Cube beamsplitters are used in a wide range of applications that require reduced ghosting, matching path lengths, and restricted beam shift. Polarizing beamsplitter (PBS) cube assemblies (Figure 1) prove useful when reduced mechanical vibration or a small form factor is needed. These products are found in phase shifting interferometers, pico-projectors, head-up displays (HUD), and head mounted displays (HMD).
Figure 1. Cube beamsplitter application examples.
MacNeille cubes are often used in PBS applications, but for low f/# applications, the contrast ratio and the color balance between blocking and passing state transmittance are not quite suitable. In contrast, wire grid polarizers or WGP provide excellent broadband performance in low f/# applications.
With a WGP integrated along the hypotenuse of a cube beamsplitter, the Moxtek ICE Cube™ offers an excellent broadband PBS in low f/# applications that demand high contrast and efficiency, without the angular performance variations seen in MacNeille beamsplitters.
Transmission Mode Performance Results at Low f/#
To demonstrate the transmission mode performance in low f/# applications, a white light LED source was collimated and allowed to pass through an iris then a high-contrast pre-analyzer, in order to choose an input polarization state, as shown in Figure 2(a). Then, the beam was directed towards Moxtek (bottom row) and MacNeille (top row) cube beamsplitters, reducing the blocking state transmittance on a screen.
A camera was used to image the blocking state transmittance (Ts) and reflectance (Rs) (Figures 2b, 2e). As soon as the pre-analyzer was rotated to 90°, the passing state reflectance (Rp) and transmittance (Tp) were again imaged with the camera (Figures 2c, 2f). The pre-analyzer was turned through around 45° to provide equal intensity in the transmitted and reflected beams to approximate a PBS configuration.
Figure 2. Low f/# performance comparison between polarizing beam splitter cubes. Measurement schematic (a) and results for MacNeille cube (b-d) and Moxtek ICE Cube (e-g) showing blocking state (b, e), passing state (c, f), and beamsplitting (d, g) configurations.
When compared to the MacNeille cube, the ICE Cube has enhanced color balance and minimum amount of leakage in the blocking state (Ts), as shown in Figure 2. Both cubes were characterized at variable angles with collimated light from the UV to the SWIR, and considerable improvement was seen in the Moxtek PBS cube, which accomodated angular deviations with very little variation in performance.
Using the setup from Figure 2, transmittance measurements were also performed using a detector (ThorLabs 5120C) in place of the screen. For a cone angle of ±16°, the Moxtek ICE Cube exhibited a contrast ratio between blocking and passing state transmittance (Tp/Ts) of 1470, whereas the MacNeille cube had a contrast ratio of just 23.
When compared to the raw transmittance without the cube, passing state transmittance (Tp) for the MacNeille and Moxtek cubes were 78.6% and 74.8%, respectively. For a smaller cone angle of ±4.4°, the Tp and contrast ratio of the MacNeille cube improved to just 81.7% and 103 respectively, while the Tp and contrast ratio for the ICE Cube increased to 81.4% and 6460, in that order.
With the help of the portable fiber spectrometer, the white LED emission spectra was collected, showing an intense peak at 450 nm, and broadband emission in the green and red wavelengths (Figure 3).
Figure 3. White light LED emission spectrum
LCOS Display Application Results
LCOS or DMD reflective panels are used in HUD, HMD, and pico-projectors. In order to replicate an LCOS display configuration by means of a PBS cube, a mirror and a broadband quarter waveplate were located in the transmitted beam path just after the cube, and the beam, which was retro reflected, was orthogonally separated, allowed to travel through a clean-up polarizer, and ultimately imaged onto a screen or quantified by means of a detector (Figure 4).
Figure 4. Performance comparison between PBS cubes for low f/# display applications. Measurement schematic (a) and results for Moxtek ICE Cube (b) and MacNeille cube (c) showing the off-state with Maltese Cross.
Optimization of the quarter-waveplate orientation altered the p-polarization state travelling through the cube and into spherical polarized light. This in turn becomes polarized upon reflection, and after passing back through the quarter-waveplate, it is changed to s-polarized light.
Following reflection off the beamsplitter, the s-polarization travels through the clean-up polarizer and creates an on-state. In contrast, aligning the waveplate’s slow or fast axis with the p-polarization state does not change the linear polarization state. Most of the retro-reflection off of the mirror travels through the cube, and the clean-up polarizer filters out the Rp light, creating the off-state image.
The off-state leakage of the Moxtek and MacNeille cubes is illustrated in Figure 4, parts (b) and (c), for an f/# of 1.74. For a smaller cone angle of ±4.4°, the quantified contrast ratio between on and off states for the MacNeille and Moxtek cubes were 4,940 and 6,600, respectively, whereas system efficiencies were 55.8% and 41.2%. For f/# of 1.74, the small detector would fail to capture a large part of the beam, and hence, a collimating lens would need to be inserted before the detector to achieve significant results.
