Description
Thorlabs PF2011-F01 Ø50.8 mm (2 in) Zerodur® UV-Enhanced Aluminum Mirror, 250 – 450 nm
Product Overview
In a deep-UV relay line, the parameter that ends a working session is rarely reflectance. A mirror returning 88 percent instead of 90 percent costs two percent of the pulse energy, and the process window absorbs that without complaint. A mirror whose figure moves enough to walk the beam a few hundred micrometres off the aperture at the fourth fold stops the session, because the operator now has to re-align four mounts that were aligned against a datum, and the alignment itself has to be re-verified with a tool that may not fit the space. The cost of that event is measured in hours, not in percent.
The PF2011-F01 is built for that ordering of priorities. It is a round mirror 50.8 mm (2 in) in diameter and 12.0 mm (0.47 in) thick, cut from Zerodur® and coated with a UV-enhanced aluminum film carrying an MgF₂ overcoat. Average reflectance is 90 percent from 250 to 450 nm. The front surface is figured to λ/10 at 632.8 nm, surface quality is 40-20 scratch-dig, the clear aperture is 90 percent of the diameter, that is 45.7 mm, parallelism between the two faces is no worse than 3 arcmin, the part is specified for angles of incidence from 0° to 45°, and the model number is laser-engraved on the fine-ground back surface so a mounted optic can still be identified without being removed from its holder.
The choice of substrate is the reason this page exists rather than a generic aluminum-mirror page. Zerodur’s coefficient of thermal expansion is 0 ± 0.100 x 10⁻⁶/°C, and Thorlabs states the comparison directly: this is significantly smaller than the value for fused silica, 0.55 x 10⁻⁶/°C. The ratio between the two numbers is 5.5. In the ultraviolet that ratio matters more than it does in the visible, for a reason that has nothing to do with how good the coating is.
Ultraviolet photons carry several times the energy of near-infrared photons: 4.96 eV at 250 nm against 1.17 eV at 1064 nm, a factor of four. Every material in the beam path therefore sits closer to its absorption edge, and every percent of absorption becomes heat in a place where the geometry is least able to shed it. That heat does two separate things, and only one of them is usually discussed. The first is a rise in the average temperature of the part, which expands it uniformly; a 20 K rise stretches a 50.8 mm fused silica body by about 0.56 µm at the rim and a Zerodur body by about 0.10 µm, and neither number changes the figure of a free-standing plate, though both change how hard the retaining ring is pushing. The second is a temperature difference across the part, and that is what bends it. A gradient through the 12.0 mm thickness curves the surfaces; the curvature is what turns absorbed power into pointing error, and the coefficient of thermal expansion is exactly the constant that converts one into the other.
An order-of-magnitude estimate makes the ratio concrete. For a plate of thickness t carrying a linear through-thickness gradient ΔT, the induced curvature is of order α·ΔT/t, so the surface slope over a half-aperture of 25.4 mm is roughly (25.4 mm)·α·ΔT/(2t). With a modest 5 K gradient and the fused silica coefficient, that slope is about 1.2 arcsec, and the reflected beam deviates by twice that, near 2.4 arcsec, which displaces a spot by roughly 116 µm after 10 m of propagation. With the Zerodur coefficient the same gradient produces about 0.22 arcsec of surface slope, 0.44 arcsec of beam deviation and roughly 21 µm of displacement at 10 m. This is an estimate from the standard thin-plate relation, not a manufacturer specification, and real mounts, real convection and real beam profiles will not reproduce it exactly. But the scaling is not in dispute: the same absorbed power, the same cooling, the same coating, and a five-fold difference in how much pointing error leaves the mirror.
Zerodur is a registered trademark of Schott.
Key Features
- Zerodur® substrate with a near-zero expansion coefficient. The specified value is 0 ± 0.100 x 10⁻⁶/°C against 0.55 x 10⁻⁶/°C for fused silica. What the specification buys is not “no expansion” but “expansion small enough that a thermal gradient produces a figure change below the alignment budget of a typical UV relay”.
