Materials Science · Size, surfaces and nanoscale behaviour

Confinement and low-dimensional structures

Useful background: Atoms, bonding, structure and defects; Measuring and checking material claims. Each worked case explains the quantities it uses.

Back to the chapter

After reading, you can

Explain low-dimensional behaviour relative to a relevant physical length.
Explain how boundary conditions select standing-wave states and how their energy spacing depends on width in a stated ideal model.
Distinguish the number of confined directions from the number of effectively extended directions in wells, wires and dots.
Compare a calculated energy spacing with supplied thermal and broadening scales without treating that comparison as proof of an observed quantum effect.
Separate electronic confinement from boundary scattering of heat carriers and choose evidence that can distinguish the mechanisms.

Dimensions can restrict motion

A thin film, wire or small particle can confine carriers or other excitations. Quantum confinement becomes relevant when dimensions approach characteristic electronic length scales. Mechanical and thermal size effects involve their own scales and mechanisms.

A five-nanometre film can be very broad in its plane. Its electrons need not respond in the same way through its thickness and along its plane. Nor must a change in its heat conduction have the same origin as a change in its electronic spectrum. “Small” describes geometry; a useful explanation names the motion, boundaries and physical scale involved.

An allowed electronic state is a quantum arrangement the system can support, whether or not an electron occupies it. Boundaries can change the allowed states. Occupation, scattering and how the state couples to a measurement are additional questions. Charge carriers and electronic energy bands supplies this allowed-versus-occupied distinction; here we make one boundary restriction concrete.

A boundary condition selects which wave shapes fit

Use an ideal particle of electron mass m_e moving in one coordinate z, between z = 0 and z = a. Potential energy is zero inside and infinite outside. Infinite barriers mean the particle cannot penetrate them, so its wave amplitude must vanish at both ends. This is a homogeneous, noninteracting model that ignores the actual atomic lattice. It is not a literal description of every thin film.

A standing wave has a spatial shape whose nodes stay fixed. Here the permitted shapes are proportional to u_n(z) = sin(nπz/a), with n = 1, 2, 3, … . The positive integer n counts half-wavelengths fitting between the ends: a = nλ_n/2, hence λ_n = 2a/n. Sin is the sine function and its argument is in radians. The choice n = 0 makes the amplitude zero everywhere, so it cannot represent a particle in the box.

The amplitude can have either sign. Probability density is proportional to its squared magnitude, u_n², and is never negative; these dimensionless shapes have not been normalised as probability densities. The plotted wave is neither an electron’s path nor the height of a bent film. For n = 2, there is an additional node at z = a/2 even though the particle is still confined over the entire interval.

The wave-momentum relation gives momentum magnitude p_n = h/λ_n, where h is Planck’s constant. With nonrelativistic kinetic energy E = p²/(2m_e), substitution gives E_n = h²n²/(8m_e a²). Both the wavelength and the energy therefore follow from the boundary condition; a smaller interval forces shorter wavelengths and larger energy spacings.

For the examples, round α = h²/(8m_e) to 0.3760 eV nm², so E_n = αn²/a² when a is in nm and energy in eV. One electronvolt is 1.602176634 × 10⁻¹⁹ J; a nanometre is 10⁻⁹ m. The lowest two energies are E₁ = α/a² and E₂ = 4α/a², so ΔE₂₁ = E₂ − E₁ = 3α/a². All calculations below use this declared rounded α.

E₁ is measured above this model’s zero potential, not an observed semiconductor band edge. ΔE₂₁ is a single-particle level spacing, not automatically a band gap, absorption peak or emitted photon energy. In a semiconductor, an effective band mass can replace m_e: it describes how energy changes with momentum in that band, rather than an atom’s mass. Finite barriers let the amplitude extend into neighbouring material and change how wave shapes match at the boundary. Electron–hole interactions and selection rules for coupling to light can also matter. A match to the inverse-square trend alone does not verify those assumptions.

Count the directions that remain effectively extended

A quantum well confines the chosen carrier in one direction and leaves two effectively extended; it is called electronically two-dimensional. A quantum wire confines two and leaves one extended, so it is one-dimensional. A quantum dot confines all three, giving discrete states in every direction and the label zero-dimensional. These names count effectively extended motion, not the physical number of dimensions occupied by the object.