For low f/#, the non-normal incidence beam that interacts with the quarter-waveplate could restrict system performance, as shown in Figure 3. Foctek Photonics supplied the quarter waveplate, Acktar Black Metal Velvet provided the beam dump, the mirror was New Focus #5108, while Moxtek provided the UVD240A clean-up polarizer. The Moxtek ICE Cube’s beamsplitting function, where the prism hypotenuse plane defines the p- and s-polarization states, is shown in Figure 5.
Figure 5. Moxtek ICE Cube beamsplitter schematic showing embedded wire grid polarizer as well as passing state transmittance (Tp) and blocking state reflectance (Rs). The weaker passing state reflectance (Rp) and blocking state transmittance (Ts) can be further removed using clean-up polarizers.
The reflected beam has both s-and p-polarization states (Rs and Rp), and the passing state transmittance is p-polarized with high purity (Tp). With the help of a clean-up polarizer or pre-analyzer, the p-polarized reflectance can be removed. Table 1 shows the ICE Cube opto-mechanical design and environmental performance details.
Table 1. ICE Cube features
||1 inch cube (inquire for other dimensions)
||embedded wire grid along hypotenuse
±15-20° polar (elevation)
||< 3 arcmin
||5.5 W/cm2 and 90°C max
*Performance can be extended into short wavelength infrared (SWIR) by omitting AR-coating on cube faces.
In near-eye display applications, large angular field of view and broadband performance are critical factors, whilst the highest luminous flux and operating temperature holds significance in standard projection display applications. Angular deviation and transmitted wavefront distortion (TWD) measurements are essential in interferometry applications.
Variable Angle Performance
Schematics of an Agilent spectrometer with variable angle Universal Measurement Accessory are shown in Figure 6. Part (a) shows the source-side, grating-based monochromater equipped with rotatable polarizing pre-analyzer.
Figure 6. (a) Simplified measurement schematic in normal incidence (0°) configuration. (b) Inside view of Agilent Universal Measurement Accessory. (c) Measurement configurations for an azimuthal sample angle of -θ
The pre-analyzer enabled sample characterization using p- and s-polarized light. Fixed to a central stage, the beamsplitter sample could rotate around 360°. The detector was placed on an extended arm, which could remain at the home position (marked A) to calculate normal-incidence transmittance, or turn around to the 90° position (marked B) to calculate reflectance.
When the sample was rotated in a clockwise (+θ) or counter-clockwise (-θ) direction, the detector continued to stay in the home position A’ for the variable angle transmittance measurement, but shifted to the 90 + θ position (B’) or 90 - θ position for counter-clockwise and clockwise rotations, in that order.
In these measurements, the instrument beam was well-collimated, with cone angles of ±1° and ±2° in the lateral and vertical directions, as shown in Figure 6 (b). In Figure 6 (c), the cube’s azimuthal rotation by an angle of -θ is shown.
Two sets of corresponding stacked detectors were integrated in the Agilent spectrometer with Universal Measurement Accessory. For a third of time period, the beam was directed to the back of the detector through a three stage chopper to facilitate a baseline drift correction. It then fully blocked the beam for another third of the time to allow a dark reference correction. For the rest of the time, the three stage chopper directed the beam towards the front detector and sample to allow the preferred measurement.
A mixed polarization state in the fixed rear beam relied on the source distribution and the grating and internal mirrors of the spectrometer; however, the front beam exhibited a different polarization state and beam path.
The pre-analyzer was used to select a single orientation, which was mostly p- or s-polarization. Large jumps can occur in the raw baseline spectra of the source at the detector and grating crossover points. To reduce this jump, a Hanle depolarizer can be placed in the front beam path of the Agilent auto-polarizer position, but for larger beam angles, a small jump in the cube spectra can still be seen at the detector crossover.
Variable Angle Performance Results
The source is extended and broadband in many beamsplitter applications, rendering beam collimation unsuitable. This provides a combination of rays from different azimuthal and polar angles that interact with the beamsplitter.
MacNeille beamsplitting cubes feature a stack of thin films which are coated along their hypotenuses so as to improve a Brewster-like asymmetry in reflectance between p- and s-polarization states, but such an arrangement fails to provide a large angular field of view and broadband performance at the same time.
In contrast, the Moxtek ICE Cube uses the ProFlux® sub-wavelength aluminum Nanowire® grid design that isolates the beam polarizations at the surface of the wire grid through an anisotropic reflection and absorption mechanism. This arrangement offers both superior contrast and reliable passing state transmittance for broadband applications requiring large angular field.
