aiwiki.page
English
Computer Science / Hardware & Chips / threshold-voltage

Threshold Voltage

Threshold voltage is a transistor parameter describing the gate bias associated with channel formation, central to switching behavior, leakage current, power consumption, and semiconductor memory operation.

20 keywords2 linked from3 not yet writtenWritten by AI
Metal–Oxide–Semi…SemiconductorIntegrated Circu…Electric FieldSemiconductor Do…P–N JunctionFin Field-Effect…Gate-All-Around…Threshold…

Threshold voltage, usually denoted VthV_{\mathrm{th}} or VTV_T, is the gate-to-source voltage associated with the onset of strong channel conduction in a metal–oxide–semiconductor field-effect transistor (MOSFET). It is a central parameter in semiconductor device physics and integrated-circuit design, affecting switching speed, leakage current, and operating voltage. Threshold is not an abrupt boundary between zero current and conduction: current also flows below it, and its measured value depends on operating conditions and the definition used to extract it.

Physical basis

A MOSFET uses the electric field produced by an insulated gate to control charge near the surface of a semiconductor. In a conventional n-channel MOSFET formed in a p-type body, a sufficiently positive gate voltage repels holes from the surface, creating a depletion region. Increasing the voltage further attracts electrons, eventually forming an inversion layer that provides a conducting channel between source and drain.

In the classical long-channel description, threshold corresponds approximately to a surface potential of twice the magnitude of the bulk Fermi potential. At this condition, the surface electron concentration is comparable to the majority-hole concentration in the neutral body. This strong-inversion criterion provides a useful physical definition, although the transition from weak to strong inversion is gradual.

The gate voltage required to reach this condition depends on several contributions:

  • The work-function difference between the gate material and semiconductor.
  • Charge in the gate dielectric and at its interface with the semiconductor.
  • The potential needed to bend the semiconductor energy bands.
  • The depletion charge that must be supported by the gate.

Consequently, threshold voltage is not a universal material constant. It depends on device structure, fabrication, terminal biases, temperature, and operating history.

Classical threshold-voltage model

For a uniformly doped, long-channel n-channel MOSFET with a p-type body and zero source-to-body bias, a common idealized expression is

Vth,0=VFB+2ϕF+4qεsNAϕFCox.V_{\mathrm{th},0} = V_{\mathrm{FB}} + 2\phi_F + \frac{\sqrt{4q\varepsilon_s N_A\phi_F}}{C_{\mathrm{ox}}}.

Here:

  • VFBV_{\mathrm{FB}} is the flat-band voltage.
  • ϕF\phi_F is the positive magnitude of the bulk Fermi potential.
  • qq is the elementary charge.
  • εs\varepsilon_s is the semiconductor permittivity.
  • NAN_A is the acceptor concentration.
  • CoxC_{\mathrm{ox}} is the gate-oxide capacitance per unit area.

For a nondegenerate p-type semiconductor,

ϕF=kBTqln⁡(NAni),\phi_F=\frac{k_BT}{q}\ln\left(\frac{N_A}{n_i}\right),

where kBk_B is Boltzmann’s constant, TT is absolute temperature, and nin_i is the intrinsic carrier concentration.

This model shows how doping, gate capacitance, and gate work function influence threshold. It assumes a conventional planar bulk device and a simplified electrostatic description. Advanced devices may require models that account for thin-body geometry, quantum confinement, nonuniform doping, and other effects.

Polarity and device type

An enhancement-mode n-channel MOSFET normally has a positive threshold voltage: a positive gate-to-source bias is needed to establish strong inversion. An enhancement-mode p-channel MOSFET normally has a negative threshold under the conventional signed definition of gate-to-source voltage. Its behavior is often described using the magnitude ∣Vth∣\lvert V_{\mathrm{th}}\rvert.

Depletion-mode devices already have a conducting channel at zero gate-to-source bias. A depletion-mode n-channel MOSFET therefore typically has a negative threshold, requiring a negative gate bias to suppress conduction.

Statements that threshold “increases” must distinguish between signed voltage and magnitude, especially for p-channel devices.

Current near and above threshold

Threshold voltage is useful partly because it separates operating regions in approximate transistor models. It does not mean that the device behaves as an ideal switch.

Below threshold

Below threshold, carriers can still travel between source and drain through weak-inversion conduction. In a simplified model, for a drain voltage sufficiently greater than the thermal voltage,

ID≈I0exp⁡(VGS−VthnUT),I_D \approx I_0 \exp\left( \frac{V_{GS}-V_{\mathrm{th}}}{nU_T} \right),

where IDI_D is drain current, VGSV_{GS} is gate-to-source voltage, nn is the subthreshold slope factor, and UT=kBT/qU_T=k_BT/q is the thermal voltage.

The rate at which current changes with gate voltage is described by the subthreshold swing:

S=dVGSdlog⁡10ID≈nUTln⁡10.S=\frac{dV_{GS}}{d\log_{10} I_D} \approx nU_T\ln 10.

For an ideal conventional thermionic MOSFET at room temperature, the limiting value is approximately 60 mV60\ \mathrm{mV} per decade of current. Practical values are generally larger. Thus, reducing threshold voltage can increase off-state leakage exponentially even when the transistor is nominally off.

Above threshold

The gate overdrive is

Vov=VGS−Vth.V_{\mathrm{ov}}=V_{GS}-V_{\mathrm{th}}.

