Hückel method

The method has several characteristics:

• It limits itself to conjugated molecules.
• Only π electron molecular orbitals are included because these determine much of the chemical and spectral properties of these molecules. The σ electrons are assumed to form the framework of the molecule and σ connectivity is used to determine whether two π orbitals interact. However, the orbitals formed by σ electrons are ignored and assumed not to interact with π electrons. This is referred to as σ-π separability. It is justified by the orthogonality of σ and π orbitals in planar molecules. For this reason, the Hückel method is limited to systems that are planar or nearly so.
• The method is based on applying the variational method to linear combination of atomic orbitals and making simplifying assumptions regarding the overlap, resonance and Coulomb integrals of these atomic orbitals. It does not attempt to solve the Schrödinger equation, and neither the functional form of the basis atomic orbitals nor details of the Hamiltonian are involved.
• For hydrocarbons, the method takes atomic connectivity as the only input; empirical parameters are only needed when heteroatoms are introduced.
• The method predicts how many energy levels exist for a given molecule, which levels are degenerate and it expresses the molecular orbital energies in terms of two parameters, called α, the energy of an electron in a 2p orbital, and β, the interaction energy between two 2p orbitals (the extent to which an electron is stabilized by allowing it to delocalize between two orbitals). The usual sign convention is to let both α and β be negative numbers. To understand and compare systems in a qualitative or even semi-quantitative sense, explicit numerical values for these parameters are typically not required.
• In addition the method also enables calculation of charge density for each atom in the π framework, the fractional bond order between any two atoms, and the overall molecular dipole moment.

The delocalization energy, π-bond orders, and π-electron population are chemically significant parameters that can be gleaned from the orbital energies and coefficients that are the direct outputs of Hückel theory.[18] These are quantities strictly derived from theory, as opposed to measurable physical properties, though they correlate with measurable qualitative and quantitative properties of the chemical species. Delocalization energy is defined as the difference in energy between that of the most stable localized Lewis structure and the energy of the molecule computed from Hückel theory orbital energies and occupancies. Since all energies are relative, we set ${\displaystyle \alpha =0}$ without loss of generality to simplify discussion. The energy of the localized structure is then set to be 2β for every two-electron localized π-bond. The Hückel energy of the molecule is ${\displaystyle \sum _{i}n_{i}E_{i}}$, where the sum is over all Hückel orbitals, ${\displaystyle n_{i}}$ is the occupancy of orbital i, set to be 2 for doubly-occupied orbitals, 1 for singly-occupied orbitals, and 0 for unoccupied orbitals, and ${\displaystyle E_{i}}$ is the energy of orbital i. Thus, the delocalization energy, conventionally a positive number, is defined as

${\displaystyle E_{\mathrm {deloc.} }={\Bigg |}{\Big (}\sum _{i}n_{i}E_{i}{\Big )}-2\beta \times (\#\ \mathrm {of} \ \mathrm {localized} \ \pi \ \mathrm {bonds} ){\Bigg |}}$.

In the case of benzene, the occupied orbitals have energies (again setting ${\displaystyle \alpha =0}$) 2β, β, and β. This gives the Hückel energy of benzene as ${\displaystyle 2\times 2\beta +2\times \beta +2\times \beta =8\beta }$. Each Kekulé structure of benzene has three double bonds, so the localized structure is assigned an energy of ${\displaystyle 2\beta \times 3=6\beta }$. The delocalization energy, measured in units of ${\displaystyle |\beta |}$, is then ${\displaystyle |8\beta -6\beta |=2|\beta |}$.

