Electronic Chemical Potential and Thermoelectric Transport: Fermi Level, Density of States, and Carrier Optimization

How electronic chemical potential controls carrier populations and thermoelectric transport, connecting Fermi–Dirac statistics, density of states, band structure, Seebeck coefficient, and conductivity.

THERMODYNAMICS OF FUNCTIONAL MATERIALS · THERMOELECTRIC TRANSPORT 15

Electronic Chemical Potential and Thermoelectric Transport: Fermi Level, Density of States, and Carrier Optimization

The electronic chemical potential is one of the most important thermodynamic quantities for understanding thermoelectric materials. It determines the equilibrium occupation of electronic states and therefore influences carrier concentration, electrical conductivity, and thermopower. In a semiconductor, changing the chemical potential relative to the band edges can shift the material from a lightly doped regime to a degenerate regime, substantially changing its transport behaviour.

For functional oxides, the electronic chemical potential is also coupled to defect chemistry. Oxygen vacancies, aliovalent dopants, cation nonstoichiometry, and changes in oxygen partial pressure can alter the electron or hole population. Consequently, thermoelectric optimization requires more than adjusting a nominal dopant concentration: it requires understanding how composition and processing control the chemical potential and the resulting transport distribution.

This article develops that connection from Fermi–Dirac statistics through density-of-states models and transport coefficients. It focuses on how the electronic chemical potential governs thermoelectric properties, how it can be estimated from measurements or calculations, and why a high density of states or high carrier concentration alone does not guarantee a high thermoelectric power factor.

1. What is the electronic chemical potential?

The chemical potential of electrons, commonly denoted by (), is the change in the thermodynamic free energy when the number of electrons changes, under the specified thermodynamic constraints:

[ e=(){T,P,} ]

In electronic-structure discussions, the Fermi level is often used to describe the electronic chemical potential at equilibrium. At finite temperature, the chemical potential determines the Fermi–Dirac occupation probability of available states.

The chemical potential is not itself a carrier concentration. Rather, it sets the occupation statistics from which electron and hole concentrations are calculated when the density of states and band structure are known.

For a semiconductor, the position of () relative to the conduction-band minimum (E_C) and valence-band maximum (E_V) is particularly important:

• () closer to (E_C) generally favours electron population.
• () closer to (E_V) generally favours hole population.
• () inside a band corresponds to a degenerate electronic regime in the simplest band picture.

2. Fermi–Dirac statistics

The probability that a state with energy (E) is occupied by an electron is

[ f(E,,T)= ]

where (k_B) is Boltzmann’s constant and (T) is absolute temperature.

The function approaches a step at (E=) as temperature approaches zero. At finite temperature, the occupation changes smoothly over an energy range of order several (k_BT).

The derivative

[ - ]

is concentrated around the chemical potential. This weighting function is central to electronic transport because states far below () are almost fully occupied, while states far above () are almost empty. In linear-response transport, the states near () usually make the dominant contribution to changes in occupation and transport.

3. From density of states to carrier concentration

The electronic density of states (D(E)) describes the number of available electronic states per unit energy and, depending on convention, per unit volume.

The electron concentration is obtained by integrating the occupied states in the conduction band:

[ n=_{E_C}^{}D_C(E)f(E,,T),dE ]

The hole concentration is obtained by integrating the unoccupied states in the valence band:

[ p=_{-}^{E_V}D_V(E)[1-f(E,,T)],dE ]

These integrals show that carrier concentration depends on both the chemical potential and the density of states. A chemical-potential shift can produce a different carrier concentration in two materials if their band structures or effective densities of states differ.

For a multiband material, the total density of states and carrier populations must include all relevant bands and valleys.

4. Nondegenerate semiconductor limit

When the chemical potential lies several (k_BT) away from a band edge and the occupation is sufficiently dilute, the Fermi–Dirac distribution can be approximated by the Boltzmann distribution.

For electrons,

[ n=N_C]

For holes,

[ p=N_V]

Here (N_C) and (N_V) are the effective densities of states in the conduction and valence bands.

