4N/4 and 3N/43N/4, the susceptibility of carbon nanobelts near TT = 0 is expressed as

χ\displaystyle\chi \displaystyle\approx χ0cos2(πN)sin(πN)(ζN4(2)(β)+ζ3N4(2)(β)),Nmod 4=0,\displaystyle-\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)}\left(\zeta_{\frac{N}{4}}^{(2)}(\beta)+\zeta_{\frac{3N}{4}}^{(2)}(\beta)\right),\hskip 8.53581ptN\,\mbox{mod}\,4=0, (37)
=\displaystyle= 2χ0cos2(πN)sin(πN)γkBT,Nmod 4=0.\displaystyle 2\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)}\frac{\gamma}{k_{\mbox{\tiny B}}T},\hskip 8.53581ptN\,\mbox{mod}\,4=0. (38)

To summarize, if NN mod 4 = 0, the susceptibility diverges in proportion to T1T^{-1} in the limit of TT \rightarrow 0. If NN mod 4 \neq 0, the susceptibility at TT = 0 is given by

χT=0\displaystyle\chi_{T=0} =\displaystyle= {χ0cos2(πN)sin(πN)sin(π2N),N mod 4 = 1, 3,χ02cos2(πN)sin2(πN),N mod 4 = 2.\displaystyle\left\{\begin{tabular}[]{ll}$-\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)\sin\left(\frac{\pi}{2N}\right)}$,&$N$ mod 4 = 1, 3,\\ $-\chi_{0}\frac{2\cos^{2}\left(\frac{\pi}{N}\right)}{\sin^{2}\left(\frac{\pi}{N}\right)}$,&$N$ mod 4 = 2.\\ \end{tabular}\right. (41)

This asymptotic behavior is valid only when the chemical potential is zero. More generally, μϵn,N(0)\mu-\epsilon_{n,N}(0) determines the difference between divergence and convergence of χ\chi near TT = 0. For example, the condition that divergence occurs in the limit of TT \rightarrow 0 is replaced with cos(ξn)\cos(\xi_{n}) = μ\mu/γ\gamma.

III.3 Asymptotic behavior of magnetic susceptibility at high temperatures

The magnetic susceptibility converges to zero as the temperature increases, as shown in Fig. 3. We consider the asymptotic behavior of the magnetic susceptibility at high temperatures. In the limit of TT \rightarrow \infty, the parameter β\beta converges to zero. Thus, we analyze the Taylor series of ζn\zeta_{n} \equiv ζn(1)\zeta_{n}^{(1)} + ζn(2)\zeta_{n}^{(2)} near β\beta = 0. The expressions for the Taylor series of G1(x)G_{1}(x) and G2(x)G_{2}(x) are as follows:

G1(x)\displaystyle G_{1}(x) =\displaystyle= 12+k=0a2k+1x2k+1,\displaystyle\frac{1}{2}+\sum_{k=0}^{\infty}a_{2k+1}x^{2k+1}, (42)
G2(x)\displaystyle G_{2}(x) =\displaystyle= k=0b2kx2k,\displaystyle\sum_{k=0}^{\infty}b_{2k}x^{2k}, (43)

where

ak\displaystyle a_{k} =\displaystyle= 1k!j=1ki=1j(1)i2j+1(ji)ik,\displaystyle-\frac{1}{k!}\sum_{j=1}^{k}\sum_{i=1}^{j}\frac{(-1)^{i}}{2^{j+1}}\left(\begin{tabular*}{7.11317pt}[]{c}$j$\\ $i$\end{tabular*}\right)i^{k}, (46)
b2k\displaystyle b_{2k} =\displaystyle= (1+2k)4k+1a2k+1.\displaystyle(1+2k)4^{k+1}a_{2k+1}. (47)

The details of the calculations for this series expansion is described in Appendix. Substituting βcos(ξn)\beta\cos(\xi_{n}) into xx in the Taylor series in Eqs. (42) and (43) yields