A geometrically thin film or narrow wire needs actual confining boundaries and relevant energy scales before those electronic labels apply. In a finite specimen every direction can have discrete states in principle, but the spacings along a very long direction may be too dense to matter for the comparison. Shorter carrier wavelengths, a different band mass or weak barriers can change the interpretation at the same geometric size.

For a separable rectangular infinite box with the same mass in each direction, the energies add: E = α(n_x²/L_x² + n_y²/L_y² + n_z²/L_z²). The positive whole numbers n_x, n_y and n_z label independent standing-wave shapes. Raising only one index from 1 to 2 costs 3α/L_j² in that direction. This additive model excludes coupling between coordinates and is not a general atomistic Hamiltonian.

For an illustrative film with L_z = 5.00 nm and L_x = L_y = 1000 nm, the first excitation in z costs 0.04512 eV. The corresponding x or y excitation costs 1.128 × 10⁻⁶ eV, forty thousand times smaller. It is therefore possible for thickness modes to be distinct while the in-plane spacings behave as an almost continuous set over a much broader energy range. State the energy and measurement range when saying “extended”.

Electronic dimensionality in a specified confinement regime. Every real object still has three spatial dimensions; the labels do not follow from shape alone.
Model structure	Confined directions	Effectively extended directions
Quantum well	One: thickness	Two: in the plane
Quantum wire	Two: transverse section	One: along the wire
Quantum dot	Three	Zero

Original ideal-box teaching diagram. Three separately offset panels show dimensionless amplitude shapes sin(nπz/a) for n one, two and three, with equal displayed amplitude scale and zero at both barriers. They contain one, two and three half-wavelengths respectively; the wavelengths are 2a, a and 2a/3. The n two and three curves have internal nodes, and negative amplitude is not negative probability. These are stationary wave shapes, not electron trajectories or film displacements, and are not normalised probability densities. The lower projected boxes distinguish a well, with z confined and x,y effectively extended; a wire, with y,z confined and x effectively extended; and a dot, with x,y,z confined and no effectively extended direction. Dashed orange edges are the short confined directions; solid blue edges are the assumed extended directions. Dimensions are schematic, and shape alone does not establish electronic confinement. The labels 2D, 1D and 0D count effectively extended electronic motion over a stated energy range. Every object still occupies three-dimensional space. No actual nanostructure image or measurement is represented.

Compare the spacing with the energy scales of the question

In the one-coordinate model, a = 5.00 nm gives E₁ = 0.01504 eV and E₂ = 0.06016 eV, hence ΔE₂₁ = 0.04512 eV. Doubling a to 10.0 nm divides both energies and their difference by four: ΔE₂₁ becomes 0.01128 eV. Halving a does not merely double the spacing. A measured result need not follow this law if material state, barrier or band description changes with size.

Thermal occupation changes over an energy scale k_B T, where k_B is Boltzmann’s constant and T is absolute temperature. Use k_B = 8.617333262 × 10⁻⁵ eV/K, rounded for these calculations. Then k_B T is 0.0258520 eV at 300 K and 0.00663535 eV at 77 K. It is an energy scale, not a statement that every electron has energy k_B T; a degenerate electron population needs its actual occupation description.

For a deliberately limited comparison, supply broadening scale Γ = 0.0100 eV and call a spacing promising for further resolution tests if ΔE₂₁ ≥ max(k_B T, Γ). Max means choose the larger of the two scales. This is a declared screening rule, not a sufficient spectral-resolution theorem, a probability of seeing a line or a definition of whether the states exist. Lifetimes, inhomogeneous sizes, detector response, occupation and coupling can all matter.

At 300 K the scale selected by this rule is 0.0258520 eV. The 5 nm model spacing passes, while 10 nm does not. At 77 K the supplied Γ dominates, so the comparison is with 0.0100 eV; both pass. Solving 3α/a² ≥ E_scale gives a ≤ √(3α/E_scale): about 6.60553 nm for the 300 K rule and 10.6207 nm for the 77 K rule. There is no universal nanometre cutoff here; it depends on mass, boundary and the selected energy scales.

The standing-wave example makes the wavelength link explicit: the lowest mode has wavelength 2a. For a different specified kinetic energy E*, the same free-particle model gives λ = h/√(2m_e E*), so different occupied energies imply different relevant wavelengths. Geometry should be compared with the actual electronic description, not only with a generic “nano” label.

