Dispersion-Compensating Prism Pairs for Ultrafast Lasers

  • Four Materials to Choose From
  • Designed for Use at Brewster's Angle
  • 700 to 900 nm Designed Spectral Range


Application Idea

AFS-FS Prisms
on a KM100B/M

Related Items

Please Wait

Dispersion-Compensating Prism Pair
An ultrafast-laser-dispersing prism pair with apex angle α and separation d can be used to spatially disperse different wavelength components and introduce wavelength-dependent phase changes. To learn more, please see the Tutorial tab.

Zemax Files
Click on the red Document icon next to the item numbers below to access the Zemax file download. Our entire Zemax Catalog is also available.
Optic Cleaning Tutorial


  • Designed Spectral Range: 700 - 900 nm
  • Apex Angle Tolerance: ±15 arcmin
  • Unpolished Sides: Flat and Ground
  • Surface Flatness: λ/4 at 633 nm
  • Use to Compensate for Dispersion
  • Fabricated from CaF2, Fused Silica, SF10, or N-SF14

These Dispersion-Compensating Prism Pairs are designed to compensate for the pulse broadening effects that occur in ultrafast laser systems without introducing significant power losses. Four large-index-of-refraction prism materials are offered to provide various group velocity dispersion (GVD) values, thereby compensating for different pulse dispersion conditions.

Prism pairs such as these have multiple applications. They can be used to create an optical filter; simply insert a knife edge to attenuate the short- or long-wavelength components. If the prism pair is placed inside a laser resonator, they can be used for wavelength tuning. Alternatively, prism pairs can be used for dispersion compensation. Each time light passes through the optical elements within a laser cavity, positive (commonly referred to as normal) dispersion occurs. The prism pair also contributes positive dispersion. However, this positive dispersion is somewhat offset by the extra distance that the longer wavelengths travel through the second prism (refer to the figure to the right). With a careful choice of prism pair geometry it is possible to create negative (commonly referred to as anomalous) dispersion that compensates for all other elements within the cavity.

For Thorlabs' full selection of optics for ultrafast applications, please see the Ultrafast Optics tab.

Item # Material GVDa
(at 800 nm)
(at 800 nm)
(Apex Angle)
ΘBd Refractive Index
(at 800 nm)
AFS-CAF CaF2e -5 fs2/cm -12 fs3/cm 69.9° 12.3 17.2 ~55.0° 1.430 40-20
AFS-FS Fused Silica -16.5 fs2/cm -20 fs3/cm 69.1° 12.4 17.0 ~55.6° 1.453 20-10
AFS-SF10 SF10e -97.5 fs2/cm -388 fs3/cm 60.6° 13.0 15.1 ~59.7° 1.711 20-10
AFS-SF14 N-SF14 -113.5 fs2/cm -473 fs3/cm 59.6° 13.0 14.9 ~60.2° 1.743 20-10
  • Typical Group Velocity Dispersion (see the Graphs and Tutorial tabs for additional information)
  • Third-Order Dispersion (see the Graphs and Tutorial tabs for additional information)
  • Apex Angle Tolerance: ±15 arcmin
  • Brewster's Angle
  • Click Link for Detailed Specifications on the Substrate Material

The graphs below represent calculated Group Velocity Dispersion and Third-Order Dispersion values for prism pairs with a separation distance of 10 mm. Please see the Tutorial tab for more information.

Group Velocity Dispersion Graph
Click to Enlarge

Click Here for Raw Data
Third-Order Dispersion Graph
Click to Enlarge

Click Here for Raw Data

Prism Pair Dispersion Calculations

The broadband frequency spectrum of ultrashort laser pulses brings rise to the necessity for dispersion compensation. Whenever an ultrashort pulse passes through a positive dispersive material, the pulse becomes “stretched” due to the frequency dependency of the refractive index of a material. In order to maintain pulse shape and duration, a negative dispersive system is required. While there are various methods for producing dispersion control, such as grating pairs or chirped mirrors; prism pairs offer a simple, tunable, and low loss solution for dispersion compensation. Through adjusting the relative geometry of the prism pair, the dispersion can be adjusted from positive through negative values. Please see below for sample results.

