Glan-Type Prisms
This is a continuation from the previous tutorial - prism polarizers.
Most prisms used at the present time are of the Glan type. Although they require considerably more calcite than Nicol types of comparable size, they are optically superior in several ways.
- Since the optic axis is perpendicular to the prism axis, the index of the extraordinary ray differs by a maximum amount from that of the ordinary ray. Thus, a wider field angle or a smaller L/A ratio is possible than with Nicol types.
- The light is nearly uniformly polarized over the field; it is not for Nicol types.
- There is effectively no lateral displacement in the apparent position of an axial object viewed through a (perfectly constructed) Glan-type prism. Nicol types give a lateral displacement.
- Since off-axis wander results in images which have astigmatism when the prism is placed in a converging beam, Glan types have slightly better imaging qualities than Nicol types.
Two other often-stated advantages of Glan-type prisms over Nicol types appear to be fallacious.
One is that the slanting end faces of Nicol-type prisms have higher reflection losses than the square-ended faces of Glan types. Since the extraordinary ray vibrates in the plane of incidence and hence is in the \(p\) direction, increasing the angle of incidence toward the polarizing angle should decrease the reflection loss.
However, the index of refraction for the extraordinary ray is higher in Nicol-type prisms (Glan types have the minimum value of the extraordinary index), so the reflection losses are actually almost identical in the two types of prisms.
The second ‘‘advantage’’ of Glan-type prisms is that the slanting end faces of the Nicol type supposedly induce elliptical polarization. This widely stated belief probably arises because in converging light the field in Nicol-type polarizers is not uniformly polarized, an effect which could be misinterpreted as ellipticity.
It is possible that strain birefringence could be introduced in the surface layer of a calcite prism by some optical polishing techniques resulting in ellipticity in the transmitted light, but there is no reason why Nicol-type prisms should be more affected than Glan types.
Glan-Thompson-Type Prisms
Glan-Thompson-type prisms may be either cemented or air-spaced.
Since, as was mentioned previously, an air-spaced Glan-Thompson-type prism is called a Glan-Foucault or simply a Glan prism, the name Glan-Thompson prism implies that the prism is cemented. Both cemented and air-spaced prisms, however, have the same basic design. The cemented prisms are optically the better design for most applications and are the most common type of prisms in use today.
The Glan-Thompson prism is named for P. Glan, who described an air-spaced Glan-Thompson-type prism in 1880, and for S. P. Thompson, who constructed a cemented version in 1881 and modified it to its present square-ended design in 1882.
These prisms are also sometimes called Glazebrook prisms because R. T. Glazebrook demonstrated analytically in 1883 than when rotated about its axis, this prism gives the most uniform rotation of the plane of polarization for a conical beam of incident light.
The cut in a Glan-Thompson-type prism is made parallel to the optic axis, which may either be parallel to two sides, as in Fig. 3a, or along a diagonal. The end faces are always perpendicular to the axis of the prism and contain the optic axis.

The extinction ratio obtainable with a good Glan-Thompson-type prism equals or
exceeds that of any other polarizer. Ratios of 5 parts in 100,000 to 1 part in 1 million can be expected, although values as high as 1 part in \(3\times10^7\) have been reported for small selected apertures of the prism.
The small residuals result mainly from imperfections in the calcite or from depolarization by scattering from the prism faces, although if the optic axis is not strictly in the plane of the end face, or if the optic axes in the two halves of the prism are not accurately parallel, the extinction ratio will be reduced.
Also, the extinction ratio may depend strongly upon which end of the prism the light is incident. When prisms are turned end for end, changes in the extinction ratio of as much as a factor of 6 have been reported.
When measuring the extinction ratio, it is essential that none of the unwanted ordinary ray, which is internally reflected at the interface and absorbed or scattered at the blackened side of the prism, reach the detector.
