This document provides an introduction to k.p perturbation theory, which is used to model the band structure of semiconductors. It begins by describing the Bloch theorem and how the Hamiltonian is written in terms of k-dependent and k-independent terms. For a simple two-band model, k.p perturbation theory yields a Dirac-like dispersion with both parabolic and linear terms. The linear term resembles a Dirac cone with an effective velocity proportional to the k.p matrix element between the bands. Historically, k.p theory explained the observation of linear dispersions in narrow-gap semiconductors like InSb. Defects introduce free carriers, and the Fermi level is typically within 100 me
This document provides an introduction to k.p perturbation theory, which is used to model the band structure of semiconductors. It begins by describing the Bloch theorem and how the Hamiltonian is written in terms of k-dependent and k-independent terms. For a simple two-band model, k.p perturbation theory yields a Dirac-like dispersion with both parabolic and linear terms. The linear term resembles a Dirac cone with an effective velocity proportional to the k.p matrix element between the bands. Historically, k.p theory explained the observation of linear dispersions in narrow-gap semiconductors like InSb. Defects introduce free carriers, and the Fermi level is typically within 100 me
This document provides an introduction to k.p perturbation theory, which is used to model the band structure of semiconductors. It begins by describing the Bloch theorem and how the Hamiltonian is written in terms of k-dependent and k-independent terms. For a simple two-band model, k.p perturbation theory yields a Dirac-like dispersion with both parabolic and linear terms. The linear term resembles a Dirac cone with an effective velocity proportional to the k.p matrix element between the bands. Historically, k.p theory explained the observation of linear dispersions in narrow-gap semiconductors like InSb. Defects introduce free carriers, and the Fermi level is typically within 100 me
This document provides an introduction to k.p perturbation theory, which is used to model the band structure of semiconductors. It begins by describing the Bloch theorem and how the Hamiltonian is written in terms of k-dependent and k-independent terms. For a simple two-band model, k.p perturbation theory yields a Dirac-like dispersion with both parabolic and linear terms. The linear term resembles a Dirac cone with an effective velocity proportional to the k.p matrix element between the bands. Historically, k.p theory explained the observation of linear dispersions in narrow-gap semiconductors like InSb. Defects introduce free carriers, and the Fermi level is typically within 100 me
p theory Electrons in a solid are described by the single electron Schrödinger equation with a potential V:
𝑝𝑝2 𝐻𝐻𝐻𝐻 = � + 𝑉𝑉� 𝜓𝜓 = 𝐸𝐸𝐸𝐸 2𝑚𝑚
If V is a periodic potential with symmetries that obey a crystal’s given symmetry, the wavefunction 𝜓𝜓 has to obey Bloch’s theorem: �⃗ 𝜓𝜓𝑛𝑛,𝑘𝑘 (𝑟𝑟⃗) = 𝑒𝑒 𝑖𝑖𝑘𝑘𝑟𝑟⃗ 𝑢𝑢𝑛𝑛,𝑘𝑘 (𝑟𝑟⃗)
Here 𝑘𝑘 is the electron wavenumber and n is the band index. 𝑢𝑢𝑛𝑛,𝑘𝑘 (𝑟𝑟⃗) should also have the same periodicity as the crystal. When 𝐻𝐻 acts on this wavefunction, one obtains: 𝑝𝑝2 �⃗ 𝐻𝐻𝜓𝜓𝑛𝑛,𝑘𝑘 (𝑟𝑟⃗) = � + 𝑉𝑉� 𝑒𝑒 𝑖𝑖𝑘𝑘𝑟𝑟⃗ 𝑢𝑢𝑛𝑛,𝑘𝑘 (𝑟𝑟⃗) 2𝑚𝑚
For simplicity, and without loss of generality, let us work in 1D where 𝑝𝑝 = −𝑖𝑖ℏ𝜕𝜕𝑥𝑥 : 1 2 𝑖𝑖𝑖𝑖𝑖𝑖 𝑝𝑝̂ 𝐻𝐻𝜓𝜓𝑛𝑛,𝑘𝑘 (𝑥𝑥) = 𝑝𝑝 𝑒𝑒 𝑢𝑢𝑛𝑛,𝑘𝑘 (𝑥𝑥) + 𝑉𝑉𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 (𝑥𝑥) = �ℏ𝑘𝑘𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 + 𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑝𝑝. � 𝑢𝑢𝑛𝑛,𝑘𝑘 � + 𝑉𝑉𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 2𝑚𝑚 2𝑚𝑚 1 �ℏ(𝑝𝑝̂ . 𝑘𝑘)𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 + ℏ2 𝑘𝑘 2 𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 + 𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑝𝑝2 𝑢𝑢𝑛𝑛,𝑘𝑘 + ℏ𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 (𝑘𝑘. 𝑝𝑝̂ )𝑢𝑢𝑛𝑛,𝑘𝑘 � + 𝑉𝑉𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 = 𝐸𝐸𝐸𝐸 𝑖𝑖𝑖𝑖𝑖𝑖 𝑢𝑢𝑛𝑛,𝑘𝑘 2𝑚𝑚 Note that k is a simple vector that characterizes the periodicity of the wavefucntion, where 𝑝𝑝̂ is the momentum operator (i.e. 𝑝𝑝̂ ≠ ℏ𝑘𝑘). We can rewrite now rewrite the expression for the Hamiltonian, by rearranging terms as:
𝐻𝐻0 is the free electron part of the Hamiltonian, and 𝐻𝐻𝑘𝑘 is the k-dependent crystal Hamiltonian. This second term will be treated perturbatively.
