Electron mobility

This article is about the mobility for electrons and holes in metals and semiconductors. For the general concept, see Electrical mobility.

In solid-state physics, the electron mobility characterizes how quickly an electron can move through a metal or semiconductor, when pulled by an electric field. In semiconductors, there is an analogous quantity for holes, called hole mobility. The term carrier mobility refers in general to both electron and hole mobility in semiconductors.

Electron and hole mobility are special cases of electrical mobility of charged particles in a fluid under an applied electric field.

When an electric field E is applied across a piece of material, the electrons respond by moving with an average velocity called the drift velocity, $\,v_{d}$. Then the electron mobility μ is defined as

$\,v_{d}=\mu E$.

Electron mobility is almost always specified in units of cm²/(V·s). This is different from the SI unit of mobility, m²/(V·s). They are related by 1m²/(V·s) = 10⁴cm²/(V·s).

Conductivity is proportional to the product of mobility and carrier concentration. For example, the same conductivity could come from a small number of electrons with high mobility for each, or a large number of electrons with a small mobility for each. For metals, it would not typically matter which of these is the case, since most metal electrical behavior depends on conductivity alone. Therefore mobility is relatively unimportant in metal physics. On the other hand, for semiconductors, the behavior of transistors and other devices can be very different depending on whether there are many electrons with low mobility or few electrons with high mobility. Therefore mobility is a very important parameter for semiconductor materials. Almost always, higher mobility leads to better device performance, with other things equal.

Semiconductor mobility depends on the impurity concentrations (including donor and acceptor concentrations), defect concentration, temperature, and electron and hole concentrations. It also depends on the electric field, particularly at high fields when velocity saturation occurs. It can be determined by the Hall effect, or inferred from transistor behavior.

Introduction

Drift velocity in an electric field

Main article: Drift velocity

Without any applied electric field, in a solid, electrons (or, in the case of semiconductors, both electrons and holes) move around randomly. Therefore, on average there will be no overall motion of charge carriers in any particular direction over time.

However, when an electric field is applied, each electron is accelerated by the electric field. If the electron were in a vacuum, it would be accelerated to faster and faster velocities (called ballistic transport). However, in a solid, the electron repeatedly scatters off crystal defects, phonons, impurities, etc. Therefore, it does not accelerate faster and faster; instead it moves with a finite average velocity, called the drift velocity. This net electron motion is usually much slower than the normally occurring random motion.

In a semiconductor the two charge carriers, electrons and holes, will typically have different drift velocities for the same electric field.

Quasi-ballistic transport is possible in solids if the electrons are accelerated across a very small distance (as small as the mean free path), or for a very short time (as short as the mean free time). In these cases, drift velocity and mobility are not meaningful.

Definition and units

See also: Electrical mobility

The electron mobility is defined by the equation:

$\,v_{d}=\mu E$.

where:

E is the magnitude of the electric field applied to a material,

v_d is the magnitude of the electron drift velocity (in other words, the electron drift speed) caused by the electric field, and

µ is the electron mobility.

The hole mobility is defined by the same equation. Both electron and hole mobilities are positive by definition.

Usually, the electron drift velocity in a material is directly proportional to the electric field, which means that the electron mobility is a constant (independent of electric field). When this is not true (for example, in very large electric fields), the mobility depends on the electric field.

The SI unit of velocity is m/s, and the SI unit of electric field is V/m. Therefore the SI unit of mobility is (m/s)/(V/m) = m²/(V·s). However, mobility is much more commonly expressed in cm²/(V·s) = 10⁻⁴ m²/(V·s).

Mobility is usually a strong function of material impurities and temperature, and is determined empirically. Mobility values are typically presented in table or chart form. Mobility is also different for electrons and holes in a given semiconductor.

