TIRF Microscopy: Penetration Depth, Setup, and What Goes Wrong

Total internal reflection fluorescence (TIRF) microscopy illuminates only a thin slab of sample immediately above the coverslip. That is its entire value proposition: everything deeper in the cell is never excited, so it never contributes background. But the single most common error in published TIRF work is not an alignment mistake — it is quoting a penetration depth calculated from theory as though it were the depth actually achieved. It is not, and the gap is large enough to change conclusions.

Start here: your calculated penetration depth is optimistic

The evanescent field decays exponentially away from the glass–water interface, with a characteristic depth given by d = λ₀ / (4π √(n₂²sin²θ − n₁²)), where θ is the incidence angle, n₂ the glass index and n₁ the sample index. Plug in numbers and you get a reassuringly small figure. Measure it, and you get something bigger.

Brunstein and Oheim’s calibration study measured decay depths against index-matched fluorescent-bead standards and found the short-range component consistently exceeded prediction: 139 ± 20 nm measured versus 134 nm calculated at 67.5°, 109 ± 16 nm versus 100 nm at 70.0°, and 91 ± 13 nm versus 84 nm at 72.5° (488 nm excitation, NA 1.49, glass n = 1.52, cytoplasm-mimicking medium n = 1.374). More importantly, they found a genuinely non-evanescent component amounting to 13–15.5% of total excitation across the tested angles — light that is not confined at all. A separate methods review reports a case where the calculated depth was roughly 82 nm while the empirically determined depth was 164 nm.

Where does the stray light come from? Fiolka and colleagues traced it primarily to the objective itself, not the sample. Beam quality degrades measurably as the beam is pushed to the periphery of the back aperture: M² rose from about 2.5 at normal incidence to 3 at θ = 64°, and the Strehl ratio fell from about 0.8 to 0.6 over the same range. Cell-induced scattering of the evanescent wave was detectable but not dominant, and coverslip surface roughness (RMS 0.24–0.37 nm bare, 1.4–5.4 nm wet) produced no detectable scattering contribution. Their figures are consistent with Mattheyses and Axelrod’s earlier estimate of 10–15% far-field contribution at the interface, with proportionally more at greater depths.

What to do about it: do not report a single computed depth. Report the polar angle with its uncertainty, the assumed local refractive index with its uncertainty, and a calibrated effective depth — and state explicitly how depth uncertainty affects your conclusion. If your claim depends on distinguishing 80 nm from 160 nm, calibrate against a bead or gel standard before you make it.

The preconditions: NA, refractive index, and the critical angle

Total internal reflection happens only above the critical angle, θ_c = sin⁻¹(n₁/n₂). This immediately sets hard hardware requirements.

• The objective NA must exceed the sample refractive index. For an aqueous buffer at n ≈ 1.33 you need NA > 1.33; for cell cytoplasm at n ≈ 1.38 you need NA > 1.38. This is why TIRF objectives are specified at NA ≥ 1.45, with 1.49 and 1.50 common and 1.65–1.70 available for extreme angles.
• The margin above the critical angle is what you actually steer. With glass at n = 1.52 and cytosol at n = 1.38, θ_c is about 65°; with a 100× 1.65 NA objective on n = 1.78 glass it drops to about 51°. With only a 1.4 NA objective, just a few percent of the peripheral area of the back aperture is usable — you have almost no adjustment range. Higher NA buys working margin, not just resolution.
• Coverslip and immersion oil indices must match. Coverslips are specified around n = 1.523; standard immersion oil is n = 1.515–1.518 at 20 °C. Because oil index is temperature-dependent, live-cell work at 37 °C drifts away from the design point — which is why many TIRF objectives carry a correction collar. Mismatch produces spherical aberration and unwanted reflections, both of which feed the stray-light problem above.

There is a subtlety practitioners routinely miss: the relevant n₁ is not the buffer, it is whatever is actually in contact with the glass. At a focal adhesion the cell membrane sits directly on the coverslip with no aqueous film between, so the local index is nearer 1.38 than 1.33 and the critical angle shifts. The same field can contain regions above and below the critical angle at a fixed illumination angle. Sweeping the angle systematically, rather than setting it once at the start of a session, is the practical remedy. For the underlying optics of high-NA immersion systems, see our guide to microscope objective selection.

Realistically, the shallowest depth you can reach with a 1.45–1.49 NA objective on a cell interface is on the order of 70–80 nm, and the usual working range is 60–100 nm, extending to around 200 nm at shallower angles. Fluorophores within roughly 100–200 nm of the coverslip are the ones efficiently excited.