ζn\displaystyle\zeta_{n} =\displaystyle= cos(ξn)+k=0cn,kβ2k+1,\displaystyle\cos\left(\xi_{n}\right)+\sum_{k=0}^{\infty}c_{n,k}\beta^{2k+1}, (48)

where

cn,k\displaystyle c_{n,k} =\displaystyle= 2a2k+122k+1cos2k+2(ξn)+b2kcos2k+2(ξn)b2kcos2k(ξn).\displaystyle 2a_{2k+1}2^{2k+1}\cos^{2k+2}\left(\xi_{n}\right)+b_{2k}\cos^{2k+2}\left(\xi_{n}\right)-b_{2k}\cos^{2k}\left(\xi_{n}\right). (49)

Let ckc_{k} be n=1Ncn,k\sum_{n=1}^{N}c_{n,k}. The summation of ζn\zeta_{n} is written as

n=1Nζn\displaystyle\sum_{n=1}^{N}\zeta_{n} =\displaystyle= {cN21βN1+𝒪(βN),N = even,cN1β2N1+𝒪(β2N),N = odd.\displaystyle\left\{\begin{tabular}[]{ll}$c_{\frac{N}{2}-1}\beta^{N-1}+\cal O\mit(\beta^{N})$,&$N$ $=$ \mbox{even,}\\ $c_{N-1}\beta^{2N-1}+\cal O\mit(\beta^{2N})$,&$N$ $=$ \mbox{odd.}\end{tabular}\right. (52)

The details of the calculation of the summation are provided in the Appendix. Using cN/21c_{N/2-1} = 2N2aN12N^{2}a_{N-1} and cN1c_{N-1} = 4N2a2N14N^{2}a_{2N-1}, the asymptotic behavior of the magnetic susceptibility at high temperatures is expressed as

χ\displaystyle\chi \displaystyle\approx 2χ0cos2(πN)sin(πN)N2aN1βN1+𝒪(βN),N=even,\displaystyle-2\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)}N^{2}a_{N-1}\beta^{N-1}+\cal O\mit(\beta^{N}),\hskip 15.649pt\,N=\mbox{even,} (53)
χ\displaystyle\chi \displaystyle\approx 4χ0cos2(πN)sin(πN)N2a2N1β2N1+𝒪(β2N),N=odd.\displaystyle-4\chi_{0}\frac{\cos^{2}\left(\frac{\pi}{N}\right)}{\sin\left(\frac{\pi}{N}\right)}N^{2}a_{2N-1}\beta^{2N-1}+\cal O\mit(\beta^{2N}),\hskip 8.53581pt\,N=\mbox{odd}. (54)

The results show that the absolute value of the magnetic susceptibility decreases with the power laws TN+1T^{-N+1} and T2N+1T^{-2N+1} for NN = even and odd, respectively. Figure 4 shows the magnetic susceptibility for NN = 4, 5, and 6 in the log-log scale. The circles shows the exact values, and the dashed lines are straight lines determined from Eqs. (53) and (54). The susceptibility decreases more rapidly with size particularly for odd NN.

Refer to caption
Figure 4: Temperature dependence of the magnetic susceptibility normalized with χ0\chi_{0} at high temperatures in the double logarithm scale. Solid circles show the exact values, and the dashed lines indicate the asymptotes.

Because the free energy is represented as a function of the magnetic flux density, the dependence of physical quantities other than magnetic susceptibility on the magnetic field HH, can also be calculated. For example, the magnetocaloric effect Oliveira2010 of a ring under adiabatic change can be calculated in terms of the free energy as follows:

TH=TCH2FTB.\displaystyle\frac{\partial T}{\partial H}=\frac{T}{C_{H}}\frac{\partial^{2}F}{\partial T\partial B}. (55)

where CHC_{H} the heat capacity at constant magnetic field.