Rearranging the electron expression gives

[ =E_C-k_BT() ]

and the hole expression gives

[ =E_V+k_BT() ]

These equations are useful for estimating the chemical potential from a known carrier concentration when the nondegenerate approximation and effective-density-of-states model are justified.

They should not be applied to heavily doped or degenerate semiconductors without using the full Fermi–Dirac integrals.

5. Degenerate carrier populations and Fermi integrals

When the chemical potential approaches or enters a band, the Boltzmann approximation becomes inaccurate. Carrier concentrations are then calculated using Fermi–Dirac integrals.

For a three-dimensional parabolic conduction band,

[ n=N_C F_{1/2}() ]

where

[ = ]

and the normalized Fermi integral may be defined as

[ F_j()= _0^ ,d ]

The exact prefactor convention for (N_C) and (F_j) must be used consistently. In the nondegenerate limit, (F_{1/2}()) approaches (()), recovering the Boltzmann expression.

The degenerate regime is common in heavily doped thermoelectric semiconductors. It requires a model that accounts for Fermi-level filling rather than assuming that carriers occupy only the dilute tail of a band.

6. Density of states in a parabolic band

For a three-dimensional parabolic conduction band with an isotropic effective mass (m^*), the density of states has the energy dependence

[ D_C(E)= ()^{3/2} ]

for (EE_C), when spin degeneracy is included in the conventional expression.

The effective density of states is

[ N_C= 2()^{3/2} ]

for a single isotropic band with the usual spin degeneracy convention. Valley or band degeneracy introduces additional factors.

A larger density-of-states effective mass can increase the number of states near the band edge. This may support a larger thermopower at a given carrier concentration in a suitable model. However, a large effective mass can also reduce carrier velocity and mobility. Thermoelectric performance therefore depends on both the density of states and the transport properties of the states.

7. Chemical potential and the Seebeck coefficient

In the nondegenerate single-parabolic-band approximation, the electron Seebeck coefficient can be expressed as

[ S_n=- ]

where (e>0) is the elementary charge and (A_n) depends on the scattering model and dimensionality.

For holes,

[ S_p=+ ]

The equations show how the chemical potential affects thermopower. Moving () closer to a band edge generally reduces the magnitude of the nondegenerate thermopower, while increasing the carrier concentration.

These expressions are approximations, not universal identities. Their constants depend on the assumed band structure and energy dependence of the relaxation time.

8. Transport distribution: why the density of states is not enough

The electronic transport distribution function combines the density of states with carrier velocity and scattering time. In a simplified isotropic relaxation-time description,

[ (E)=D(E)v^2(E)(E) ]

The transport integrals are

[ L_n= _{-}^{} (E)(E-)^n (-)dE ]

and the Seebeck coefficient for a single electron channel is

[ S=- ]

The electrical conductivity is proportional to the zeroth transport integral, with prefactors determined by the convention used for ((E)) and the transport geometry.

This framework clarifies a key point: the density of states counts available states, but electrical transport weights those states by their ability to carry current. States with high density but very low group velocity may contribute less to conductivity than their density alone suggests.

9. Chemical potential and electrical conductivity

For a single carrier type,

[ =ne_e ]

for electrons, where (e) here denotes the carrier mobility, not the electronic chemical potential. To avoid ambiguity, mobility is often written as ({}):

[ =ne_{} ]

For holes,

[ =pe_{,p} ]

Changing the electronic chemical potential can change the carrier concentration and therefore conductivity. However, doping and defect formation can also change mobility by introducing ionized-impurity scattering, grain-boundary barriers, disorder, or secondary phases.

Thus, an increase in carrier concentration does not necessarily produce a proportional increase in conductivity.

10. The thermoelectric optimization problem

The power factor is

[ ]

Increasing carrier concentration generally increases conductivity but often decreases the magnitude of the Seebeck coefficient. The optimum power factor therefore occurs at a balance between these competing trends.