- UV-enhanced aluminum with an MgF₂ overcoat, doing two jobs at once. Thorlabs describes the layer plainly: the UV-enhanced aluminum coating has an overcoat of MgF₂ to increase its average reflectance in the ultraviolet. The mechanism is worth understanding, because it explains both the 90 percent figure and the handling rules. Fresh aluminum reflects well across the ultraviolet, but aluminum in air grows a native oxide within seconds, and that oxide is strongly absorbing below 300 nm. The MgF₂ layer is deposited in vacuum immediately after the metal, so it seals the aluminum before the oxide can form, and being a low-index material over a high-index metal it also acts as a matching layer that lifts the average reflectance across the band. It is not a cosmetic protective film; it is part of the optical design, and damaging it changes the reflectance, not just the appearance.
- Ravg of 90 percent from 250 to 450 nm. In the visible, 90 percent would be a mediocre number. In the ultraviolet it is a good one, because aluminum’s intrinsic reflectance falls toward the blue end of its useful range and every interface in front of the metal costs a little more than it would at 633 nm. Read the figure as an average across a 200 nm wide band, not as a guaranteed value at any single wavelength.
- λ/10 figure at 632.8 nm over a two-inch aperture, with 40-20 surface quality. The figure is the number that governs wavefront quality after reflection; the scratch-dig grade governs how much scatter the surface adds, and scatter is more visible in the ultraviolet than in the visible because a given defect scatters a larger fraction of a shorter wavelength. The coated part is specified at 40-20.
- Damage thresholds published at both 266 nm and 355 nm, at two different spot sizes. The 266 nm value is 0.25 J/cm² with 10 ns pulses at 10 Hz over a Ø0.150 mm spot; the 355 nm value is 0.3 J/cm² with 10 ns pulses at 10 Hz over a Ø0.381 mm spot. Converting those fluences to the energy in a single pulse gives about 44 µJ at 266 nm and about 342 µJ at 355 nm. The two numbers differ by a factor of nearly eight, and the fluence numbers differ by only 20 percent, because the test spots differ by a factor of 2.54 in diameter and 6.5 in area. Comparing the two fluences as if they described the same beam is the most common misreading of a damage-threshold table.
- A 12.0 mm body, which is a thermal component as much as a mechanical one. Heat entering the coated face has to diffuse through 12 mm of glass-ceramic before it can leave through the mount. For a material of this diffusivity the through-thickness diffusion time is on the order of three minutes, so the body low-pass filters short thermal transients instead of transmitting them to the back face as a step. The same thickness makes the part stiff enough that a hand-tightened retaining ring holds the figure without printing a ring-shaped error into it.
Technical Specifications
| Parameter | Specification |
|---|---|
| Part Number | PF2011-F01 |
| Coating | UV-enhanced aluminum |
| Overcoat | MgF₂ (magnesium fluoride; also written MgF2) |
| Spectral Range | 250 nm to 450 nm |
| Average Reflectance | Ravg of 90 percent from 250 to 450 nm |
| Substrate | Zerodur® (substrate grade) |
| Coefficient of Thermal Expansion | 0 ± 0.100 x 10⁻⁶/°C (fused silica reference: 0.55 x 10⁻⁶/°C) |
| Diameter | 50.8 mm (2 in), tolerance +0.0 / -0.1 mm |
| Thickness | 12.0 mm (0.47 in), tolerance ±0.2 mm |
| Parallelism | 3 arcmin or better |
| Front Surface Flatness | λ/10 at 632.8 nm |
| Surface Quality | 40-20 scratch-dig |
| Clear Aperture | 90 percent of diameter (45.7 mm) |
| Angle of Incidence | 0° to 45° |
| Back Surface | Fine ground, model number laser-engraved |
| Pulsed Damage Threshold (266 nm) | 0.25 J/cm² at 266 nm, 10 ns, 10 Hz, Ø0.150 mm |
| Pulsed Damage Threshold (355 nm) | 0.3 J/cm² at 355 nm, 10 ns, 10 Hz, Ø0.381 mm |
| CW Damage Threshold (1064 nm) | 300 W/cm at 1.064 µm, Ø0.044 mm |
| CW Damage Threshold (10.6 µm) | 500 W/cm at 10.6 µm, Ø0.339 mm |
| Mounting | Two-inch optic mount or kinematic mount with retaining ring |
Two notes on how to read that table. The four damage thresholds belong to the coating, not to the size of the optic: the same coating is deposited on the one-inch and half-inch versions of this mirror and its measured threshold does not change with diameter or substrate, because the coating process does not change. What changes with diameter is how much energy the part has to carry. And the thresholds are measured on clean optics. A fingerprint, a dust particle or an outgassed film on the surface lowers the threshold substantially, which is why cleanliness in a UV system is a performance requirement rather than housekeeping.