Relation to conductivity

There is a simple relation between mobility and electrical conductivity. Let n be the number density of electrons, and let μ_e be their mobility. In the electric field E, each of these electrons will move with the velocity vector $-\mu _{e}\mathbf {E}$, for a total current density of $ne\mu _{e}\mathbf {E}$ (where e is the elementary charge). Therefore, the electrical conductivity σ satisfies:

$\sigma =ne\mu _{e}$.

This formula is valid when the conductivity is due entirely to electrons. In a p-type semiconductor, the conductivity is due to holes instead, but the formula is essentially the same: If p is the density of holes and μ_h is the hole mobility, then the conductivity is

$\sigma =pe\mu _{h}$.

If a semiconductor has both electrons and holes, the total conductivity is

$\sigma =e(n\mu _{e}+p\mu _{h}).$

Examples

Typical electron mobility for Si at room temperature (300 K) is 1400 cm²/ (V·s) and the hole mobility is around 450 cm²/ (V·s).

Very high mobility has been found in several low-dimensional systems, such as two-dimensional electron gases (2DEG) (35,000,000 cm²/(V·s) at low temperature), carbon nanotubes (100,000 cm²/(V·s) at room temperature) and more recently, graphene (200,000 cm²/ V·s at low temperature). Organic semiconductors (polymer, oligomer) developed thus far have carrier mobilities below 10 cm²/(V·s), and usually much lower.

Electric field dependence and velocity saturation

Main article: Velocity saturation

At low fields, the drift velocity v_d is proportional to the electric field E, so mobility μ is constant. This value of μ is called the low-field mobility.

As the electric field is increased, however, the carrier velocity increases sublinearly and asymptotically towards a maximum possible value, called the saturation velocity v_sat. For example, the value of v_sat is on the order of 1×10⁷ cm/s for both electrons and holes in Si. It is on the order of 6×10⁶ cm/s for Ge. This velocity is a characteristic of the material and a strong function of doping or impurity levels and temperature. It is one of the key material and semiconductor device properties that determine a device such as a transistor's ultimate limit of speed of response and frequency.

This velocity saturation phenomenon results from a process called optical phonon scattering. At high fields, carriers are accelerated enough to gain sufficient kinetic energy between collisions to emit an optical phonon, and they do so very quickly, before being accelerated once again. The velocity that the electron reaches before emitting a phonon is:

${\frac {m^{*}v_{emit}^{2}}{2}}\approx \hbar \omega _{phonon(opt.)}$

where ω_phonon(opt.) is the optical-phonon angular frequency and m* the carrier effective mass in the direction of the electric field. The value of E_{phonon (opt.)} is 0.063 eV for Si and 0.034 eV for GaAs and Ge. The saturation velocity is only one-half of v_emit, because the electron starts at zero velocity and accelerates up to v_emit in each cycle. (This is a somewhat oversimplified description.)

Velocity saturation is not the only possible high-field behavior. Another is the Gunn effect, where a sufficiently high electric field can cause intervalley electron transfer, which reduces drift velocity. This is unusual; increasing the electric field almost always increases the drift velocity, or else leaves it unchanged. The result is negative differential resistance.

In the regime of velocity saturation (or other high-field effects), mobility is a strong function of electric field. This means that mobility is a somewhat less useful concept, compared to simply discussing drift velocity directly.

Relation between scattering and mobility

Recall that by definition, mobility is dependent on the drift velocity. The main factor determining drift velocity (other than effective mass) is scattering time, i.e. how long the carrier is ballistically accelerated by the electric field until it scatters (collides) with something that changes its direction and/or energy. The most important sources of scattering in typical semiconductor materials, discussed below, are ionized impurity scattering and acoustic phonon scattering (also called lattice scattering). In some cases other sources of scattering may be important, such as neutral impurity scattering, optical phonon scattering, surface scattering, and defect scattering.

