GLOBAL KNOWLEDGE ARK

Materials Science · Size, surfaces and nanoscale behaviour

Defects, contamination and measurement artefacts

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

  • Distinguish genuine nanoscale response from changed specimen state and artefacts.
  • Separate the conducting core, outer geometry and terminal-contact contributions in a bounded wire model.
  • Interpret matched exposure, imaging-history and background controls without claiming a unique microscopic cause.
  • State which extra observations would test a proposed surface or defect mechanism.
01

A small amount can dominate a small object

Surface contamination, adsorbed water, oxidation or electron-beam damage can alter a nanoscale specimen. Contacts can contribute much of a measured electrical resistance. A claimed size trend may therefore combine intrinsic size dependence with preparation and measurement effects.

A measured difference can be real while its proposed explanation is wrong. An oxide-rich shell is part of the specimen’s actual state, even when it is unwanted. A contact contribution belongs to the defined device measurement. Calling either an artefact means that it distorts a particular intended inference; it does not make the observation imaginary. Define the target first: resistance of the complete contacted device, response of its conducting interior, or a size-dependent electronic mechanism.

A defect is a departure from a chosen ordered reference, such as a missing atom, a foreign atom or an interface with different local bonding. A trap is an electronic state that can capture and release charge. Defects, adsorbed species and nearby trapped charge can alter carrier populations, scattering or the electrostatic environment of a narrow conducting region. A fluctuating trap occupancy can therefore produce a fluctuating current even if the external diameter is unchanged. Doping, defects and interfaces develops the distinction between chemical amount, charged centres and mobile carriers.

There is no universal sign. A changed surface may deplete carriers, supply charge, add scattering sites or alter injection at a contact. In particular, a dielectric covering is not automatically an electrically improved surface: the new interface can introduce traps. A nominally identical treatment can also leave different coverage or history. Distinguish a material-specific mechanism from the general claim that surfaces matter.

The following dimensions, resistances, exposure records and count records are original hypothetical inputs. They do not specify a real metal or semiconductor. From measured transport to device behaviour supplies the contact and sensing-span background; Microscopy, diffraction and structure evidence explains why image contrast, chemical identity, scale and resolution answer different questions.

02

A measured outside diameter is not always the conducting diameter

Isolate a classical, diffusive, low-field wire model at fixed temperature. A uniform cylindrical conducting core has length L = 1.00 µm and effective resistivity ρ_core = 2.00 × 10⁻⁷ Ω m. Its concentric shell is insulating in this model; no parallel current flows through the support. The final outside diameter D and shell thickness h are independently supplied geometric quantities. The conducting diameter is d = D − 2h, because the shell occupies both sides of a diameter. These are final dimensions, not a model of oxide growth or material consumption.

The core area is A_core = πd²/4. For a uniform Ohmic core, R_core = ρ_core L/A_core = 4ρ_core L/(πd²). Ohmic here means voltage is proportional to current within the specified small range. One nanometre is 10⁻⁹ m, so a diameter in nanometres must be converted before using this SI expression. The model presumes a local resistivity description; it is not a ballistic or coherent quantum-wire calculation.

Suppose the same h = 2.00 nm covers wires with outside diameters 40.0 and 20.0 nm. Their conducting diameters are 36.0 and 16.0 nm. Their core area fractions, A_core/A_outer = (d/D)², are 0.810 and 0.640. The same absolute shell thickness occupies a larger fraction of the smaller cross-section. Their conditional core resistances are about 196.488 and 994.718 Ω. The ratio is (36/16)² = 5.0625, although outside diameter alone would suggest a factor of four.

An analyst who measures R_core correctly but uses the outside area reports ρ_app = R_core A_outer/L. Substitution gives ρ_app = ρ_core[D/(D − 2h)]². Thus the reported values are about 2.46914 × 10⁻⁷ and 3.12500 × 10⁻⁷ Ω m. Apparent resistivity rises by 26.5625% as outside diameter halves in this comparison, even though the supplied core resistivity has not changed. The figure shows this conditional geometry effect, not measured material data.

Knowing D alone cannot decide between a smaller conducting core and a changed core resistivity: resistance contains their ratio. A nonconducting surface depletion region could also differ from a chemically identified oxide thickness. The measured electrical boundary and the structural or chemical boundary need not coincide. Real roughness, nonuniform shells, leakage, heating, carrier depletion and scattering can require a different model; do not silently assign every observed nanowire the constant ρ_core used here.

