DLC Coatings: Internal Stress, Adhesion Tests, Thickness Limits

Diamond-like carbon (DLC) is an amorphous carbon-based film in which carbon atoms carry both diamond-type sp3 and graphite-type sp2 bonds, normally deposited on the workpiece by physical vapour deposition (PVD) or chemical vapour deposition (CVD). It does not belong to the same process family as the liquid coatings this site usually discusses: a paint flows and cross-links in air into a film tens of micrometres thick, whereas a DLC film is built atom by atom under ion bombardment at sub-micrometre to a few micrometres. The two are complementary on machine surfaces, which is exactly why coating engineers need the numbers and the failure gates of DLC.

TL;DR　Three lines carry the whole argument. One: internal stress. As-deposited compressive stress commonly reaches 8-10 GPa and accumulates with thickness; once the critical value is passed the film buckles and detaches in sheets, so making it thick is not a display of process capability — only after stress is brought down to 2-4 GPa does a single layer become usable at 0.5-5 µm. Two: the tests are not interchangeable. ISO 26443:2023 gives a qualitative Rockwell indentation class of 0-3, ASTM C1624-22 and ISO 20502:2016 give scratch critical loads, and GB/T 47254-2026 (identical to ISO 4821:2022) gives dynamic adhesion under lubrication. They measure different things and cannot be converted into one another; two of them state explicitly that they do not measure intrinsic interfacial bond strength. Three: the environment is a variable, not a backdrop. The same a-C:H film reads 0.03-0.05 in dry nitrogen, about 0.30 initially and 0.10-0.30 at steady state in 60% RH, and 0.021 or 0.027 in humid nitrogen after run-in. One further note for specifiers: DLC is not a barrier layer — pinholes must be handled by sealing or an undercoat.