The π-bond orders derived from Hückel theory are defined using the orbital coefficients of the Hückel MOs. The π-bond order between atoms j and k is defined as

${\displaystyle \mathrm {BO} _{\pi }(j,k)=\sum _{i}n_{i}c_{j}^{(i)}c_{k}^{(i)}}$,

where ${\displaystyle n_{i}}$ is again the orbital occupancy of orbital i and ${\displaystyle c_{j}^{(i)}}$ and ${\displaystyle c_{k}^{(i)}}$ are the coefficients on atoms j and k, respectively, for orbital i. For benzene, the three occupied MOs, expressed as linear combinations of AOs ${\displaystyle \phi _{i}}$, are:[19]

${\displaystyle \Psi (A_{2u})={\frac {1}{\sqrt {6}}}(\phi _{1}+\phi _{2}+\phi _{3}+\phi _{4}+\phi _{5}+\phi _{6})}$, [${\displaystyle E=2\beta }$];

${\displaystyle \Psi (E_{1g}^{(x)})={\frac {1}{\sqrt {12}}}(2\phi _{1}+\phi _{2}-\phi _{3}-2\phi _{4}-\phi _{5}+\phi _{6})}$, [${\displaystyle E=\beta }$];

${\displaystyle \Psi (E_{1g}^{(y)})={\frac {1}{2}}(\phi _{2}+\phi _{3}-\phi _{5}-\phi _{6})}$, [${\displaystyle E=\beta }$].

Perhaps surprisingly, the π-bond order formula gives a bond order of

${\displaystyle 2{\Bigg (}{\frac {1}{\sqrt {6}}}{\Bigg )}{\Bigg (}{\frac {1}{\sqrt {6}}}{\Bigg )}+2{\Bigg (}{\frac {2}{\sqrt {12}}}{\Bigg )}{\Bigg (}{\frac {1}{\sqrt {12}}}{\Bigg )}+2(0){\Big (}{\frac {1}{2}}{\Big )}={\frac {2}{3}}}$

for the bond between carbons 1 and 2.[20] The resulting total (σ + π) bond order of ${\displaystyle 1{\frac {2}{3}}}$ is the same between any other pair of adjacent carbon atoms. This is more than the naive π-bond order of ${\displaystyle {\frac {1}{2}}}$ (for a total bond order of ${\displaystyle 1{\frac {1}{2}}}$) that one might guess when simply considering the Kekulé structures and the usual definition of bond order in valence bond theory. The Hückel definition of bond order attempts to quantify any additional stabilization that the system enjoys resulting from delocalization. In a sense, the Hückel bond order suggests that there are four π-bonds in benzene instead of the three that are implied by the Kekulé-type Lewis structures. The "extra" bond is attributed to the additional stabilization that results from the aromaticity of the benzene molecule. (This is only one of several definitions for non-integral bond orders, and other definitions will lead to different values that fall between 1 and 2.)

The π-electron population is calculated in a very similar way to the bond order using the orbital coefficients of the Hückel MOs. The π-electron population on atom j is defined as

${\displaystyle n_{\pi }(j)=\sum _{i}n_{i}[c_{j}^{(i)}]^{2}}$.

The associated Hückel Coulomb charge is defined as ${\displaystyle q_{j}=N_{\pi }(j)-n_{\pi }(j)}$, where ${\displaystyle N_{\pi }(j)}$ is the number of π-electrons contributed by a neutral, sp²-hybridized atom j (we always have ${\displaystyle N_{\pi }=1}$ for carbon).

For carbon 1 on benzene, this yields a π-electron population of

${\displaystyle 2{\Bigg (}{\frac {1}{\sqrt {6}}}{\Bigg )}{\Bigg (}{\frac {1}{\sqrt {6}}}{\Bigg )}+2{\Bigg (}{\frac {2}{\sqrt {12}}}{\Bigg )}{\Bigg (}{\frac {2}{\sqrt {12}}}{\Bigg )}+2(0)(0)=1}$.

Since each carbon atom contributes one π-electron to the molecule, this gives a Coulomb charge of 0 for carbon 1 (and all other carbon atoms), as expected.

In the cases of benzyl cation and benzyl anion shown above,

${\displaystyle q_{j}(\mathrm {CH} _{2}^{+})=N_{\pi }(j)-n_{\pi }(j)=1-0.43=+0.57}$ and ${\displaystyle q_{j}(\mathrm {CH} _{2}^{-})=N_{\pi }(j)-n_{\pi }(j)=1-1.57=-0.57}$,

${\displaystyle q_{j}(\mathrm {C} _{o,p}^{+})=N_{\pi }(j)-n_{\pi }(j)=1-0.86=+0.14}$ and ${\displaystyle q_{j}(\mathrm {C} _{o,p}^{-})=N_{\pi }(j)-n_{\pi }(j)=1-1.14=-0.14}$.