Illumination uniformity: fringes, shadows, and azimuthal spinning

A single stationary laser spot at the back focal plane gives you a coherent beam entering from one side. Three artifacts follow, and all three masquerade as biology:

1. Interference fringes from coherent scattering and diffraction at every surface in the path — lenses, mirrors, the coverslip itself.
2. Directional shadowing and streaking emanating from the side the laser enters, cast by optically dense organelles and refractive-index variation inside the cell.
3. Forward scattering of the evanescent wave, which brightens structures downstream of dense objects.

The fix is azimuthal beam spinning (spTIRF): rotate the focused spot around the TIRF annulus at the back focal plane so that each azimuthal angle’s fringe and shadow pattern averages out within one camera exposure. Implementations use galvanometer mirrors, acousto-optic deflectors, or a digital micromirror device; one published system spins at 200 Hz against 5 ms exposures. The measured benefit is real: intensity coefficient of variation across the field dropped from 0.31 ± 0.11 to 0.12 ± 0.05 with spinning, and frame-to-frame excitation constancy of about ±1% of the mean has been reported over multi-second recordings, versus roughly ±30% pixel variation for the stationary case.

The important caveat: spinning fixes directional non-uniformity. It is not a cure for non-evanescent far-field excitation, which is angle-symmetric and survives averaging. If you intend to compare intensities between objects in a field — the whole basis of vesicle-docking and axial-position measurements — azimuthal spinning is close to mandatory, but it does not license skipping depth calibration.

Prism-based vs objective-based TIRF

Objective-type TIRF, where the laser enters through the periphery of the objective’s back aperture, dominates commercial instruments. It leaves the sample completely accessible for perfusion, patch clamping, or force measurement, and collects fluorescence efficiently. Its costs are the ones described above: coupling the laser into the rear aperture is finicky, angle control is less direct than with a prism, and the objective is itself the main source of stray excitation.

Prism-based TIRF couples the beam through a prism on the opposite side of the sample. Angle is easy to vary over a wide range and the excitation path never passes through the collection optics, so far-field contamination is far lower — which is exactly why much of the quantitative single-molecule biophysics literature is prism-based. The trade-off is physical access: the prism sits where you would want your pipette, perfusion line, or heated chamber. Choose prism when clean, calibratable confinement matters more than sample manipulation; choose objective-type when you need to touch the sample or you are imaging live cells.

Single-molecule TIRF: the surface is the experiment

For in vitro single-molecule work — smFRET, single-molecule pull-down (SiMPull), photobleaching step counting — TIRF hardware is the easy part. Surface chemistry decides whether the data are interpretable.

• Passivation. Coverslips are cleaned aggressively (KOH sonication, or piranha for objective-type work) then coated with methoxy-PEG carrying a small fraction of biotin-PEG for specific capture; published preparations use roughly 2% biotinylated PEG, or mPEG:biotin-PEG at about 40:1 molar ratio. Good passivation is quantitative: fewer than 0.01 non-specifically adsorbed molecules per µm² above background.
• Density. Target roughly 0.1–0.2 molecules per µm² in the field. Denser and diffraction-limited spots overlap, corrupting both intensity and step counting; sparser and you spend the session collecting statistics.
• Stoichiometry by photobleaching steps. Because each fluorophore bleaches in a single discrete event, counting steps in a spot’s intensity trace counts subunits. This only works if labelling is near-complete and the density rule above holds; incomplete labelling biases counts downward, and overlapping spots bias them upward.
• Controls, not optional. Substitute control antibodies for capture and detection reagents, run lysates lacking bait or prey, and test the PEG surface with labelled protein and no capture reagent. SiMPull is sensitive enough that non-specific binding you would never see on a blot dominates a single-molecule field.

Photobleaching is a resource you spend, not a nuisance you tolerate — step counting depends on it, while trajectory length depends on suppressing it. The trade-offs are covered in our guide to photobleaching and phototoxicity in live-cell imaging. Note also that TIRF does not beat the diffraction limit; it improves axial confinement and contrast, but lateral resolution is still governed by the optical resolution limit. Localisation precision on a well-separated single emitter can be far finer than that limit, which is a different claim entirely.

When TIRF is the wrong tool

TIRF sees the ventral membrane and the few tens of nanometres above it. It cannot see the apical membrane, the nucleus, the Golgi, or anything mid-cytoplasm — not poorly, but at all. It also requires the structure of interest to be genuinely adherent to glass; a cell that is loosely attached, or a tissue slice, has no usable interface.