1. Key data first

• Naming: ISO 20523:2017 Carbon based films — Classification and designations (published September 2017) sets classification, designations and short names. It applies to films whose predominant constituent is carbon, deposited by PVD or CVD, and covers amorphous carbon-based films (also called DLC), CVD diamond, graphite and polymer-like films. It does not apply to fullerenes, carbon nanotubes or graphene, and its definitions refer only to non-gradient layers; carbon bound as a carbide inside a-C:Me or a-C:H:Me is not counted toward the carbon total. Britain adopted it as BS ISO 20523:2017 (in force 2017-09-30), France as NF A91-056 / NF ISO 20523:2018, and it cites VDI 2840:2012.
• Hardness window: a widely circulated secondary paraphrase of the ISO 20523 classes gives a-C 15-25 GPa, ta-C 30-50 GPa, a-C:H 10-20 GPa, a-C:H:Me 15-30 GPa and polymer-like <5 GPa; an open review instead lists hard a-C:H at 10-40 GPa and soft films below 5 GPa. Measured values: pulsed-laser-deposited ta-C at 55-65 GPa Vickers with 800-900 GPa dynamic indentation modulus; an industry feasibility report lists ta-C at 40-80 GPa and 400-800 GPa. The spread is real and is addressed in section 6.
• Bonding and hydrogen: the same secondary table lists a-C at sp3/sp2 0.2-0.8 with hydrogen <1 at%; ta-C above 0.8 with <1 at%; a-C:H at 0.1-0.5 with 20-40 at%; a-C:H:Me at 0.3-0.6 with 10-30 at%; polymer-like above 40 at%. PLD and filtered-arc films are reported with sp3 up to 85%, and the industrial reading is 80-87%, approaching 90%.
• Internal stress: as-deposited PLD ta-C sits at 8-10 GPa; the industrial reading for ta-C is 5-12 GPa; ultra-thin FCVA ta-C, grown from 7.6 nm to 50 nm, shows both residual compressive stress and sp3 fraction falling then rising, with a minimum of 3.9 GPa at 29 nm. Thermal or pulsed-laser annealing yields near stress-free films.
• Thickness ceiling: open literature states plainly that DLC is difficult to grow thick because stress rises with thickness and the film tends to detach after a critical thickness; average 2 µm films are reported on a variety of substrates, and the single-layer working window is 0.5-5 µm with a ±5% uniformity target. The test standards cut off in the same place: ASTM C1624-22 covers 0.10-30 µm with a 200 µm radius stylus and is not recommended below <0.1 µm or above 30 µm, while ISO 26443:2023 applies to coatings ≤20 µm.
• Deposition energy and rate: ion plating uses carbon ion energies of 100-750 eV and grows at about 1 µm/h; filtered cathodic vacuum arc (FCVA) needs 40-100 eV to exceed 80% sp3 at 1-10 nm/s; PLD needs fluence above 6 J/cm², mean kinetic energy above 30 eV and substrate below 100 ℃; rf-PECVD was run at 0.11-1.07 W/cm² and 0.67-40 Pa with methane as the sole carbon source.
• Friction against atmosphere: rf-PECVD a-C:H (surface roughness 1-4 nm) sliding against silicon nitride gave 0.03-0.05 in dry nitrogen; in 60% RH the initial value was about 0.30 and steady state 0.10-0.30; in humid nitrogen (0.7-40 %RH and 4.2-40 %RH) a run-in period brought a-C and a-C:H down to 0.021 and 0.027, independent of relative humidity. Reviews quote DLC around 0.1, comparable with PTFE, lower still for hydrogenated and fluorinated grades, while hydrogen-free a-C shows comparatively high friction in inert gas.
• Wear-rate magnitude: 50 nm ta-C on an Al2O3-TiC ceramic measured 51 GPa at 25 nm indentation depth and a wear rate of 1.8×10⁻⁴ mm³/(N·m) at a 37 GPa contact pressure; a-C:H wear rates span 10⁻⁸ to 10⁻⁶ mm³/(N·m) in dry nitrogen and 2.0×10⁻⁹ to 8.9×10⁻⁸ mm³/(N·m) at 60% RH.
• Adhesion verdicts: ISO 26443:2023 uses a Rockwell diamond indenter for a qualitative read; the load follows the substrate — 1471.5 N (scale C) above 54 HRC, 981 N (scale D) up to 54 HRC, 588.6 N (scale A) for ceramics and cemented carbide — with a loading time of 1-8 s, a hold of (4±2) s, at least 3 points and optical microscopy at no less than 100×. Classes 0-3: 0 no cracking or delamination, 1 cracking without adhesive delamination (both acceptable), 2 partial adhesive delamination (report the percentage), 3 complete adhesive delamination (report r/a). ASTM C1624-22 draws a Rockwell C diamond stylus with a 120° included angle and 200 µm tip radius across the surface, defining the normal force that produces a given damage level as the critical scratch load.
• The Chinese frontier: GB/T 47254-2026 Plain bearings — Dynamic adhesion test method for DLC coated parts under lubricated conditions was issued on 2026-02-27 and implemented on 2026-06-01, adopted identically from ISO 4821:2022 under SAC/TC236. It combines a ball-on-disc contact with acoustic emission under lubrication, states that results do not apply to unlubricated conditions, and defines DLC through ISO 20523. Its introduction records the industry reality that, with no official method available, vehicle makers and suppliers had to validate bonding on purpose-built rigs using real engine parts. Annex B links the method to engine-rig results.
• Pinholes and sealing: after 50 nm atomic layer deposition of Al2O3 and Ta2O5 nanolaminates or mixtures closed the defects of CrN and DLC coatings, the current density of CrN-coated steel dropped by over two orders of magnitude and neutral salt spray pitting was delayed from 2 h to 168 h; adhesion of the sealing layer on DLC was weaker, yet the salt-spray delay was still clear, which confirms that pinholes were being sealed. Read across to paint: DLC carries the wear and scuffing duty, not the corrosion barrier.
• Temperature and surface quality: reviews place graphitisation above roughly 300 ℃; the industrial stress-relief window is quoted at 200-400 ℃ with pulsed bias of -50 V to -200 V pulling stress from about 10 GPa to 2-4 GPa, and transition layers of 2-10 nm (Si, SiC, TiC or Ti/Cr). Surface figures: a-C:H roughness 1-4 nm, FCVA films below Ra 1 nm and smoother than the bare silicon wafer at 0.412 nm.