As can be demonstrated through Fourier analysis, ultrashort pulses (small Δt) innately possess a broadband frequency spectrum (large Δf). However, due to the wavelength-dependent nature of the index of refraction, longer wavelengths travel faster through an optical material than short wavelengths. Consequently, femtosecond pulses are subject to pulse broadening as a result of group velocity dispersion (GVD). These effects can be reversed by using a prism pair such as those offered here to provide a negative dispersive system in order to shorten the duration of an ultrashort laser pulse.

To achieve the specified GVD values while also maximizing transmission, light should be incident on the first prism at Brewster's Angle (ΘB). As shown in the schematic above, the first prism is used to separate the various wavelength components. Then, a second prism is positioned such that the various wavelengths of refracted light rays will propagate parallel to each other upon exiting. Furthermore, by including another pair of prisms (4 prisms total), the final rays can be made to propagate collinearly with each other as well as with the path of the incident rays [1]. This feature can be particularly useful when installing a dispersion compensator into an existing optical system.

The apex angle in the ULD prism pairs is chosen such that the input and output angles are both near Brewster's Angle (ΘB) in order to minimize the reflection of p-polarized light.

The frequency-dependent exit angle Θf(ω) is given by

where the variables n, ΘB, and α correspond to the frequency-dependent refractive index of the prism material, angle of incidence with the surface of the first prism, and apex angle of the prism, respectively.

The accumulated phase φ(ω) in a prism pair is equal to

where the variables ω, d, Θf, and Θfshort correspond to frequency, prism separation, exit angle (frequency dependent), and exit angle of the shortest transmitted wavelength, respectively.

The first, second, and third derivatives of the accumulated phase with respect to frequency are proportional to the Group Delay (GD), the Group Velocity Dispersion (GVD) and the Third-Order Dispersion (TOD) of the prism pair, respectively.

The graphs below represent calculated GVD and TOD values for prisms separated by a distance (d) of 10 mm and with an angle of incidence equivalent to the Brewster angle of the material (ΘB, see Specs tab for approximate values).

Group Velocity Dispersion Graph
Click to Enlarge

Click Here for Raw Data
Third-Order Dispersion Graph
Click to Enlarge

Click Here for Raw Data
    [1] R. L. Fork, O. E. Martinez, and J. P. Gordon, "Negative dispersion using pairs of prisms," Opt. Lett. 9, 150-152 (1984).

Selection Guide for Prisms

Thorlabs offers a wide variety of prisms, which can be used to reflect, invert, rotate, disperse, steer, and collimate light. For prisms and substrates not listed below, please contact Tech Support.

Beam Steering Prisms

Prism Material Deviation Invert Reverse or Rotate Illustration Applications
Right Angle Prisms N-BK7, UV Fused Silica, Calcium Fluoride, or Zinc Selenide 90° 90° No 1

90° reflector used in optical systems such as telescopes and periscopes.

180° 180° No 1

180° reflector, independent of entrance beam angle.

Acts as a non-reversing mirror and can be used in binocular configurations.

TIR Retroreflectors
(Unmounted and Mounted)
and Specular Retroreflectors
(Unmounted and Mounted)
N-BK7 180° 180° No Retroreflector

180° reflector, independent of entrance beam angle.

Beam alignment and beam delivery. Substitute for mirror in applications where orientation is difficult to control.

Unmounted Penta Prisms
Mounted Penta Prisms
N-BK7 90° No No 1

90° reflector, without inversion or reversal of the beam profile.

Can be used for alignment and optical tooling.

Roof Prisms N-BK7 90° 90° 180o Rotation 1

90° reflector, inverted and rotated (deflected left to right and top to bottom).

Can be used for alignment and optical tooling.

Unmounted Dove Prisms
Mounted Dove Prisms
N-BK7 No 180° 2x Prism Rotation 1

Dove prisms may invert, reverse, or rotate an image based on which face the light is incident on.

Prism in a beam rotator orientation.

180° 180° No 1

Prism acts as a non-reversing mirror.

Same properties as a retroreflector or right angle (180° orientation) prism in an optical setup.