The chemical potential provides a useful variable for describing this balance. In band-transport calculations, one can evaluate (S) and () as functions of () at a fixed temperature, then identify the chemical-potential range that maximizes (S^2).

This is a model-based optimization. In a real material, changing () by doping may also change the band structure, scattering, phase stability, and defect populations. The theoretical optimum is therefore not automatically a directly accessible composition.

11. Worked example: estimating chemical potential from carrier concentration

Consider a hypothetical nondegenerate n-type semiconductor at 800 K. Suppose its effective density of states is

[ N_C=1.0^{{20} \mathrm{cm}{-3}} ]

and its electron concentration is

[ n=1.0^{{19} \mathrm{cm}{-3}} ]

Using

[ E_C-=k_BT() ]

and (k_B=8.6173^{{-5} \mathrm{eV,K}{-1}}),

[ k_BT=(8.6173^{-5})(800) =0.06894  ]

Therefore,

[ E_C-(10) ]

[ ]

Within the assumed nondegenerate model, the chemical potential lies approximately 0.159 eV below the conduction-band edge.

This numerical result is illustrative. The assumed carrier concentration and effective density of states must be experimentally or computationally justified for a real material, and the nondegenerate condition should be checked.

12. How doping shifts the chemical potential

Aliovalent substitution can change the electron or hole population by changing charge balance. In a simple donor-doped semiconductor, increasing the ionized donor concentration can move the chemical potential toward the conduction band.

In oxides, the actual response depends on defect compensation. A nominal donor may be compensated by cation vacancies, oxygen interstitials, or other acceptor defects. The electron concentration can therefore be much smaller than the nominal dopant concentration.

A useful analysis combines:

• Defect formation energies and chemical-potential constraints.
• Charge-neutrality equations.
• Dopant ionization and compensation.
• Density-of-states calculations.
• Hall or other carrier measurements, where applicable.

The chemical potential is the link between the defect-equilibrium problem and the electronic occupation problem.

13. Oxygen partial pressure and electronic chemical potential in oxides

Oxygen activity can change the carrier population in transition-metal and other nonstoichiometric oxides. A representative oxygen-loss reaction is

[ ]

The equilibrium concentration of oxygen vacancies and electrons depends on temperature, oxygen chemical potential, and the other defect species present.

As the electron concentration changes, charge neutrality requires a corresponding change in the electronic chemical potential. In a simple semiconductor model, this can shift () relative to the conduction-band edge and change both conductivity and thermopower.

The relationship is not always monotonic. Defect association, changes in cation valence, phase transformations, and mobility changes can alter the response. Oxygen-pressure-dependent transport should therefore be interpreted using a defect model appropriate to the specific oxide.

14. Multiband materials and band convergence

Some thermoelectric materials have several bands close in energy. Their contributions to conductivity and thermopower must be combined.

For parallel carrier channels (i), the total Seebeck coefficient is

[ S_{}= {_i _i} ]

while the total conductivity is

[ _{}=_i_i ]

A second band can improve thermoelectric performance if it contributes useful conductivity and increases the energy asymmetry of the combined transport distribution. Band convergence is one strategy for increasing the density of states near the chemical potential without necessarily imposing the same mobility penalty as simply increasing a single-band effective mass.

However, the effect depends on band velocities, degeneracy, scattering, and the relative positions of the bands. A larger total density of states does not guarantee a larger power factor.

15. Estimating chemical potential from electronic-structure calculations

Electronic-structure calculations can provide the band structure and density of states. At a specified temperature and electron count, the chemical potential is obtained by enforcing the electron-number constraint:

[ N_e= D(E)f(E,,T),dE ]

The calculation adjusts () until the integrated occupied density of states matches the required electron count.

For transport calculations, the electronic structure is combined with a transport model or a calculated transport distribution. The resulting (S(,T)) and ((,T)) can be used to explore trends and estimate an optimal chemical-potential range.

Important limitations include:

• Exchange-correlation approximations and band-gap errors.
• Treatment of strongly correlated transition-metal states.
• Uncertainty in scattering times and their energy dependence.
• Disorder and defect configurations not represented by the ideal crystal.
• Temperature-dependent structural and magnetic effects.
• Differences between a rigid-band approximation and actual chemical doping.