Ideal-box first excitation ΔE₂₁, using α = 0.3760 eV nm². These are model calculations, not measured spectra. The screen uses the supplied Γ = 0.0100 eV.
Width a	ΔE₂₁ (eV)	300 K screen	77 K screen
5.00 nm	0.04512	Passes	Passes
10.0 nm	0.01128	Does not pass	Passes
50.0 nm	0.0004512	Does not pass	Does not pass

A shorter scattering path is a different size mechanism

A mean free path is a characteristic distance travelled between the scattering events relevant to the chosen transport process. It is different from wavelength. Boundaries can redirect carriers and change transport even when electronic level spacings remain small compared with the relevant energy range. Surface roughness, specularity and interfaces affect how strongly a boundary scatters; merely reaching a surface is not a universal fixed loss.

Phonons are quantised collective lattice vibrations. They can transport thermal energy without being electronic charge carriers. In a simple isotropic kinetic model, lattice thermal conductivity is κ = (1/3) C_V vℓ, where C_V is heat capacity per volume, v is a representative propagation speed and ℓ is a representative heat-transport mean free path. The dimensions are [J/(m³ K)] × [m/s] × [m] = W/(m K). This single-speed, single-path description compresses a real spectrum of modes.

For a separate fictional heat-carrier record, stipulate C_V = 1.50 × 10⁶ J/(m³ K), v = 3000 m/s and bulk ℓ_bulk = 100 nm. Take independent bulk and diffuse-boundary scattering rates with an assigned boundary path d, so 1/ℓ_eff = 1/ℓ_bulk + 1/d. The reciprocal paths have units of inverse length. This is a supplied rate-addition approximation with geometry factor one, not a universal rule that a real wire’s path equals its diameter.

With d = 50.0 nm, ℓ_eff = 33.3333 nm and κ = 50.0 W/(m K), compared with bulk-model 150 W/(m K). At d = 200 nm, ℓ_eff = 66.6667 nm and κ = 100 W/(m K). This difference follows from added scattering while holding C_V and v unchanged; no electronic level quantisation or change of bonding was needed in the calculation. A long, steady heat-flow gauge and the stipulated effective-path approximation are assumed; a ballistic or strongly mode-dependent problem needs another description.

At the same numerical 50 nm scale, the separate ideal electron-box spacing is only 0.0004512 eV and misses both supplied energy screens, while this phonon-path model has a large thermal change. These are two independently supplied teaching models, not a joint prediction for one identified material. They demonstrate why a thermal size effect does not by itself establish electronic confinement.

An actual 2012 study by Feser and colleagues compared thermal transport with the controlled roughness of silicon nanowire arrays. Its reported association between reduced thermal conductivity and Raman-line broadening did not supply a unique confinement explanation; the authors explicitly identified a mismatch with phonon-confinement models and left the broadening mechanism unresolved. Preserve that limitation instead of relabelling every size-dependent thermal observation as a quantum electronic effect.

Two original independent teaching models share a figure but do not predict one identified material. The upper plot uses ideal electron-box gap ΔE21=1.128/a² in electronvolts with a in nanometres. Width spans two to 50 nanometres linearly; gap has a logarithmic axis with equal intervals for factors of ten. At five nanometres the circle marks 0.04512 eV, at ten the square marks 0.01128 eV, and at 50 the diamond marks 0.0004512 eV. The dashed 300 K comparison level is 0.0258520 eV; the dotted 77 K level is 0.0100 eV because the supplied broadening Γ dominates there. These are declared screening levels, not measured lines or proof of observability. Crossovers are about 6.60553 and 10.6207 nm. The lower plot shows the separate stipulated isotropic phonon model with bulk mean free path 100 nm, representative speed 3000 metres per second and volumetric heat capacity 1.50 million joules per cubic metre per kelvin. Independent boundary scattering is assigned inverse path 1/d. Its thermal conductivity is 150d/(d+100) watts per metre per kelvin with d in nanometres. The points at d = 50, 100 and 200 nm give 50, 75 and 100 W/(m K); the dashed bulk-model reference is 150. Neither this single-path approximation nor a thermal conductivity decrease establishes electronic confinement, and neither curve is experimental data.