k.p perturbation theory: a toy model with 2 bands
A perturbation treatment is the simplest way to compute the band dispersion of a material using the k.p formalism. For a given semiconductor system with 2 bands (conduction and valence) an optical spectroscopy measurement can allow on to extract the value of the energy gap. With this knowledge acquired beforehand, k.p theory allows a computation of the band dispersion in the vicinity of the band edges. Assume we have a system constituted of 2 bands located at ±Eg/2 (mid gap in the zero-energy point) Physically speaking, this means that the eigenvalues of the k-independent Hamitlonian for these two bands are ±Eg/2. 𝐸𝐸𝑔𝑔 𝐻𝐻0 𝑢𝑢𝑛𝑛,0 = 𝑢𝑢 2 𝑛𝑛,0 Applying first order degenerate perturbation yields: 𝐸𝐸𝑛𝑛 +< 𝑢𝑢𝑛𝑛,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑛𝑛,0 > < 𝑢𝑢𝑛𝑛,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑚𝑚,0 > � � < 𝑢𝑢𝑚𝑚,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑛𝑛,0 > 𝐸𝐸𝑚𝑚 +< 𝑢𝑢𝑚𝑚,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑚𝑚,0 >
Assuming the two bands have opposite parity, as is the case with most semiconductors. The operator 𝑝𝑝̂ acting on each Bloch function will yield its derivative, so that the 𝑝𝑝̂ dependent diagonal terms vanish by symmetry:
ℏ2 𝑘𝑘2 ℏ ℏ2 𝑘𝑘2 < 𝑢𝑢𝑛𝑛,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑛𝑛,0 >= + 𝑘𝑘 < 𝑢𝑢𝑛𝑛,0 |𝑝𝑝 � |𝑢𝑢𝑛𝑛,0 >= 2𝑚𝑚 𝑚𝑚 2𝑚𝑚 Also, the off-diagonal are particularly interesting to consider since they yield k-linear terms: ℏ ℏ < 𝑢𝑢𝑛𝑛,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑚𝑚,0 >=< 𝑢𝑢𝑚𝑚,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑛𝑛,0 >= 𝑘𝑘 < 𝑢𝑢𝑛𝑛,0 |𝑝𝑝 � |𝑢𝑢𝑚𝑚,0 >= 𝑘𝑘. 𝑃𝑃 𝑚𝑚 𝑚𝑚 Here 𝑃𝑃 ≡< 𝑢𝑢𝑛𝑛,0 |𝐻𝐻𝑘𝑘 |𝑢𝑢𝑚𝑚,0 > is simply a real number referred to as the k.p matrix element. We will later examine its true meaning in a more realistic context. In this situation, it simply takes into account the (‘repulsive?’) interaction between the two bands considered. So in the end, we have to diagonalize the following Hamiltonian to find the band structure of the material in the vicinity of the band edge: 𝐸𝐸𝑔𝑔 ℏ2 𝑘𝑘 2 ℏ + 𝑘𝑘. 𝑃𝑃 𝑢𝑢 𝑢𝑢 ⎛2 2𝑚𝑚 𝑚𝑚 ⎞ � 𝑛𝑛,𝑘𝑘 � = 𝐸𝐸𝑛𝑛,𝑘𝑘 � 𝑛𝑛,𝑘𝑘 � ℏ 𝐸𝐸𝑔𝑔 ℏ 𝑘𝑘2 2 𝑢𝑢 𝑚𝑚,𝑘𝑘 𝑢𝑢 𝑚𝑚,𝑘𝑘 ⎝ 𝑚𝑚 𝑘𝑘. 𝑃𝑃 − + 2 2𝑚𝑚 ⎠ So we find,
The energy has a parabolic term and a term that resembles a Dirac dispersion with velocity, 𝑃𝑃 𝑣𝑣 = 𝑚𝑚 Typically, in semiconductors, the 2 bands result from cation/anion s and px,y orbitals, so that,
< 𝑢𝑢𝑛𝑛,0 |𝑝𝑝�|𝑢𝑢𝑚𝑚,0 > < 𝑆𝑆|𝑝𝑝�|𝑓𝑓(𝑋𝑋, 𝑌𝑌) >
= 𝑚𝑚 𝑚𝑚 The approximate order of magnitude of this matrix element is well known to be 𝑣𝑣𝑐𝑐 ~106m/s or ℏ𝑣𝑣𝑐𝑐 ≈ 6.6𝑒𝑒𝑒𝑒Å. Conversely, since m is the free-electron mass, the k2 term coefficient is about 3.7eVÅ2 . In Fig. 1 we plot 𝐸𝐸±,𝑘𝑘 for different values of 𝐸𝐸𝑔𝑔 /2. The Dirac-like shape of the band structure is indeed remarkable.