The mathematics of the Hückel method is based on the Ritz method. In short, given a basis set of n normalized atomic orbitals ${\displaystyle \{\phi _{i}\}_{i=1}^{n}}$, an ansatz molecular orbital ${\displaystyle \psi _{g}=N(c_{1}\phi _{1}+\cdots +c_{n}\phi _{n})}$ is written down, with normalization constant N and coefficients ${\displaystyle c_{i}}$ which are to be determined. In other words, we are assuming that the molecular orbital (MO) can be written as a linear combination of atomic orbitals, a conceptually intuitive and convenient approximation (the linear combination of atomic orbitals or LCAO approximation). The variational theorem states that given an eigenvalue problem ${\displaystyle {\hat {H}}|\psi ^{(i)}\rangle =E^{(i)}|\psi ^{(i)}\rangle }$ with smallest eigenvalue ${\displaystyle E^{(0)}}$ and corresponding wavefunction ${\displaystyle \psi ^{(0)}}$, any normalized trial wavefunction ${\displaystyle \psi _{g}}$ (i.e., ${\textstyle \langle \psi _{g}|\psi _{g}\rangle =\int _{\mathbb {R} ^{3}}\psi _{g}^{*}\,\psi _{g}\,dV=1}$ holds) will satisfy

${\displaystyle {\mathcal {E}}[\psi _{g}]=\langle \psi _{g}|{\hat {H}}|\psi _{g}\rangle =\int _{\mathbb {R} ^{3}}\psi _{g}^{*}\,{\hat {H}}\psi _{g}\,dV\geq E^{(0)}}$,

with equality holding if and only if ${\displaystyle \psi _{g}=\psi ^{(0)}}$. Thus, by minimizing ${\displaystyle E(c_{1},\ldots ,c_{n})={\mathcal {E}}[\psi _{g}]}$ with respect to coefficients ${\displaystyle c_{i}}$ for normalized trial wavefunctions ${\displaystyle \psi _{g}(c_{1},\ldots ,c_{n})}$, we obtain a closer approximation of the true ground-state wavefunction and its energy.

To start, we apply the normalization condition to the ansatz and expand to get an expression for N in terms of the ${\displaystyle c_{i}}$. Then, we substitute the ansatz into the expression for E and expand, yielding

${\displaystyle E(c_{1},\ldots ,c_{n})=N^{2}{\Big [}\sum _{i=1}^{n}c_{i}^{2}H_{ii}+\sum _{1\leq i\neq j\leq n}c_{i}c_{j}H_{ij}{\Big ]}}$, where ${\displaystyle N={\Big [}\sum _{i=1}^{n}c_{i}^{2}S_{ii}+\sum _{1\leq i\neq j\leq n}c_{i}c_{j}S_{ij}{\Big ]}^{-1/2}}$,

${\displaystyle S_{ij}=\langle \phi _{i}|\phi _{j}\rangle =\int _{\mathbb {R} ^{3}}\phi _{i}^{*}\,\phi _{j}\,dV}$, and ${\displaystyle H_{ij}=\langle \phi _{i}|{\hat {H}}|\phi _{j}\rangle =\int _{\mathbb {R} ^{3}}\phi _{i}^{*}\,{\hat {H}}\phi _{j}\,dV}$.

In the remainder of the derivation, we will assume that the atomic orbitals are real. (For the simple case of the Hückel theory, they will be the 2p_z orbitals on carbon.) Thus, ${\displaystyle S_{ij}=S_{ji}^{*}=S_{ji}}$, and because the Hamiltonian operator is hermitian, ${\displaystyle H_{ij}=H_{ji}^{*}=H_{ji}}$. Setting ${\displaystyle \partial {E}/\partial {c_{i}}=0}$ for ${\displaystyle i=1,\ldots ,n}$ to minimize E and collecting terms, we obtain a system of n simultaneous equations

${\displaystyle \sum _{j=1}^{n}c_{j}(H_{ij}-ES_{ij})=0\quad (i=1,\cdots ,n)}$.