IV Magnetic susceptibility of a belt

Electron hopping between two rings is possible in a belt. We distinguish between the upper and lower rings with integers kk = 1 and 2, respectively. Each site in the belt is labeled by jj (11 \leq jj \leq NN) and kk. The total number of sites is 2NN. Unlike the rings in Section 2, NN must be even for the belt. The Hamiltonian of a belt is expressed as

HBelt\displaystyle H_{\mbox{\tiny Belt}} =\displaystyle= γk=12j=1N(eiϕjcj+1,kcj,k+eiϕjcj,kcj+1,k)\displaystyle-\gamma\sum_{k=1}^{2}\sum_{j=1}^{N}(e^{i\phi_{j}}c_{j+1,k}^{{\dagger}}c_{j,k}+e^{-i\phi_{j}}c_{j,k}^{{\dagger}}c_{j+1,k}) (58)
γj = 1j oddN(cj,2cj,1+cj,1cj,2),\displaystyle-\gamma^{\prime}\sum_{\begin{tabular}[]{l}\scriptsize{$j$ $=$ $1$}\\ \scriptsize{$j$ odd}\\ \end{tabular}}^{N}(c_{j,2}^{{\dagger}}c_{j,1}+c_{j,1}^{{\dagger}}c_{j,2}),

where γ-\gamma^{\prime} is the transfer integral between the odd sites of the upper and lower rings. We assume that the transition between even sites is prohibited.

The energy eigenvalues of HBeltH_{\mbox{\tiny Belt}} normalized by γ\gamma are given by

λn,l,N(α,t)\displaystyle\lambda_{n,l,N}(\alpha,t) =\displaystyle= 12(plt+qlt2+16cos2[2nπN+α2sin(2πN)]),\displaystyle\frac{1}{2}\left(-p_{l}t+q_{l}\sqrt{t^{2}+16\cos^{2}\left[\frac{2n\pi}{N}+\frac{\alpha}{2}\sin\left(\frac{2\pi}{N}\right)\right]}\right), (59)
n=1,2,,N/2,\displaystyle\hskip 170.71652ptn=1,2,\ldots,N/2,

where tt = γ/γ\gamma^{\prime}/\gamma, and {pl,ql}\{p_{l},q_{l}\} = {1,1}\{-1,-1\}, {1,1}\{-1,1\}, {1,1}\{1,-1\}, and {1,1}\{1,1\} for ll = 1, 2, 3, and 4, respectively where tt = γ/γ\gamma^{\prime}/\gamma, and plp_{l} =11 and 1-1 for ll = 1, and 2, respectively. The components of eigenstates labeled with (l,j,k)(l,j,k) and plp_{l}, ψn,l,j,k\psi_{n,l,j,k} are expressed as

ψn,l,j,1\displaystyle\psi_{n,l,j,1} =\displaystyle= Cn,lvn,l,j,\displaystyle C_{n,l}v_{n,l,j}, (60)
ψn,l,j,2\displaystyle\psi_{n,l,j,2} =\displaystyle= plCn,lvn,l,j+N,\displaystyle p_{l}C_{n,l}v_{n,l,j+N}, (61)

where Cn,lC_{n,l} is a normalized constant determined from the condition that l,j(|ψn,l,j,1|2+|ψn,l,j,2|2)\sum_{l,j}(|\psi_{n,l,j,1}|^{2}+|\psi_{n,l,j,2}|^{2}) = 1, and vn,l,jv_{n,l,j} is defined by

vn,l,j\displaystyle v_{n,l,j} =\displaystyle= {λn,l,N(α,t)eiθn,j,j=odd,λn,N(α)eiθn,j,j=even.\displaystyle\left\{\begin{tabular}[]{ll}$\lambda_{n,l,N}(\alpha,t)e^{i\theta_{n,j}}$,&$j\,=\mbox{odd}$,\\ $\lambda_{n,N}(\alpha)e^{i\theta_{n,j}}$,&$j\,=\mbox{even}.$\end{tabular}\right. (64)

In contrast to the eigenstate presented in Section 2, these expressions of the eigenstates do not form an orthogonal basis but they satisfy the Schrödinger equation as follows. If jj is even, the state after applying HBeltH_{\mbox{\tiny Belt}} to the eigenstate ψn\psi_{n} is written as

