Materials Science · Size, surfaces and nanoscale behaviour

Nanoparticles, nanowires and thin films

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

Describe nanoscale objects through geometry, surfaces and support.
Distinguish a particle core, surface layer and cluster, and state what a reported size actually describes.
Use number and volume accounts to explain why a small population of larger particles can matter.
Explain how wire–wire junctions and film–substrate boundaries can change an assembled specimen’s response.
Compare matched quantities and choose evidence that separates an interface or support effect from a changed interior material.

A particle, wire and film have different boundaries

Nanoparticles have small dimensions in several directions; nanowires are elongated; thin films are supported or freestanding layers. Surface chemistry, substrate constraint and interface area differ among them. Their dimensions alone do not specify the full material state.

Consider three specimens containing the same nominal chemical phase: dispersed particles, long wires crossing one another, and a continuous layer on a substrate. An electron travelling through the wire assembly must cross junctions that are absent from an isolated wire. A signal from the supported layer may include the substrate. The chemical name and smallest dimension do not tell us which whole specimen was measured.

Morphology means shape and organisation. For a particle, record its dimensions and shape; for a wire, diameter and length; for a film, thickness, lateral extent, continuity and support. A thin deposited layer can consist of disconnected islands, touching grains or a continuous solid. A wire can contain grain boundaries, an oxide or a molecular coating. None of these arrangements is specified by one diameter.

The geometric labels zero-, one- and two-dimensional usually count the long directions: none for a compact particle, one for an elongated wire and two for a thin sheet or film. Every object still occupies three-dimensional space. These shape labels do not prove electronic quantum confinement. Confinement and low-dimensional structures explains the separate comparison with wavelengths and energy scales; here we follow the object’s actual boundaries and connections.

Describe the specimen before comparing a property. These are geometric categories, not guaranteed transport regimes.
Form	Dimensions to retain	Boundaries and organisation that can matter
Compact nanoparticle	Several small dimensions; actual shape and distribution	Core–coating and coating–medium interfaces; isolated particles or clusters
Nanowire	Diameter, length and aspect ratio = length/diameter	Side surface, end contacts, internal defects and wire–wire junctions in an assembly
Thin film	Thickness and lateral extent; continuity	Two outer faces when freestanding; a buried interface when supported; islands, grains or continuous layer

The core is only one part of a dispersed particle

A ligand is a molecule or molecular group attached to a surface. A ligand layer can alter surface chemical reactivity and how particles interact with a liquid or with each other. Charged surface groups can produce repulsion between similarly charged particles; ions in the liquid can screen that interaction. Overlapping molecular layers can also resist close approach through steric exclusion: the layers occupy space and lose accessible molecular arrangements when crowded. The outcome depends on coverage, molecular structure and medium, not just the name “coated”.

A well-dispersed suspension contains separated particles moving through the medium. Clustering joins several particles into a larger object without necessarily growing each crystalline core. Dissolution and reprecipitation can instead change core sizes; fusion can erase boundaries between touching cores. A larger measured object therefore does not uniquely mean that individual crystals grew. A stabilising ligand can also impede a desired particle–particle electrical contact. Improved dispersion is a particular response, not a guarantee of improved conduction after assembly.

Use an original ideal geometry: a spherical core has diameter d = 10.0 nm and a uniform shell of radial thickness δ = 2.00 nm. The outside diameter is d_out = d + 2δ = 14.0 nm, because the layer occurs on both sides. The core volume fraction within that outer sphere is f_core = d³/(d + 2δ)³ = (10/14)³ = 0.364431, or 36.4431%. The remaining 63.5569% is shell volume in this stipulated geometry. These are volume fractions, not mass fractions; core and shell densities have not been supplied. The circular cross-section in the figure is not an area representation of those volume percentages.

An electron-microscopy image may resolve the dense core much more clearly than a low-contrast organic shell. Dynamic light scattering instead estimates a hydrodynamic diameter from diffusion through a liquid: the diameter of an equivalent sphere with the same diffusion rate under its model. Solvation, surface layers, shape, interactions and clusters can matter. It is not automatically a direct core diameter. A 14 nm hydrodynamic result beside a 10 nm core image can be consistent with a coating, but the two numbers alone do not prove this uniform 2 nm shell. They can also sample different populations or preparation states.

A number mean can hide most of the solid volume

Now use a separate fictional population whose complete core-size record is supplied: 99 spherical cores have diameter 10.0 nm and one has diameter 50.0 nm. All have the same phase and density. Ignore coatings for this account and assume every particle is counted once. These are exact teaching inputs, not an experimental sampling claim.