A rigid-band shift is useful for screening trends, but it assumes that doping changes the Fermi level without substantially changing the underlying bands. This assumption can fail in complex oxides.

16. Experimental routes to constrain the chemical potential

The chemical potential is not usually measured directly by a routine thermoelectric transport experiment. It is constrained through complementary measurements and models.

Useful methods include:

• Hall effect: estimates carrier sign and concentration under an appropriate Hall-factor model.
• Seebeck coefficient: constrains the energy dependence of transport and, with a validated band model, the chemical-potential position.
• Electrical conductivity: provides information on carrier concentration and mobility together.
• Photoelectron spectroscopy: probes occupied electronic states and work-function or valence-band information, subject to surface sensitivity and calibration.
• Optical spectroscopy: constrains band gaps and electronic transitions.
• X-ray absorption spectroscopy: can provide information on oxidation state and unoccupied electronic states.
• First-principles calculations: estimate band structure, density of states, and defect-related changes.

No single method provides a complete description. Combining transport, spectroscopy, composition analysis, and defect thermodynamics gives a more reliable picture.

17. Practical workflow for thermoelectric oxide research

A useful sequence for investigating a doped oxide is:

1. Establish the phase composition and crystal structure after synthesis.
2. Determine the relevant defect chemistry and likely charge-compensation mechanisms.
3. Measure Seebeck coefficient and electrical conductivity over the same temperature range.
4. Where possible, measure Hall response or carrier concentration independently.
5. Examine oxygen-pressure dependence if oxygen nonstoichiometry is relevant.
6. Compare the data with a single-band or multiband transport model.
7. Evaluate whether changes in thermopower arise primarily from chemical-potential shifts, scattering changes, or phase/defect transformations.
8. Calculate the power factor and assess reproducibility after thermal cycling.

This workflow helps avoid assigning every change in (S) to a change in carrier concentration when the transport distribution may also have changed.

18. Common mistakes

• Confusing chemical potential with carrier mobility: they are different physical quantities.
• Treating the Fermi level as a direct measurement of carrier concentration: the density of states is also required.
• Using Boltzmann statistics in a degenerate regime: use Fermi–Dirac integrals when appropriate.
• Equating nominal dopant concentration with free-carrier concentration: compensation can be substantial in oxides.
• Optimizing density of states alone: velocity and scattering determine how effectively states contribute to transport.
• Assuming a rigid-band model always applies: chemical substitution can change the bands and introduce disorder.
• Inferring a unique chemical potential from Seebeck data without a model: different band and scattering models can produce similar thermopower.
• Ignoring atmosphere-dependent defect chemistry: oxygen activity can change both carrier populations and phase stability.

19. Further reading

These articles develop related concepts and experimental methods:

• Thermodynamics of the Seebeck Effect: Chemical Potential, Carrier Statistics, and Thermoelectric Transport
• Kelvin Relations in Thermoelectric Materials
• Thermoelectric Transport: Seebeck Coefficient, Electrical Conductivity, and Power Factor
• Defect Formation Energy and Equilibrium Concentrations in Oxides
• Chemical Potential and Activity in Solid Materials
• Oxygen Chemical Potential and pO2-Dependent Defect Chemistry
• Oxygen Nonstoichiometry Measurement in Functional Oxides

20. Final perspective

The electronic chemical potential controls the occupation of electronic states and links thermodynamic equilibrium to measurable transport properties. Through the density of states and Fermi–Dirac statistics, it determines carrier populations; through the transport distribution, it influences thermopower and electrical conductivity.

For thermoelectric oxides, the most useful interpretation connects chemical potential to defect equilibria, oxygen activity, band structure, and scattering. Optimizing thermoelectric performance is therefore not simply a matter of maximizing carrier concentration or density of states. It requires engineering the chemical potential and the energy-dependent transport of carriers while preserving the desired phase and material stability.