Nanometres are not a universal switch

For electronic quantum confinement, boundaries restrict allowed electronic states, so compare size with a relevant electronic wavelength or confinement length. A size comparable to a mean free path can instead change boundary scattering without establishing quantum confinement. Mechanical and thermal comparisons require their own relevant lengths. A material can show a size effect in one property while retaining nearly bulk behaviour in another. Identify the confinement direction rather than labelling every small object equivalent.

To test a confinement explanation, identify the relevant direction and electronic states, independently measure the geometry, and predict which energy differences should change under matched material and boundary conditions. Compare actual resolved features, not merely a colour change or a low transport value. Finite barriers, defects, contamination, strain and changing composition are consequential alternatives.

A thermal observation needs the heat carriers, surface/interface scattering and measurement geometry examined separately. Microscopy, diffraction and structure evidence explains why structural contrasts and line widths have more than one possible origin. Evidence improves when an added observation distinguishes those alternatives rather than just repeating the same signal at another size.

Worked examples

Work through the case

A film is very thin but broad in two directions. Is confinement necessarily equal in all directions?

Identify the short dimension.
Compare it with the relevant physical length.
The broad directions can remain effectively extended.

No. Dimensionality is direction- and mechanism-dependent.

Double the width and recalculate the spacing

Use the ideal electron box with α = 0.3760 eV nm². Compare widths 5.00 and 10.0 nm: find E₁, E₂ and ΔE₂₁, then apply the supplied Γ = 0.0100 eV screens at 300 K and 77 K. Have you predicted a measured absorption peak?

At 5 nm, divide α by 25: E₁ = 0.01504 eV and E₂ = 4E₁ = 0.06016 eV. Their difference is 0.04512 eV.
At 10 nm, divide by 100: E₁ = 0.003760 eV, E₂ = 0.01504 eV and ΔE₂₁ = 0.01128 eV. All are one quarter of the corresponding 5 nm values.
At 300 K the screen is max(0.0258520,0.0100) = 0.0258520 eV, so only 5 nm passes. At 77 K it is max(0.00663535,0.0100) = 0.0100 eV, so both pass.
The rule identifies a comparison worth testing under its assumptions. A real optical spectrum also needs state occupation, allowed transition coupling, line shape and material/barrier evidence; the calculated level spacing is not automatically an absorption peak.

The first spacing falls from 0.04512 to 0.01128 eV when width doubles. Which comparison passes depends on the supplied thermal and broadening scales; no measured spectrum has been predicted.

A rectangular wire can have different transverse gaps

Use the separable electron-box model with L_x = 5.00 nm, L_y = 10.0 nm and L_z = 1000 nm. Starting from indices (1,1,1), which one-index excitation costs least, and which transverse directions pass the Γ = 0.0100 eV screen at 77 K versus 300 K?

The x increment is 3α/5² = 0.04512 eV. The y increment is 3α/10² = 0.01128 eV. The z increment is 3α/1000² = 1.128 × 10⁻⁶ eV.
The least-cost one-index excitation is along the long z direction: (1,1,2). Among transverse excitations, (1,2,1) costs less than (2,1,1), because L_y is larger than L_x.
At 77 K the declared scale is 0.0100 eV; both transverse gaps exceed it, while z does not. At 300 K the scale is 0.0258520 eV; only the x increment exceeds it.
Geometric wire shape is unchanged. The low-dimensional approximation relevant to an observation depends on its energy and temperature range, not a requirement that every transverse spacing be equally conspicuous. This is still the ideal separable model.

The long-direction excitation is least costly. At 77 K both transverse gaps pass the supplied screen; at 300 K only the narrower x direction does.

Change heat transport without invoking an electronic gap

For the separate fictional phonon model, use C_V = 1.50 × 10⁶ J/(m³ K), v = 3000 m/s and ℓ_bulk = 100 nm. Compare boundary paths d = 50.0 and 200 nm using the stated rate-addition rule. What has this established about electronic confinement?

For d = 50 nm, ℓ_eff = 1/(1/100 + 1/50) = 33.3333 nm. Convert to metres before κ = C_Vvℓ_eff/3, giving 50.0 W/(m K).
For d = 200 nm, ℓ_eff = 1/(1/100 + 1/200) = 66.6667 nm, giving 100 W/(m K). Without the boundary term, the model gives 150 W/(m K).
The conductivity ratios are 1/3 and 2/3 of this bulk-model value. They follow from supplied independent scattering paths and unchanged C_V,v; boundary specularity or a changed mode spectrum could invalidate that approximation.
No electronic energy level, state occupation or optical transition was measured in this calculation. The separate electron-box result at 50 nm misses both energy screens, so the phrase size effect alone has not selected an electronic-confinement mechanism.

