Dynamic multilevel spiral phase plate generator

The design and characterisation of a reconfigurable multi-level spiral phase plate is shown. The device is based on a pie-shape liquid-crystal structure with 24 slices driven by custom electronics that allow independent excitation control of each electrode. The electrooptical cell was manufactured using maskless laser ablation lithography and has shown an unprecedented high fill factor. The topological charge can be dynamically changed between 1, 2, 3, 4, 6, 8 and 12. The device has been calibrated and characterised at 632.8 nm but can be employed at any wavelength in the visible and near infrared spectrum, just modifying the driving parameters of the electrodes. The experimental results have been compared to predictions derived from simulations. An excellent correspondence between theoretical and experimental result has been found in all cases.

Conventional SLMs are generally hampered by a pixelated structure that generates aliasing and reduces the fill factor of the device leading to a loss of efficiency. Less generic electro-optical reconfigurable LC multi-level SPP devices for generating optical vortex based on discrete 25 or interconnected [26][27][28] electrodes, with reduced interpixel space have been demonstrated.
This work reports the design, simulation and manufacturing of a reconfigurable LC multi-level SPP, with minimal interpixel-space and maximal direct control of the individual switching sectors. The LC cell is based on the previously reported device 25 with several novelties: the electrodes pattern is created using a maskless laser ablation 29,30 , resulting in a very high fill factor of electrodes; the manufactured device consist in 24 pie slices sections which is twice as many as any commercially available SPP; a custom made cost-efficient, pulse-width modulation driver was developed and manufactured for device driving; finally the manufacturing protocol makes it possible to increase the number of sections arbitrarily, without limitations in neither the fabrication process nor in the driver.

Methods
Manufacturing. An LC cell consists of two opposing transparent ITO electrodes treated with an alignment agent, separated by spacers. The gap between the electrodes filled with an LC and sealed.
A CAD program was used to design the electrode layout. A 50 × 50 mm 2 ITO covered 1.1 mm thick glass slide was used for the patterned bottom electrode and single electrode of 30 × 20 mm 2 ITO glass slide was used as counter electrode The active area is 10 × 10 mm 2 . The active electrode pattern consists of 24 pie slices. Each pie slice is connected to a contact pad measuring 2 × 4 mm 2 (Fig. 3a).
The ITO was engraved by CAD controlled ablation system (Lasing S.A., Madrid, ES) fitted with a 300 mW, 349 nm laser (Explorer Laser Spectrum Physics, Mountain View, US), using a back-scribing approach, writing through the glass 30 . The resulting ITO electrode separations were in the order of 1.5-2.0 µm (Fig. 3b).
The minimal electrode separation is important particularly in the centre of the active area, where all the cutting lines meet. The design was made so that different cutting parameters could be employed in different sectors.
Cells were mounted using rubbed polyimide as alignment layer and MDA-98-1600 as LC. The polyimide layer induces a homogenous alignment of the LC molecules parallel to the cell surface in the direction of the rubbing. The LC a is positive nematic, which will tend to align parallel to any applied electrical field. The degree of switching depends on the torque applied by the electrical field and the strength of the polyimide anchoring. The stronger the field applied, the more the molecules will tend to orient perpendicularly to the cell surface; once the field is removed the LC will relax back into the state dictated by the alignment layer.  The cell thickness (d) was ensured using 9 µm wide silicon spheres as spacers in the sealing gasket. The cells were filled by capillarity action. After the LC filling, each cell was wired and electrically tested. The cells were connected to a multichannel voltage generator, described below, and were assessed in white light between crossed polarizers to check the quality of the sample (Fig. 4).
A driver to control the vortex generator was manufactured. An Arduino UNO clone (Funduino) was employed as a USB-SPI interface for transferring a 24 × 12 bit array to a 24-channel 12-bit pulse-width-modulation (PWM) LED driver (TLC5947, Texas Instruments implemented in the Adafruit 1429 board). The 24 output channels of this driver were used as control inputs in six four-channel high speed analogue switches (ADG5412 SPST, Analog Devices Inc.) connecting, or not, the individual SPP electrodes to a 6 V, 10 kHz square waveform signal generated by an external arbitrary waveform generator (Stanford Instruments DS345) (Fig. 5). The refresh rate of the LED driver was measured to be 1.4 ms, which is far shorter than the relaxation time of a 9 µm thick liquid crystal cell.
The electronic driver, and the user interface allows for the generation of the desired multilevel discrete SPP profiles needed for the generation of the vortices (Fig. 6).  Characterization. To establish the relationship between applied PWM duty cycle and phase retardation, the device was situated with the alignment direction (the switching plane) at 45° between two crossed polarizers, at 0° and 90° respectively, and illuminated by a He-Ne laser with a wavelength λ ( ) of 632.8 nm. In this setup the intensity varies with the phase difference δ ( ) caused by the plano birefringence   derived, employing Jones matrix formalism (see for instance Caputo et al. 31 ). In this study, both camera dark current, and polariser imperfections have been ignored, since the resulting intensity variation is normalised. To ensure a CCD gamma factor of 1, only the green pixels were employed in the intensity calculations. A digital camera (Nikon D500) fitted with a macro-lens was used for the light intensity reading. Raw image data for each separate colour pixel on the CMOS were captured.
In order to illuminate most of the device, the laser beam was expanded to 8 mm diameter using a beam expander.
Four centric regions of the device were chosen arbitrarily to be characterized, each region covering approximately 1/10 of the device. One common excitation signal was applied to the whole device during the characterization, and the intensity readings of each region were captured and averaged for each of the colour pixels separately.
During the calibration the illumination over the sample was not perfectly uniform (Fig. 7a), however, this does not affect data acquisition, since only relative intensity variations are relevant.
The birefringence evolution as a function of the PWM duty cycle (dc) was found to be well described by a function δ = + − · Ae C B dc over a full 2π range δ π π ∈ ( [3 ; ]). The agreement between the fit (A = 17.0, B = −5.38 and C = 0.15) and the experimental data is shown in Fig. 7b.
The fitted curve was employed to calculate the PWM for each retardation level used in the different topologies.