2. Naming: what ISO 20523 actually covers

2.1 The standard addresses the material layer, not the coating system

One sentence in the scope of ISO 20523:2017 is routinely overlooked: the document concerns the material of carbon based films, not the whole coating — and a real component carries a main functional layer plus additional layers above or below it. Layers whose composition or properties change through the thickness are called gradient layers, and the definitions apply only to non-gradient layers. In practice this means naming "the DLC on this part" is imprecise: quotations and acceptance documents should list main layer + transition layer + underlayer + top layer one by one, the same discipline used for thermal sprayed zinc-aluminium systems.

2.2 Four designations, and the difference is hydrogen and metal

The common designations are a-C (hydrogen-free amorphous carbon), a-C:H (hydrogenated), ta-C (tetrahedral amorphous carbon, high sp3, virtually no hydrogen) and a-C:Me / a-C:H:Me (metal-bearing, the metal usually present as carbide). The boundary clause matters: films with additional elements fall inside the document only when carbon is the predominant constituent, and carbon already tied up as carbide is not counted — heavily metal-loaded films therefore drift out of the DLC definition. Historically the term "diamond-like" came from early ion-beam deposits of hard, insulating, wear-resistant amorphous carbon.

2.3 Table 1 — the class window and what it means in engineering

The hardness column is a public paraphrase of the classification table, and the page itself warns that it is not the standard text; measured literature for the same class also reports 40-80 GPa. Treat it as an order of magnitude, never as an acceptance limit — which is the subject of section 6.

3. Mechanism: amorphous, yet diamond-like

3.1 Short-range order without long-range order

Carbon follows two well-known bonding routes: graphite, a hexagonal layer of sp2 bonds, and diamond, a cubic lattice of tetrahedral sp3 bonds. What is unusual about DLC is not that it contains diamond crystallites but that it locks a large share of carbon atoms into local tetrahedral coordination with no long-range periodicity. Hardness comes from that local sp3 bonding, not from grain boundaries or precipitates. Because it is not crystalline it has no cleavage planes, so crack propagation meets more resistance than in polycrystalline diamond. The hardness-versus-toughness tension discussed for nano ceramic coatings shows up in DLC as a different imbalance: internal stress.

3.2 What hydrogen does — softening and lubricating at once

Hydrogen caps the dangling bonds of the sp3 network and effectively shortens it, so the higher the hydrogen content the more polymer-like and the softer the film, with friction falling as well. Reviews state that adding hydrogen to the hydrocarbon feed raises the polymeric content, lowers the friction coefficient and smooths the coating; DLC surfaces are generally terminated by a hydrogen layer and adsorb water vapour, which further changes friction. That is the root cause of one formulation giving 0.03 in one atmosphere and 0.30 in another. The other half of the story is structural change under sliding: friction converts a-C:H into a material dominated by sp2 bonds with higher short-range order, and the amount of wear corresponds to how far that conversion progresses. The lubricating layer and the sacrificial layer are the same thing.

3.3 Doping and multilayers pushing the boundary back

Silicon, tungsten, fluorine, nitrogen and metal loading are used to extend practical thickness or service range. Published combinations include a-C:H:W multilayers, a-C:H:Cr multilayers, single a-C:H and a-C:H:Si layers made by radio-frequency reactive magnetron sputtering or plasma assisted CVD on hardened and tempered tool steel, evaluated by salt spray to DIN 50021-SS. Corrosion of silicon-doped multilayer DLC is documented as sensitive both to the deposition cycle and to the medium (NaCl, H2SO4, HCl, NaOH). Multilayers serve a second purpose: stress relief, since alternating softer layers interrupt the build-up of accumulated stress.

4. Four deposition routes and their windows

4.1 PECVD / PACVD — low temperature, smooth, works on plastics

Parallel-plate rf reactors dominate industrial practice, with carbon feeds ranging from acetylene, ethylene, methane and propane to hexane, benzene and cyclohexane, all vapour-pressure adequate; laboratory windows sit near 0.11-1.07 W/cm² and 0.67-40 Pa. Because deposition can run cold, plastics become coatable — scratch-resistant DLC on polycarbonate lenses is the standard example — at the cost of high hydrogen and hardness in the 10-20 GPa band. Chinese literature typically grows DLC on metal from acetylene with argon as auxiliary gas and evaluates both potentiodynamic polarisation and salt spray.