[HBeltψn]l,j,k\displaystyle[H_{\mbox{\tiny Belt}}\psi_{n}]_{l,j,k} =\displaystyle= γ(eiϕj1ψn,l,j1,k+eiϕjψn,l,j+1,k),\displaystyle-\gamma(e^{i\phi_{j-1}}\psi_{n,l,j-1,k}+e^{-i\phi_{j}}\psi_{n,l,j+1,k}), (65)
=\displaystyle= γ(Cn,leiϕj1λn,l,Neiθn,k+Cn,leiϕjλn,l,Neiθn,k),\displaystyle-\gamma(C_{n,l}e^{i\phi_{j-1}}\lambda_{n,l,N}e^{i\theta_{n,k}}+C_{n,l}e^{-i\phi_{j}}\lambda_{n,l,N}e^{i\theta_{n,k}}),
=\displaystyle= γλn,l,N(λn,Neiθn,k),\displaystyle-\gamma\lambda_{n,l,N}(\lambda_{n,N}e^{i\theta_{n,k}}),
=\displaystyle= γλn,l,Nψn,l,j,k.\displaystyle-\gamma\lambda_{n,l,N}\psi_{n,l,j,k}.

If jj is odd and kk = 1, we have

[HBeltψn]l,j,1\displaystyle[H_{\mbox{\tiny Belt}}\psi_{n}]_{l,j,1} =\displaystyle= γ(eiϕj1ψn,l,j1,1+eiϕjψn,l,j+1,1+tψn,l,j,2),\displaystyle-\gamma(e^{i\phi_{j-1}}\psi_{n,l,j-1,1}+e^{-i\phi_{j}}\psi_{n,l,j+1,1}+t\psi_{n,l,j,2}), (66)
=\displaystyle= γ(λn,N(λn,Neiθn,k)+pltλn,l,Neiθn,k),\displaystyle-\gamma(\lambda_{n,N}(\lambda_{n,N}e^{i\theta_{n,k}})+p_{l}t\lambda_{n,l,N}e^{i\theta_{n,k}}),
=\displaystyle= γλn,l,Nψn,l,j,k.\displaystyle-\gamma\lambda_{n,l,N}\psi_{n,l,j,k}.

Here, to obtain the last equation, we used the following equation:

λn,l,N2+pltλn,l,Nλn,N2=0.\displaystyle\lambda_{n,l,N}^{2}+p_{l}t\lambda_{n,l,N}-\lambda_{n,N}^{2}=0. (67)

Similarly, the Schrödinger equation is satisfied for the case where jj is odd and kk = 2.

Figure 5 shows the normalized energy eigenvalues of the belts for NN = 4 and 6 at tt = 0.2. For NN = 4, the energy eigenvalues of (n,l)(n,l) = (1,2)(1,2) and (1,4)(1,4) are zero at α\alpha = 0, and this degeneracy remains even when the transfer of electrons between rings occurs.

Refer to caption
Figure 5: Energy eigenvalues normalized of belts with γ\gamma for (a) NN = 4 and (b) NN = 6. The ratio of γ\gamma^{\prime} to γ\gamma is 0.2.

The first and second derivatives of the normalized energy eigenvalues are, respectively, given by

λn,l,N(α,t)α|α=0\displaystyle\left.\frac{\partial\lambda_{n,l,N}(\alpha,t)}{\partial\alpha}\right|_{\alpha=0} =\displaystyle= 2plqlsin(2πN)sin(4πnN)t2+8+8cos(4πnN),\displaystyle\frac{2p_{l}q_{l}\sin\left(\frac{2\pi}{N}\right)\sin\left(\frac{4\pi n}{N}\right)}{\sqrt{t^{2}+8+8\cos\left(\frac{4\pi n}{N}\right)}}, (68)
2λn,l,N(α,t)2α|α=0\displaystyle\left.\frac{\partial^{2}\lambda_{n,l,N}(\alpha,t)}{\partial^{2}\alpha}\right|_{\alpha=0} =\displaystyle= 2plqlsin2(2πN){6+(8+t2)cos(4πnN)+2cos(8πnN)}(t2+8+8cos(4πnN))32.\displaystyle\frac{2p_{l}q_{l}\sin^{2}\left(\frac{2\pi}{N}\right)\left\{6+(8+t^{2})\cos\left(\frac{4\pi n}{N}\right)+2\cos\left(\frac{8\pi n}{N}\right)\right\}}{(t^{2}+8+8\cos\left(\frac{4\pi n}{N}\right))^{\frac{3}{2}}}.