Elastic scattering means that energy is (almost) conserved during the scattering event. Some elastic scattering processes are scattering from acoustic phonons, impurity scattering, piezoelectric scattering, etc. In acoustic phonon scattering, electrons scatter from state k to k', while emitting or absorbing a phonon of wave vector q. This phenomenon is usually modeled by assuming that lattice vibrations cause small shifts in energy bands. The additional potential causing the scattering process is generated by the deviations of bands due to these small transitions from frozen lattice positions.

Ionized impurity scattering

Semiconductors are doped with donors and/or acceptors, which are typically ionized, and are thus charged. The Coulombic forces will deflect an electron or hole approaching the ionized impurity. This is known as ionized impurity scattering. The amount of deflection depends on the speed of the carrier and its proximity to the ion. The more heavily a material is doped, the higher the probability that a carrier will collide with an ion in a given time, and the smaller the mean free time between collisions, and the smaller the mobility. When determining the strength of these interactions due to the long-range nature of the Coulomb potential, other impurities and free carriers cause the range of interaction with the carriers to reduce significantly compared to bare Coulomb interaction.

If these scatterers are near the interface, the complexity of the problem increases due to the existence of crystal defects and disorders. Charge trapping centers that scatter free carriers form in many cases due to defects associated with dangling bonds. Scattering happens because after trapping a charge, the defect becomes charged and therefore starts interacting with free carriers. If scattered carriers are in the inversion layer at the interface, the reduced dimensionality of the carriers makes the case differ from the case of bulk impurity scattering as carriers move only in two dimensions. Interfacial roughness also causes short-range scattering limiting the mobility of quasi-two-dimensional electrons at the interface.

Lattice (phonon) scattering

At any temperature above absolute zero, the vibrating atoms create pressure (acoustic) waves in the crystal, which are termed phonons. Like electrons, phonons can be considered to be particles. A phonon can interact (collide) with an electron (or hole) and scatter it. At higher temperature, there are more phonons, therefore increased phonon scattering which tends to reduce mobility.

Piezoelectric scattering

Piezoelectric effect can occur only in compound semiconductor due to their polar nature. It is small in most semiconductors but may lead to local electric fields that cause scattering of carriers by deflecting them, this effect is important mainly at low temperatures where other scattering mechanisms are weak. These electric fields arise from the distortion of the basic unit cell as strain is applied in certain directions in the lattice.

Surface roughness scattering

Surface roughness scattering caused by interfacial disorder is short range scattering limiting the mobility of quasi-two-dimensional electrons at the interface. From high-resolution transmission electron micrographs, it has been determined that the interface is not abrupt on the atomic level, but actual position of the interfacial plane varies one or two atomic layers along the surface. These variations are random and cause fluctuations of the energy levels at the interface, which then causes scattering.

Alloy scattering

In compound (alloy) semiconductors, which many thermoelectric materials are, scattering caused by the perturbation of crystal potential due to the random positioning of substituting atom species in a relevant sublattice is known as alloy scattering. This can only happen in ternary or higher alloys as their crystal structure forms by randomly replacing some atoms in one of the sublattices (sublattice) of the crystal structure. Generally, this phenomenon is quite weak but in certain materials or circumstances, it can become dominant effect limiting conductivity. In bulk materials, interface scattering is usually ignored.

Inelastic scattering

During inelastic scattering processes, significant energy exchange happens. As with elastic phonon scattering also in the inelastic case, the potential arises from energy band deformations caused by atomic vibrations. Optical phonons causing inelastic scattering usually have the energy in the range 30-50 meV, for comparison energies of acoustic phonon are typically less than 1 meV but some might have energy in order of 10 meV. There is significant change in carrier energy during the scattering process. Optical or high-energy acoustic phonons can also cause intervalley or interband scattering, which means that scattering is not limited within single valley.

Electron–electron scattering

Due to the Pauli exclusion principle, electrons can be considered as non-interacting if their density does not exceed the value 10¹⁷ cm⁻³ or electric field value 10³ V/cm. However, significantly above these limits electron–electron scattering starts to dominate. Long range and nonlinearity of the Coulomb potential governing interactions between electrons make these interactions difficult to deal with.