The number mean gives each particle equal weight: mean(d)_number = (99 × 10 + 1 × 50)/100 = 10.4 nm. Only 1% of the particles are in the larger group. Yet sphere volume is proportional to d³. The larger particle has (50/10)³ = 125 times the core volume of a small one. Its fraction of total core volume is 50³/(99 × 10³ + 50³) = 125000/224000 = 0.558036, about 55.8%. Because density is common, the same fraction applies to core mass in this example.

The number mean is correct, but it answers a different question from the volume account. If a response scales with core mass, the one larger particle cannot be treated as a negligible contribution merely because it is rare by number. A reaction confined to accessible surfaces, a magnetic signal, or an optical signal can have another weighting and mechanism. Do not replace their physics by volume weighting simply because that calculation is available.

Report what the size statistic represents and how the population was observed. Images of a finite selected field can miss another population or merge overlapping objects. A light-scattering distribution is weighted through a scattering and diffusion model; it is not the supplied count histogram. Transforming it to a number distribution introduces additional assumptions. Microscopy, diffraction and structure evidence explains why contrast, sampled fields and measured quantities belong with the interpretation.

Original ideal geometry and complete fictional particle count, not experimental data. The upper circular drawing is a cross-section of a spherical 10 nanometre core with a uniform 2 nanometre radial shell: the outside diameter is 14 nanometres. Core and outside radii are drawn in the exact ratio 5 to 7, but the following volume bar uses cubes, not projected circle areas. Blue core occupies 36.4431 percent and hatched shell 63.5569 percent of the ideal outer volume. The lower count contains exactly 99 blue circles representing 10 nanometre spherical cores and one orange-outlined circle representing a 50 nanometre core; their drawn diameters have ratio one to five. Position in the count display does not describe a real dispersion. The number mean is 10.4 nanometres. Blue small cores make up 99 percent of number but 44.1964 percent of total core volume. The single larger core makes up 1 percent of number but 55.8036 percent of core volume. The number and volume bars run from zero to 100 percent at 4.8 pixels per percentage point. Density is common, so core mass has the same fractions as core volume here. Hatched orange regions identify the shell or larger group within their separately labelled panels; they are not a scattering intensity or hydrodynamic measurement.

An assembly must connect the wires

A long wire can provide an internal conducting route, but a wire network needs connected routes from one electrode to the other. Alignment, wire length, gaps and junction quality alter those routes. A crossing visible in a projection may still be separated by a coating or may lie at a different depth. Touching geometry and a low electrical junction resistance are different evidence.

To isolate the connection effect, stipulate a fictional low-field, isothermal ohmic network with two independent identical paths in parallel between ideal electrodes. Each path contains two wire segments of 50.0 Ω each and one wire–wire junction of 400 Ω. There are no other paths, leakage, shared internal segments or electrode contact drops. The rectangles in the figure represent circuit contributions, not physical wire lengths or a random-network geometry.

A single path has R_path = 50 + 400 + 50 = 500 Ω. Both paths see the same applied voltage. Their currents add, so the network conductance is G = 1/R_path + 1/R_path = 2/500 S, and R_network = 1/G = 250 Ω. Conductance G = I/V is measured in siemens, S = A/V; it is not the material conductivity σ in S/m. At 0.100 V the whole network carries 0.400 mA. In either path the junction accounts for 400/500 = 80% of the voltage drop.

Suppose only the junction resistance changes to 40.0 Ω. The path becomes 140 Ω and the network 70.0 Ω; its conductance rises by 250/70 = 3.57143 times. The wire interiors are unchanged. Halving each wire-segment resistance instead, while retaining the 400 Ω junction, gives 450 Ω per path and 225 Ω for the network: only an 11.1111% conductance increase. A high-conductivity individual wire does not remove the need to cross a poor junction.

Actual 2024 work on nanomaterial networks combined size-dependent direct-current comparisons with impedance measurements to distinguish particle and junction contributions under its model. It illustrates why independent junction information matters. Its fitted material values are not the invented numbers here, and the two-path construction is not its full porous-network model. A real network can contain shared paths, uneven junctions and non-ohmic or temperature-dependent regions. A lower measured network resistance alone cannot locate the improvement.

