Motorized Zoom + APO for Semiconductor Wafer Inspection

This article is also available in 简体中文.

Which semiconductor inspection tasks suit a zoom + APO chain?

A production station needs four tiers: wafer or die-map overview, defect classification, IC surface review of bond pads and passivation, and microstructure work on grains, MEMS features and thin films. One zoom body plus one APO objective covers all four, because the body changes total magnification continuously while the APO objective, a finite-conjugate compound lens, sets the numerical aperture (NA) and the resolution floor.

• Semiconductor defect inspection — particles, scratches and residues, where continuous zoom lets an operator jump from a flagged die-map coordinate to a sub-millimeter field.
• Wafer inspection — whole-wafer, die-level and edge review. Field of view is sensor size divided by total magnification: on a 2/3″ sensor (8.8×6.6 mm) the field is 6.3×4.7 mm at 1.4×, 0.98×0.73 mm at 9× and 0.39×0.29 mm at 22.5× (computed by multiplication, not yet verified for this combination).
• IC surface inspection — bond pads, solder bumps, laser marking and passivation pinholes, where roughly 1 μm of resolution at a 34 mm standoff usually separates pass from fail.
• Microstructure imaging — grain boundaries, TSV walls and thin films, where a 50× APO at NA 0.42 gives 1 μm nominal resolution with only 1.6 μm of depth of field, moving the effort from optics to focus stability.

A 3.45 μm camera at 9× resolves about 0.38 μm on the sample, finer than the 5 μm nominal of a 2× APO objective.

Which body and objective combination fits each tier?

No published zoom + APO combination table exists, so every figure below is arithmetic, not a specification. Ranges assume a 1× camera adapter, itself contested: the catalogue writes the TV tube factor as 0.65× in one place and 0.67× in another.

All total-magnification figures are computed by multiplication, not yet verified for this combination. Field is not free: objective field runs from Φ12 mm at 2× to Φ0.48 mm at 50×. Working distance is not monotonic either — 5× has the longest standoff at 45 mm, 50× the shortest at 20.5 mm — so that spread drives Z travel on a multi-objective station. See the magnification and vignetting reference.

What does motorization add on an inspection tool?

The zoom body is the only continuously variable element here; the objective is a fixed multiplier. Motorizing the body turns the chain into a programmable magnification axis, which is where the operational gains sit rather than in raw resolution.

• Preset positions and repeat framing. RS-232 control lets the host store magnification positions as recipe steps, so a review station returns to the same framing for A/B comparison or stitching without an operator's hand on the ring.
• A concrete example. The LZ-650104 motorized continuous zoom lens spans 0.7×–4.5× (6.5:1) at 82±2 mm working distance, C-mount, 2/3″ format.

The 82±2 mm figure belongs to the body alone; the manual 6.5X body is specified at 87±2 mm, and the two must not be mixed. Repeat-positioning tolerance is not published — ask for it rather than assuming it scales with the zoom ratio. See the motorized zoom lens guide.

How do you lay out a 20–45 mm working-distance station?

APO working distance runs from 20.5 mm (50×) to 45 mm (5×), a delta of about 24.5 mm, so the budget follows the objective: the 20.5 mm standoff constrains ring illumination, probes and fixture height.

• Z travel. Allow at least 25 mm to swap between the 2×–50× objectives on one axis, and do not assume shorter working distance follows higher magnification — 5× is longest at 45 mm, 50× shortest at 20.5 mm.
• Focus budget. At 1.6 μm of depth of field at 50× and 3.5 μm at 10× and 20×, wafer bow, stage tilt and thermal drift eat the same budget, so plan on active Z focus or tiles small enough to stay inside the depth of field.
• Sensor and image circle. The 2/3″ bodies pair with a 2/3″ sensor; a larger sensor needs a larger adapter — in the 4K-series boundary table, 1″ needs at least 1×, 4/3″ at least 1.5×, φ32 mm 2×.
• Low-end field uniformity. In a 2026-04 internal test (2448×2048, 2/3″-class camera) one test combination showed a circular illumination field of Ø1774 px — 72.5% of sensor width, 78.6% of the diagonal — with corner brightness at 0 (−100% corner-to-center); at higher magnification the corners recovered and the difference narrowed to about −82%. The images carried no magnification or combination annotation, so this describes that test combination only. The official selection table's ~18 low-end vignetting marks cover 0.25×–2× attachment lenses, not APO objectives, so they do not transfer.

What must engineering verify before you quote a number?

Five items decide whether a semiconductor recipe works.

• High-magnification MTF and resolution. Neither catalogue nor datasheets publish MTF, distortion or resolution for a zoom + APO combination; near the top of a 5.8×–75× range the limiting element may be the zoom body, not the objective.
• Combined field of view and vignetting. An official zoom + APO combination table does not exist, so no field-of-view or vignetting figure for a composed chain can be quoted as a specification.
• Parfocality across the zoom range. Holding focus while the body zooms depends on conjugate matching between the finite-conjugate objective and the body's intermediate image, which is not published.
• Working distance of the assembly. The objective's working distance is the design input, but a complete assembly's standoff needs confirmation, not inference.
• Repeat-positioning tolerance. How accurately an RS-232 preset returns to the same magnification needs a measured figure before it enters a metrology recipe.

Every combination value here is arithmetic. Send the defect size, sample stack and camera pixel pitch and the combination can be confirmed against your sensor instead of an assumption — contact the optics team.