| Test Condition | Fluence | Test Spot | Energy in One Pulse |
|---|---|---|---|
| 266 nm, 10 ns, 10 Hz | 0.25 J/cm² | Ø0.150 mm | approx. 44 µJ |
| 355 nm, 10 ns, 10 Hz | 0.3 J/cm² | Ø0.381 mm | approx. 342 µJ |
Applications
- Excimer beam delivery at 248 nm. 248 nm sits inside the specified band with 2 nm of margin at the short end. Folding mirrors in an excimer cabinet see a beam with poor spatial coherence, a large area and a duty cycle that heats the optic steadily rather than in bursts, which is precisely the regime where the substrate coefficient decides whether the pointing stays put over a shift.
- Third-harmonic Nd:YAG chains at 355 nm. A tripled YAG line typically contains several folds between the harmonic separator and the work surface. Each fold multiplies whatever angular error the previous mirror introduces, so the cumulative pointing budget is tighter than any single mirror’s specification suggests.
- UV inspection and metrology illumination. Wafer and photomask inspection tools use 266 nm and 355 nm light for resolution; here the mirror is in the illumination path rather than in a high-fluence focus, so scatter and figure matter more than damage threshold.
- UV fluorescence and spectroscopy. Excitation paths that fold a 250 to 450 nm beam into a sample chamber benefit from a coating whose average reflectance is specified across the whole band rather than at one line, because the excitation wavelength is often changed by swapping a filter.
- UV direct-write and stereolithography. Projection heads that steer a 355 nm beam through a scan lens use a fold mirror immediately before the lens, where any angular drift becomes a position error in the focal plane at the full lever arm of the working distance.
Selection Guide
The first decision is the band, and the band boundary is sharp. Below 450 nm, the UV-enhanced aluminum is the only coating in this family that is specified; the protected aluminum and protected silver parts start at 450 nm and are not characterised below it, and the gold parts start at 800 nm. Above 450 nm the UV-enhanced coating is still reflective but its average specification no longer applies, so there is no reason to pay for it. If the working wavelength is 248 nm, this is the part. If it is 193 nm, this is not the part, and no coating in the PF2011 family is specified there.
The second decision is aperture. The clear aperture is 45.7 mm, and at the 45° end of the incidence range the footprint of a round beam on the surface stretches by a factor of about 1.41 in one axis. Sizing the beam so that its 1/e² diameter stays inside 45.7 mm divided by that factor leaves margin for the beam walking as the laser warms up, which it will. A beam clipped at the rim does not simply lose power; it acquires a diffraction fringe that propagates and appears as structure in the far field.
The third decision is whether the substrate is worth the premium, and the honest answer depends on the application. If the mirror sits in a stable, low-power path and the ambient temperature is controlled, a fused silica part with the same coating will perform to the same reflectance and the same figure, because the coating is identical and the figure is polished to the same λ/10. If the path is long, the fold count is high, the laser runs at high average power in the ultraviolet, or the enclosure temperature is not controlled, the substrate is what separates a mirror that holds alignment from one that has to be re-trimmed. The Zerodur premium buys stability, not reflectance.
The fourth decision is mounting, and it interacts with the third. A low-expansion substrate is defeated by a mount that imposes its own expansion on the optic. Aluminium kinematic mounts move roughly twenty times more per kelvin than Zerodur does, so a rigid three-point clamp on an aluminium body will bend a Zerodur mirror long before the mirror would have bent itself. Use a mount with nylon-tipped or Delrin-tipped retainers, tighten by hand, and leave the optic free to breathe in the radial direction.