Wedge Prisms N-BK7 Models Available from 2° to 10° No No 1

Beam steering applications.

By rotating one wedged prism, light can be steered to trace the circle defined by 2 times the specified deviation angle.

No No Wedge Prism Pair

Variable beam steering applications.

When both wedges are rotated, the beam can be moved anywhere within the circle defined by 4 times the specified deviation angle.

Coupling Prisms Rutile (TiO2) or GGG Variablea No No Coupling Prism

High index of refraction substrate used to couple light into films.

Rutile used for nfilm > 1.8

GGG used for nfilm < 1.8

  • Depends on Angle of Incidence and Index of Refraction

Dispersive Prisms

Prism Material Deviation Invert Reverse or Rotate Illustration Applications
Equilateral Prisms F2, N-F2,
Calcium Fluoride,
or Zinc Selenide
Variablea No No

Dispersion prisms are a substitute for diffraction gratings.

Use to separate white light into visible spectrum.

Dispersion Compensating Prism Pairs Fused Silica, Calcium Fluoride, SF10, or N-SF14 Variable Vertical Offset No No Dispersion-Compensating Prism Pair

Compensate for pulse broadening effects in ultrafast laser systems.

Can be used as an optical filter, for wavelength tuning, or dispersion compensation.


Pellin Broca Prisms N-BK7,
UV Fused Silica,
or Calcium Fluoride
90° 90° No 1

Ideal for wavelength separation of a beam of light, output at 90°.

Used to separate harmonics of a laser or compensate for group velocity dispersion.

  • Depends on Angle of Incidence and Index of Refraction

Beam Manipulating Prisms

Prism Material Deviation Invert Reverse or Rotate Illustration Applications
Anamorphic Prism Pairs N-KZFS8 or
Variable Vertical Offset No No 1

Variable magnification along one axis.

Collimating elliptical beams (e.g., laser diodes)

Converts an elliptical beam into a circular beam by magnifying or contracting the input beam in one axis.

Axicons (UVFS, ZnSe) UV Fused Silica
or Zinc Selenide
Variablea No No 1

Creates a conical, non-diverging beam with a Bessel intensity profile from a collimated source.

  • Depends on Prism Physical Angle

Polarization Altering Prisms

Prism Material Deviation Invert Reverse or Rotate Illustration Applications
Glan-Taylor, Glan-Laser, and α-BBO Glan-Laser Polarizers Glan-Taylor:

α-BBO or Calcite
p-pol. - 0°

s-pol. - 112°a
No No Glan-Taylor Polarizer

Double prism configuration and birefringent calcite produce extremely pure linearly polarized light.

Total Internal Reflection of s-pol. at the gap between the prism while p-pol. is transmitted.

Rutile Polarizers Rutile (TiO2) s-pol. - 0°

p-pol. absorbed by housing
No No Rutile Polarizer Diagram

Double prism configuration and birefringent rutile (TiO2) produce extremely pure linearly polarized light.

Total Internal Reflection of p-pol. at the gap between the prisms while s-pol. is transmitted.


Double Glan-Taylor Polarizers Calcite p-pol. - 0°

s-pol. absorbed by housing
No No Glan-Taylor Polarizer

Triple prism configuration and birefringent calcite produce maximum polarized field over a large half angle.

Total Internal Reflection of s-pol. at the gap between the prism while p-pol. is transmitted.

Glan Thompson Polarizers Calcite p-pol. - 0°

s-pol. absorbed by housing
No No Glan-Thompson Polarizer

Double prism configuration and birefringent calcite produce a polarizer with the widest field of view while maintaining a high extinction ratio.

Total Internal Reflection of s-pol. at the gap between the prism while p-pol. is transmitted.

Wollaston Prisms and
Wollaston Polarizers
Quartz, Magnesium Fluoride, α-BBO, Calcite, Yttrium Orthovanadate Symmetric
p-pol. and
s-pol. deviation angle
No No Wollaston Prism

Double prism configuration and birefringent calcite produce the widest deviation angle of beam displacing polarizers.

s-pol. and p-pol. deviate symmetrically from the prism. Wollaston prisms are used in spectrometers and polarization analyzers.