Variable angle passing state transmittance from 0° to ±30° angle of incidence are compared in Figure 7. When compared to the MacNeille Cube, the AR-coated Moxtek Cube provided better performance in terms of varying angle of incidence. This reduces design considerations and enables a better usage of the luminous output from extended and broadband sources.
Figure 7. Variable angle passing state transmittance (Tp) performance comparison for a typical Moxtek ICE Cube and competing beamsplitters for 0° to ±30° angle of incidence. (a) AR-coated Moxtek ICE Cube performance. (b) MacNeille Cube performance.
The Agilent Universal Measurement Accessory and the Harrick variable angle transmittance accessory, placed in a CARY 5000 UV-Vis-IR spectrometer, were used to determine the unwanted leakage through the polarizing beamsplitter cubes (Ts).
The latter product allowed an increased beam cone angle, reduced noise, and more optimized detectors. To determine the blocking state transmittance (Ts), the transmission axis of each beamsplitter was crossed with a high contrast pre-analyzer — the tripled Moxtek UVT240 polarizer.
Variable angle contrast ratio (Tp/Ts) performance from 0° to ±30° angle of incidence for polar and azimuthal angles is compared in Figure 8. When compared to the MacNeille Cube, the AR-coated Moxtek Cube provided a better performance with varying angle of incidence. This would enhance the contrast ratio from extended and broadband sources.
Figure 8. Variable angle contrast ratio (Tp/Ts) performance for a typical Moxtek ICE Cube and competing beamsplitters for 0° to ±30° angle of incidence. (a) AR-coated Moxtek ICE Cube performance. (b) MacNeille Cube performance. Dashed and dotted lines are for azimuthal and polar angles, respectively.
A clean-up polarizer can be used to remove redundant reflectance (Rp) from the p-polarized state, but in order to maintain uniform color balance and brightness, a flat response in s-polarized reflectance (Rs) across a wide range of angles and wavelengths was required.
Figure 9 compares the passing state reflectance and blocking state reflectance (Rp/Rs) from 0° to ±30° azimuthal angle of incidence. When compared to the MacNeille Cube, the Moxtek Cube once again provided more reliable performance with varying angle of incidence, leading to minimum color shift.
Figure 9. Variable angle reflectance (Rp, Rs) for a typical Moxtek ICE Cube and competing beamsplitters from 0° to ±30° azimuthal angles of incidence (a) AR-coated Moxtek ICE Cube performance. (b) MacNeille Cube performance. Dashed lines with red color map indicate Rs while dotted grayscale lines denote Rp.
Overall efficiency is a critical factor for a polarizing beamsplitter. The efficiency for 0° to ±30° azimuthal angle of incidence is compared in Figure 10.
Figure 10. Variable angle efficiency (Tp x Rs) for a typical Moxtek ICE Cube and competing beamsplitters from 0° to ±30° azimuthal angles of incidence. (a) AR-coated Moxtek ICE Cube performance. (b) MacNeille Cube performance. Darker colored curves strike the hypotenuse closer to normal incidence, while lighter colors are more glancing angle.
The Moxtek ICE Cube (Figure 11 and 12) had flat spectral performance with varying angle of incidence, while the MacNeille cube showed poor spectral uniformity in beamsplitting efficiency for all angles, except normal incidence. This can improve contrast and color uniformity in display applications, and can even improve fringe visibility for the Moxtek Cube in broadband interferometric applications.
Figure 11. ICE Cube-C
Figure 12. Moxtek ICE cube
The Moxtek ICE Cube contains a pair of glass prisms that are joined to the wire grid plate through optical cement. This setup eliminates the use of the ICE Cube in high flux and high temperature applications. An integrated plate wire grid polarizer in the ICE Cube uses inorganic materials that are akin to Moxtek’s visible spectrum polarizer products.
These products provide sustained performance in high humidity and high temperature projection display applications. Also, the integrated Nanowire® design of the Moxtek ICE Cube series beamsplitter protects against environmental contamination and handling damage, providing better reliability in humid conditions.
The Moxtek ICE Cube polarizing beamsplitter delivers excellent broadband performance across a broad angular aperture, without any of the angular performance variations and color shifts that are evident in MacNeille cube beamsplitters. The Moxtek ICE Cube can integrate angular deviations with reduced performance variation, which corresponds to a 20-40° field angle, thereby allowing better use of light when poorly collimated sources are used.
The Moxtek ICE Cube, with Nanowire® grid polarizer, is perfect for applications where large angular fields are needed. To realize greater efficiency, flatter angular response and extended broadband performance, a plate-style wire grid polarizing beamsplitter like PBF and PBS can be used, provided ghost images and matching path lengths are not a major concern.
Moxtek / Youtube
This information has been sourced, reviewed and adapted from materials provided by MOXTEK, Inc.
For more information on this source, please visit MOXTEK, Inc.