Conducting core versus outer areaOriginal conditional geometry and resistivity model, not measured material data. Upper transverse sections have a common length scale. A forty-nanometre outside diameter contains a thirty-six-nanometre conducting core; a twenty-nanometre outside diameter contains a sixteen-nanometre conducting core. Both have a two-nanometre radial insulating shell. Hatched outer rings distinguish shell from solid core. Core area fractions are 0.810 and 0.640. The lower graph plots apparent resistivity calculated using outside area, in units of ten to the minus seven ohm metres, against outside diameter from twenty to sixty nanometres. Supplied core resistivity is two in those units and remains constant. A solid blue curve with circles uses a two-nanometre shell; a dashed orange horizontal line with a square uses zero shell. The finite-shell curve decreases from 3.125 at twenty nanometres through about 2.46914 at forty nanometres toward 2.29592 at sixty nanometres. Both curves use a uniform one-micrometre diffusive Ohmic core and an insulating shell. These are conditional geometric predictions, not a universal nanoscale resistivity law.The outside outline can overstate conducting areaOriginal geometry · transverse sections at a common scaleD = 40.0 nmd = 36.0 nmCore area: 81.0% of outside areaD = 20.0 nmd = 16.0 nmCore area: 64.0% of outside areaHatched ring: h = 2.00 nm on each radial side.h = 2.00 nmh = 0 (reference)ρ_app (10⁻⁷ Ω m) · outside area used2.002.252.502.753.003.252030405060Outside diameter D (nm)ρ_core = 2.00 × 10⁻⁷ Ω m for every point.A changed inferred value need not be a changed core property.
Original conditional geometry and resistivity model, not measured material data. Upper transverse sections have a common length scale. A forty-nanometre outside diameter contains a thirty-six-nanometre conducting core; a twenty-nanometre outside diameter contains a sixteen-nanometre conducting core. Both have a two-nanometre radial insulating shell. Hatched outer rings distinguish shell from solid core. Core area fractions are 0.810 and 0.640. The lower graph plots apparent resistivity calculated using outside area, in units of ten to the minus seven ohm metres, against outside diameter from twenty to sixty nanometres. Supplied core resistivity is two in those units and remains constant. A solid blue curve with circles uses a two-nanometre shell; a dashed orange horizontal line with a square uses zero shell. The finite-shell curve decreases from 3.125 at twenty nanometres through about 2.46914 at forty nanometres toward 2.29592 at sixty nanometres. Both curves use a uniform one-micrometre diffusive Ohmic core and an insulating shell. These are conditional geometric predictions, not a universal nanoscale resistivity law.
03

Use the same span to separate a changed core from changed contacts

Consider a separate matched before/exposure dossier for the 40.0 nm outside-diameter wire. Length remains 1.00 µm. All readings use the same specified temperature and a verified local linear current range, with heating negligible. An ideal independent voltage measurement spans the entire same core length included in the terminal measurement. It draws negligible current and does not perturb the specimen. There is no extra unsensed wire or lead contribution. These conditions are stipulated, rather than guaranteed by attaching more probes.

Define R_terminal = V_terminal/I and R_core = V_core/I for the same imposed current and the same core span. Within this series model, R_terminal = R_core + R_contacts, where R_contacts combines the two terminal junction contributions. Their individual values are not separately identified. The table gives electrical observations rounded to 0.1 Ω and independently supplied core diameters. An inert time/handling control matches the baseline at that reporting precision; the reactive exposure has a smaller core and larger readings.

The inferred combined contact contribution is 236.5 − 196.5 = 40.0 Ω before exposure and in the inert control. It is 348.7 − 248.7 = 100.0 Ω after reactive exposure. The terminal rise of 112.2 Ω therefore comprises a 52.2 Ω core rise and a 60.0 Ω contact rise within the stipulated model. An explanation in which only the contacts changed cannot account for the independent rise across the same core span. Conversely, attributing the entire terminal rise to the core would also be wrong.

The core diameter changes from 36.0 to 32.0 nm. Constant ρ_core and L predict a resistance ratio (36/32)² = 1.265625 and a final R_core near 248.680 Ω, consistent with the rounded 248.7 Ω record. Reconstructing ρ_core from each measured core resistance and its own area gives about 2.00 × 10⁻⁷ Ω m in both cases. This shows that a changed conducting area can account for the core rise; it does not independently prove unchanged microscopic scattering.