4.2 FCVA — high sp3 and high stress arrive together

A cathodic arc on graphite is passed through a radial magnetic field that removes droplets, so only carbon ions reach the surface; ion energy of 40-100 eV pushes sp3 above 80%, growth runs at 1-10 nm/s and the surface falls below Ra 1 nm. High energy and high stress are two ends of one curve: the baseline industrial scheme sits near 10 GPa and must be pulled to 2-4 GPa with a -50 V to -200 V pulsed bias. The opposite experimental finding is instructive: filtered arc at floating potential supplies enough energy for atomic intermixing at the surface without the large internal stress seen at higher energy, and delivered higher critical loads than a-C:H. Adhesion does not always sit on the high-energy side.

4.3 PLD and ion plating — two sources of ultra-hard films

Pulsed laser deposition ablates graphite with a 248 nm excimer laser; fluence above 6 J/cm², mean kinetic energy above 30 eV and substrate below 100 ℃ are needed for high sp3, giving fully amorphous films with sp3 up to 85% and as-deposited stress of 8-10 GPa. Thermal or pulsed-laser annealing then yields near stress-free films several micrometres thick with good adherence on silicon and cemented carbide, at 55-65 GPa Vickers. Ion plating offers the 100-750 eV window, about 1 µm/h growth and Knoop hardness near 6000 kg/mm² with very low hydrogen.

4.4 Table 2 — route comparison

5. Internal stress is the gate that sets thickness

5.1 Where the stress comes from

Two effects superpose. High-energy ions hammer the growing surface into a metastable dense structure, so compressive stress is frozen in as the network forms; and the coefficient-of-thermal-expansion mismatch between film and steel, ceramic or cemented carbide adds a second layer during cool-down. The introduction of GB/T 47254-2026 states the causality directly: DLC hard coatings are prepared at elevated temperature, thermal expansion differs between coating and engine component, internal stress is therefore high, and quality control of the bond between coating and substrate after deposition becomes decisive. Open reviews add that voids and microstructure absorb gases, which pushes stress further.

5.2 Critical thickness and buckling

Stress is in-plane, so its stored energy grows with thickness, and the only way to release it is to buckle, crack and detach in sheets. The literature says exactly this — thick films are hard to grow and detach beyond a critical thickness — and reports stable average films of 2 µm on many substrates. Translated into engineering: the single layer lives at 0.5-5 µm; going thicker requires multilayers, gradient layers or a stress buffer; and the thickness ceilings of the test standards (indentation ≤20 µm, scratch 0.10-30 µm) sit right across that working band.

5.3 Four stress-reduction routes and their side effects

• Pulsed bias: -50 V to -200 V, taking stress from about 10 GPa to 2-4 GPa — but the energy change shifts the sp3 ratio, so hardness must be re-confirmed.
• Low-temperature annealing: 200-400 ℃ releases part of the stress without large sp3 loss — capped by the graphitisation boundary near 300 ℃.
• Multilayer and gradient design: alternating ta-C and a-C, or 2-10 nm of Si, SiC, TiC or Ti/Cr — paid for in cycle time and batch consistency.
• Substrate preparation: Ar+ ion etch cleaning and interfacial mixing. The substrate itself matters: reviews report good adhesion on silicon but difficulty on gold, copper, tungsten and stainless steel, and measurements show a-C:H bonding better on silicon while ta-C bonds better on a conducting ceramic.

Substrate cleanliness is the same problem paint faces: soluble salts and interfacial contamination after blasting are prerequisites for DLC as well — see soluble salt contamination of painted substrates.

