1. Introduction

Arguably, the ideal near-eye display will be a monolithic, fully transparent ‘smart glass’ with 3D cues including accommodation. Current accommodative near-eye displays are far from monolithic and only achieve transparency by using beam splitters. Future direct-view, transparent holographic display portend significant fabrication complexity. Leaky mode modulators show promise as a low complexity solution for flat-screen holographic video display technologies, including near eye displays. We envision a geometry, like the one shown in Fig. 1(a), in which surface acoustic waves form variable lenses travel across a waveguide on a clear substrate in front of the viewer.

 

Fig. 1 Leaky-mode modulator layouts. (a) The traditional leaky-mode modulator with end exiting light. (b) The back side emitting leaky-mode modulator. (c) Target design; transparent, holographic, near-eye display.

Download Full Size | PDF

Current leaky-mode devices have been prevented from realizing this geometry because light exits the side, or smallest dimension of the substrate, as shown in Fig. 1(a) [1–5]. In this paper we demonstrate bottom-exit leaky mode devices [Fig. 1(b)], which bring us closer to our target transparent, holographic, flat-screen and near-eye displays [Fig. 1(c)].

2. Background

A leaky mode device operates as follows: TE light enters a proton exchanged waveguide on lithium niobate. Light trapped in the waveguide propagates head-on toward a counter-propagating pattern surface acoustic waves (SAW) generated by a transducer. This SAW pattern is encoded with holographic information. The guided light SAWs interact over an interaction length to couple light into a leaky mode. The leaky mode light is of orthogonal polarization and exits the device through the edge of the substrate to form a holographic image [6]. These devices enjoy the advantages of low fabrication complexity (usually fewer than three mask steps), polarization rotation of signal light, and frequency control of color. The edge of the device is small, usually between one half and three millimeters. The smallness of this dimension limits the usable interaction length because the projection of interaction length is windowed by the device edge. For leaky mode devices, larger interaction lengths lead to greater diffraction efficiencies. In flat screen form-factors where leaky mode devices are physically multiplexed, the larger the interaction length, the smaller the number of leaky mode devices needed to populate a given display area. In direct-view near-eye geometries, the interaction length defines the viewer’s clear aperture. Finally, the limitation of interaction length also has significant consequences for the maximum achievable angular and spatial resolution, as explained below.

The primary method of creating displays with leaky mode devices has been to optically multiplex the output of an end-exiting modulator using a telescope and scanning mirror, as shown Fig. 2(a). The resulting scanned aperture holographic video display will then typically be used to create holographic stereograms or fully computed holographic images. In both of these cases, the interaction length of is of particular importance. A larger interaction length leads to a larger number of addressable views in holographic stereograms as shown in Fig. 2(b). In fully computed holograms a larger interaction length corresponds to a smaller point spread function (PSF) and a greater depth resolution as shown in Fig. 2(c).

 

Fig. 2 The traditional method to use acousto-optic modulators (AOM) for holographic video displays. (A) The AOM is projected through a telescopic magnification, which is scanned across the display aperture. (B) The number of stereoscopic views is dependent upon overall angular output. (C) In fully computed holography, the point-spread function (PSF) dictates resolution parameters of the display.

Download Full Size | PDF

The mathematical relationship for the maximum number of independent views of a holographic stereogram, or maximum number of scannable points in a fully computed holographic display, by a leaky mode device is given as,

In contrast to the end exiting device, the backside emission device trades several advantages and disadvantage. To facilitate these comparisons, and simplify fabrication, the bottom-exit device was made in two sections. The first section is a traditional side-exit device, the second section is a bottom exit stage. The first section was characterized with our automated characterization apparatus [8,9]. The apparatus was upgraded for this test to include a rotational instead of linear actuator to directly measure output angles. Next, the second stage was attached and the resulting bottom-exit device was characterized. From this data the internal angle of the light was calculated by Snell’s Law. Figure 3(a) shows the k-map of the end-exit device. This range of internal angles was then applied to the backside emission device, shown in Fig. 3(b). Due to the near parallel nature of the output against the back of the sample, the addition of a high spatial frequency grating is required. The influence of this grating is shown by the $K_{grating}$ offset of $K_{light}$ in the k-map of this part of the figure. As can be seen in the k-map, the angular sweep of the output is significantly diminished.

 

Fig. 3 The k-map output of the end-exiting leaky-mode modulator (a) and the back side output layout (b).

Download Full Size | PDF

3. Design and fabrication

To create the required high spatial frequency grating in the back side emission device, we used surface relief gratings and induced internal gratings (by femtosecond laser micromachining). Of these two approaches, surface relief gratings have been the first to produce positive results and this method will be the focus of the remainder of the paper. The end-exit device was fabricated according to the recipe described in [10]. The process for fabricating the surface relief grating is described below.