Rochon Prisms Magnesium Fluoride
Yttrium Orthovanadate
Ordinary Ray: 0°

Extraordinary Ray: deviation angle
No No

Double prism configuration and birefringent MgF2 or YVO4 produce a small deviation angle with a high extinction ratio.

Extraordinary ray deviates from the input beam's optical axis, while ordinary ray does not deviate.

Beam Displacing Prisms Calcite 2.7 or 4.0 mm Beam Displacement No No Beam Displacing Prism

Single prism configuration and birefringent calcite separate an input beam into two orthogonally polarized output beams.

s-pol. and p-pol. are displaced by 2.7 or 4.0 mm. Beam displacing prisms can be used as polarizing beamsplitters where 90o separation is not possible.

Fresnel Rhomb Retarders N-BK7 Linear to circular polarization

Vertical Offset
No No Fresnel Rhomb Quarter Wave

λ/4 Fresnel Rhomb Retarder turns a linear input into circularly polarized output.

Uniform λ/4 retardance over a wider wavelength range compared to birefringent wave plates.

Rotates linearly polarized light 90° No No Fresnel Rhomb Half Wave

λ/2 Fresnel Rhomb Retarder rotates linearly polarized light 90°.

Uniform λ/2 retardance over a wider wavelength range compared to birefringent wave plates.

  • S-polarized light is not pure and contains some P-polarized reflections.

Beamsplitter Prisms

Prism Material Deviation Invert Reverse or Rotate Illustration Applications
Beamsplitter Cubes N-BK7 50:50 splitting ratio, 0° and 90°

s- and p- pol. within 10% of each other
No No Non-polarizing Beamsplitter

Double prism configuration and dielectric coating provide 50:50 beamsplitting nearly independent of polarization.

Non-polarizing beamsplitter over the specified wavelength range.

Polarizing Beamsplitter Cubes N-BK7, UV Fused Silica, or N-SF1 p-pol. - 0°

s-pol. - 90°
No No Polarizing Beamsplitter Cube

Double prism configuration and dielectric coating transmit p-pol. light and reflect s-pol. light.

For highest polarization use the transmitted beam.