Original exact two-path circuit model of a nanowire assembly, not a random network or a drawing of wire lengths. Two ideal electrodes share a voltage of 0.100 volt. Two independent identical paths connect them. Each path contains a blue rectangular 50 ohm wire segment, an orange-hatched rectangular 400 ohm junction and another blue 50 ohm wire segment. Each path is 500 ohms; the two paths together are 250 ohms. At 0.100 volt each path carries 0.200 milliampere and the total is 0.400 milliampere. The voltage-allocation bar shows 10 millivolts on the first wire, 80 on the junction and 10 on the last wire, with lengths proportional to voltage drop. The lower resistance bars start at zero, with 1.6 pixels per ohm and an axis from zero to 300 ohms. Original network resistance is 250 ohms. Changing only each junction to 40 ohms gives 70 ohms overall. Changing only each wire segment to 25 ohms with the original junctions gives 225 ohms overall. These are separate changed cases, not sequential processing or measurements. Hatched junctions and solid wire contributions distinguish component types beyond colour. The result depends on the supplied number and independence of paths, ideal electrode contacts and fixed isothermal ohmic values.

A supported film is a coupled specimen

A freestanding film has no substrate contact over its free area; both faces meet the surrounding medium. A supported film instead has a buried interface. Bonding can transfer in-plane deformation; a film that would expand freely may be prevented from following that free shape. If an attached thin film expands more than a much thicker substrate would over the same temperature rise, a well-bonded approximately flat interior can be compressed relative to its free state. Slip, cracking, delamination, curvature and structural relaxation can change that result. Thermal expansion and thermal stress explains the difference between free expansion and constraint; no universal film stress or failure follows from thickness alone.

A substrate can also contribute directly to a measurement without changing the film. Use a separate exact fictional electrical comparison at one fixed temperature. A continuous uniform film has thickness t = 10.0 nm, length L = 1.00 mm and width W = 1.00 mm. It has supplied film conductance G_f = 0.500 mS, where 1 mS = 10⁻³ S. The film and substrate each span the same pair of ideal electrodes and share the same uniform in-plane potential gradient. Their interface is assumed not to introduce another channel or transfer limitation. Contacts are ideal, the state stays isothermal and no other current flows.

The equal voltage makes these two paths parallel: I_total = I_f + I_s = (G_f + G_s)V, hence G_total = G_f + G_s. Support A contributes G_s,A = 0.0500 mS; support B contributes G_s,B = 2.00 mS. The total conductances are 0.550 and 2.50 mS, although G_f is identical. At the same 0.100 V, currents are 0.0550 and 0.250 mA. Neither total divided by the film’s dimensions is automatically a film property.

For the uniform film alone, G_f = σ_f Wt/L, so σ_f = G_f L/(Wt) = 5.00 × 10⁴ S/m. Assigning all of support A’s total to the film would give an apparent 5.50 × 10⁴ S/m; assigning support B’s would give 2.50 × 10⁵ S/m. The changed apparent ranking is a current-path effect in this example, not evidence of different film composition or quantum confinement.

A matched bare support can constrain its contribution, but subtraction needs a transfer assumption. If the support retains its bare-state conductance after coating, subtracting 0.0500 or 2.00 mS recovers 0.500 mS. If deposition changes the support or interface, it need not. For instance, the same 2.50 mS coated total could contain a 0.700 mS film plus a changed 1.80 mS support. Subtracting the former 2.00 mS bare value would still return 0.500 mS while misassigning the actual film. A control is informative through its physical model, not through subtraction alone.

Original hypothetical in-plane parallel-path comparison. Film thickness and true film conductance are common; all contacts, geometry and states satisfy the stipulated model.
Support	G_f (mS)	G_s (mS)	G_total (mS)	Apparent σ if all assigned to film (S/m)
A	0.500	0.0500	0.550	5.50 × 10⁴
B	0.500	2.00	2.50	2.50 × 10⁵

Original hypothetical in-plane parallel-path film/support comparison at fixed temperature, not experimental data. The cross-section shows a continuous film on a much thicker substrate, with thicknesses and contacts schematic rather than to scale. Two ideal electrodes access both channels and impose the same in-plane voltage. A solid arrow identifies the film path and a dashed arrow the support path. Film thickness is 10 nanometres and both lateral dimensions are 1 millimetre. Its conductance is 0.500 millisiemens in both supplied unchanged-support cases. Support A contributes 0.0500 millisiemens, giving total 0.550 millisiemens. Support B contributes 2.00 millisiemens, giving total 2.50 millisiemens. The stacked conductance bars start at zero and use 192 pixels per millisiemens, with solid blue film contributions and orange-hatched support contributions; their axis runs zero to 2.50 millisiemens. The film conductivity alone is 50000 siemens per metre under the defined uniform model. The lower alternative bar is a distinct hypothetical decomposition of the same 2.50 millisiemens total: a 0.700 millisiemens film plus a changed 1.80 millisiemens support. Subtracting the former bare support value of 2.00 millisiemens would not reveal that change. This counterexample does not assert that either actual support changed; it shows why a control needs a justified transfer model.