To create surface relief gratings the sample is coated with KL5302 photoresist at 3000 RPM for 1 minute. The resist is then exposed with an interference lithography setup with a laser (λ = 325 nm, 35 μW/cm²) for 8 minutes. The exposed resist is then developed away, and 50 nm of nickel is deposited over the patterned resist. The sample is then soaked in Acetone to lift off the nickel on top of the resist, then put in a reactive ion etcher (300 W, 3.1 sccm Ar, 100 sccm CF4, 100 mTorr). This etching process is done in 10 minute blocks for 50 minutes to reach a depth of 200 nm. The time for this step could be reduced by using an ICP RIE, as reported in [11]. The remainder of the nickel mask is then etched away. This process is shown in Fig. 4, and produced a ~300 nm period surface relief grating.

 

Fig. 4 The fabrication process of the surface relief grating is described. (a) Resist is spun on the sample, (b) exposed, (c) developed, (d) nickel is deposited, (e) the nickel on the resist is lifted off, (f) devices are etched in an RIE, (g) device after etching, and (h) the nickel is etched off. (i) The completed output coupler, attached to an active device.

Download Full Size | PDF

4. Results

Output from the two devices was recorded and is shown in Fig. 5. The input power for the RF transducer was set to 250 mW and the laser was set to 100 mW. The total angular sweep of the end exiting device was measured as ~16° (left), with the backside emitting device (right) having an output sweep of ~2°. The measured output from the back-exiting device was corrected for the increased aperture size in relation to the radius of measurement. This data shows that the only functional change between the two devices is the angular sweep capability of each.

 

Fig. 5 Measured output for the end-exiting modulator (left) and the back-exiting device (right). Data for the back-exiting device was corrected for output aperture width.

Download Full Size | PDF

Due to the change in orientation of the device, in relation to output angle, and the transparency of the device, light can be viewed directly, resolving into virtual focal points. A camera was placed to view the output of the device, as shown in Fig. 6(a). The camera was then focused to the plane of the device, 4 feet beyond the device and 10 feet beyond the device, Fig. 6(b). Greater distances of focus were not attempted due to testing space constraints. The grids and distances in the images were provided via computer displays. The resolved output of the device matches with the expected point-spread function, both of which are shown in Fig. 6(c).

 

Fig. 6 The test setup for gathering data from the back-exiting modulator. The device was placed in view of the camera, and the camera was focused to the plane of the device, 4 feet, and ten feet beyond the device to the virtual focus at each depth (a). Images from each distance are shown (b), with a breakout to show detail of the point spread function (c). For a video of the above test, see Visualization 1.

Download Full Size | PDF

Noise in the system is caused primarily by light scattered and uncoupled at each coupling stage. This light can be minimized by improving the efficiency of each stage. The surface relief gratings currently function at ~20% efficiency but through optimization could reach 30%. However, there is a theoretical 100% efficiency if volume gratings are introduced [12]. The leaky-mode coupler only operates at ~10% efficiency but could be optimized to perform at 90% efficiency, as shown in [13]. However, these steps may be unnecessary, as Visualization 1 shows, the scattered light of the device is negligible at the virtual focus of the device.

5. Conclusion

The bottom exit leaky-mode modulator offers a new platform for near-eye displays that approaches our transparent, monolithic ideal. The device reported in this paper possesses a roughly 1cm clear aperture with a variable analog acoustic lens which sweeps the optical field at 400kHz and has a variable focus which transitions smoothly from 4ft to 10ft and beyond. The quality of focus can be improved by increasing the interaction length which is no longer tied to the substrate thickness. Increased interaction length also leads to a larger visual field preventing tunnel vision. The interaction length can be maximized by reducing waveguide loss and optimizing channel device.

The primary disadvantage of bottom-exit devices is that the angular sweep is greatly reduced compared to that of the end-exit device. Thankfully, wearable devices have lower required effective angles. The human eye only uses a small range of angles to identify high spatial frequency data. This range is the foveola, and maps between 1 and 1.2°, depending on individual anatomy [14]. As such, a display device worn over a single eye need only cover the angular range over which the human eye can accurately distinguish detail to be functionally viable. However, lower angular range does limit the visual extent of written images. A future paper will address two methods for overcoming this limitation.

The next step toward creating a near eye display with this device will be to determine how best to vertically multiplex the device output. The most straightforward path would be to simply fabricate enough channels to write all video lines simultaneously however, this should not be necessary as a single channel has enough bandwidth by itself to write a high definition image at ten depth planes at video rates. A more elegant solution would be multiplex one or a small number of channels optically while maintaining the monolithic, transparent form factor. Once accomplished, this device will enable next generation near-eye technologies.