King and Talim found that they had to use two 4-mm-diameter apertures and a distance of 80 mm between the photomultiplier detector and prism to eliminate the \(o\)-ray scattered light. With no limiting apertures and a 20-mm distance, their measured extinction ratio was in error by a factor of 80.
The field angle of the prism depends both on the cement used between the two halves and on the angle of the cut, which is determined by the L/A ratio. Very large field angles can be obtained with Glan-Thompson prisms. For example, if the L/A ratio is 4, the field angle can be nearly 42°. Normally, however, smaller L/A ratios are used. The most common types of cemented prisms are the long form, having an L/A ratio of 3 and a field angle of 26°, and the short form, having an L/A ratio of 2.5 and a field angle of 15°.
Transmission
In Fig. 5 the transmission of a typical Glan-Thompson prism is compared with curves for a Glan-Taylor prism and a Nicol prism.

Thompson, B, Glan-Taylor, and C, Nicol prism. In the visible and near infrared regions the Glan-Thompson has the best energy throughput. In the near ultraviolet the Glan-Thompson may still be superior because the Glan-Taylor has such an extremely small field angle that it may cut out most of the incident beam.
The Glan-Thompson is superior over most of the range, but its transmission decreases in the near ultraviolet, primarily because the cement begins to absorb.
Its usable transmission range can be extended to about 2500 Å by using an ultraviolet-transmitting cement. Highly purified glycerin, mineral oil, castor oil, and Dow Corning DC-200 silicone oil, which because of its high viscosity is not as subject to seepage as lighter oils, have been used as cements in the ultraviolet, as have dextrose, glucose, and gedamine (a urea formaldehyde resin in butyl alcohol).
Transmission curves for 1-mm thicknesses of several of these materials are shown in Fig. 6, along with the curve for Canada balsam, a cement formerly widely used for polarizing prisms in the visible region.

Gedamine, one of the best of the ultraviolet-transmitting cements, has an index of refraction \(n_D=1.465\) and can be fitted to the dispersion relation
\[\tag{4}n=1.464+\frac{0.0048}{\lambda^2}\]
where the wavelength \(\lambda\) is in micrometers.
Figure 7 shows ultraviolet transmission curves for Glan-Thompson prisms with L/A ratios of 2.5 and 3 which are probably cemented with \(n\)-butyl methacrylate, a low-index polymer that has largely replaced Canada balsam.
Better ultraviolet transmission is obtained with a Glan-Thompson prism cemented with DC-200 silicone oil. Air-spaced prisms can be used to nearly 2140 Å in the ultraviolet, where calcite begins to absorb strongly. Transmission curves for two such prisms are shown in Fig. 7.
The Glan-Taylor, which is an air-spaced prism of the Lippich design, has a higher ultraviolet transmission than the Glan-Foucault, an air-spaced Glan-Thompson prism. The reason for this difference is that multiple reflections occur between the two halves of the Glan-Foucault prism, resulting in a lowered transmission, but are largely absent in the Glan-Taylor design.
The infrared transmission limit of typical Glan-Thompson prisms is about 2.7 μm although they have been used to 3 μm. The same authors report using a 2.5-cm-long Glan-Thompson prism in the 4.4- to 4.9-μm region.

Field Angle
Since many prism polarizers are used with lasers that have parallel beams of small diameter, field-angle effects are not as important as previously when extended area sources were used.
Other Glan-Thompson-Type Prisms
Other types of Glan-Thompson-type prisms include the Ahrens prism (two Glan-Thompson prisms placed side-by-side), Glan-Foucault prism (an air-spaced Glan-Thompson prism), Grosse prism (an air-spaced Ahrens prism), and those constructed of glass and calcite.
Lippich-Type Prisms
Lippich (1885) suggested a polarizing-prism design similar to the Glan-Thompson but with the optical axis in the entrance face and at right angles to the intersection of the cut with the entrance face (Fig. 3b).
For this case, the index of refraction of the extraordinary ray is a function of angle of incidence and can be calculated from Eq. (1) [refer to the prism polarizers tutorial] after \(\phi\), the complement of the angle of refraction of the wave normal is determined from Eq. (2) [refer to the prism polarizers tutorial].