Relation between mobility and scattering time

A simple model gives the approximate relation between scattering time (average time between scattering events) and mobility. It is assumed that after each scattering event, the carrier's motion is randomized, so it has zero average velocity. After that, it accelerates uniformly in the electric field, until it scatters again. The resulting average drift mobility is:

$\mu ={\frac {q}{m^{*}}}{\overline {\tau }}$

where q is the elementary charge, m* is the carrier effective mass, and τ is the average scattering time.

If the effective mass is anisotropic (direction-dependent), m* is the effective mass in the direction of the electric field.

Matthiessen's rule

Normally, more than one source of scattering is present, for example both impurities and lattice phonons. It is normally a very good approximation to combine their influences using "Matthiessen's Rule" (developed from work by Augustus Matthiessen in 1864):

${\frac {1}{\mu }}={\frac {1}{\mu _{\rm {impurities}}}}+{\frac {1}{\mu _{\rm {lattice}}}}$.

where µ is the actual mobility, $\mu _{\rm {impurities}}$ is the mobility that the material would have if there was impurity scattering but no other source of scattering, and $\mu _{\rm {lattice}}$ is the mobility that the material would have if there was lattice phonon scattering but no other source of scattering. Other terms may be added for other scattering sources, for example

${\frac {1}{\mu }}={\frac {1}{\mu _{\rm {impurities}}}}+{\frac {1}{\mu _{\rm {lattice}}}}+{\frac {1}{\mu _{\rm {defects}}}}+\cdots$.

Matthiessen's rule can also be stated in terms of the scattering time:

${\frac {1}{\tau }}={\frac {1}{\tau _{\rm {impurities}}}}+{\frac {1}{\tau _{\rm {lattice}}}}+{\frac {1}{\tau _{\rm {defects}}}}+\cdots$.

where τ is the true average scattering time and τ_impurities is the scattering time if there was impurity scattering but no other source of scattering, etc.

Matthiessen's rule is an approximation and is not universally valid. This rule is not valid if the factors affecting the mobility depend on each other, because individual scattering probabilities cannot be summed unless they are independent of each other. The average free time of flight of a carrier and therefore the relaxation time is inversely proportional to the scattering probability. For example, lattice scattering alters the average electron velocity (in the electric-field direction), which in turn alters the tendency to scatter off impurities. There are more complicated formulas that attempt to take these effects into account.

Temperature dependence of mobility

Typical temperature dependence of mobility

	Si	Ge	GaAs
Electrons	∝T −2.4	∝T −1.7	∝T −1.0
Holes	∝T −2.2	∝T −2.3	∝T −2.1

With increasing temperature, phonon concentration increases and causes increased scattering. Thus lattice scattering lowers the carrier mobility more and more at higher temperature. Theoretical calculations reveal that the mobility in non-polar semiconductors, such as silicon and germanium, is dominated by acoustic phonon interaction. The resulting mobility is expected to be proportional to T ^−3/2, while the mobility due to optical phonon scattering only is expected to be proportional to T ^−1/2. Experimentally, values of the temperature dependence of the mobility in Si, Ge and GaAs are listed in table.

As ${\frac {1}{\tau }}\propto \left\langle v\right\rangle \Sigma$, where $\Sigma$ is the scattering cross section for electrons and holes at a scattering center and $\left\langle v\right\rangle$ is a thermal average (Boltzmann statistics) over all electron or hole velocities in the lower conduction band or upper valence band, temperature dependence of the mobility can be determined. In here, the following definition for the scattering cross section is used: number of particles scattered into solid angle dΩ per unit time divided by number of particles per area per time (incident intensity), which comes from classical mechanics. As Boltzmann statistics are valid for semiconductors $\left\langle v\right\rangle \sim {\sqrt {T}}$.