Figure 6 shows the temperature dependence of the magnetic susceptibility of the belts with tt = 1 for NN = 4, 6, and 10. In contrast to Fig. 3, the susceptibility at absolute zero for NN = 4 is finite. We recall that the second term in the left-hand side of Eq. (23) causes the divergence for rings. In the case of the belt with NN = 4 and tt = 0, the derivative of the energy eigenvalues with nn = 2 and 3 near α\alpha = 0 is not zero (see Fig. 2), whereas, if tt is not zero, the first derivative of the energy eigenvalues of the belt with NN = 4 for (n,l)(n,l) = (1,2)(1,2) and (1,4)(1,4) converges to zero (see Fig. 5), as α\alpha approaches zero. Thus, the second term vanishes in the limit of α\alpha \rightarrow 0, and the divergence does not occur. This finiteness of magnetic susceptibility for belts holds for any size, if NN is a multiple of four.

Refer to caption
Figure 6: Temperature dependence of the magnetic susceptibilities of the belts normalized with γ\gamma for NN = 4, 6, and 10.

Interestingly, the sign of susceptibility for NN = 10 changes from negative to positive as the temperature increases. Figure 7(a) shows the phase diagram of the sign in a plane of the transfer integral tt and the normalized temperature for NN = 10. The sign is always negative for small tt values, independent of temperature. Above a critical value of tt near 0.76, the sign changes from negative to positive as the temperature increases from absolute zero. Conversely, the sign of the belt with NN = 4 changes from positive to negative as the temperature increases from absolute zero as shown in Fig. 7(b).

Refer to caption
Figure 7: Phase diagram of the belts divided by the sign of the magnetic susceptibility for (a) NN = 10 and (b) NN = 4.

V Conclusion

We showed that the magnetic properties of carbon nanobelts strongly depend on the divisibility of the number of sites by four. If the number of sites in the ring is a multiple of four, the magnetic susceptibility diverges as the temperature approaches absolute zero. This singularity occurs when the following conditions are satisfied. First, the energy eigenvalues converges to zero as the magnetic field decreases to zero. Second, their first derivative with respect to the magnetic field at BB = 0 is finite. If the chemical potential μ\mu is not zero, the first condition should be written as the condition that the energy eigenvalues converge to the chemical potential as the magnetic field decreases. As we assume that μ\mu = 0, the divergence occurs for NN = 4. The number of distinct energy eigenvalues increases as the number of sites increases. Thus, the divergence can occur for various values of the chemical potential as NN increases. For example, if μ\mu = γ\gamma, the divergence occurs for NN = 6 (see Fig. 2(c)).

The energy eigenvalues of the belts change depending on the parameter tt, which is a ratio of the transfer integral along the circumference and width directions. For a belt consisting of two carbon rings with NN = 4, the derivative of energy eigenvalues at zero becomes zero by the presence of the transfer integral along the width direction. It follows that the divergence at absolute zero does not occurs. However, this does not imply that no divergence occur for tt \neq 0. The magnetic susceptibility of the belt with NN = 6 diverges if the parameter tt satisfying γλn,N(0,t)\gamma\lambda_{n,N}(0,t) = μ\mu exists.

This study focuses primarily on the interaction between electrons and an external magnetic field, and the many-body effect of electrons on the magnetic properties is considered indirectly via Pauli’s exclusion principle. To describe the electron systems in the carbon nanobelt more accurately, the interactions between electrons should be considered. The most important interaction is likely the Coulomb interaction between electrons, which may be considered using the Hubbard model, whose interaction is described by cjcjcjcjc_{j\uparrow}^{{\dagger}}c_{j\uparrow}c_{j\downarrow}^{{\dagger}}c_{j\downarrow}, where \uparrow and \downarrow denote spin up and spin down, respectively. For graphene nanoribbonsKumar2020; Lou2024, more realistic transport results have been obtained using the Hubbard model Hancock2010 .

