Deal–Grove model¶

A linear–parabolic growth model for thermal oxide thickness in which oxidant moves from gas to surface, diffuses through existing oxide, and reacts at the substrate interface under steady flux.

Version: v1 · 2026-09-28 · History
Domain-specific #: 8878
Domain group: Applied Sciences & Engineering
Origin domain: Engineering & Design (beyond software)
Subdomains: Semiconductor Device Fabrication, Thermal Oxidation → Engineering & Design (beyond software)

Core Idea¶

The Deal–Grove model predicts thermal oxide growth by treating oxidant supply as three serial processes: transport from ambient gas to the surface, diffusion through the existing oxide, and reaction with substrate at the buried interface. The oxide grows where oxidant is consumed at that interface.

Under steady state, the same flux passes through all three steps. Gas transfer, Fickian diffusion, and first-order interface reaction act like thickness-dependent resistances. Converting interfacial flux to thickness yields the familiar linear–parabolic growth law.

Thin oxides emphasize interface reaction and show approximately linear growth; as oxide thickens, its diffusion resistance increases and growth becomes approximately parabolic. Parameters depend on material, oxidant, pressure, and temperature. Deviations at very small scales or during transients mark limits rather than mere fitting error.

How would you explain it like I'm…

Growing Crust Slowdown

When you heat some materials in the right gas, a thin skin grows on them, a bit like rust. The gas has to push through the skin that is already there to reach the material underneath, and the new skin grows at the bottom, where the gas meets the material. So at first the skin grows fast, but the thicker it gets, the longer the trip, and the slower it grows. The Deal–Grove model is the rule that tells you how fast that skin grows.

Three-Step Oxide Growth

Engineers grow thin oxide layers on materials by heating them in an oxygen-carrying gas. The Deal–Grove model says the oxygen has three jobs to do in a row: get from the gas to the surface, travel through the oxide layer already there, and react with the material at the hidden bottom surface. The new oxide forms at that bottom, not on top. Since all three steps happen one after another, the slowest step sets the pace. When the layer is thin, the reaction is the slow part and thickness grows steadily; when the layer is thick, squeezing through it becomes the slow part, and growth slows down more and more.

Linear–Parabolic Oxide Growth

The Deal–Grove model predicts thermal oxide growth by treating the supply of oxidant as three processes in series: gas-phase transport to the surface, diffusion through the existing oxide, and chemical reaction with the substrate at the buried interface where oxide actually forms. In steady state the same flux of oxidant must pass through all three steps, so each acts like a resistance in a series circuit, and the diffusion resistance grows as the oxide thickens. Converting the flux arriving at the interface into a growth rate gives the linear–parabolic growth law. For thin oxides the interface reaction dominates and thickness grows roughly linearly with time; for thick oxides diffusion dominates and growth becomes roughly parabolic, with thickness rising like the square root of time. The rate parameters depend on the material, the oxidant, the pressure, and the temperature, and the model breaks down for very thin films or during transients.

The Deal–Grove model describes thermal oxidation as a steady-state, three-step serial process. Oxidant moves from the ambient gas to the oxide surface (gas-phase mass transfer), diffuses through the existing oxide by Fick's law, and undergoes a first-order reaction with the substrate at the buried oxide–substrate interface, which is where new oxide forms. Because in steady state the same flux passes through every stage, the stages combine like series resistances, with the diffusion term proportional to the current oxide thickness. Converting the interfacial flux into a rate of thickness increase yields the linear–parabolic growth law. In the thin-oxide limit, interface reaction controls the rate and growth is approximately linear in time; in the thick-oxide limit, diffusion controls and growth is approximately parabolic. The rate parameters depend on material, oxidant, pressure, and temperature. Departures at very small thicknesses or during transients indicate the limits of the model's assumptions rather than just imperfect parameter fitting.

Structural Signature¶

Sig role-phrases:

ambient oxidant. Supplies the chemical species and boundary concentration driving growth. Constitutive input. If altered: Changing wet versus dry ambient changes parameters and growth rate.
gas-to-surface transport. Moves oxidant from bulk ambient to the oxide surface. Series transport step. If altered: Neglect is valid only when its resistance is demonstrably small.
oxide diffusion. Carries oxidant through an existing layer with resistance increasing with thickness. Constitutive thickness-dependent step. If altered: Without it the parabolic thick-oxide regime disappears.
interface reaction. Consumes oxidant at the oxide–substrate boundary to form new oxide. Identity-bearing conversion. If altered: Surface deposition without substrate-interface consumption is another growth model.
steady equal flux. Links all stages and converts flux into oxide-thickness rate. Constitutive approximation. If altered: Strong transients or nonuniformity violate the derivation.

What It Is Not¶

Not oxide deposition. Oxidant crosses the film and reacts at the substrate interface.
Not pure diffusion. Interface reaction contributes, especially for thin oxide.
Not a universal nanoscale law. Early growth and other mechanisms can violate assumptions.
Not parameter-free. Transport, diffusion, reaction, and oxidant concentration depend on process conditions.

Scope of Application¶

The model applies to thermal oxidation process analysis where a compact film, interfacial growth, and quasi-steady serial transport are reasonable.

