Chemical vapor deposition (CVD): Question bank — Hardware deep dive

Before we start, three words we lean on constantly — make sure each is a picture in your head, not just a symbol:

True or false — justify

True or false: CVD deposits film by physically spraying atoms onto the wafer.

False — that describes PVD. In CVD the film is produced by a chemical reaction of gaseous precursors at the surface; spraying has no chemistry.

True or false: In the reaction-limited regime, growth rate rises exponentially with temperature.

True — here rate

\approx k_{s} C_{g}

and

k_{s} = k_{0} e^{- E_{a} / k T}

, so heating steepens the climb over the

E_{a}

hill and rate shoots up.

True or false: In the transport-limited regime, doubling the wafer temperature roughly doubles the growth rate.

False — transport is set by

h_{g}

, which barely changes with

T

(gas diffusion is weakly temperature-dependent). Rate stays nearly flat with

T

True or false: Lower chamber pressure makes reactant reach the surface faster.

True — gas diffusivity

D \propto 1/ P

and

h_{g} \propto D

, so dropping pressure widens the "doorway"

h_{g}

True or false: Because low-pressure CVD raises

h_{g}

, it makes the process transport-limited.

False — raising

h_{g}

makes transport fast (a big conductance), so the slow step becomes

k_{s}

: it pushes you into reaction-limited control, which is exactly why LPCVD is uniform.

True or false: The combined flux

F = \frac{k _{s} h _{g}}{k _{s} + h _{g}} C_{g}

can be larger than either

k_{s} C_{g}

h_{g} C_{g}

False — it is a series combination (

R = 1/ k_{s} + 1/ h_{g}

), so

F

is always smaller than the flux either step alone would allow; two resistances in series can only slow you down.

True or false: PECVD's advantage is that it deposits at low temperature.

True — plasma supplies the reaction energy electrically instead of thermally, so films grow below ~400 °C, safe over already-deposited metal.

True or false: A conformal film is one that is very thick.

False — conformal means equal thickness everywhere, including trench sidewalls and bottoms; it's about uniformity of coverage, not amount. See Step Coverage & Conformality.

Why questions

Why is the "slowest step" idea central to understanding CVD?

Transport and reaction happen in series, so the overall rate is capped by whichever conductance is smaller — fixing the fast step does nothing until you address the bottleneck.

Why does temperature choose the regime rather than pressure?

k_{s}

grows exponentially with

T

while

h_{g}

is nearly

T

-flat, so heating swings the ratio

k_{s} / h_{g}

over many orders of magnitude — that swing is what flips you from reaction- to transport-limited.

Why do engineers deliberately run in the reaction-limited regime for thickness uniformity?

There, rate

\approx k_{s} C_{g}

depends on temperature (which is easy to hold uniform) and not on gas flow, so even uneven flow across a batch of wafers gives even film thickness.

Why can CVD coat inside deep trenches while PVD cannot?

Reactant is a gas that diffuses to and reacts on every exposed surface; PVD arrives along straight physical paths (line-of-sight), so shadowed sidewalls starve.

Why is CVD film purity controllable by gas flows?

Composition is set by the ratio of precursor gases fed in, since the film is literally the reaction product — turn a knob on flow, change the stoichiometry.

Why does Thermal Oxidation differ from CVD even though both grow

SiO_{2}

Thermal oxidation consumes the silicon wafer itself as one reactant (oxide grows into the substrate); CVD deposits oxide onto the surface from external gases, consuming no wafer.

Edge cases

Edge case: What is the growth rate if the surface reaction is completely shut off,

k_{s} = 0

F = \frac{k _{s} h _{g}}{k _{s} + h _{g}} C_{g} \to 0

— no reaction, no film, no matter how fast gas arrives. Transport alone deposits nothing.

Edge case: What happens to

F

k_{s} \to \infty

(infinitely fast chemistry)?

F \to h_{g} C_{g}

: every molecule that reaches the surface reacts instantly, so gas delivery is the sole limit — pure transport-limited behaviour.

Edge case: If

h_{g} = k_{s}

, what fraction of

C_{g}

survives to the surface?

C_{s} = \frac{h _{g}}{h _{g} + k _{s}} C_{g} = \frac{1}{2} C_{g}

— the two resistances split the concentration drop evenly, and

F = \frac{1}{2} k_{s} C_{g}

Edge case: A trench with aspect ratio near zero (a flat open surface) — does conformality still matter?

Barely — with no sidewalls or shadowing, even line-of-sight PVD coats evenly, so CVD's conformality advantage only pays off as aspect ratio grows.

Edge case: At extremely low pressure,

h_{g}

becomes huge — is growth then unlimited?

No — once

h_{g} ≫ k_{s}

you are fully reaction-limited, so rate saturates at

k_{s} C_{g}

and only rises with temperature; raising

h_{g}

further changes nothing.

Recall One-sentence master key

Everything above collapses to: find the smaller of $h_{g}$ and $k_{s}$ — that's your bottleneck, and it tells you whether $T$ (reaction-limited) or flow/pressure (transport-limited) controls the film.

Cross-links to revisit if any answer felt shaky: Chemical vapor deposition (CVD) · Arrhenius Equation · Boundary Layer (Fluid Dynamics) · Step Coverage & Conformality · Physical Vapor Deposition (PVD) · Thin Film Metrology.

Question bank — Chemical vapor deposition (CVD)

True or false — justify

Spot the error

Why questions

Edge cases