6. Verification: what is measured, under which standard, and what it does not prove

6.1 Hardness and modulus — why the numbers disagree

The general requirements for instrumented indentation are in ASTM E2546-15 (DOI 10.1520/E2546-15): where sample size or condition makes conventional methods unusable, it measures thin coatings quantitatively, covering nano, micro and macro indentation (pointing to ISO 14577-1) — and it excludes deliberate lateral displacement, i.e. scratch testing. Two effects cause misreading. First, depth and substrate influence: the same 50 nm ta-C film reports 51 GPa at 25 nm depth. Second, tip blunting: hard ta-C blunts the diamond tip, so the area function was recalibrated after every indentation, and the authors proposed a tip-geometry-independent hardness slope ratio against fused silica. The conclusion: hardness numbers taken at different depths, with different calibrations and by different methods (Vickers, dynamic indentation, instrumented indentation) are not directly comparable — a supplier's 80 GPa and a customer's 45 GPa can both be correct. The same method-versus-method gap appears for automotive clearcoat scratch resistance.

6.2 Adhesion test one — the Rockwell indentation class of ISO 26443:2023

Edition 2 appeared in October 2023, replacing the 2008 edition; Europe adopted EN ISO 26443:2023 and Germany published DIN EN ISO 26443:2024-02, the line running back to DIN CEN/TS 1071-8:2004. A Rockwell diamond cone indenter is pressed into the coating and the zone around the indent is read under an optical microscope. It is a qualitative adhesion assessment for coatings up to 20 µm (also usable for some metallic coatings) and is not suitable for elastic coatings on hard substrates. Machines follow ISO 6508-1 and ISO 6508-2, and the indenter profile must be inspected optically at no less than 200×. Its value is speed and low cost for batch sampling; its limit is that it yields a class, not a number.

6.3 Adhesion tests two — scratch: ASTM C1624-22 and ISO 20502:2016

ASTM C1624-22 covers hard (Vickers ≥5 GPa), thin (≤30 µm) ceramic coatings — carbides, nitrides, oxides, diamond and diamond-like carbon are named — on metal and ceramic substrates at ambient temperature, drawing a Rockwell C diamond stylus with a 120° angle and 200 µm radius at constant speed under constant or rising normal force, with tangential force and acoustic emission as secondary data. It states plainly that it does not measure intrinsic interfacial bond strength, only an engineering value for the coating-substrate system under given stylus, loading-rate and displacement-rate conditions; polymer and ductile metal coatings are outside it, and so, unmodified, are films below 0.1 µm or above 30 µm. Two extensions are allowed: lubricated or elevated-temperature testing, and testing after high-temperature, oxidative or corrosive exposure to judge retained properties.

On the ceramics side the international yardstick is ISO 20502:2016 Fine ceramics — Determination of adhesion of ceramic coatings by scratch testing (ISO/TC 206, replacing DIN EN 1071-3:2005, adopted in Europe as EN ISO 20502:2016). Its discipline goes well beyond "an Lc number": coatings ≤20 µm, force range 1-100 N; a Rockwell C diamond stylus verified per ISO 6508-2, replaced every 200 tests or on any damage; substrate Ra no greater than 0.5 µm, otherwise normal-force fluctuation reaches 0.1 N; coatings free of oil, grease and moisture; recommended atmosphere (22±2) ℃ / (50±10)% RH; a single result is not representative, so at least 5 valid tests; critical normal force must exceed 1 N at 0.1 N resolution, and measurement uncertainty commonly reaches 20% at 95% confidence with 5-10% operator bias. Three modes are defined — progressive PFST (fast screening), constant-force CFST (statistical damage) and multi-pass MPST (simulating fatigue). On terminology the standard is insistent: "critical load" is not accepted, because failure is caused by force, hence critical normal force Lc. Calibration recommends TiN/Si or CrN/Si reference specimens, with BCR-692 certified reference material for checking the whole instrument state.

Reading C1624 next to 20502 shows how vague the word adhesion has become in DLC purchasing: one caps thickness at 30 µm and lists acoustic emission as secondary, the other caps at 20 µm and writes ≥5 tests and 20% uncertainty into the method. A technical condition must state which standard, which failure level, and at what roughness and humidity it was measured; otherwise a figure such as Lc = 50 N cannot be compared. The overview page nano wear-resistant hard coatings handles mechanisms and selection across many material families; this page pins the naming, stress, thickness and acceptance numbers along the DLC line alone, so the two are complementary rather than repetitive.