In this fictional dossier, a separate structural/compositional observation resolves a smaller crystalline conducting core and an oxide-rich outer layer in the reacted set; the inert set does not show that change. That evidence supports investigating an exposure-induced surface change. It still does not isolate an atomic trap species or establish quantum confinement. The control narrows the explanations under matched conditions; finite resolution and any unmodelled probe or geometry effect remain relevant.

Original hypothetical matched exposure record. Common outside diameter 40.0 nm and core span 1.00 µm. Resistances are observations rounded to 0.1 Ω; the later geometric prediction is a separate consistency check.
RecordCore diameter d (nm)R_core (Ω)R_terminal (Ω)
Before36.0196.5236.5
Inert time/handling control36.0196.5236.5
Reactive exposure32.0248.7348.7
Matched exposure and same-span resistance componentsOriginal fictional matched exposure dossier, not published measurements. The upper two transverse sections have the same forty-nanometre outside diameter and common length scale. Before exposure the conducting core diameter is thirty-six nanometres; after reactive exposure it is thirty-two nanometres. Orange diagonal hatching identifies the nonconducting shell and blue outlines identify the core. The inert time and handling control retains the before geometry at reporting precision. The lower stacked horizontal bars show independently measured core resistance as solid blue and inferred combined contact resistance as orange with diagonal hatching. Before and inert control each have 196.5 ohms core and 40.0 ohms contacts, total 236.5 ohms. Reactive exposure has 248.7 ohms core and 100.0 ohms contacts, total 348.7 ohms. All voltage observations span the same one-micrometre uniform core with ideal noninvasive sensing, negligible heating and no extra unsensed wire or lead contribution. The terminal rise of 112.2 ohms divides into 52.2 ohms core and 60.0 ohms contacts. Separate composition and core-diameter observations are part of this stipulated dossier; the bars alone do not identify oxide chemistry or one defect mechanism.A matched control can separate contributionsSame outside diameter · independently supplied core geometryBeforeD = 40.0 nm; d = 36.0 nmReactive exposureD = 40.0 nm; d = 32.0 nmInert time/handling control: same as Before at reported precision.Same core span: 1.00 µm · ideal interior voltage sensingResistance (Ω) · core + inferred combined contacts0100200300400Before236.5Inert control236.5Reactive exposure348.7CoreCombined contactsBefore/control: core 196.5 Ω; contacts 40.0 Ω.After: core 248.7 Ω; contacts 100.0 Ω.112.2 Ω terminal rise = 52.2 Ω core + 60.0 Ω contacts.A contacts-only change cannot explain the interior observation.Different evidence is still needed for a unique microscopic cause.
Original fictional matched exposure dossier, not published measurements. The upper two transverse sections have the same forty-nanometre outside diameter and common length scale. Before exposure the conducting core diameter is thirty-six nanometres; after reactive exposure it is thirty-two nanometres. Orange diagonal hatching identifies the nonconducting shell and blue outlines identify the core. The inert time and handling control retains the before geometry at reporting precision. The lower stacked horizontal bars show independently measured core resistance as solid blue and inferred combined contact resistance as orange with diagonal hatching. Before and inert control each have 196.5 ohms core and 40.0 ohms contacts, total 236.5 ohms. Reactive exposure has 248.7 ohms core and 100.0 ohms contacts, total 348.7 ohms. All voltage observations span the same one-micrometre uniform core with ideal noninvasive sensing, negligible heating and no extra unsensed wire or lead contribution. The terminal rise of 112.2 ohms divides into 52.2 ohms core and 60.0 ohms contacts. Separate composition and core-diameter observations are part of this stipulated dossier; the bars alone do not identify oxide chemistry or one defect mechanism.
04

An agreement tests a model but does not identify every mechanism

Geometry, carriers and scattering can change together. If the exposed core’s resistivity instead rose by 25% while its diameter stayed at 36.0 nm, the core resistance would also rise by 25%. One resistance ratio cannot distinguish those cases. The additional diameter observation matters because it measures a different part of the explanation. Its uncertainty and the relation between the imaged and electrically sampled region matter too.

A particular trap mechanism needs more discriminating evidence than a higher resistance. For example, a changed time-dependent current or gate-sweep response can motivate a trapping hypothesis, while spatial composition or a resolved shell can test a chemical/geometric account. None is an automatic one-to-one fingerprint: nearby dielectric or support traps can influence the channel as well as defects in the wire itself. Explain which alternatives an observation can reject and which remain compatible.