When ${\displaystyle i\neq j}$, ${\displaystyle S_{ij}}$ and ${\displaystyle H_{ij}}$ are called the overlap and resonance (or exchange) integrals, respectively, while ${\displaystyle H_{ii}}$ is called the Coulomb integral, and ${\displaystyle S_{ii}=1}$ simply expresses the fact that the ${\displaystyle \phi _{i}}$ are normalized. The n × n matrices ${\displaystyle [S_{ij}]}$ and ${\displaystyle [H_{ij}]}$ are known as the overlap and Hamiltonian matrices, respectively.

By a well-known result from linear algebra, nontrivial solutions ${\displaystyle (c_{1},c_{2},\ldots ,c_{n})}$ to the above system of linear equations can only exist if the coefficient matrix ${\displaystyle [H_{ij}-ES_{ij}]}$ is singular. Hence, ${\displaystyle E}$ must have a value such that the determinant of the coefficient matrix vanishes:

${\displaystyle \mathrm {det} ([H_{ij}-ES_{ij}])=0}$. (*)

This determinant expression is known as the secular determinant and gives rise to a generalized eigenvalue problem. The variational theorem guarantees that the lowest value of ${\displaystyle E}$ that gives rise to a nontrivial (that is, not all zero) solution vector ${\displaystyle (c_{1},c_{2},\ldots ,c_{n})}$ represents the best LCAO approximation of the energy of the most stable π orbital; higher values of ${\displaystyle E}$ with nontrivial solution vectors represent reasonable estimates of the energies of the remaining π orbitals.

The Hückel method makes a few further simplifying assumptions concerning the values of the ${\displaystyle S_{ij}}$ and ${\displaystyle H_{ij}}$. In particular, it is first assumed that distinct ${\displaystyle \phi _{i}}$ have zero overlap. Together with the assumption that ${\displaystyle \phi _{i}}$ are normalized, this means that the overlap matrix is the n × n identity matrix: ${\displaystyle [S_{ij}]=\mathbf {I} _{n}}$. Solving for E in (*) then reduces to finding the eigenvalues of the Hamiltonian matrix.

Second, in the simplest case of a planar, unsaturated hydrocarbon, the Hamiltonian matrix ${\displaystyle \mathbf {H} =[H_{ij}]}$ is parameterized in the following way:

${\displaystyle H_{ij}={\begin{cases}\alpha ,&i=j;\\\beta ,&i,j\ \ {\text{adjacent}};\\0,&{\text{otherwise}}.\end{cases}}}$ (**)

To summarize, we are assuming that: (1) the energy of an electron in an isolated C(2p_z) orbital is ${\displaystyle H_{ii}=\alpha }$; (2) the energy of interaction between C(2p_z) orbitals on adjacent carbons i and j (i.e., i and j are connected by a σ-bond) is ${\displaystyle H_{ij}=\beta }$; (3) orbitals on carbons not joined in this way are assumed not to interact, so ${\displaystyle H_{ij}=0}$ for nonadjacent i and j; and, as mentioned above, (4) the spatial overlap of electron density between different orbitals, represented by non-diagonal elements of the overlap matrix, is ignored by setting ${\displaystyle S_{ij}=0\ \ (i\neq j)}$, even when the orbitals are adjacent.

This neglect of orbital overlap is an especially severe approximation. In actuality, orbital overlap is a prerequisite for orbital interaction, and it is impossible to have ${\displaystyle H_{ij}=\beta }$ while ${\displaystyle S_{ij}=0}$. For typical bond distances (1.40 Å) as might be found in benzene, for example, the true value of the overlap for C(2p_z) orbitals on adjacent atoms i and j is about ${\displaystyle S_{ij}=0.21}$; even larger values are found when the bond distance is shorter (e.g., ${\displaystyle S_{ij}=0.27}$ ethylene).[21] A major consequence of having nonzero overlap integrals is the fact that, compared to non-interacting isolated orbitals, bonding orbitals are not energetically stabilized by nearly as much as antibonding orbitals are destabilized. The orbital energies derived from the Hückel treatment do not account for this asymmetry (see Hückel solution for ethylene (below) for details).