6.4 Adhesion test three — dynamic adhesion under lubrication

GB/T 47254-2026 and ISO 4821:2022 fill the gap the first two cannot reach: a lubricated rolling-sliding contact in plain-bearing service. The apparatus is ball-on-disc with acoustic emission — the elastic wave released when the coating fails is captured, so bonding strength can be assessed without running a real engine, and Annex B establishes correlation with engine-rig outcomes. Normative references include ISO 683-1:2017 (bearing steels), ISO 1101 and ISO 21920-2 for geometry and surface texture (GB/T 1182-2018 MOD). Its limit must be quoted: results do not apply to unlubricated conditions. The group standard T/CSAE 103-2019 for low-friction DLC on engine parts and GB/T 6462-2021 for metallographic coating thickness both appear on testing-house lists (secondary reading); confirm current validity with the issuing bodies. For the tribology pairing, MoS2 dry-film lubricant coatings offer a useful contrast on torque factor.

6.5 Friction and wear — atmosphere is a variable

The spread within one film family is enough to overturn a selection. Three measured threads: dry nitrogen 0.03-0.05; 60% RH starting near 0.30 and settling at 0.10-0.30, with humidity raising both friction and wear; humid nitrogen at 0.7-40 %RH or 4.2-40 %RH giving 0.021 and 0.027 for a-C and a-C:H once a run-in period is introduced, independent of humidity. The counter-intuitive part: hydrogen-free a-C shows relatively high friction in inert gas, and "bone dry" is not necessarily better — a little water or hydrogen participates in achieving ultra-low friction, which is why the authors call humidity the key. For customers: a friction figure must be delivered together with atmosphere, counter-body, load, sliding speed and run-in history, or it is not comparable.

6.6 Corrosion — pinholes mean it is not a barrier

The chemical inertness of carbon films is genuine, but continuity is harder the thinner the film. The ALD sealing study gives auditable numbers: a conformal 50 nm Al2O3 and Ta2O5 nanolaminate follows the surface and grows into the pinholes, current density of CrN-coated steel falls by more than two orders of magnitude, and neutral salt spray pitting moves from 2 h to 168 h; on DLC the seal adheres less well yet the salt-spray delay remains clear. Corrosion of silicon-doped multilayer DLC varies with deposition cycle and medium, which makes the process cycle itself a corrosion parameter. For corrosion engineers the correct reading is: DLC takes the wear and scuffing load, while the barrier belongs to the underlayer or seal — compare ALD nano coatings for barrier and moisture sealing and sealing of micro-arc oxidation ceramic films.

6.7 Table 3 — verification methods side by side

6.8 Table 4 — how DLC and liquid coatings divide the work

7. Temperature and chemical limits — when not to choose DLC

7.1 Graphitisation above about 300 ℃

Reviews list a clear drawback: DLC tends to graphitise above roughly 300 ℃. Three consequences follow. Stress-relief annealing should stay on the safe side of that line (industry quotes 200-400 ℃, and the actual value depends on film grade and atmosphere); service temperature approaching it should not be assumed to leave friction and hardness unchanged; and any welding, thermal spraying or subsequent baking that crosses the line forces DLC to be applied last in the sequence.

7.2 Friction changes the structure itself

Raman spectroscopy is the reading tool: the dispersion of peak positions and intensities with excitation wavelength from 785 to 229 nm was measured across a wide range of disordered and amorphous carbons (DOI 10.1103/PhysRevB.64.075414), which underpins the routine use of Raman to judge sp2/sp3 structure. The sliding experiments already state the practical conclusion: friction converts a-C:H into sp2-dominated, more ordered material, and wear corresponds to the extent of that conversion. DLC's lubricating interface is therefore continuously regenerated and continuously consumed, and life depends on that balance rather than on initial hardness.

