Written and maintained by CASRAI Editorial Board
Last updated
The number on an objective’s barrel — 40x, 63x, 100x — tells you almost nothing about whether you can actually resolve the structure you’re trying to see. Resolution, the smallest distance between two points that still shows up as two points rather than one blur, is set by physics long before magnification enters the picture, and it is degraded further, in a real image, by three things a spec sheet doesn’t mention: how you sample the image, how well the immersion medium matches the specimen, and how you illuminate the sample in the first place. This guide works through each cost in order — the diffraction limit itself, then sampling, numerical aperture (NA), wavelength, and refractive-index mismatch — so you can tell whether a blurry or “empty” image is a real resolution ceiling or something you’re leaving on the table.
nn
The diffraction limit: what it actually says
n
Light diffracts around the edges of any finite aperture, including an objective’s front lens. Because of this, a microscope cannot image a point source as a point — it images it as a small diffraction pattern (the Airy pattern), and two point sources close enough together produce overlapping patterns that blur into one. The distance at which they’re still just distinguishable is the diffraction-limited resolution, and two closely related formulas describe it:
n
- n
- Abbe’s criterion: r = λ / (2 × NA)
- Rayleigh’s criterion: r = 0.61 × λ / NA
n
n
n
where r is the smallest resolvable distance, λ is the wavelength of light used, and NA is the numerical aperture of the objective (for a well-matched system, the condenser’s NA contributes too — see below). Both formulas say the same thing two ways: resolution improves as wavelength gets shorter or as NA gets larger, and it cannot improve past what those two numbers allow, no matter how much you magnify the resulting image. A widely cited real-world example: at a mid-spectrum wavelength of 550 nm and an NA of 1.35–1.40 (a good oil-immersion objective), the diffraction limit works out to roughly 0.24–0.25 µm — about the width of a large virus, and well below what a bare eye or a webcam-grade sensor could ever register as separate points regardless of screen zoom.
n
This is why cranking up magnification past the point where the image has already reached the diffraction limit produces empty magnification: a bigger, softer blur with no new information in it, not more resolved detail. Magnification makes existing resolution visible to the eye or the pixel grid; it cannot create resolution the optics never captured.
nn
Numerical aperture: the lever that costs the most
n
NA appears in both formulas and, of the two variables, it’s the one you have the most practical control over. NA is defined by the objective’s front lens geometry and the refractive index of the medium between the lens and the specimen: NA = n × sin(θ), where n is that medium’s refractive index and θ is the half-angle of the maximum cone of light the objective can collect. Two consequences follow directly:
n
- n
- Higher NA means more collected light and better resolution — not more magnification. A 40x/0.75 dry objective resolves less than a 60x/1.4 oil objective even though the numbers suggest the oil objective is “only” 1.5x more powerful; the resolution difference tracks NA, not the x-number.
- Immersion media raise the achievable ceiling because they raise n. Air has n ≈ 1.0, capping dry-objective NA below 1.0 in practice; water (n ≈ 1.33) and glycerol (n ≈ 1.47) push the ceiling higher; oil (n ≈ 1.515, matched to glass) is what gets you into the 1.3–1.4+ NA range where sub-micron diffraction-limited resolution becomes possible at all.
n
n
n
NA is also a two-sided number in a properly configured widefield or transmitted-light system: the objective’s NA sets the collection side, but the condenser’s NA sets the illumination side, and Rayleigh’s full form (r = 1.22λ / (NAobj + NAcond)) uses both. Underfilling the condenser aperture — a stopped-down aperture diaphragm, or a condenser that’s never been properly centered and focused — caps the achievable resolution regardless of how good the objective is, which is exactly why correct Köhler illumination alignment is a resolution prerequisite, not just a contrast or evenness nicety.
nn
Wavelength: the other half of the formula, with real tradeoffs
n
Shorter wavelengths resolve finer detail — blue light (~450–480 nm) resolves better than red (~620–700 nm) at the same NA, which is part of why UV and, further still, electron microscopy (using an electron’s much shorter effective wavelength) reach resolutions no visible-light system can touch; see the separate discussion of where diffraction-limited light microscopy stops being enough and transmission electron microscopy or atomic force microscopy take over. Within visible-light fluorescence work, though, wavelength isn’t a free dial: it’s fixed by the fluorophore’s emission spectrum, and shorter-wavelength excitation generally means more phototoxicity and faster photobleaching. In practice, wavelength gets chosen for the label and the biology, and NA and sampling are where the achievable resolution actually gets won or lost on a given experiment.
nn
Sampling: the resolution the optics deliver is not automatically the resolution your image records
n
A diffraction-limited optical system and a well-matched immersion medium can deliver, say, 0.24 µm resolution to the image plane — and a camera or scanning system with pixels or scan steps too coarse to register that detail will still throw it away. This is the Nyquist sampling problem applied to microscopy: to faithfully record a feature of size r without aliasing or loss, the sampling interval (effective pixel size at the specimen plane, or the scan step size in a laser-scanning system) needs to be meaningfully smaller than r — the commonly used target is roughly 2–2.3 samples per resolved unit, not one. Undersample and you silently discard resolution the optics actually captured; the image looks softer than the system is capable of, and no amount of post-processing recovers information the sensor never recorded. Oversample well past that ratio and you gain nothing but larger files, longer scan times, and (in live or photosensitive samples) more light dose than the experiment needs for no informational benefit — the goal is matching the sampling rate to the optical resolution, not maximizing pixel count.