If you need single-molecule contrast deeper in the cell, highly inclined and laminated optical sheet (HILO) illumination is the natural step: the beam is refracted at an angle just below critical, producing a thin inclined sheet that penetrates microns into the sample rather than nanometres, at the cost of the near-total background rejection TIRF gives you. Most objective-type TIRF systems can do HILO simply by backing the angle off. For general optical sectioning over whole cells and thicker specimens, confocal microscopy remains the workhorse, and the speed-versus-photodamage trade between spinning-disk and point-scanning confocal is usually the live-cell decision. TIRF and confocal are complements, not competitors: confocal sections at roughly a micron, TIRF at roughly a tenth of that but only at one surface.

A minimum reporting standard

If you publish TIRF data, state: objective magnification and NA; excitation wavelength; the polar incidence angle with its uncertainty; the refractive indices assumed for glass and sample and how the sample index was obtained; whether illumination was stationary or azimuthally spun; and the effective penetration depth with its uncertainty, together with how it was calibrated. Without those, a reader cannot tell whether an intensity difference reflects axial position, illumination non-uniformity, or stray far-field excitation — and neither can you.

Frequently asked questions

How deep does TIRF actually image?

Practically, 60–100 nm at steep angles and up to roughly 200 nm at shallower ones, with about 70–80 nm being the realistic floor for a 1.45–1.49 NA objective on a cell interface. Calibrated measurements typically run larger than the theoretical value for the same angle, and 10–15% of excitation is not evanescent at all.

What numerical aperture do I need for TIRF?

Strictly, NA must exceed the refractive index of the medium above the glass — about 1.33 for buffer, about 1.38 for cytoplasm. In practice buy NA ≥ 1.45; 1.49 and 1.50 are the common choices, and 1.65–1.70 objectives exist for extreme angles at the cost of specialised glass and immersion media. A 1.4 NA objective technically reaches TIR on buffer but leaves almost no angular adjustment range.

Can I do TIRF with a normal coverslip?

You need the specified thickness for your objective and a refractive index around 1.52 matched to your immersion oil (n = 1.515–1.518 at 20 °C). For single-molecule work the bigger issue is cleanliness and passivation, not the glass grade — uncleaned coverslips carry fluorescent contamination and non-specific binding that ruin single-molecule fields.

Why is my TIRF image striped or shadowed on one side?

That is the signature of stationary single-spot illumination: interference fringes from coherent scattering, plus shadows cast by dense organelles downstream of the entry direction. Azimuthal beam spinning averages these out within the exposure and has been shown to reduce field intensity variation from a coefficient of variation of about 0.31 to about 0.12.

Is TIRF a super-resolution technique?

No. TIRF improves axial confinement and contrast; lateral resolution is still diffraction-limited. It is, however, the standard illumination scheme underneath several single-molecule localisation methods, because its low background is what makes individual emitters detectable in the first place.

TIRF or HILO for imaging inside cells?

HILO, without question — TIRF physically cannot excite anything more than a couple of hundred nanometres from the coverslip. HILO tilts the beam just below the critical angle to form a thin inclined sheet that reaches micrometres into the cell, trading some background rejection for access to the nucleus and cytoplasm. Most objective-type TIRF systems can switch between the two by changing the angle.

Does the temperature of my live-cell chamber matter?

Yes. Immersion-oil refractive index is temperature-dependent, so an objective aligned at room temperature is not aligned at 37 °C. Use the correction collar if the objective has one, and re-check the angle after the chamber has equilibrated rather than during warm-up.

References

• Fish KN. Total Internal Reflection Fluorescence (TIRF) Microscopy. Current Protocols (PMC9522316).
• Brunstein M, Oheim M. Calibrating Evanescent-Wave Penetration Depths for Biological TIRF Microscopy. Biophysical Journal (PMC6732530).
• Fiolka R et al. Eliminating Unwanted Far-Field Excitation in Objective-Type TIRF, Part I: Identifying Sources of Nonevanescent Excitation Light. Biophysical Journal (PMC4026778).
• Ellefsen KL, Dynes JL, Parker I. Spinning-Spot Shadowless TIRF Microscopy. PLOS ONE (PMC4550233).
• Jain A et al. Single-molecule pull-down for studying protein interactions. Nature Protocols (PMC3654178).
• Nikon MicroscopyU. Total Internal Reflection Fluorescence (TIRF) Microscopy.
• Leica Microsystems Science Lab. Total Internal Reflection Fluorescence (TIRF) Microscopy.
• Tokunaga M, Imamoto N, Sakata-Sogawa K. Highly inclined thin illumination enables clear single-molecule imaging in cells. Nature Methods (2008).