Various carbon nanobelts and rings shapes have been proposed and synthesized. For example, Möbius carbon nanobelts, with a twisted structure are smoothly connected at the front and back Segawa2022 ; Heine2020 . Their magnetic properties may be significantly different from that the belts considered in this study.

Acknowledgments

This research was supported by the Ministry of Education, Culture, Sports, Science and Technology, through a Grant-in-Aid for Scientific Research(C), MEXT KAKENHI Grant Number 21K04895.

References

  • (1) Elser V and Haddon R C 1987 Magnetic behavior of icosahedral C60 Phys. Rev. A 36 4579. /10.1103/PhysRevA.36.4579
  • (2) Coffey D and Trugman S A 1992 Magnetic properties of undoped C60C_{60} Phys. Rev. Lett 69 176. /10.1103/PhysRevLett.69.176
  • (3) Ajiki H and Ando T 1993 Magnetic properties of carbon nanotubes J. Phys. Soc. Jpn. 62 2470. /10.1143/JPSJ.62.2470
  • (4) Lu J P 1995 Novel magnetic properties of carbon nanotubes ıPhys. Rev. Lett. 74 1123. /10.1103/PhysRevLett.74.1123
  • (5) Chen R B, Chang C P, Hwang J S, Chuu D S and Lin M F 2005 Magnetization of finite carbon nanotubes J. Phys. Soc. Jpn. 74. /10.1143/jpsj.74.1404
  • (6) Sepioni M, Nair R R, Rablen S, Narayanan J, Tuna F, Winpenny R, Geim A K and Grigorieva I V 2010 Limits on intrinsic magnetism in graphene Phys. Rev. Lett. 105 207205. /10.1103/PhysRevLett.105.207205
  • (7) Espinosa-Ortega T, Lukýanchuk I A and Rubo Y G 2013 Magnetic properties of graphene quantum dots Phys. Rev. B 87, 205434. /10.1103/PhysRevB.87.205434
  • (8) Povie G, Segawa Y, Nishihara T, Miyauchi Y and Itami K 2017 Synthesis of a carbon nanobelt Science 356 172. /10.1126/science.aam8158
  • (9) Segawa Y, Yagi A , Ito H and Itami K 2016 A theoretical study on the strain energy of carbon nanobelts Org. Lett. 18 1430. /10.1021/acs.orglett.6b00365
  • (10) Imoto D, Yagi A and Itami K 2023 Carbon nanobelts: Brief history and perspective Precis. Chem. 9 516. /10.1021/prechem.3c00083
  • (11) Li Y, Kono H, Maekawa T, Segawa Y, Yagi A and Itami K 2021 Chemical synthesis of carbon nanorings and nanobelts Acc. Mater. Res 2 681. /10.1021/accountsmr.1c00105
  • (12) Wang M-W, Fan W, Li X, Liu Y, Li Z, Jiang W, Wu J and Wang Z 2022 Molecular carbons: How far can we go? ACS Nano 17 20734. /10.1021/acsnano.3c07970
  • (13) Wang Y, Zhou Y and Du K 2024 Enumeration, nomenclature, and stability rules of carbon nanobelts J. Chem. Inf. Model. 64 1261. /10.1021/acs.jcim.3c02051
  • (14) Lin J, Wang S, Zhang F, Yang B, Du P, Chen C, Zang Y and Zhu D 2022 Highly efficient charge transport across carbon nanobelts Sci. Adv. 8 eade4692. 10.1126/sciadv.ade4692
  • (15) Li K, Eres G, Howe J, Chuang Y-J, Li X, Gu Z, Zhang L, Xie S and Pan Z 2013 Self-assembly of graphene on carbon nanotube surfaces Sci. Rep. 3 2353. /10.1038/srep02353
  • (16) Goerbig M O 2011 Electronic properties of graphene in a strong magnetic field Rev. Mod. Phys. 89 1193. /10.1103/RevModPhys.83.1193