FIG 1. Dispersion obtained from the 2-band k.p method for different values of the half-energy gap. A minor different between Dirac fermions resulting from k.p treatments and ideal Dirac cones is the presence of both linearly dispersing components and parabolic components. You will examine the implications of the parabolic components on the dispersion in the homework assignment.
Linear dispersions in condensed-matter
Linear Dirac-like dispersions have been known to occur in semiconductors as early as the mid-1950s. One of this first treatments of the band structure of a semiconductor (InSb) came in a paper by Evan O. Kane (G.E. Research Lab, N.Y.) who used a k.p formalism applied to the bands of III-V materials. Kane noticed that while band structure of InSb is parabolic at low energies and quite similar to that of Ge for example, it asymptotically became linear far away from the band edge. From a practical point-of-view, most semiconductors host usually host a non-negligible amount of defects such as atomic vacancies and anti-site defects. The most immediate impact of such defects is the fact they result in free carriers. As long as the amount of defects is small enough to not cause a renormalization of the band structure, but large enough to yield free carriers, one can represent their impact by a finite carrier filling (or emptying) of the energy levels of the system. Typically, one finds between 1014 and 1019cm-3 carriers per cm-3 in most semiconductors. A record low carrier concentration for Ge for example was 109cm- 3 but is exceptionally difficult to reach. A record low for HgTe is 1012cm-3 but is also extremely challenging to reach. In general, these records require the use of industrial grade synthesis equipment an operation with highly optimized growth parameters and cleanliness. Thus overall the range of carrier densities that typically occur in materials gives a range of Fermi energies that are possible an interesting namely, ~0meV and ~100meV above the mid-gap. Most optical and electronic applications almost require the Fermi level to be in this range. If we fix Ef to 300meV in Fig. 1 again, one can easily notice that for larger gap systems, the band edge becomes more and more parabolic. Thus, a general consequence of k.p theory with the assumptions used above is that narrow gap semiconductors always have a Dirac-like ‘non-parabolic’ dispersion.
Consequences on transport and optical properties
The most immediate consequence on transport an optical properties is the fact that the effective mass becomes energy dependent. As we saw in HW1, one can write the effective mass as: −1 ∗ 𝑑𝑑 2 𝐸𝐸𝑓𝑓 2 𝑚𝑚 = ℏ � 2 � 𝑑𝑑𝑘𝑘
Here 𝐴𝐴𝑓𝑓 = 𝜋𝜋𝑘𝑘𝑓𝑓2 is the area of the Fermi surface
With, the Fermi energy for the conduction band given by:
It is interesting to note that for typical narrow-gap (and even ‘mid-gap’) systems, the free electron term is negligible, and we can comfortably use the following formula for the two band k.p effective mass: 𝐸𝐸𝑔𝑔 𝑚𝑚∗ = 2𝑣𝑣2 Recalling the resemblance between this formula rewritten as: 𝐸𝐸𝑔𝑔 = 𝑚𝑚∗ 𝑣𝑣 2 2𝑣𝑣 2 and, Einstein’s energy-mass relation, one can easily get an idea about the exciting parallels that can be made between condensed-matter and high-energy physics.
(Lecture Notes in Physics 839) Luis Alvarez-Gaumé, Miguel A. Vázquez-Mozo (Auth.) - An Invitation To Quantum Field Theory - Springer-Verlag Berlin Heidelberg (2012)
(Lecture Notes in Physics 839) Luis Alvarez-Gaumé, Miguel A. Vázquez-Mozo (Auth.) - An Invitation To Quantum Field Theory - Springer-Verlag Berlin Heidelberg (2012)