Series Comparison
The same 50.8 mm Zerodur® body is offered with seven metallic coatings. They differ in spectral range, in average reflectance, and in whether the metal is bare or carries a dielectric overcoat.
| Code | Coating | Spectral Range | Average Reflectance | Overcoat |
|---|---|---|---|---|
| F01 | UV-enhanced aluminum | 250-450 nm | Ravg 90% | MgF₂ |
| G01 | Protected aluminum | 450 nm-20 µm | Ravg 90% (450 nm-2 µm), 95% (2-20 µm) | SiO₂ |
| P01 | Protected silver | 450 nm-20 µm | Ravg 97% (450 nm-2 µm), 95% (2-20 µm) | SiO₂ |
| M01 | Protected gold | 800 nm-20 µm | Ravg 96% | SiO₂ |
| M02 | MIR enhanced gold | 2-20 µm | Ravg 98% | SiO₂ |
| M03 | Unprotected gold | 800 nm-20 µm | Ravg 97% | None |
| L01 | Unprotected gold, line optimised | 10.6 µm line | Ravg 99% | None |
Two structural facts about this table are worth more than the numbers. First, the F01 is the only member with an MgF₂ overcoat; every other protected member uses SiO₂, which is more durable but absorbs at the short end of the ultraviolet, and that single material choice is why the ultraviolet band stops at this one code. Second, the reflectance figures in the right-hand columns are averages over bands of very different widths, so they cannot be sorted into a ranking. A 97 percent average over 450 nm to 2 µm and a 96 percent average over 800 nm to 20 µm describe different spans of spectrum, and the part that looks better on the row does not necessarily deliver more energy at your wavelength.
FAQ
Q1. Can I use this mirror at 248 nm, and can I use it at 193 nm?
At 248 nm, yes. The specified band starts at 250 nm and 248 nm is 2 nm below that edge, well inside the tolerance of a production coating run and far inside the range where an excimer line actually operates, since a KrF laser’s emission is broad and its line-narrowed output sits within a fraction of a nanometre of 248.3 nm. Treat the 250 nm figure as the guaranteed boundary and 248 nm as the working wavelength it was designed around. At 193 nm, no. That is an ArF wavelength, and it is 57 nm below the specified edge. Aluminum’s reflectance collapses in that region and the MgF₂ layer itself becomes the limiting absorber, so no coating in this family is specified at 193 nm, and a part sold for 193 nm will be a different design with a fluoride stack and usually a purged housing. Using this mirror at 193 nm will not damage it instantly, but the reflectance will be far below 90 percent and will degrade as the coating is exposed.
Q2. Why is the flatness quoted at 632.8 nm when I work at 266 nm?
Because the surface error is a physical height, and a height measured in waves depends on the wavelength used to measure it. A surface that deviates by λ/10 at 632.8 nm deviates by 0.063 µm; at 266 nm the same physical deviation is λ/4.2, and the reflected wavefront error is twice the surface error, so it becomes about λ/2.1. That does not mean the mirror is worse in the ultraviolet; it means the same surface is a larger fraction of a shorter wavelength. This is why a UV relay with several folds needs its wavefront budget computed at the working wavelength, not at 633 nm.
Q3. The two pulsed thresholds are 0.25 and 0.3 J/cm². Which one applies to my beam?
Neither, directly. Those are the measured thresholds at 266 nm and 355 nm with 10 ns pulses, and they were measured over test spots of Ø0.150 mm and Ø0.381 mm respectively. If your beam is smaller than the test spot, the threshold in J/cm² is not conservative in the way operators usually assume, because defect-initiated damage scales with the probability of finding a defect inside the beam area, and a smaller spot samples fewer defects. If your beam is larger, the fluence limit applies but the total energy on the part is much higher than in the test. Use the published value as an upper bound, derate it for pulse duration, repetition rate and any contamination, and treat 50 percent of the published fluence as a working ceiling for a production tool.
Q4. Does the MgF₂ overcoat make the mirror cleanable?