Confinement and low-dimensional structures develops allowed-state restrictions and their characteristic scales. Surface scattering is a different mechanism, and the geometry calculation here requires no changed quantum levels at all. Agreement with an inverse-diameter trend, or with one confinement formula after fitting unknown parameters, does not by itself establish the physical boundary, carriers or measurement coupling assumed by that formula.

05

The act of observing can add another specimen history

An electron beam can transfer energy and charge to a specimen, alter bonds or activate defects. It can also change a surface by transforming adsorbed precursor molecules into deposits. Deposition and removal compete, depending on the material, precursor supply, vacuum and irradiation conditions; beam exposure does not have one universal effect. Published aluminium/gold film experiments also show that the beam can affect deformation and local geometry. Such evidence is a reason to compare observation histories, not a safe-energy rule for every specimen.

Keep electron number separate from exposure area. For a constant beam current I_b, actual unblanked time t_b and uniformly irradiated area A_b, charge is Q_b = I_b t_b. With elementary charge e = 1.602176634 × 10⁻¹⁹ C, incident electron number is N = Q_b/e and area fluence is Φ = N/A_b. Fluence counts incident electrons per area; it is not absorbed energy or a calculated defect concentration. Current outside the stated region, beam blanking and nonuniform scanning can change this simple account.

For four frames, each with 0.100 s of unblanked exposure at 1.00 pA over 100 × 100 nm, t_b = 0.400 s and Q_b = 4.00 × 10⁻¹³ C. The incident number is about 2.49660 × 10⁶ electrons, so Φ is about 249.660 electrons/nm². Delivering the same number over 50 × 50 nm quadruples the fluence to about 998.641 electrons/nm². Counting frames without area, current and actual exposure time would miss that change. Equal fluence still need not produce equal change at different energies, dose rates or specimen states.

Compare a repeatedly imaged region with a matched region receiving its first observation only at the final stage, while accounting for the final observation’s own exposure. A beam-off interval and a matched time/handling control can help distinguish elapsed time from irradiation history. A change localised to the previously illuminated region is evidence for a beam-related effect, but may reflect deposition, charging or structural change. Independent contrast, composition or response evidence is needed to distinguish those mechanisms.

A support can also contribute signal without changing the specimen. In a separate stipulated additive expected-count model, matched specimen-plus-support readings are 120 before and 180 after a condition change, while matched support-only readings are 80 and 140. Subtraction gives a specimen contribution of 40 in each condition. The raw total rose by 50%, but the entire rise is accounted for by the supplied background change. This arithmetic does not establish uncertainty, chemical identity or absence of all specimen changes. Real random counts need an uncertainty model, and a blank must match the relevant support, acquisition and response conditions.

06

Use discriminating controls

Vary contact geometry, exposure or measurement method where relevant. Inspect structure before and after testing when the method can change it. Record detection limits and avoid treating a clean-looking image as chemical proof of a pristine surface.

Choose the control for the causal question. For a contact-only claim, measure an appropriate independent interior voltage over a known matched span. For a surface-geometry claim, compare conducting dimensions and composition before and after matched exposure. For an imaging-history claim, compare exposure, time and location rather than simply taking another frame. For a background claim, acquire an appropriate blank and state the subtraction model.

A useful conclusion names the measured object, its state, the voltage or signal definition, the comparison and the remaining ambiguity. Here the fictional exposure record rejects a contacts-only explanation under its stated model and is compatible with a smaller conducting core plus changed contacts. It does not establish a new intrinsic resistivity law, a unique defect identity, a cleaning method or a universally safe measurement procedure. Calibration, uncertainty and repeatability explains why repeat agreement alone cannot supply the missing accuracy or uncertainty evidence.

Worked examples

Work through the case

Resistance rises as wires become thinner. Does that alone establish quantum confinement?

  1. Consider contact resistance and surface scattering.
  2. Check oxidation and actual conducting area.
  3. Seek evidence distinguishing mechanisms.

No. Several size-correlated effects can produce the trend.

A false intrinsic-resistivity trend from the wrong area

Use L = 1.00 µm, ρ_core = 2.00 × 10⁻⁷ Ω m and h = 2.00 nm. When outside diameter changes from 40.0 to 20.0 nm, calculate the core-resistance ratio and the apparent-resistivity ratio obtained from outside areas. Has core resistivity changed?