In the latter equation, \(\beta\), the angle the normal to the surface makes with the optic axis, is 90° since the optic axis is parallel to the entrance face. Since the directions of the ray and the wave normal no longer coincide, the ray direction must be calculated from Eq. (3) [refer to the prism polarizers tutorial].
Lippich prisms are now little-used because they have small field angles, except for two; the air-spaced Lippich, often called a Glan-Taylor prism, and the Marple-Hess prism (two Glan-Taylor prisms back-to-back).
Glan-Taylor Prism
The Glan-Taylor prism, first described in 1948 by Archard and Taylor, has substantial advantages over its Glan-Thompson design counterpart, the Glan-Foucault prism. Since air-spaced prisms have a very small field angle, the light must be nearly normally incident on the prism face, so that the difference in field angles between the Glan-Taylor and Glan-Foucault prisms (caused by the difference in the refractive index of the extraordinary ray) is negligible.
The major advantages of the Glan-Taylor prism are that its calculated transmission is between 60 and 100 percent higher than that of the Glan-Foucault prism and the intensity of multiple reflections between the two sides of the cut, always a principal drawback with air-spaced prisms, is reduced to less than 10 percent of the value for the Glan-Foucault prism.
The calculated and measured transmittances of a Glan-Taylor prism are in reasonable agreement, but the measured transmittance of a Glan-Foucault prism (Fig. 7) may be considerably higher than its theoretical value. Even so, the transmission of the Glan-Taylor prism is definitely superior to that of the Glan-Foucault prism, as can be seen in Fig. 7. Extinction ratios of better than 1 part in \(10^3\) are obtainable for the Glan-Taylor prism.
A final advantage of the Glan-Taylor prism is that it can be cut in such a way as to
conserve calcite. Archard and Taylor used the Ahrens method of spar cutting described by Thompson and found that 35 percent of the original calcite rhomb could be used in the finished prism.
In a modified version of the Glan-Taylor prism becoming popular for laser applications, the cut angle is increased, the front and back faces are coated with antireflection coatings, and portions of the sides are either covered with absorbing black glass plates or highly polished to let the unwanted beams escape.
The effect of increasing the cut angle is twofold: a beam normally incident on the prism face will have a smaller angle of incidence on the cut and hence a smaller reflection loss at the cut than a standard Glan-Taylor prism, but, at the same time, the semi-field angle will be reduced throughout most of the visible and near-infrared regions.
A new type of air-spaced prism has a very high transmittance for the extraordinary ray. It resembles the Glan-Taylor prism in that the optic axis is parallel to the entrance face and at right angles to the intersection of the cut with the entrance face. However, instead of striking the prism face at normal incidence, the light is incident at the Brewster angle for the extraordinary ray (54.02° for the 6328-Å helium-neon laser wavelength), so that there is no reflection loss for the \(e\) ray at this surface.
Since the ordinary ray is deviated about 3° more than the extraordinary ray and its critical angle is over 4° less, it can be totally reflected at the cut with tolerance to spare while the extraordinary ray can be incident on the cut at only a few degrees beyond its Brewster angle.
Thus this prism design has the possibility of an extremely low light loss caused by reflections at the various surfaces. A prototype had a measured transmission of 0.985 for the extraordinary ray at 6328 Å. If the prism is to be used with light sources other than lasers, its semi-field angle can be calculated.
A major drawback to the Brewster angle prism is that since the light beam passes
through a plane-parallel slab of calcite at nonnormal incidence, it is displaced by an amount that is proportional to the total thickness of the calcite. Some of the prisms are made with glass in place of calcite for the second element. In this case, the beam will usually be deviated in addition to being displaced. Measurements on a calcite-glass prototype at 6328 Å showed that the output beam was laterally displaced by several millimeters with an angular deviation estimated to be less than 0.5°.