The eigenvalues of ${\displaystyle \mathbf {H} }$ are the Hückel molecular orbital energies ${\displaystyle E_{1},\ldots ,E_{n}}$, expressed in terms of ${\displaystyle \alpha }$ and ${\displaystyle \beta }$, while the eigenvectors are the Hückel MOs ${\displaystyle \Psi _{1},\ldots ,\Psi _{n}}$, expressed as linear combinations of the atomic orbitals ${\displaystyle \phi _{i}}$. Using the expression for the normalization constant N and the fact that ${\displaystyle [S_{ij}]=\mathbf {I} _{n}}$, we can find the normalized MOs by incorporating the additional condition

${\displaystyle \sum _{i=1}^{n}c_{i}^{2}=1}$.

The Hückel MOs are thus uniquely determined when eigenvalues are all distinct. When an eigenvalue is degenerate (two or more of the ${\displaystyle E_{i}}$ are equal), the eigenspace corresponding to the degenerate energy level has dimension greater than 1, and the normalized MOs at that energy level are then not uniquely determined. When that happens, further assumptions pertaining to the coefficients of the degenerate orbitals (usually ones that make the MOs orthogonal and mathematically convenient[22]) have to be made in order to generate a concrete set of molecular orbital functions.

If the substance is a planar, unsaturated hydrocarbon, the coefficients of the MOs can be found without appeal to empirical parameters, while orbital energies are given in terms of only ${\displaystyle \alpha }$ and ${\displaystyle \beta }$. On the other hand, for systems containing heteroatoms, such as pyridine or formaldehyde, values of correction constants ${\displaystyle h_{\mathrm {X} }}$ and ${\displaystyle k_{\mathrm {X-Y} }}$ have to be specified for the atoms and bonds in question, and ${\displaystyle \alpha }$ and ${\displaystyle \beta }$ in (**) are replaced by ${\displaystyle \alpha +h_{\mathrm {X} }\beta }$ and ${\displaystyle k_{\mathrm {X-Y} }\beta }$, respectively.

In the Hückel treatment for ethylene, we write the Hückel MOs ${\displaystyle \Psi \,}$ as a linear combination of the atomic orbitals (2p orbitals) on each of the carbon atoms:

${\displaystyle \ \Psi =c_{1}\phi _{1}+c_{2}\phi _{2}}$.

Applying the result obtained by the Ritz method, we have the system of equations

${\displaystyle {\begin{bmatrix}H_{11}-ES_{11}&H_{12}-ES_{12}\\H_{21}-ES_{21}&H_{22}-ES_{22}\\\end{bmatrix}}{\begin{bmatrix}c_{1}\\c_{2}\\\end{bmatrix}}=0}$,

where:

${\displaystyle H_{ij}=\langle \phi _{i}|{\hat {H}}|\phi _{j}\rangle }$ and

${\displaystyle S_{ij}=\langle \phi _{i}|\phi _{j}\rangle }$.

(Since 2p_z atomic orbital can be expressed as a pure real function, the * representing complex conjugation can be dropped.) The Hückel method assumes that all overlap integrals (including the normalization integrals) equal the Kronecker delta, ${\displaystyle S_{ij}=\delta _{ij}\,}$, all Coulomb integrals ${\displaystyle H_{ii}\,}$ are equal, and the resonance integral ${\displaystyle H_{ij}\,}$ is nonzero when the atoms i and j are bonded. Using the standard Hückel variable names, we set

${\displaystyle H_{11}=H_{22}=\alpha \,}$,

${\displaystyle H_{12}=H_{21}=\beta \,}$,

${\displaystyle S_{11}=S_{22}=1\,}$, and

${\displaystyle S_{12}=S_{21}=0\,}$.

The Hamiltonian matrix is

${\displaystyle \mathbf {H} ={\begin{bmatrix}\alpha &\beta \\\beta &\alpha \\\end{bmatrix}}}$.