7.3 Lubricant chemistry has to be rig-verified

Behaviour in additive-containing oils is governed by interfacial chemistry, so "film acceptable" is not "system acceptable". That is precisely the gap GB/T 47254-2026 closes: oil condition, ball-on-disc contact and acoustic-emission judgement inside one method, with a correlation annex against engine rigs. Selection documents should carry oil grade, load spectrum and surface texture (the ISO 21920-2 parameter system) into the validation plan.

8. Failure triage and acceptance checklist

• Sheet detachment with lifted edges: suspect accumulated stress beyond the critical value first — check whether the layer exceeds the 0.5-5 µm single-layer window, whether multilayers or buffers were designed, and whether the substrate is one of the metals on which adhesion is known to be difficult.
• Delamination around an indentation rim (class 2/3): when reading ISO 26443:2023 confirm that load and substrate hardness match (1471.5 N above 54 HRC) and exclude indenter damage; the result must carry the delaminated area fraction or r/a.
• Scattered critical loads: check stylus radius against film thickness (200 µm suits 0.10-30 µm), the loading and displacement rates reported, and read tangential force with acoustic emission.
• Friction too high or unstable: record the atmosphere (dry nitrogen, ambient RH, humid nitrogen) and run-in history before discussing film grade; never compare two data sheets whose atmospheres are unstated.
• Early pitting in salt spray: that is a pinhole problem, not a failure of carbon inertness — evaluate a 50 nm-class seal or an underlayer and re-measure interfacial adhesion.
• Hardness disputes: agree indentation depth, calibration method and reference blocks, use a tip-geometry-independent slope ratio if needed, and state whether tip blunting occurred.
• Writing the technical condition: name the main layer per ISO 20523, list transition, under and top layers, attach a thickness method (metallographic cross-section), an adhesion standard with the required class or Lc level, and a service temperature ceiling — never "DLC coated" alone.

9. Frequently asked questions

Q1　Can DLC be used as an anti-corrosion coating?

Not as a barrier layer. Carbon films are chemically inert, but real films contain pinholes: in the ALD sealing study, salt-spray pitting moved from 2 h to 168 h precisely because those pinholes were closed. The sound division is barrier in the underlayer or seal, DLC as the wear and scuffing face — the same logic as mechanical durability of superhydrophobic coatings.

Q2　Why can DLC not be applied thick?

Because compressive stress accumulates with thickness. As-deposited ta-C sits at 8-10 GPa (industry reading 5-12 GPa); beyond a critical thickness it buckles and detaches. Pulsed bias, low-temperature annealing, alternating multilayers and transition layers bring it to 2-4 GPa, at the price of re-confirming sp3 and hardness. The usable single layer therefore stops at 0.5-5 µm.

Q3　Why do published hardness figures differ so much?

Three effects stack: substrate influence at a given indentation depth (51 GPa at 25 nm in a 50 nm film), diamond-tip blunting by the hard film that forces recalibration after every indent, and method differences (Vickers, dynamic indentation, instrumented indentation). One and the same ta-C film can legitimately carry 30-50 GPa, 40-80 GPa and 55-65 GPa at once.

Q4　Which friction figure should drive selection?

The one taken in your atmosphere. Dry nitrogen 0.03-0.05, 60% RH steady state 0.10-0.30, humid nitrogen after run-in 0.021 and 0.027 — all are true. Note also that hydrogen-free a-C runs comparatively high in inert gas, so "drier is better" does not hold.

Q5　Can ISO 26443, ASTM C1624, ISO 20502 and GB/T 47254 results be compared?

Not converted. They are a dry ambient qualitative indentation class, a dry ambient scratch critical load, a ceramic-scratch critical normal force with ≥5 tests, and a lubricated dynamic adhesion test respectively, over different thickness ranges and environments; two of them state they cannot reach intrinsic bond strength. They suit horizontal comparison within one method, not arithmetic conversion.

Q6　Can DLC be coated onto plastics?

Yes — that is the value of the low-temperature routes, with scratch-resistant polycarbonate lenses as the public example. But plastic conductivity of heat, stiffness and surface energy change the stress distribution and the failure mode: reviews warn that ceramic-type films on low-modulus polymer substrates usually fail in bending through the thickness before they fail cohesively.

