Polished metals sit in the 0.05–0.30 emissivity band, far below the 0.95 preset shipped on most handheld infrared thermometers, so the default setting under-reads bare metal surfaces by tens of degrees C unless corrected [S1][S5].
Emissivity is the ratio of a surface's radiated IR energy to that of a perfect black-body emitter (rated 1.0); a polished metal coupon reflects most incident IR and emits very little, which is why an infrared thermometer pointed at bare aluminium, copper, or stainless returns a cold-looking number [S1].
Why the 0.95 Default Fails on Bare Metal
The factory 0.95 preset on consumer-grade IR thermometers is calibrated for organic, painted, or oxidised surfaces, where most materials (skin, wood, paper, plastic, asphalt) cluster between 0.88 and 0.97 [S1][S2]. Shiny metals break that assumption: ice measures 0.97, but polished aluminium, copper, and stainless can drop to 0.03–0.20 depending on finish and wavelength, and the thermometer software, assuming 0.95, treats the weak real signal as if it came from a high-emitter, which mathematically produces a falsely low temperature [S1].
The error is not symmetric. If the real emissivity is 0.10 and the unit is set to 0.95, the indicated temperature can be 30–80 °C below the true surface temperature on parts that actually sit at 100–300 °C, and the gap widens as temperature rises because the Stefan–Boltzmann signal is non-linear in T [S1]. That is the reason the EEVblog user checking a NEMA11 stepper motor casing was steered away from the bare alloy and towards a sticker or paint mark [S3].
Three Practical Methods to Get a Clean Reading
Method 1: look up the value in an emissivity table and dial it into the meter. ThermoWorks, Fluke, and other vendors publish free tables listing polished aluminium at roughly 0.05, polished copper at 0.05–0.10, polished stainless at 0.10–0.16, and heavily oxidised steel or cast iron at 0.80–0.95; the operator reads off the row that matches the alloy, surface finish, and IR wavelength of the specific instrument, then enters the number on adjustable units [S2][S5].
Method 2: paint, tape, or soot a small spot on the target so the thermometer reads the coating instead of the metal. Matte black paint, electrical insulation tape, Kapton, and a candle-soot layer all sit at 0.93–0.98, close enough to the preset that a fixed-0.95 unit becomes usable; the trade-off is that the spot must be larger than the meter's field of view, the coating adds a small thermal resistance (a few tenths of a °C at steady state), and the surface gets contaminated, so this method is the standard go-to for one-off motor-casing and heatsink checks [S3][S5].
Method 3: use a contact probe as a calibration anchor. Sandwich a thermocouple or RTD bead against the same spot, take a contact reading, adjust the IR emissivity setting until the IR display matches the contact value, and lock that setting in for repeat scans; this is the workflow the EEVblog community recommends for sub-100 °C motor and heatsink work where a 5–6 °C error is acceptable [S3][S4].
Decision Criteria: Chart Lookup vs. Surface Modification vs. Contact Calibration
Choosing between the three methods comes down to four criteria: required accuracy, surface access, whether the part can be contaminated, and whether the measurement is one-off or repetitive. For one-off checks on accessible, non-critical surfaces, paint or tape is the lowest-effort path; for repetitive production scans on the same alloy, a chart lookup with a stored preset gives the fastest repeat cycle; for highest accuracy, a contact-probe cross-check is the only way to defeat reflected-IR errors from nearby hot or cold bodies [S3][S4][S5].
The decision matrix lines up roughly as: (1) Chart lookup: fast, non-contact, no surface damage, accuracy typically ±2–5 % of reading on a known alloy. (2) Paint/tape/soot: fastest in the field, accuracy comparable to the preset's ±2 % spec, but leaves a mark and adds a thin thermal barrier. (3) Contact cross-check: highest accuracy, often within ±1 °C, but slower and limited to points where a probe can be bonded or pressed [S3][S4][S5].
Field Checks and Limits Engineers Should Know
Two sanity tests separate a real reading from a reflection error. First, the ice-bath and boiling-water test (0 °C and 100 °C reference sources) confirms the instrument is fundamentally accurate before any emissivity adjustment; if the unit fails those, no emissivity setting will save the measurement [S4]. Second, the moving-background test: pass a hot or cold object behind the suspect metal target and watch the IR reading; if the indicated temperature shifts with the background, the surface is acting as a reflector rather than an emitter, and the reading is meaningless without surface modification or contact cross-check [S3][S4].
Emissivity also drifts with temperature, surface oxidation, viewing angle, and wavelength. A polished coupon at 25 °C can shift several percentage points by 500 °C, an oxide layer grown in service can move a "polished stainless" reading from 0.15 to 0.85 within weeks of heat exposure, and most handheld IR thermometers operate in the 8–14 µm long-wave band where published tables are most reliable, while short-wave instruments used on hot metal may need a different table entry [S1][S5]. For applications where the target is molten, oxidised, or part of a moving production line, switching to a bimetallic-strip reference probe at a known fixed point, or hard-mounting a calibrated infrared pyrometer with adjustable emissivity and a 1–2 % repeatability spec, is the engineering-grade answer.
Trackable signals to watch between now and the next calibration cycle: emissivity-table revisions from ThermoWorks, Fluke, and NIST (the ThermoWorks reference was last refreshed 2025-09) [S2], and any new handheld units shipping with user-settable presets down to 0.10 as standard rather than as an unlocked "advanced" menu, which would close the gap that today forces operators to paint or tape shiny-metal test points.
For the relevant spec sheets and selection criteria, see infrared level.
Related analysis: Engineering talent shortage 2026: power and semiconductors competing for the same pool.