The conditional thermal conductivities are 50.0 and 100 W/(m K). They demonstrate a boundary-scattering mechanism in the toy model, with no electronic-confinement conclusion.

Try the reasoning

Does a 100 nm dimension prove quantum confinement for every material?
Show the answer and reasoning
No.
The relevant electronic scale varies.
Can size alter phonon transport without the same electronic effect?
Show the answer and reasoning
Yes.
The carriers and length scales differ.
In the one-coordinate infinite box, what wavelength fits n = 3 when a = 6.00 nm? Why is n = 0 not a zero-energy occupied state?
Show the answer and reasoning
λ₃ = 2a/3 = 4.00 nm. For n = 0 the amplitude vanishes everywhere, so there is no normalisable particle state.
Three half-wavelengths occupy the interval: 3(4/2) = 6 nm. A physical state needs nonzero total probability. The standing-wave shape is not the particle’s trajectory.
Change the ideal box width from 10.0 to 7.50 nm with α unchanged. Find the new ΔE₂₁ and its ratio to the old value. Does it pass the supplied 300 K screen?
Show the answer and reasoning
ΔE₂₁ = 0.0200533 eV, ratio 16/9 ≈ 1.77778. It remains below 0.0258520 eV, so it does not pass that screen.
Use 3α/a² = 1.128/56.25. The inverse-square ratio is (10/7.5)², not merely 10/7.5. Failure of this heuristic screen does not mean the ideal model’s discrete states cease to exist.
In the same mathematical box model, double the particle mass while keeping width 5.00 nm, barriers and all other assumptions fixed. What happens to ΔE₂₁ and to the 300 K screen?
Show the answer and reasoning
The spacing halves to 0.02256 eV and misses the supplied 0.0258520 eV screen.
α = h²/(8m) is inversely proportional to mass. The geometry alone therefore does not fix the energy spacing. A real semiconductor’s effective-band-mass description requires its own justified model; this exercise only changes a parameter in the supplied equation.
At 77 K, reduce the stipulated Γ from 0.0100 to 0.00200 eV. What scale now controls the same spacing screen? Find its largest allowed a with α = 0.3760 eV nm².
Show the answer and reasoning
k_B T ≈ 0.00663535 eV now dominates, giving a ≤ about 13.0384 nm.
Take the maximum, not the smaller or the sum: max(0.00663535,0.00200). Then a_max = √(1.128/0.00663535), using the unrounded k_B T for the displayed limit. Improving the supplied broadening below the thermal scale no longer changes this particular heuristic threshold; it is not a complete line-resolution theory.
In the fictional phonon model, let d = ℓ_bulk = 100 nm. Find ℓ_eff and κ. Does equality of size and mean free path also specify an electronic wavelength?
Show the answer and reasoning
ℓ_eff = 50.0 nm and κ = 75.0 W/(m K). No electronic wavelength follows.
Two equal reciprocal-path contributions add, halving the effective path. With C_V and v fixed, κ halves relative to 150 W/(m K). A mean free path describes scattering; electronic wavelength needs the electronic state or energy and its appropriate model.
A series of smaller particles changes colour and shows a broader structural line. Is the inverse-square electron-box explanation now established? Name one useful discriminating comparison.
Show the answer and reasoning
No. Independently check size, composition and boundary/state changes, then compare resolved electronic features with an appropriate size-dependent model under matched conditions.
Colour can include absorption, scattering and chemistry; a broadened structural line can reflect instrument response, strain or coherent extent. Neither alone supplies the specified electronic level spacing. A controlled spectral/structural comparison should predict different outcomes for a confinement explanation and a consequential alternative.

What to carry forward

Compare size with a relevant physical scale.
Specify direction and property.
Confining boundaries select permitted wave shapes; the infinite-box level spacing scales as 1/(ma²) under its explicit assumptions.
Well, wire and dot labels count effectively extended directions over a stated energy range, not the physical dimensions of matter.
Thermal energy, broadening, occupation and measurement coupling affect which discrete-state changes become evident.
Boundary scattering of phonons or electrons can change transport without establishing the same quantum-confinement effect.

References

Teaching resources and research supporting this lesson’s explanations and examples.