Posted Comments:
Zhongcheng Wang  (posted 2023-12-30 00:22:50.093)
Hello, may I ask, the main difference between these four prism pairs is the difference in dispersion compensation? Why are prism pairs with flat dispersion curves and smaller absolute values more expensive? Is it because something else has an advantage in terms of pulse width compression?
jpolaris  (posted 2024-01-02 06:34:14.0)
Thank you for contacting Thorlabs. The difference in price between the dispersion-compensating prisms that we offer is primarily attributable to the procurement costs for the various substrates. For instance, CaF2 is more expensive as a raw material than UV fused silica, SF10, and N-SF14. In terms of specifications, the CaF2 prism has a better surface quality than the prisms made from the other substrates. When making a decision on which prism to use, you would need to determine how much compensation is needed, as well as to consider the amount of space available in your setup. The accumulated phase, and therefore also GD, GVD, and TOD, is dependent on the prism spacing and substrate material. We discuss this in greater detail at the following link: https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=3242&tabname=Tutorial
Patrick H  (posted 2023-07-11 14:31:02.993)
Is the SF-14 also usable for dispersion compensation at 1064nm?
cdolbashian  (posted 2023-07-21 02:02:14.0)
Thank you for reaching out to us Patrick. While we do not spec performance of these components with this particular wavelength, compensation can be achieved, and tuned, by changing the inter-prism spacing, as well as the incident angles on each prism. Fundamentals of the design and operation of such a prism pair can be seen in the paper which we reference in our "tutorial" tab of that product page: [1] R. L. Fork, O. E. Martinez, and J. P. Gordon, "Negative dispersion using pairs of prisms," Opt. Lett. 9, 150-152 (1984)
Fu min  (posted 2022-06-22 10:44:47.873)
I want to konw the Dispersion formula of SF10,and its reflractive index.
cdolbashian  (posted 2022-06-24 03:30:42.0)
Thank you for reaching out to us with this inquiry. I have contacted you directly with the appropriate information.
Paul Bittorf  (posted 2021-11-29 10:14:55.527)
Dear Sir or Madam, for a new ultrafast laser setup we plan to include multiple prism pairs for the fundamental (~800 nm), second and third harmonic for dispersion compensation. On your site, you present the prism pairs designed for the wavelength range of 700 - 900 nm, which is only perfect for the fundamental. Can we use the same prisms as well for the second (~400 nm) and third (~266 nm) harmonic or is there a big difference in the refractive index for such wavelengths and therefore a huge tradeoff? Also, does the material have an important impact besides the lenght of the compensator? Thank you for answering. Best regards, Paul Bittorf
cdolbashian  (posted 2021-12-22 04:51:04.0)
Thank you for reaching out to us Paul. These prisms can indeed be used for other wavelengths, though you would need to adjust the angle of the prisms and the inter-prism spacing. The governing equations and referenced research paper can be found on the "Tutorial" tab on this page. The material, N-BK7, has a known index of refraction which helps to determine the spacing and angle. Please feel free to reach out to us in the future with any other inquiries.
John Linden  (posted 2021-06-29 14:55:53.713)
Dear Rep. I am looking for a solution to correct beam ellipticity for UV 343nm with pulsewidth of ~250fs. is this product sufficient? I think the anamorphic prism pair is a good product but not sure of the compatability with ultrafast UV laser. Also it seems UV is not the designed wavelength for the AFS-FS as well. Appreciate any recomendation.
cdolbashian  (posted 2021-07-20 01:09:05.0)
Thank you for contacting us at Thorlabs. Given a 250fs pulsewidth for a laser in the UV range, a sufficient method for correcting beam ellipticity would be a pair of UV-fused silica, plano-convex lenses, rather than this prism pair. The function of this prism pair is used to correct for pulse broadening, which occurs when ultrafast pulses pass through a dispersive material. For your pulse duration, and the thickness of a standard lens of UV fused silica, the pulse broadening should be negligible. I have reached out to you directly to discuss different options.
lokirune  (posted 2018-08-26 23:48:22.507)
Hi. I bought this, but the size is smaller than I expected ( even I looked at the spec before I bought), so I couldn't put this in my system. Do you have other options with bigger size and coating for a lower wavelength(300nm)? If not, is it OK to buy just 2 right angle prisms? Thanks.
YLohia  (posted 2018-08-27 05:59:43.0)
Hello, thank you for contacting Thorlabs. We do have the ability to offer custom prism pairs, depending on your specific requirements and order quantity. Custom item orders should be placed through techsupport@thorlabs.com. We will reach out to you directly to discuss your requirements. Though an equilateral prism would likely work better, a right angle prism would work as well as long as you are prepared to solve the equations shown on the Tutorial tab of this page.
lokirune  (posted 2018-05-22 22:54:09.68)
about the AFS-CAF, is it OK to use with high power pulsed laser ? ( for example, 5ns 20W 532nm) Thanks.
YLohia  (posted 2018-05-23 08:59:13.0)