Results
The 24 slices of the SPP allow for seven different topological charges [1, 2, 3, 4, 6, 8 and 12], employing 24 different excitation levels, deduced from the preceding calibration. The excitation of the different topologies was first imaged between crossed polarisers at 45° to the switching plane as employed for the calibration (Fig. 8 top row). The darkest areas correspond to minimum δ sin 2 , i.e. where the introduced phase difference δ π ≈ 2 , and the brightest areas correspond to δ π ≈ or δ π ≈ 3 . Subsequently the second polariser was removed, and the SPP was placed such that the switching plane was aligned with the impinging polarization. The beam expander and the macro objective were removed, and the naked light beam originating in a conventional HeNe laser tube, with no focusing elements, was shone directly onto the camera CMOS chip.
The resulting intensity distributions can be seen in Fig. 8 (middle row). The zero-power density, singularity in the centre of the beam with a radius proportional to the topological charge, is clearly visible illustrating the expected functioning of the device. No significant power variations were detected when changing the sign of the topology or rotating the SPP profile.
All the camera and laser parameters were kept constant during the capture of the switching pattern and the diffraction patterns but changed in between the two sets of data. Thus, the intensity reading from image to image is directly comparable horizontally, but not vertically.
The diffraction pattern showed not only the principal diffraction ring, but also a weaker higher order diffraction ring (most visible in the image of the diffraction pattern of topology 8 (Fig. 8 middle row). This unexpected pin-hole like diffraction pattern motivated a comparison of the experimental results with a simulation of the setup.
A simple train of optical elements were cascaded mathematically to reach a description similar to the one proposed by others 28,32 : a Fourier transform of a gaussian power distribution of the initial laser, is filtered by a circular aperture and the SSP before being transformed back by an inverse Fourier transform into the power distribution depicted in the last row in Fig. 8. The simulation showed that the PIN-hole pattern was intrinsic to optical setup, and not a consequence of imperfections in the manufactured SPP. The major divergence between the measured and the simulated diffraction pattern is seen in the approximation to the initial gaussian distribution of the incident laser, and the actual measured power distribution.
The first row of Fig. 8 shows that imperfections were present in the presented SSP especially in the periphery of the device. The imperfections in the upper part of the device are due to a damaged alignment layer, probably caused by the manual handling of the device. The visible imperfection is approximately 5 mm from the device centre, which means that it does not affect the employed laser which, as seen in the first photo of the second row, has a 1.7 mm waist. In this latter image, it can be seen that the impinging light laser beam can only loosely be described as a gaussian beam. Nonetheless, visually high contrast ratios are achieved in the images of the diffraction rings for all the topologies. The reason is that all the spatial intensity variation in the SSP and in the laser source is averaged out over the whole diffraction pattern. Simulations introducing random noise in both the laser intensity profile, as well as transmission and phase variation in the SSP, resulted in a reduction of the contrast of the diffraction patterns, in a direct proportion to the area, and intensity affected.
The simulation model allowed for evaluation of more complex devices. The neatness of diffraction ring depends on the number of pie slices that makes up the SSP. The closer to an infinite number of slices in each 0-2π period, i.e. a continuous variation, the neater the generation of the diffraction rings, as can be appreciated in the simulations shown in Fig. 9.

Conclusions
A 24 pie-slice reconfigurable SPP with a custom-built driver has been manufactured and successfully tested. The electrodes were obtained using laser ablation. The device represents, to the knowledge of the authors, the SSP device with the highest degree of programming freedom (full independent switching control over each of the 24 electrodes), and the highest fill factor, with less than a ¼ mm 2 (12 cuts each 10 mm long and 2 µm wide) of interpixel space in a 10 mm diameter (78 mm 2 ) active area. The device was designed to deliver retardations far larger than a full wavelength at the wavelength used in the characterization (632.8 nm). Consequently, the same device potentially could be employed NIR, with the same electronic driver and controlling software.