The matrix equation that needs to be solved is then

${\displaystyle {\begin{bmatrix}\alpha -E&\beta \\\beta &\alpha -E\\\end{bmatrix}}{\begin{bmatrix}c_{1}\\c_{2}\\\end{bmatrix}}=0}$,

or, dividing by ${\displaystyle \beta }$,

${\displaystyle {\begin{bmatrix}{\frac {\alpha -E}{\beta }}&1\\1&{\frac {\alpha -E}{\beta }}\\\end{bmatrix}}{\begin{bmatrix}c_{1}\\c_{2}\\\end{bmatrix}}=0}$.

Setting ${\displaystyle x:={\frac {\alpha -E}{\beta }}}$, we obtain

${\displaystyle {\begin{bmatrix}x&1\\1&x\\\end{bmatrix}}{\begin{bmatrix}c_{1}\\c_{2}\\\end{bmatrix}}=0}$. (***)

This homogeneous system of equations has nontrivial solutions for ${\displaystyle c_{1},c_{2}}$ (solutions besides the physically meaningless ${\displaystyle c_{1}=c_{2}=0}$) iff the matrix is singular and the determinant is zero:

${\displaystyle {\begin{vmatrix}x&1\\1&x\\\end{vmatrix}}=0}$.

Solving for ${\displaystyle x}$,

${\displaystyle x^{2}-1=0\,}$, or

${\displaystyle x=\pm 1\,}$.

Since ${\displaystyle E=\alpha -x\beta }$, the energy levels are

${\displaystyle E=\alpha -\pm 1\times \beta }$, or

${\displaystyle E=\alpha \mp \beta }$.

The coefficients can then be found by expanding (***):

${\displaystyle c_{2}=-xc_{1}\,}$and

${\displaystyle c_{1}=-xc_{2}\,}$.

Since the matrix is singular, the two equations are linearly dependent, and the solution set is not uniquely determined until we apply the normalization condition. We can only solve for ${\displaystyle c_{2}}$ in terms of ${\displaystyle c_{1}}$:

${\displaystyle c_{2}=-\pm 1\times c_{1}\,}$, or

${\displaystyle c_{2}=\mp c_{1}\,}$.

After normalization with ${\displaystyle c_{1}^{2}+c_{2}^{2}=1}$, the numerical values of ${\displaystyle c_{1}}$ and ${\displaystyle c_{2}}$ can be found:

${\displaystyle c_{1}={\frac {1}{\sqrt {2}}}}$ and ${\displaystyle c_{2}=\mp {\frac {1}{\sqrt {2}}}}$.

Finally, the Hückel molecular orbitals are

${\displaystyle \Psi _{\mp }=c_{1}\phi _{1}+c_{2}\phi _{2}={\frac {1}{\sqrt {2}}}\phi _{1}\mp {\frac {1}{\sqrt {2}}}\phi _{2}={\frac {\phi _{1}\mp \phi _{2}}{\sqrt {2}}}\,}$.

The constant β in the energy term is negative; therefore, ${\displaystyle E_{+}=\alpha +\beta }$ with ${\textstyle \Psi _{+}={\frac {1}{\sqrt {2}}}(\phi _{1}+\phi _{2})\,}$ is the lower energy corresponding to the HOMO energy and ${\displaystyle E_{-}=\alpha -\beta }$ with ${\textstyle \Psi _{-}={\frac {1}{\sqrt {2}}}(\phi _{1}-\phi _{2})\,}$ is the LUMO energy.

If, contrary to the Hückel treatment, a positive value for ${\displaystyle S:=S_{12}=S_{21}}$ were included, the energies would instead be

${\displaystyle E_{\pm }={\frac {\alpha \pm \beta }{1\pm S}}}$,

while the corresponding orbitals would take the form

${\displaystyle \Psi _{\pm }={\sqrt {\frac {1}{2\pm 2S}}}\phi _{1}\pm {\sqrt {\frac {1}{2\pm 2S}}}\phi _{2}}$.

An important consequence of setting ${\displaystyle S>0}$ is that the bonding (in-phase) combination is always stabilized to a lesser extent than the antibonding (out-of-phase) combination is destabilized, relative to the energy of the free 2p orbital. Thus, in general, 2-center 4-electron interactions, where both the bonding and antibonding orbitals are occupied, are destabilizing overall. This asymmetry is ignored by Hückel theory. In general, for the orbital energies derived from Hückel theory, the sum of stabilization energies for the bonding orbitals is equal to the sum of destabilization energies for the antibonding orbitals, as in the simplest case of ethylene shown here and the case of butadiene shown below.