Preparation changes the object

Ligands, oxides, solvents and supports can affect stability and measured response. Report size distribution and morphology rather than only a mean diameter. A substrate-supported film measurement may include the substrate’s electrical, mechanical or thermal contribution.

Choose a comparison that makes the alternatives predict different observations. For a coated particle, pair core-sensitive structural dimensions with an independently constrained surface or solution-state response; drying can itself alter clusters. For wires, distinguish a connected topology from the resistance of its junctions and interiors. For films, identify whether the support changes the film’s state, contributes a parallel signal, or both. An apparently identical nominal size does not hold those variables fixed.

The relevant response also needs its own denominator and direction. Conductance of an assembled network, conductivity inferred for a defined uniform film, core mass fraction and hydrodynamic diameter are different quantities. A photograph or a reproducible total signal does not silently convert one into another. From measured transport to device behaviour follows actual voltage spans and operating state; Specimens, dimensions and test conditions develops controlled specimen comparisons.

Interpretation choices for supplied observations; these are discriminating questions, not a laboratory procedure or a guarantee that one method identifies the cause.
Observation	Alternatives still consistent with it	What a separating comparison must establish
Larger hydrodynamic size; similar resolved cores	Changed shell/solvation, clustering, or different sampled populations	Core structure, surface/medium state and population identity under matched conditions
Lower wire-network resistance	Changed junctions, interiors, connected paths or electrode contacts	Which paths and voltage drops changed while the other contributions were held or measured
Different supported-film signal at the same nominal thickness	Changed film state, support contribution, interface or conducting thickness	Current/signal path and whether the support control transfers to the coated specimen

Worked examples

Work through the case

Two 10 nm films have the same composition but different substrates. Must their measured response match?

Compare strain and interfacial chemistry.
Check substrate contributions to the method.
Assess the film’s actual structure.

No. The supported system can differ.

Follow the same particles with two different weights

For the supplied 99 cores of 10 nm and one core of 50 nm, compare the number mean diameter with the large-particle fraction of total core mass. What has and has not been established?

Each particle counts once in the number mean: (99 × 10 + 50)/100 = 10.4 nm. The large-particle number fraction is 1/100 = 1%.
The common factor π/6 in every spherical core volume cancels from the fraction. Small-core total is proportional to 99 × 10³ = 99000 nm³; the large core is proportional to 50³ = 125000 nm³.
The large fraction is 125000/(99000 + 125000) = 0.558036. Equal core density makes this also 55.8036% of core mass.
The supplied complete population establishes this mass account only. It does not prove that a sampled real image counted every particle, that the larger object is a fused core rather than a cluster, or that any optical/electrical response follows mass.

The number mean is 10.4 nm, but the one larger core contains about 55.8% of core mass under the stated shape and equal-density assumptions.

Change the junction before ranking the wire interiors

For the two-path network, compare lowering each 400 Ω junction to 40 Ω with halving each 50 Ω wire-segment resistance. Keep the other contributions fixed in each separate change.

Original path resistance is 2 × 50 + 400 = 500 Ω. Two independent equal paths in parallel give 250 Ω.
Junction-only change gives 2 × 50 + 40 = 140 Ω per path, hence 70 Ω overall. At the same voltage, the current and conductance increase by 250/70 = 25/7 ≈ 3.57143.
Interior-only change gives 2 × 25 + 400 = 450 Ω per path, hence 225 Ω overall. Conductance changes by 250/225 = 10/9: an increase of 11.1111%, not a doubling.
The larger benefit belongs to this junction-dominated stipulated network. It neither ranks all nanowire materials nor proves that a real treatment changed only the junctions.

The two separate changes give 70 and 225 Ω. A junction-limited assembly can respond much more to its connections than to a large improvement of its interiors.

Recover the quantity that belongs to the film

Two otherwise matched 10 nm films have total in-plane conductances 0.550 mS on A and 2.50 mS on B. Under the explicitly unchanged-support parallel model, recover the film conductance and conductivity. What additional assumption does a bare-support subtraction need?