Hello, thank you for contacting Thorlabs. Unfortunately, we do not currently have test data for our uncoated CaF2 substrates since we do not manufacture the raw material itself. The actual damage threshold is also dependent on factors such as beam diameter and repetition rate of the laser (is 20 W the peak power or the average power?). That being said, CaF2 is general considered to be a "high damage threshold" material and is used in high power pulsed applications so it is likely for this to work for you, though we cannot provide any guarantees due to the lack of data.
user  (posted 2017-05-30 16:01:15.873)
Will these prisms work at longer wavelengths - e.g. in the range 2-3 micrometers? Do you offer any AR coated options please?
tfrisch  (posted 2017-06-22 09:09:11.0)
Hello, thank you for contacting Thorlabs. We can offer these with other coatings, but at longer wavelengths, the distance between the prisms will need to be increased to compensate for the lower dispersion of the substrate. Please contact TechSupport@Thorlabs.com to discuss the requirements of your custom application.
ethewalt  (posted 2015-11-05 13:20:59.727)
The GVD graph on the dispersion-compensating prism pair page needs to be fixed. The labels for fused silica and calcium fluoride are interchanged, and the horizontal gridlines don't correspond to the GVD axis labels.
besembeson  (posted 2015-11-05 01:48:40.0)
Response from Bweh at Thorlabs USA: Thanks for bringing this to our attention. We are going to fix the graphs on the website.
user  (posted 2015-08-25 22:07:19.71)
There is a typo in the formula for theta_f in the tutorial. There is an "arcsin" and an "asin". Choose one :)
myanakas  (posted 2015-08-28 03:48:27.0)
Response from Mike at Thorlabs. Thank you for your feedback. This typo has been corrected and we apologize for any inconvenience this may have caused.
tianzhushang  (posted 2013-08-22 14:50:37.537)
I've bought the AFS-SF14,How could I set the length between the two prisms and Can you e-mail me the data with the method to calculate it?
jlow  (posted 2013-09-17 17:06:00.0)
Response from Jeremy at Thorlabs: We have a sample calculations PDF linked in the "Specs" tab on our website. Our web team is looking to re-post a revised, updated version which should be clearer. We will contact your directly to discuss about this in more detail.
357510589  (posted 2012-10-16 09:50:54.14)
bdada  (posted 2011-10-07 18:19:00.0)
Response from Buki at Thorlabs: Thank you for your feedback. We will be happy to provide you with this data or the appropriate method to calculate it. We need the part number of the prism pair you are using and the configuration of the prisms. We also have MathCad files for some of our dispersion prisms that we have emailed to you. Please contact TechSupport@thorlabs.com if you have further questions.
avle  (posted 2011-09-26 23:55:36.0)
Hi there, I was wondering how we would calculate the negative dispersion parameter for a given prism set, with known optical properties and geometric configuration of the prism pair? The dispersion parameter in units of ps/nm Thanks!
jjurado  (posted 2011-06-24 18:27:00.0)
Response from Javier at Thorlabs to lionel.tombez: Thank you very much for contacting us. The AFS-SF10 dispersion compensation prism pairs will certainly transmit well at 1.1um. The prisms are cut at the same angle as Newport's prisms, so the performance will be identical.
lionel.tombez  (posted 2011-06-23 09:08:08.0)
Can one use these prisms (SF10) to compress 1.1 um pulses ? I dont get the difference with Newport prism compressor product, which is of the same material (SF10) but specified up to 2.3um. Thanks Customer Email: lionel.tombez@unine.ch This customer would like to be contacted.
Greg  (posted 2011-01-11 10:23:19.0)
A response from Greg at Thorlabs to william.t.lotshaw: Thank you for pointing out this error. The graphs were incorrect and have been fixed.
william.t.lotshaw  (posted 2011-01-10 11:22:58.0)
The GVD and TOD curves for fused silica and CaF2 appear to be reversed in the graphs tab if the values you list in the overview for the Brewster dispersion compensation prisms is correct
Javier  (posted 2010-06-17 10:39:21.0)
Response from Javier at Thorlabs to jessew: our specifications list the total TOD for the prism compressor. We derived the GVD and TOD values based on accumulated phase of the prism compressor. I can send you a document which outlines the derivation for these values.
jessew  (posted 2010-06-16 23:08:50.0)
Im unclear on how the numbers for TOD were derived for the prism pairs. Is the material TOD listed, or total TOD for a compressor constructed out of the prism pair? For example, using the standard definition of material TOD as the third derivative of wave number k with respect to angular frequency omega, TOD for SF10 should be 0.10 fs^3/rad per micrometer of material propagation. Thanks, -Jesse Wilson
Based on your currency / country selection, your order will ship from Newton, New Jersey  
+1 Qty Docs Part Number - Universal Price Available
AFS-CAF Support Documentation
AFS-CAFCaF2 Ultrafast Laser Dispersion-Compensating Prism Pair
AFS-FS Support Documentation
AFS-FSFused Silica Ultrafast Laser Dispersion-Compensating Prism Pair
AFS-SF10 Support Documentation
AFS-SF10SF10 Ultrafast Laser Dispersion-Compensating Prism Pair
AFS-SF14 Support Documentation
AFS-SF14N-SF14 Ultrafast Laser Dispersion-Compensating Prism Pair