n
This is also where “more megapixels” as a camera upgrade pitch falls apart on its own: a sensor’s pixel count only helps resolution up to the point where pixel size at the specimen plane (physical pixel size divided by total magnification) is already fine enough relative to the diffraction limit. Past that point, added resolution has to come from NA or wavelength, not the sensor.
nn
Refractive-index mismatch: the cost that gets worse with depth
n
The NA formula above assumes a continuous, matched refractive-index path from the objective’s front lens to the focal point — immersion medium, coverslip, and the specimen’s own mounting medium all at (or very close to) the same n. Real samples routinely break that assumption: an oil-immersion objective (designed around n ≈ 1.515, matched to glass and standard mounting media) focused into an aqueous, live-cell environment (n ≈ 1.33–1.38) is a textbook refractive-index mismatch, and so is any objective’s design RI meeting a coverslip of the wrong thickness or a mounting medium chosen for something other than optical matching.
n
The consequence is spherical aberration: rays that would converge at a single point in a matched system instead converge at slightly different focal points depending on how far off-axis they entered the lens, spreading a point source’s image out rather than concentrating it. Two things follow, and both scale with the size of the mismatch:
n
- n
- Resolution and signal intensity both degrade — the effective PSF (point spread function) broadens, particularly along the optical axis, so axial resolution suffers more than lateral resolution.
- The degradation gets worse with imaging depth, not constant through the sample: the further the focal plane sits into a mismatched medium, the more the light path accumulates the aberration, which is why a thick tissue section or spheroid often looks acceptably sharp near the coverslip and visibly softer and dimmer 30–50 µm deeper, even with focus and illumination unchanged.
n
n
n
The practical fix is matching immersion type to the actual imaging environment rather than defaulting to whichever objective happens to be mounted: water-immersion or water-dipping objectives for live aqueous samples and deep tissue imaging, glycerol immersion as an intermediate option for some cleared-tissue protocols, and oil immersion reserved for genuinely RI-matched fixed/mounted preparations where it delivers its full NA advantage. A correction collar, where the objective has one, adjusts for coverslip thickness variation specifically — it does not correct for a bulk mismatch between the immersion medium and the specimen’s own refractive index.
nn
Putting it together on a real image
n
When an image looks less resolved than the objective’s rated NA and the fluorophore’s wavelength should allow, the shortfall is almost always one of the levers above, not a mysterious optics failure:
n
- n
- Diffraction limit sets the theoretical floor — check it with r = 0.61λ/NA before assuming a problem exists at all.
- NA — is the condenser aperture actually open and centered, and is the correct immersion medium in place and free of air bubbles?
- Illumination — is the system actually in Köhler alignment, or is an uncentered/unfocused condenser silently capping the effective NA?
- Sampling — is the pixel size (or scan step) at the specimen plane fine enough relative to r, roughly 2–2.3x oversampled, or is the detector throwing resolved detail away?
- Refractive-index match — does the immersion medium’s RI actually match the mounting medium and sample, and does image quality drop off with depth in a way that points to spherical aberration?
n
n
n
n
n
n
Working the list in that order — theoretical limit, then NA and illumination, then sampling, then RI matching — catches most real-world resolution shortfalls without needing to swap objectives or upgrade a camera first.
nn
Frequently asked questions
n
Does a higher-magnification objective always give better resolution?
n
No. Resolution tracks numerical aperture and wavelength, not the x-number on the barrel. A high-magnification, low-NA objective can resolve less than a lower-magnification, high-NA one; magnifying past the diffraction limit just produces empty magnification, a bigger blur with no new detail in it.
n
Why does oil immersion improve resolution?
n
Oil immersion raises the refractive index of the medium between the coverslip and the objective’s front lens to roughly match glass (n ≈ 1.515), which raises the achievable numerical aperture well above what’s possible in air. Since resolution improves as NA increases, oil immersion is what makes NA above roughly 1.0 — and sub-micron diffraction-limited resolution — achievable at all.
n
What’s the difference between resolution and magnification?
n
Magnification is how large the image appears; resolution is the smallest distance between two points that the system can still show as two separate points. Magnification makes existing resolution visible to the eye or a sensor’s pixel grid, but it cannot create resolution the optics never captured — resolution is set by wavelength and NA and cannot be recovered by zooming in further, digitally or optically.
n
How much should I oversample when choosing pixel size or scan step?
n
Match sampling to the optical resolution rather than maximizing it: roughly 2 to 2.3 samples per resolved unit (per the Nyquist criterion applied to imaging) captures what the optics deliver without wasting acquisition time, file size, or light dose on a live or photosensitive sample. Undersampling below that throws away real resolution; oversampling well past it adds cost with no further informational benefit.
n
Why does my image get worse deeper into the sample even with focus and light unchanged?
n
This is the signature of refractive-index mismatch between the immersion medium and the specimen — commonly oil immersion on an aqueous or live sample. The resulting spherical aberration accumulates with depth, so image sharpness and signal both degrade progressively the deeper the focal plane sits in the mismatched medium. Switching to a water- or glycerol-immersion objective matched to the sample’s actual refractive index is the standard fix.