The Hückel MO theory treatment of 1,3-butadiene is largely analogous to the treatment of ethylene, shown in detail above, though we must now find the eigenvalues and eigenvectors of a 4 × 4 Hamiltonian matrix. We first write the molecular orbital ${\displaystyle \Psi \,}$ as a linear combination of the four atomic orbitals ${\displaystyle \phi _{i}}$ (carbon 2p orbitals) with coefficients ${\displaystyle c_{i}}$:

${\displaystyle \ \Psi =c_{1}\phi _{1}+c_{2}\phi _{2}+c_{3}\phi _{3}+c_{4}\phi _{4}}$.

The Hamiltonian matrix is

${\displaystyle \mathbf {H} ={\begin{bmatrix}\alpha &\beta &0&0\\\beta &\alpha &\beta &0\\0&\beta &\alpha &\beta \\0&0&\beta &\alpha \\\end{bmatrix}}}$.

In the same way, we write the secular equations in matrix form as

${\displaystyle {\begin{bmatrix}\alpha -E&\beta &0&0\\\beta &\alpha -E&\beta &0\\0&\beta &\alpha -E&\beta \\0&0&\beta &\alpha -E\\\end{bmatrix}}{\begin{bmatrix}c_{1}\\c_{2}\\c_{3}\\c_{4}\\\end{bmatrix}}=0}$,

which leads to

${\displaystyle {\Big (}{\frac {\alpha -E}{\beta }}{\Big )}^{4}-3{\Big (}{\frac {\alpha -E}{\beta }}{\Big )}^{2}+1=0}$

and

${\displaystyle E_{1,2,3,4}=\alpha +{\frac {{\sqrt {5}}\pm 1}{2}}\beta ,\alpha -{\frac {{\sqrt {5}}\mp 1}{2}}\beta }$, or approximately,

${\displaystyle E_{1,2,3,4}\approx \alpha +1.618\beta ,\alpha +0.618\beta ,\alpha -0.618\beta ,\alpha -1.618\beta }$, where 1.618... and 0.618... are the golden ratios ${\displaystyle \varphi }$ and ${\displaystyle 1/\varphi }$.

The orbitals are given by

${\displaystyle \Psi _{1}\approx 0.372\phi _{1}+0.602\phi _{2}+0.602\phi _{3}+0.372\phi _{4}}$,

${\displaystyle \Psi _{2}\approx 0.602\phi _{1}+0.372\phi _{2}-0.372\phi _{3}-0.602\phi _{4}}$,

${\displaystyle \Psi _{3}\approx 0.602\phi _{1}-0.372\phi _{2}-0.372\phi _{3}+0.602\phi _{4}}$, and

${\displaystyle \Psi _{4}\approx 0.372\phi _{1}-0.602\phi _{2}+0.602\phi _{3}-0.372\phi _{4}}$.

• Möbius–Hückel concept
• Möbius aromaticity
• Tight Binding

• "Hückel method" at chem.swin.edu.au, webpage: mod3-huckel.
• N. Goudard; Y. Carissan; D. Hagebaum-Reignier; S. Humbel (2008). "HuLiS : Java Applet – Simple Hückel Theory and Mesomery – program logiciel software" (in French). Retrieved 19 August 2010.
• Rauk, Arvi. SHMO, Simple Hückel Molecular Orbital Theory Calculator. Java Applet (downloadable) Archived 2018-06-22 at the Wayback Machine.

• The HMO-Model and its applications: Basis and Manipulation, E. Heilbronner and H. Bock, English translation, 1976, Verlag Chemie.
• The HMO-Model and its applications: Problems with Solutions, E. Heilbronner and H. Bock, English translation, 1976, Verlag Chemie.
• The HMO-Model and its applications: Tables of Hückel Molecular Orbitals, E. Heilbronner and H. Bock, English translation, 1976, Verlag Chemie.