For scattering from acoustic phonons, for temperatures well above Debye temperature, the estimated cross section Σ_ph is determined from the square of the average vibrational amplitude of a phonon to be proportional to T. The scattering from charged defects (ionized donors or acceptors) leads to the cross section ${\Sigma }_{def}\propto {\left\langle v\right\rangle }^{-4}$. This formula is the scattering cross section for "Rutherford scattering", where a point charge (carrier) moves past another point charge (defect) experiencing Coulomb interaction.

The temperature dependencies of these two scattering mechanism in semiconductors can be determined by combining formulas for τ, Σ and $\left\langle v\right\rangle$, to be for scattering from acoustic phonons ${\mu }_{ph}\sim T^{-3/2}$ and from charged defects ${\mu }_{def}\sim T^{3/2}$.

The effect of ionized impurity scattering, however, decreases with increasing temperature because the average thermal speeds of the carriers are increase. Thus, the carriers spend less time near an ionized impurity as they pass and the scattering effect of the ions is thus reduced.

These two effects operate simultaneously on the carriers through Matthiessen's rule. At lower temperatures, ionized impurity scattering dominates, while at higher temperatures, phonon scattering dominates, and the actual mobility reaches a maximum at an intermediate temperature.

Measurement of semiconductor mobility

Hall mobility

Main article: Hall effect

Hall effect measurement setup for holes

Hall effect measurement setup for electrons

Carrier mobility is most commonly measured using the Hall effect. The result of the measurement is called the "Hall mobility" (meaning "mobility inferred from a Hall-effect measurement").

Consider a semiconductor sample with a rectangular cross section as shown in the figures, a current is flowing in the x-direction and a magnetic field is applied in the z-direction. The resulting Lorentz force will accelerate the electrons (n-type materials) or holes (p-type materials) in the (−y) direction, according to the right hand rule and set up an electric field ξ_y. As a result there is a voltage across the sample, which can be measured with a high-impedance voltmeter. This voltage, V_H, is called the Hall voltage. V_H is negative for n-type material and positive for p-type material.

Mathematically, the Lorentz force acting on a charge q is given by

For electrons:

${\overrightarrow {F}}_{Hn}=-q({\overrightarrow {v}}_{n}\times {\overrightarrow {B}}_{z})$

For holes:

${\overrightarrow {F}}_{Hp}=+q({\overrightarrow {v}}_{p}\times {\overrightarrow {B}}_{z})$

In steady state this force is balanced by the force set up by the Hall voltage, so that there is no net force on the carriers in the y direction. For electron,

${\overrightarrow {F}}_{y}=(-q){\overrightarrow {\xi }}_{y}+(-q)[{\overrightarrow {v}}_{n}\times {\overrightarrow {B}}_{z}]=0$

$\Rightarrow -q\xi _{y}+qv_{x}B_{z}=0$

$\xi _{y}=v_{x}B_{z}$

For electrons, the field points in the -y direction, and for holes, it points in the +y direction.

The electron current I is given by $I=-qnv_{x}tW$. Sub v_x into the expression for ξ_y,

$\xi _{y}=-{\frac {IB}{nqtW}}=+{\frac {R_{Hn}IB}{tW}}$

where R_Hn is the Hall coefficient for electron, and is defined as

$R_{Hn}=-{\frac {1}{nq}}$

Since $\xi _{y}={\frac {V_{H}}{W}}$

$R_{Hn}=-{\frac {1}{nq}}={\frac {V_{Hn}t}{IB}}$

Similarly, for holes

$R_{Hp}={\frac {1}{pq}}={\frac {V_{Hp}t}{IB}}$

From the Hall coefficient, we can obtain the carrier mobility as follows:

$\mu _{n}=(-nq)\mu _{n}(-{\frac {1}{nq}})=-\sigma _{n}R_{Hn}$

$=-{\frac {\sigma _{n}V_{Hn}t}{IB}}$

Similarly,

$\mu _{p}={\frac {\sigma _{p}V_{Hp}t}{IB}}$

Here the value of V_Hp (Hall voltage), t (sample thickness), I (current) and B (magnetic field) can be measured directly, and the conductivities σ_n or σ_p are either known or can be obtained from measuring the resistivity.