Marple-Hess Prism
If a larger field angle is required than can be obtained with a Glan-Taylor prism, a Marple-Hess prism may be used. This prism, which was first proposed in 1960 as a double Glan-Foucault by D. T. F. Marple of the General Electric Research Laboratories and modified to the Taylor design by Howard Hess of the Karl Lambrecht Corporation, is effectively two Glan-Taylor prisms back-to-back.
The analysis for this prism is made in the same way as for the Glan-Taylor prism and Lippich-type prisms in general, keeping in mind that the refractive index of the ‘‘cement’’ is 1 since the components are air-spaced.
Since the ordinary ray is totally reflected for all angles of incidence by one or the other of the two cuts, the field angle is symmetric about the longitudinal axis of the prism and is determined entirely by the angle at which the extraordinary ray is totally reflected at one of the two cuts. This angle can be readily calculated. The field angle is considerably larger than for the Glan-Foucault or Glan-Taylor prism and does not decrease as the wavelength increases.
Unlike the Glan-Foucault or Glan-Taylor prisms, which stop being efficient polarizers when the angle of incidence on the prism face becomes too large, the Marple-Hess prism continues to be an efficient polarizer as long as the axial ordinary ray is not transmitted.
If the prism is used at a longer wavelength than the longest one for which it was designed (smallest value of \(n_o\) used to determine the cut angle), the value of \(n_o\) will be still smaller and the critical angle for the axial ordinary ray will not be exceeded.
Thus the axial \(o\) ray will start to be transmitted before off-axis rays get through. When this situation occurs, it only makes matters worse to decrease the convergence angle. Thus, there is a limiting long wavelength, depending on the cut angle, beyond which the Marple-Hess prism is not a good polarizer. At wavelengths shorter than the limiting wavelength, the Marple-Hess prism has significant advantages over other air-spaced prism designs.
It is not easy to make a Marple-Hess prism, and the extinction ratio in the commercial model is given as between \(1\times10^{-4}\) and \(5\times10^{-5}\), somewhat lower than for a Glan-Taylor prism. On the other hand, even though the Marple-Hess prism has an increased L/A ratio, 1.8 as compared to 0.85 for a Glan-Taylor prism, its ultraviolet transmission is still superior to commercially available ultraviolet transmitting Glan-Thompson prisms of comparable aperture.
Frank-Ritter-Type Prisms
The third general category of Glan-type polarizing prisms is the Frank-Ritter design. Prisms of this type are characterized by having the optic axis in the plane of the entrance face, as in other Glan-type prisms, but having the cut made at 45° to the optic axis (Fig. 3c) rather than at 0°, as in Glan-Thompson prisms, or at 90°, as in Lippich prisms.
Frank-Ritter prisms are particularly popular in the Soviet Union, and over 80 percent of the polarizing prisms made there have been of this design. Usually double prisms comparable to the Ahrens modification of the Glan-Thompson are used, primarily because from a rhombohedron of Iceland spar two Frank-Ritter double prisms can be obtained but only one Ahrens of the same cross section or one Glan-Thompson of smaller cross section.
However, this apparent advantage can be illusory since Iceland spar crystals often are not obtained as rhombs. For example, if the natural crystal is in the form of a plate, it may be less wasteful of material to make a Glan-Thompson or Ahrens prism than a Frank-Ritter prism.
Optically, Frank-Ritter prisms should be similar to Glan-Thompson and Ahrens types, although the acceptance angle for a given L/A ratio is somewhat smaller since the refractive index of the extraordinary ray is larger than \(n_e\) in the prism section containing the longitudinal axis and perpendicular to the cut. In practice, the degree of polarization for a Frank-Ritter prism seems to be quite inferior to that of a good Glan-Thompson or even an Ahrens prism.
Use of Glan-Type Prisms in Optical Systems
Several precautions should be taken when using Glan-type prisms in optical systems:
- The field angle of the prism should not be exceeded.