It makes the mirror survive normal laboratory cleaning, which a bare aluminum surface would not, but it does not make it robust. MgF₂ is softer than the SiO₂ used on the visible-range aluminum part, and it is soluble in water to a small degree, so the cleaning method has to be displacement rather than abrasion: flood with clean solvent, drag a lens tissue once in one direction under the weight of the tissue only, and let the solvent evaporate. Never rub dry, never scrub a spot, and never use a cleaning solution that leaves a residue, because a residue on the outermost layer changes the reflectance of the whole stack.
Q5. How much does the Zerodur substrate actually change in a real UV setup?
It changes the part of the error budget that grows with time and with laser power, and it leaves the part that is fixed at installation alone. A fused silica mirror of the same figure starts from the same place; the difference appears after the laser has been running for twenty minutes and the enclosure has warmed, or after the room air conditioning cycles. In a single-pass fold in a temperature-controlled laboratory, the difference may be unmeasurable. In a six-fold relay inside a sealed cabinet with 10 W of ultraviolet circulating, it is the difference between a tool that holds its alignment across a shift and one that needs a touch-up at lunchtime.
Q6. Can this mirror be used at normal incidence and at 45° with the same performance?
Yes, in the sense that a metal film has no angle-dependent interference notch to fall into: the specified range of 0° to 45° is a statement that the coating behaves as a broadband metal reflector across that span, not that it has been tuned for a particular angle. What does change with angle is the footprint, which stretches by about 1.41 at 45°, and the polarization split, which grows with angle for any metal. At 45° the s- and p-reflectances differ slightly, so a system that folds a polarization-sensitive beam should keep the fold count even or characterise the pair.
Usage Precautions
- Handle by the edge only, with gloves. A fingerprint on a UV mirror is worse than on a visible one, because the organic film that remains absorbs strongly at 266 nm and 355 nm. The damage threshold quoted on this page was measured on a clean optic; a contaminated one has a lower and unknown threshold.
- Clean by displacement, never by abrasion. Flood the surface with clean solvent, drag a lens tissue once in a single direction using only the weight of the tissue, and stop. Do not scrub, do not use a dry tissue, and do not attempt to polish out a mark in the coating.
- Keep the optic purged or in a dry enclosure when the laser is running in the ultraviolet. 248 nm and 266 nm light in air generates ozone, and ozone attacks both organic materials in the mount and the surfaces of optics in the beam path. A nitrogen purge protects the coating and the mount at the same time.
- Do not over-clamp. A 12.0 mm Zerodur body is stiff, but a retaining ring tightened with a tool will still print a ring-shaped error into a λ/10 surface. Hand-tight with a nylon-tipped ring is the correct specification, and the ring should not bear on the clear aperture.
- Store in the supplied case, and do not stack optics. The case holds the part off its faces. Stacking, wrapping in paper or storing loose hardware in the same drawer puts an abrasive particle in contact with a coated surface under load, which is how most scratched mirrors in a laboratory actually got scratched.
- Re-check alignment after the first thermal cycle, not before. Mount stress relaxes and adhesives creep during the first warm-up. A mirror that reads perfectly on the interferometer at installation and 2 arcsec off after an hour is usually a mount problem, not an optic problem, and the Zerodur substrate is what lets you tell the two apart.
Disclaimer
The specifications on this page are compiled from Thorlabs’ publicly published product documentation and are provided for selection reference only. Where this page and the manufacturer’s current official datasheet differ, the datasheet prevails. Coatings, specifications and appearance are subject to change without notice. Damage thresholds are measured values under the specific test conditions stated and are not a guarantee of performance in any particular setup; the user is responsible for verifying suitability at their own wavelength, fluence, pulse duration, repetition rate and beam size. The thermal estimates on this page are engineering approximations derived from published material constants, not manufacturer specifications. Prices and lead times are confirmed by our sales team at the time of quotation.
Contact Us
If you are building a UV relay and want a second opinion on whether the substrate or the coating is your limiting factor, send us the layout: wavelengths, fold count, beam diameter, average power and the alignment tolerance at the last element. We will tell you which of the two is worth spending on.
Hotline: +86 159-2722-0467
E-mail: 859818709@qq.com | Website: www.soar-laser.cn
Single units for evaluation and volume orders are both supported, and we can quote the one-inch and half-inch versions of the same coating from the same family.









Reviews
There are no reviews yet.