Field-effect mobility

See also: MOSFET

Not to be confused with Wien effect.

The mobility can also be measured using a field-effect transistor (FET). The result of the measurement is called the "field-effect mobility" (meaning "mobility inferred from a field-effect measurement").

The measurement can work in two ways: From saturation-mode measurements, or linear-region measurements. (See MOSFET for a description of the different modes or regions of operation.)

Using saturation mode

In this technique, for each fixed gate voltage V_GS, the drain-source voltage V_DS is increased until the current I_D saturates. Next, the square root of this saturated current is plotted against the gate voltage, and the slope m_sat is measured. Then the mobility is:

$\mu =m_{sat}^{2}{\frac {2L}{W}}{\frac {1}{C_{i}}}$

where L and W are the length and width of the channel and C_i is the gate insulator capacitance per unit area. This equation comes from the approximate equation for a MOSFET in saturation mode:

$I_{D}={\frac {\mu C_{i}}{2}}{\frac {W}{L}}(V_{GS}-V_{th})^{2}.$

where V_th is the threshold voltage. This approximation ignores the Early effect (channel length modulation), among other things. In practice, this technique may underestimate the true mobility.

Using the linear region

In this technique, the transistor is operated in the linear region (or "ohmic mode"), where V_DS is small and $I_{D}\propto V_{GS}$ with slope m_lin. Then the mobility is:

$\mu =m_{lin}{\frac {L}{W}}{\frac {1}{V_{DS}}}{\frac {1}{C_{i}}}$.

This equation comes from the approximate equation for a MOSFET in the linear region:

$I_{D}=\mu C_{i}{\frac {W}{L}}\left((V_{GS}-V_{th})V_{DS}-{\frac {V_{DS}^{2}}{2}}\right)$

In practice, this technique may overestimate the true mobility, because if V_DS is not small enough and V_G is not large enough, the MOSFET may not stay in the linear region.

Doping concentration dependence in heavily-doped silicon

The charge carriers in semiconductors are electrons and holes. Their numbers are controlled by the concentrations of impurity elements, i.e. doping concentration. Thus doping concentration has great influence on carrier mobility.

While there is considerable scatter in the experimental data, for noncompensated material (no counter doping) for heavily doped substrates (i.e. $10^{18}\mathrm {cm} ^{-3}$ and up), the mobility in silicon is often characterized by the empirical relationship:

$\mu =\mu _{o}+{\frac {\mu _{1}}{1+({\frac {N}{N_{\text{ref}}}})^{\alpha }}}$

where N is the doping concentration (either N_D or N_A), and N_ref and α are fitting parameters. At room temperature, the above equation becomes:

Majority carriers:

$\mu _{n}(N_{D})=65+{\frac {1265}{1+({\frac {N_{D}}{8.5\times 10^{16}}})^{0.72}}}$

$\mu _{p}(N_{A})=48+{\frac {447}{1+({\frac {N_{A}}{6.3\times 10^{16}}})^{0.76}}}$

Minority carriers:

$\mu _{n}(N_{A})=232+{\frac {1180}{1+({\frac {N_{A}}{8\times 10^{16}}})^{0.9}}}$

$\mu _{p}(N_{D})=130+{\frac {370}{1+({\frac {N_{D}}{8\times 10^{17}}})^{1.25}}}$

These equations apply only to silicon, and only under low field.

See also

• Speed of electricity

External links

• semiconductor glossary entry for electron mobility
• Resistivity and Mobility Calculator from the BYU Cleanroom
• Online lecture- Mobility from an atomistic point of view