  • (17) Bustamante J V, Wu N J, Fermon C, Pannetier-Lecoeur M, Wakamura T, Watanabe K, Taniguchi T, Pellegrin T, Bernard A, Daddinounou S, Bouchiat V, Guéron S, Ferrier M, Montambaux G and Bouchiat H 2021 Detection of graphene’s divergent orbital diamagnetism at the Dirac point Science 374 1399. /10.1126/science.abf9396
  • (18) Ezawa M 2007 Metallic graphene nanodisks: Electronic and magnetic properties Phys. Rev. B 76 245415. /10.1103/PhysRevB.76.245415
  • (19) Liu J, Ma Z, Wright A R and Zhang C 2008 Orbital magnetization of graphene and graphene nanoribbons J. Appl. Phys. 103 103711. /10.1063/1.2930875
  • (20) Ominato Y and Koshino M 2013 Orbital magnetism of graphene flakes Phys. Rev. B 87, 115433. /10.1103/PhysRevB.87.115433
  • (21) Deyo S and Hershfield S 2021 Magnetism in graphene flakes with edge disorder Phys. Rev. B 104 014404. /10.1103/PhysRevB.104.014404
  • (22) Shi T-H and Wang M-X 2020 Zigzag hydrocarbon belts CCS Chem 2 916. /10.31635/ccschem.020.202000287
  • (23) Wu C-S, Lee P-Y and Chai J-D 2016 Electronic properties of cyclacenes from TAO-DFT Scientific Reports 6 37249. /10.1038/srep37249
  • (24) Negrin-Yuvero H, Freixas V M, Ondarse-Alvarez D, Ledesma A E, Tretiak S and Fernandez-Alberti S 2023 Photoexcited energy relaxationin a zigzag carbon nanobelt J. Phys. Chem. C 127 5449. /10.1021/acs.jpcc.2c08676
  • (25) Lemi Türker and Selçuk Gümüş 2004 Cyclacenes J. Mol. Struct. 685 1. /0.1016/j.theochem.2004.04.021
  • (26) Gleiter R, Esser B and Kornmayer S C 2009 Cyclacenes: Hoop-shaped systems composed of conjugated rings Acc. Chem. Res. 42 1108. /10.1021/ar9000179
  • (27) Boykin T B 2001 Tight-binding-like expressions for the continuous-space electromagnetic coupling Hamiltonian Am. J. Phys. 69 793. /10.1119/1.1344169
  • (28) Inui N 2023 Continuous limit of a tight-binding model on a hexagonal disk in a magnetic field: size dependence on energy level Phs. Scr. 98 105254. 10.1088/1402-4896/acfa32
  • (29) Maebuchi K and Inui N 2023 Electric current distribution induced by applied magnetic field in a bent graphene nanoribbon cantilever J. Phys. D: Appl. Phys. 57 015002. /10.1088/1361-6463/acfd38
  • (30) Wakabayashi K, Fujita M, Ajiki H and Sigrist M 1999 Electronic and magnetic properties of nanographite ribbons Phys. Rev. B 59 8271. https://doi.org/10.1103/PhysRevB.59.8271
  • (31) de Oliveira N A, and von Ranke P J 2010 Theoretical aspects of the magnetocaloric effect Phys. Rep. 489 89. /10.1016/j.physrep.2009.12.006
  • (32) Hancock Y, Uppstu A, Saloriutta K, Harju A and Puska M J 2010 Generalized tight-binding transport model for graphene nanoribbon-based systems Phys. Rev. B 42 245402. /10.1103/PhysRevB.81.245402
  • (33) Segawa Y, Watanabe T, Yamanoue K, Kuwayama M, Watanabe K, Pirillo J, Hijikata Y and Itami K 2022 Synthesis of a Möbius carbon nanobelt Nature Synth 1 535. /10.1038/s44160-022-00075-8
  • (34) Hanson-Heine M W D and Hirst J D 2020 Möbius and Hückel Cyclacenes with Dewar and Ladenburg defects J. Phys. Chem. A 124 5408. 10.1021/acs.jpca.0c04137

*

Appendix A Derivations of the Taylor series

The coefficient of the Taylor series of G1G_{1} is expressed as

ak=dkG1(x)dxk|x=0.\displaystyle a_{k}=\left.\frac{d^{k}G_{1}(x)}{dx^{k}}\right|_{x=0}. (70)