Q7　Can a worn DLC surface be touched up on site with paint?

Not to the same functional level. Re-depositing DLC needs vacuum equipment, renewed substrate cleaning and pretreatment, and the old carbon film has to be stripped; on site you can restore corrosion protection and appearance, not low friction or scuffing resistance. This is the key difference when DLC and liquid coatings share one maintenance plan.

Q8　How to weigh DLC against nano ceramic coatings or MoS2 dry films?

By load and environment. Where a micrometre-thick build, on-site application and cost control matter, take the nano ceramic or dry-film lubricant route — MoS2 is limited by humidity and oxidation while DLC is limited by atmosphere and stress, in opposite directions. Where ultra-low friction and scuffing resistance matter and the part can enter a vacuum chamber, take DLC. Metal- and tungsten-bearing grades (a-C:H:Me, a-C:H:W) sit in between, buying load capacity and thickness with multilayers.

10. Sources

• ISO 20523:2017, Carbon based films — Classification and designations (short names); BS ISO 20523:2017 in force 2017-09-30; NF A91-056 / NF ISO 20523:2018; cites VDI 2840:2012.
• ISO 26443:2023 (2nd edition, October 2023), Fine ceramics — Rockwell indentation test for evaluation of adhesion of ceramic coatings; EN ISO 26443:2023 / DIN EN ISO 26443:2024-02; replaces ISO 26443:2008; predecessor DIN CEN/TS 1071-8:2004; references ISO 6508-1 and ISO 6508-2.
• ISO 20502:2016, Fine ceramics — Determination of adhesion of ceramic coatings by scratch testing (ISO/TC 206; replaces DIN EN 1071-3:2005; EN ISO 20502:2016, BS EN ISO 20502:2016, NF A92-815, UNE-EN ISO 20502:2016; references ISO 6508-2, ISO 4288).
• ASTM C1624-22, Standard Test Method for Adhesion Strength and Mechanical Failure Modes of Ceramic Coatings by Quantitative Single Point Scratch Testing.
• ASTM E2546-15 (updated 2023), Standard Practice for Instrumented Indentation Testing, DOI 10.1520/E2546-15; with ISO 14577-1:2015.
• GB/T 47254-2026, issued 2026-02-27, implemented 2026-06-01, identical adoption of ISO 4821:2022, SAC/TC236; normative references ISO 683-1:2017, ISO 1101, ISO 21920-2 (GB/T 1182-2018 MOD).
• J. Robertson, "Diamond-like amorphous carbon", Materials Science and Engineering: R: Reports 37 (2002) 129-281, DOI 10.1016/S0927-796X(02)00005-0.
• Y. Lifshitz, "Diamond-like carbon — present status", Diamond and Related Materials 8 (1999) 1659-1676.
• A. C. Ferrari, J. Robertson, "Resonant Raman spectroscopy of disordered, amorphous, and diamondlike carbon", Physical Review B 64 (2001) 075414, DOI 10.1103/PhysRevB.64.075414.
• K. Miyoshi et al., friction and wear of plasma-deposited amorphous hydrogenated carbon films against silicon nitride in dry nitrogen and 60% RH.
• Sealing study: 50 nm Al2O3 / Ta2O5 atomic layer deposition on CrN and DLC, current density reduced by more than two orders of magnitude, NSS pitting delayed from 2 h to 168 h.
• M. Cui et al., corrosion behaviour of silicon-doped multilayer DLC, RSC Advances (2016), DOI 10.1039/c6ra05527c.
• Industry and testing-house material: ta-C property tables, FCVA process windows, T/CSAE 103-2019 and GB/T 6462-2021 (cited as secondary reading; the standard text governs). Thermal conductivity appears in secondary sources both as 1-5 W/(m·K) and as 400-1000 W/(m·K); the two conflict, so this page does not use it as a design input.

Last updated: 2026-11-02　Compiled by the technical editorial team of Kexin New Materials (Guangdong) Co., Ltd. from published standards and literature for B2B technical exchange. Figures follow the standard texts and third-party measurement; trade terms EXW / FOB only.