Silicon fabrication. Predicts thermal gate and field oxides.
Process scheduling. Estimates time to target thickness.
Parameter extraction. Separates linear and parabolic rate constants.
Temperature studies. Relates kinetics to process conditions.
Model diagnosis. Uses residuals to detect thin-film and transient limits.

Clarity¶

The model explains why one oxidation process changes apparent rate as the layer grows: the reaction site stays buried while diffusion distance increases. It distinguishes mechanistic rate constants from an arbitrary polynomial fit.

Manages Complexity¶

Gas transport, solubility, diffusion, reaction, and moving-boundary geometry are reduced to equal flux through serial resistances. That compression yields a tractable thickness law while keeping each assumption and failure regime visible.

Abstract Reasoning¶

Specify substrate, oxidant, temperature, pressure, and initial oxide thickness.
Write the gas-transfer, oxide-diffusion, and interface-reaction fluxes with consistent concentrations.
Apply steady equal flux and solve for interfacial concentration.
Convert consumed oxidant flux to moving-interface growth and integrate thickness over time.
Compare residuals by thickness regime before extrapolating beyond calibrated conditions.

Knowledge Transfer¶

The serial transport–diffusion–reaction skeleton transfers to other moving-interface problems when its assumptions hold. Deal–Grove constants and the linear–parabolic law do not transfer unchanged to deposition, porous films, or reaction distributed through a layer.

Examples¶

Canonical¶

Dry oxygen reaches a silicon wafer, dissolves at the oxide surface, diffuses through SiO2, and reacts at the Si/SiO2 interface. Equal steady flux produces rapid thin-film growth that slows as diffusion distance increases.

Mapped back: ambient oxidant → dry O2; gas-to-surface transport → boundary transfer; oxide diffusion → through SiO2 thickness; interface reaction → Si/SiO2 consumption; steady equal flux → one flux and moving boundary.

Applied / In Practice¶

A process engineer fits thin- and thick-oxide runs to extract linear and parabolic constants, schedules a target thickness, then flags systematic ultrathin deviation rather than forcing the same model through it.

Mapped back: ambient oxidant → specified furnace ambient; gas-to-surface transport → included resistance; oxide diffusion → parabolic regime; interface reaction → linear regime; steady equal flux → fit and validity test.

Structural Tensions¶

T1: mechanistic simplicity vs. thin-film accuracy. The compact steady model supports fabrication planning but misses some early-growth physics. Diagnostic: At what thickness do residuals cease to be random?

T2: interface reaction vs. diffusion resistance. One dominates thin growth and the other thick growth, with a continuous transition. Diagnostic: Which resistance controls the current process point?

T3: fitted constants vs. physical portability. Parameters summarize process conditions and cannot be moved blindly across ambients or temperatures. Diagnostic: Were constants calibrated under matching conditions?

Structural–Framed Character¶

Deal–Grove is structural-leaning. Transport, diffusion, reaction, and moving boundaries are physical; steady-state and parameterization are modeling choices. It transfers within suitable oxidation processes but remains semiconductor-specific in name and constants. Its character: a serial-resistance account of a slowing buried-interface growth front.

Structural Core vs. Domain Accent¶

Skeletal core. Supply crosses sequential resistances and is consumed at a moving interface whose growth lengthens one resistance.

Domain-bound accent. Silicon, oxide, oxidant, furnace conditions, Fick diffusion, and interfacial reaction define the model.

Why not prime. Coupled transport and reaction travel, but Deal–Grove is a named semiconductor oxidation law.

This entry is a kind of Physical-System Model.

Rate limiting. Dominant resistance changes with oxide thickness.
Moving boundary. Product growth changes the geometry governing future transport.
No canonical parent edge is asserted in the current DAG.

Relationships to Other Abstractions¶

Current abstraction Deal–Grove model Domain-specific

Parents (1) — more general patterns this builds on

Deal–Grove model is a kind of Physical-System Model Domain-specific

It is a physical model of silicon oxidation kinetics.

Hierarchy path (1) — routes to 1 parentless root

Deal–Grove model → Physical-System Model → Representation → Abstraction

Neighborhood in Abstraction Space¶

Deal–Grove model sits in a sparse region of the domain-specific corpus (75^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2551 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Parabolic oxidation law. Tell: Is the full reaction-plus-diffusion transition represented?
Thin-film deposition. Tell: Does material arrive at the outer surface or form at the buried interface?
Native oxide growth. Tell: Do process conditions satisfy the thermal model assumptions?
Reaction–diffusion model. Tell: Is the specific Deal–Grove serial steady-state structure intended?

References¶

Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Deal%E2%80%93Grove_model (revision 1370057815).
Preserved source candidate: http://optoelectronics.eecs.berkeley.edu/ey1989s2464928.pdf
Preserved source candidate: https://www.researchgate.net/publication/306273009
Preserved source candidate: https://www.iue.tuwien.ac.at/phd/hollauer/node16.html
Preserved source candidate: http://www.lelandstanfordjunior.com/thermaloxide.html

The frozen Wikipedia revision is discovery provenance. The retained source set was reviewed for identity, formal or operational relation, and scope. The encyclopedia's structural synthesis is bounded to those claims; a thin authority surface is recorded as a nonblocking source-strengthening repair rather than concealed.