The corresponding unchanged supports contribute 0.0500 and 2.00 mS. Subtracting each from its own coated total gives G_f = 0.500 mS in both cases.
Convert t = 10 nm = 10⁻⁸ m, and W = L = 10⁻³ m. Then σ_f = 0.500 × 10⁻³ × 10⁻³/(10⁻³ × 10⁻⁸) = 5.00 × 10⁴ S/m.
The uncorrected apparent conductivities would be 5.50 × 10⁴ and 2.50 × 10⁵ S/m. They assign the substrate current to a fictitious film-only section.
The subtraction requires the same support conductance, electrode access and applicable parallel-path model in bare and coated states. A deposition-induced support/interface change can defeat that transfer.

Both films give 0.500 mS and 5.00 × 10⁴ S/m only within the supplied model. The total signal and a bare-support number alone cannot prove that model transfers.

Try the reasoning

Does a mean particle size reveal rare large aggregates?
Show the answer and reasoning
No.
A distribution and sampling method are needed.
Why report surface ligands?
Show the answer and reasoning
They alter interfaces and interactions.
The bare-core model may be inappropriate.
Keep a 10.0 nm spherical core but change the uniform radial shell from 2.00 to 1.00 nm. Find the new outer diameter and core volume fraction. Does that percentage equal a measured core mass fraction or identify the hydrodynamic diameter without further assumptions?
Show the answer and reasoning
The outside diameter is 12.0 nm and the ideal core volume fraction is (10/12)³ = 57.8704%. Neither further inference follows automatically.
A radial layer occurs twice in the diameter. Core and shell densities would be needed for a mass fraction, and diffusion depends on the actual solvated/interacting object. The outer geometric diameter is a supplied shape, not an independently measured hydrodynamic result.
A new complete fictional count has 98 spherical 10 nm cores and two spherical 40 nm cores of the same density. Find the number mean diameter and the fraction of core mass in the larger group.
Show the answer and reasoning
The number mean is 10.6 nm. The larger group has 128000/226000 = 56.6372% of core mass.
The mean is (98 × 10 + 2 × 40)/100. Volume weights are 98 × 10³ and 2 × 40³, and their common sphere factor and density cancel. A number mean close to 10 nm again coexists with most core mass in a small number of larger particles.
In the original 250 Ω two-path network, one junction becomes electrically open while all wire interiors and the other path remain unchanged. What resistance is observed? Would excellent individual wire conductivity recover the lost parallel path by itself?
Show the answer and reasoning
Only one 500 Ω path remains, so the network resistance doubles to 500 Ω. The disconnected path does not reappear by improving its wire interiors alone.
An open junction carries no current in the stipulated circuit. Finite conductance of the remaining wire segments does not bridge that gap. This is a topology change, not a claim that the material resistivity doubled.
A separate network has three independent equal paths, each containing three 20 Ω wire segments and two 90 Ω junctions. Find its resistance, then the resistance if only both junctions in every path become 10 Ω.
Show the answer and reasoning
The original paths are 240 Ω, giving 80.0 Ω overall. The changed paths are 80.0 Ω, giving 26.6667 Ω overall.
Add along each path, then add the three parallel conductances. The conductance rises threefold. The number of segments and junctions must belong to the actual supplied route; one cannot transfer the earlier two-path formula without changing that count.
A coated specimen gives 2.50 mS, and its corresponding bare support previously gave 2.00 mS. Can these observations distinguish a 0.500 mS film on an unchanged 2.00 mS support from a 0.700 mS film on a changed 1.80 mS support?
Show the answer and reasoning
No. Both accounts give the same coated total; bare-support subtraction alone returns 0.500 mS without identifying which account applies.
The second account changes the support between observations. Additional independent information must constrain that change and any interface current, while preserving the relevant film state. A reproducible subtraction does not by itself establish component identity.
A well-bonded film would expand more than its much thicker substrate over a temperature rise. In the approximately flat interior with negligible slip, is the film’s constrained in-plane deformation larger or smaller than its free expansion? Does the sign alone give a stress magnitude or explain an electrical signal?
Show the answer and reasoning
It is smaller than the film’s free expansion, giving compressive mismatch relative to that free state. The sign gives neither a stress magnitude nor a unique electrical explanation.
The substrate imposes its smaller expansion in this stipulated limit. Elastic constants, actual geometry and relaxation are needed for stress, while electrical transport also depends on carriers, defects, contacts and current paths. Other constraints or curvature can require a different model.

What to carry forward

Describe geometry, distribution and boundaries.
Treat supports and surface layers as part of the specimen.
Distinguish core size, coating thickness, hydrodynamic size and clusters before comparing their numbers.
A number distribution and a mass or volume account weight the same particles differently.
Wire interiors, junctions and connected paths are separate parts of an assembly’s response.
A support can change the film’s state and can also contribute directly to the measured signal; a matched control needs a valid transfer model.

References

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