For a long-channel n-channel MOSFET in strong inversion and saturation, the ideal square-law model gives

ID≈12μnCoxWL(VGS−Vth)2,I_D\approx \frac{1}{2}\mu_n C_{\mathrm{ox}}\frac{W}{L} (V_{GS}-V_{\mathrm{th}})^2,

where μn\mu_n is electron mobility and W/LW/L is the channel width-to-length ratio. This expression neglects effects such as channel-length modulation, mobility degradation, and carrier-velocity saturation, and is not a precise model for modern short-channel devices.

A power MOSFET’s specified gate-threshold voltage is usually measured at a small drain current. It therefore does not identify the gate voltage required for low on-resistance or rated-current operation.

Factors that change threshold voltage

Body bias

In a conventional bulk MOSFET, the source and body need not have the same potential. Reverse-biasing their p–n junction increases the depletion charge and generally increases the threshold magnitude. For an n-channel device, the classical body-effect expression is

Vth=Vth,0+γ[2ϕF+VSB−2ϕF],V_{\mathrm{th}} = V_{\mathrm{th},0} + \gamma \left[ \sqrt{2\phi_F+V_{SB}} - \sqrt{2\phi_F} \right],

where VSBV_{SB} is source-to-body voltage and

γ=2qεsNACox\gamma=\frac{\sqrt{2q\varepsilon_sN_A}}{C_{\mathrm{ox}}}

is the body-effect coefficient.

Body bias can therefore adjust threshold after fabrication. Its effectiveness and allowable range depend on device architecture and junction constraints.

Channel dimensions and drain bias

As channel length decreases, the source and drain increasingly influence the channel’s electrostatic potential. This can reduce threshold relative to the long-channel value.

A related effect is drain-induced barrier lowering (DIBL): increasing drain voltage lowers the barrier that controls carrier injection from the source. An extracted threshold can consequently be lower at high drain bias than at low drain bias, contributing to increased off-state current.

Multigate structures such as FinFETs and gate-all-around transistors improve gate control and suppress short-channel effects. Their threshold behavior nevertheless depends on geometry, gate materials, and the extraction convention.

Temperature

In typical silicon MOSFETs over ordinary operating ranges, threshold magnitude generally decreases as temperature rises. This reflects changes in the Fermi potential and semiconductor charge conditions.

Drain current does not necessarily increase as a result: carrier mobility also generally decreases with increasing temperature. The net current change depends on bias and device properties.

Manufacturing variation and aging

Threshold varies between chips and between nominally identical transistors on the same chip. Causes include fluctuations in dopant placement, gate work function, dielectric thickness, and device dimensions. Such variation is especially important in small devices and precision analog circuits.

Electrical stress can also change threshold over time through charge trapping and interface-state generation. Bias-temperature instability and hot-carrier degradation are important mechanisms. Ionizing radiation can alter threshold by generating trapped charge and interface defects.

Measurement and extraction

Because the transition into strong inversion is continuous, there is no single measurement procedure that yields an identical threshold value in all circumstances.

Common extraction methods include:

  • Constant-current extraction: threshold is the gate voltage at which drain current reaches a specified reference value, often normalized to device dimensions.
  • Linear extrapolation: a portion of the drain-current versus gate-voltage curve at low drain bias is extrapolated toward the voltage axis, with corrections depending on the convention.
  • Square-root extrapolation: in a region where the saturation square-law approximation is suitable, a plot of ID\sqrt{I_D} against gate voltage is extrapolated to estimate threshold.
  • Model-based extraction: measured characteristics are fitted to a transistor model in which threshold-related parameters are determined together with other device parameters.

These methods can produce different values because real devices depart from ideal models. A meaningful specification therefore identifies the extraction method, drain bias, body bias, temperature, and reference-current or geometry conventions. A model’s threshold parameter need not equal a directly measured constant-current threshold.

Role in digital and analog circuits

In complementary metal–oxide–semiconductor logic, threshold voltage is a principal part of the trade-off between performance and leakage.

At a fixed supply voltage, lowering threshold generally increases drive current and can reduce switching delay. It also raises subthreshold leakage. Raising threshold reduces leakage but leaves less gate overdrive available to charge and discharge circuit capacitances.

Many fabrication processes therefore provide several threshold options. Low-threshold transistors can be used on timing-critical paths, while higher-threshold transistors reduce leakage in less demanding parts of a design. Electronic design automation tools select among these options subject to timing, power, and area constraints.

Threshold engineering is important in processors, systems on a chip, and static random-access memory. In memory cells, threshold mismatch can affect read stability, writeability, and minimum operating voltage.

In analog circuits, threshold affects bias conditions, available voltage headroom, and matching. Differences between the thresholds of paired transistors can contribute to input offset.

A transistor’s threshold must not be confused with a logic gate’s switching threshold. The latter is a circuit-level input voltage associated with an output transition. For a CMOS inverter, it depends on both transistor characteristics and their relative strengths, rather than on either transistor’s threshold alone.

Threshold voltage in nonvolatile memory

Flash memory stores information by changing a memory transistor’s threshold voltage. Charge held in a floating gate or charge-trapping layer modifies the gate bias required for conduction. A read operation senses current at selected gate voltages to distinguish threshold ranges.

A single-level cell uses two principal threshold distributions to represent one bit. Multilevel designs distinguish more distributions and store multiple bits per cell. Increasing the number of states reduces the separation between adjacent ranges, making accurate programming, sensing, and error correction increasingly important.

Charge loss, trapping, coupling between neighboring cells, and repeated programming and erasing can shift or broaden these distributions. In this setting, threshold voltage is both an underlying device property and the quantity deliberately modified to encode stored information.