  1. Subtract the shell on both sides: d₁ = 36.0 nm and d₂ = 16.0 nm. Fixed L and ρ_core give R₂/R₁ = (36/16)² = 5.0625.
  2. Outside areas have ratio A_outer,2/A_outer,1 = (20/40)² = 1/4. Thus ρ_app,2/ρ_app,1 = (R₂/R₁)/4 = 1.265625.
  3. The apparent values are about 2.46914 × 10⁻⁷ and 3.12500 × 10⁻⁷ Ω m. Their 26.5625% rise results from using the wrong conducting area in this model.

Core resistance rises by a factor of 5.0625 and apparent resistivity by 1.265625. The supplied core resistivity stays constant; these observations alone do not establish confinement.

Reject a contacts-only explanation with the right voltage span

Use the matched exposure table. Partition the terminal rise, then check whether the independently supplied 36.0-to-32.0 nm core change can explain the interior reading without changing ρ_core.

  1. Before exposure, combined contacts give 236.5 − 196.5 = 40.0 Ω. After exposure they give 348.7 − 248.7 = 100.0 Ω.
  2. The terminal rise is 112.2 Ω: core rises by 52.2 Ω and contacts by 60.0 Ω. The interior observation therefore contradicts a contacts-only change within this same-span series model.
  3. At constant ρ_core and L, R_after/R_before = (36/32)² = 1.265625. The independently specified geometric model gives R_after = 4(2.00 × 10⁻⁷)(10⁻⁶)/[π(32 × 10⁻⁹)²] ≈ 248.680 Ω.
  4. The geometric prediction agrees with the rounded 248.7 Ω observation. Agreement is a consistency result with the supplied area; it does not uniquely identify microscopic defects or prove every assumption.

The observed rise divides into 52.2 Ω interior and 60.0 Ω combined-contact contributions. A reduced conducting area accounts for the interior rise at reporting precision; a unique intrinsic-size mechanism is not established.

Compare a changed field of view

Four 0.100 s unblanked frames use 1.00 pA. Compare the fluence for uniform 100 × 100 nm and 50 × 50 nm fields. Does the result identify a safe number of frames?

  1. Actual total exposure is 0.400 s, giving charge 4.00 × 10⁻¹³ C. Divide by e to obtain about 2.49660 × 10⁶ incident electrons.
  2. Areas are 10,000 and 2,500 nm². The respective fluences are 249.660 and 998.641 electrons/nm², a factor of four.
  3. The calculation counts incident electrons per area. Energy transfer, reaction pathways, specimen state and dose-rate effects are not supplied, so neither a damage fraction nor a safe frame count follows.

The smaller field has four times the fluence for the same incident electron number. No safe-frame threshold or damage mechanism has been determined.

Try the reasoning

  1. Can an imaging beam change a specimen?

    Show the answer and reasoning

    Yes.

    Dose can alter structure or chemistry.

  2. Why measure blanks or supports?

    Show the answer and reasoning

    They reveal background contributions.

    The specimen signal may be small relative to them.

  3. A separate uniform core has the same L and ρ_core as the wire model, but D = 30.0 nm and h = 3.00 nm. Find d, R_core and ρ_app based on outside area. A terminal reading is 500.0 Ω over the same span with no lead contribution; infer combined contacts.

    Show the answer and reasoning

    d = 24.0 nm; R_core ≈ 442.097 Ω; ρ_app = 3.12500 × 10⁻⁷ Ω m; combined contacts ≈ 57.903 Ω.

    Two shell thicknesses must be removed from a diameter. The outside-to-core area factor is (30/24)² = 1.5625. Subtract the core contribution from 500.0 Ω only because the span and series assumptions are supplied.

  4. A report uses d = D − h for the 40.0 nm outside diameter with h = 2.00 nm. What resistance does that predict, and what is wrong with the construction?

    Show the answer and reasoning

    It uses d = 38.0 nm and predicts about 176.349 Ω, below the correct 196.488 Ω.

    The shell occupies both sides of a diameter. The wrong area is too large; its predicted resistance is about 10.25% too small. A radial thickness must not be subtracted only once from a diameter.

  5. A before/after image gives the same 40.0 nm outside diameter. Does that establish the same conducting area, or rule out an exposure-induced resistance change?