- There should be an adequate entrance aperture so that the prism does not become the limiting aperture of the optical system.
- Baffles should be placed preceding and following the prism to avoid incorrect collection of polarized light or extraneous stray light.
Common Defects and Testing of Glan-Type Prisms
Several common defects are found in the construction of Glan-type prisms and limit their performance:
- The axial beam is displaced as the prism is rotated. This defect, called squirm, results when the optic axes in the two halves of the prism are not strictly parallel. A line object viewed through the completed prism will oscillate as the prism is turned around the line of sight.
- The axial ray is deviated as the prism is rotated. This defect is caused by the two prism faces not being parallel. A residual deviation of 3 minutes of arc is a normal tolerance for a good Glan-Thompson prism; deviations of 1 minute or less can be obtained on special order.
- The optic axis does not lie in the end face. This error is often the most serious, since if the optic axis is not in the end face and the prism is illuminated with convergent light, the planes of vibration of the transmitted light are no longer parallel across the face of the prism.This effect, which in Nicol-type prisms gives rise to the Landolt fringe, is illustrated in the following practical case. For a convergent beam of light of semi-cone angle \(i\), the maximum variation of the plane of vibration of the emergent beam is \(\pm\gamma\), where, approximately, \[\tag{5}\tan\gamma=n_e\sin{i}\tan\phi\] and \(\phi\) is the angle of inclination of the optic axis to the end face, caused by a polishing error. For \(i=3°\) and \(p=5°\), the plane of vibration of the emergent beam varies across the prism face by \(\pm23\) minutes of arc. Thus, good extinction cannot be achieved over the entire aperture of this prism even if nearly parallel light is incident on it. The field angle is also affected if the optic axis is not in the end face or is not properly oriented in the end face, but these effects are small.
- The cut angle is incorrect or is different in the two halves of the prism. If the cut angle is slightly incorrect, the field angle may be decreased. This error is particularly important in Glan-Foucault or Glan-Taylor prisms, for which the angular tolerances are quite severe, and a small change in cut angle for these prisms may greatly alter the field angle. If the cut angles are different in the two halves of the prism, the field angle will change when the prism is turned end-for-end. The field angle is determined by the cut angle in the half of the prism toward the incident beam. Differences in the two cut angles may also cause a beam deviation. If the angles in the two halves differ by a small angle a that makes the end faces nonparallel, the beam will be deviated by an angle \(\delta=\alpha(n_e-1)\). If instead, the end faces are parallel and the difference in cut angle is taken up by the cement layer which has a refractive index of approximately \(n_e\), there will be no deviation. However, if the prism is air-spaced, the deviation \(\delta'\) caused by a nonparallel air film is approximately \(\delta'=\alpha{n_e}\), illustrating one reason why air-spaced prisms are harder to make then conventional Glan-Thompson prisms.
- The transmittance is different when the prism is rotated through 180°. A potentially more serious problem when one is making photometric measurements is that the transmission of the prism may not be the same in two orientations exactly 180° apart. This effect may be caused by the presence of additional light outside the entrance or exit field angle, possibly because of strain birefringence in the calcite.
Two factors which limit other aspects of polarizer performance in addition to the extinction ratio are axis wander, i. e., variation of the azimuth of the transmitted beam over the polarizer aperture, and the ellipticity of the emergent polarized beams caused by material defects in the second half of the prism.
In order to determine the cut angle, field angle, parallelism of the prism surfaces, thickness and parallelism of the air film or cement layer, and other prism parameters, one can use the testing procedures outlined by Decker et al., which require a spectrometer with a Gauss eyepiece, laser source, and moderately good polarizer. (Other testing procedures have been suggested by Archard.)
Rowell et al. have given a procedure for determining the absolute alignment of a prism polarizer. However, they failed to consider some polarizer defects, as pointed out by Aspnes who gives a more general alignment procedure that compensates for the prism defects. (There is also a response from Rowell.)
The next tutorial discusses in detail about single-mode fibers for communications.