The kk-th derivative of G1(x)G_{1}(x) is given by

dkG1(x)dxk\displaystyle\frac{d^{k}G_{1}(x)}{dx^{k}} =\displaystyle= j=1k(1)jTk,jejx(1+ex)j+1,\displaystyle\sum_{j=1}^{k}(-1)^{j}\frac{T_{k,j}e^{-jx}}{(1+e^{-x})^{j+1}}, (71)

where

Tk,j\displaystyle T_{k,j} =\displaystyle= i=0j(1)ji(ji)ik.\displaystyle\sum_{i=0}^{j}(-1)^{j-i}\binom{j}{i}i^{k}. (72)

Thus, aka_{k} is expressed as

ak\displaystyle a_{k} =\displaystyle= 1k!j=1ki=1j(1)i2j+1(ji)ik.\displaystyle-\frac{1}{k!}\sum_{j=1}^{k}\sum_{i=1}^{j}\frac{(-1)^{i}}{2^{j+1}}\binom{j}{i}i^{k}. (73)

Comparing G1(x)G_{1}(x) with G2(x)G_{2}(x), we obtain the following relation:

G2(x)\displaystyle G_{2}(x) =\displaystyle= 2dG1(2x)dx.\displaystyle 2\frac{dG_{1}(2x)}{dx}. (74)

Accordingly, the coefficient b2kb_{2k} can be expressed using a2k+1a_{2k+1} as Eq. (47) The coefficient ckc_{k} defined in Eq. (49) is written as

ck,N\displaystyle c_{k,N} =\displaystyle= 22k+2a2k+1dk+1,N+b2kdk+1,Nb2kdk,N,\displaystyle 2^{2k+2}a_{2k+1}d_{k+1,N}+b_{2k}d_{k+1,N}-b_{2k}d_{k,N}, (75)

where

dk,N\displaystyle d_{k,N} \displaystyle\equiv n=1Ncos2k(ξn).\displaystyle\sum_{n=1}^{N}\cos^{2k}(\xi_{n}). (76)

We define hk,Nh_{k,N} as

hk,N\displaystyle h_{k,N} \displaystyle\equiv N4k(2kk).\displaystyle\frac{N}{4^{k}}\binom{2k}{k}. (77)

If NN is even and kk << N/2N/2, dk,Nd_{k,N} is given by hk,Nh_{k,N}. If NN is even and k=N/2k=N/2, an additional term is required for hk,Nh_{k,N}, and dN/2,Nd_{N/2,N} is expressed as

dN2,N\displaystyle d_{\frac{N}{2},N} =\displaystyle= hN2,N+N2N1,N=even.\displaystyle h_{\frac{N}{2},N}+\frac{N}{2^{N-1}},\hskip 8.53581ptN=\mbox{even}. (78)

These change determine the exponent of decay for the magnetic susceptibility at high temperatures. Similarly, if NN is odd and kk << NN, dk,Nd_{k,N} is given by hk,Nh_{k,N}. If NN is even and k=Nk=N,

dN,N\displaystyle d_{N,N} =\displaystyle= hN,N+N22N1,N=odd.\displaystyle h_{N,N}+\frac{N}{2^{2N-1}},\hskip 8.53581ptN=\mbox{odd}. (79)

Let NN be even. Substituting Eqs. (73), (47), and (78) into the left-hand side of (75), we observe that ckc_{k} is zero for kk << N/21N/2-1. Thus, the first nonzero coefficient, cN/21c_{N/2-1} is given by

cN21,N\displaystyle c_{\frac{N}{2}-1,N} =\displaystyle= 2NNaN1{(1N)hN21,N+NhN2,N+N2N1},\displaystyle 2^{N}Na_{N-1}\left\{(1-N)h_{\frac{N}{2}-1,N}+Nh_{\frac{N}{2},N}+\frac{N}{2^{N-1}}\right\}, (80)
=\displaystyle= 2N2aN1.\displaystyle 2N^{2}a_{N-1}. (81)

Similarly, if NN is odd, ckc_{k} is zero for kk << N1N-1, and cN1c_{N-1} == 4N2a2N14N^{2}a_{2N-1}.