    Show the answer and reasoning

    No. The conducting core or electrically active boundary can change inside the same outside outline.

    The fictional exposure record is one counterexample: d changes from 36.0 to 32.0 nm. Structural contrast alone also need not locate a depleted electrical region or identify its composition.

  6. The terminal measurement spans 1.00 µm of uniform wire, but interior probes span only 0.500 µm. Can their resistance difference be labelled combined contact resistance without correction?

    Show the answer and reasoning

    No. It also contains the resistance of the unsensed half of the wire.

    Only under independently justified uniformity can the half-span resistance be doubled to estimate the whole core contribution. A local damaged segment can invalidate that extrapolation; probe count alone does not solve the span problem.

  7. A separate matched geometry measurement establishes unchanged d and L. R_core rises by 25% in its verified Ohmic range. What follows for its effective resistivity, and has the specific defect been identified?

    Show the answer and reasoning

    Within the uniform model, effective ρ_core rises by 25%; the microscopic cause remains unresolved.

    With A_core and L fixed, ρ_core = R_core A_core/L has the same ratio as R_core. Carrier, scattering, temperature or model-state evidence is still needed before assigning one mechanism; a plausible name is not that evidence.

  8. In an additive matched expected-count model, specimen-plus-support changes from 120 to 180 while support-only changes from 80 to 140. Calculate both net contributions and assess a claimed 50% specimen increase.

    Show the answer and reasoning

    Both net contributions are 40. The 50% increase belongs to the raw total, not the inferred specimen contribution in this model.

    Subtract the appropriate support at each condition. A single unchanged blank would not remove the changed background. This is model arithmetic, not a significance test or a confidence statement.

  9. A uniform 100 × 100 nm region receives 0.500 pA for 1.00 s of unblanked exposure. Calculate incident fluence. Can it be ranked as less damaging than the four-frame example solely from its smaller current?

    Show the answer and reasoning

    About 312.075 electrons/nm². Smaller current alone does not establish less change or damage.

    Charge is 5.00 × 10⁻¹³ C, giving about 3.12075 × 10⁶ electrons over 10,000 nm². Its fluence exceeds the 249.660 electrons/nm² four-frame example; neither fluence alone nor current alone specifies a universal damage response.

  10. After repeated imaging, only the illuminated region changes contrast. A matched virgin region does not show the same contrast at its first final-stage observation. Does this identify atomic displacement as the unique cause?

    Show the answer and reasoning

    It supports a local irradiation-history effect, but not that unique mechanism.

    Charging, deposited contamination and structural or chemical changes can all alter contrast. Match acquisition/support conditions, account for the final observation’s dose and seek evidence that distinguishes the proposed mechanisms.

What to carry forward

  • Control surfaces, contacts and measurement disturbance.
  • A size correlation is not a unique mechanism.
  • Use the conducting geometry and the actual sensing span; outer size and terminal resistance need not identify core resistivity.
  • Matched controls can reject an explanation without proving a unique microscopic cause.
  • Record observation history and matched background; exposure arithmetic alone is not a damage law.

References

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

  1. Integrated Nanoscale Experimentation, Prediction, and ValidationNIST · programme overview, page updated March 2025Institutional programme contextnist.govRead source
  2. Introduction to Nanoelectronics · Part 4: Two Terminal Quantum Wire DevicesMarc Baldo · MIT OpenCourseWare · Spring 2010Selected university teaching · PDFocw.mit.eduRead source
  3. Metrology for the Electrical Characterization of Semiconductor NanowiresCurt A. Richter and colleagues · IEEE Transactions on Electron Devices 55 (2008), 3086–3095Selected primary researchtsapps.nist.govRead source
  4. Electrical characterization of chemical and dielectric passivation of InAs nanowiresGregory W. Holloway, Chris M. Haapamaki, Paul Kuyanov, Ray R. LaPierre and Jonathan Baugh · 2016 preprintSelected primary researcharxiv.orgRead source
  5. Electron irradiation-based cleaning of the scanning electron microscope and its samplesAndrás E. Vladár · Micron 196–197 (2025), 103857Selected primary researchtsapps.nist.govRead source
  6. Electron Beam Induced Artifacts During in situ TEM Deformation of Nanostructured MetalsRohit Sarkar, Christian Rentenberger and Jagannathan Rajagopalan · Scientific Reports 5 (2015), 16345Selected primary researchnature.comRead source