Phase-only liquid crystal on silicon (LCOS) devices have been used as the key component in a wide range of applications, such as holographic projections^[1–4], adaptive optics in medical sciences^[5], and optical switching systems^[6–8]. Significant research effort has been made towards LCOS device assembly technology, device applications, and characterization of their performance^[9]. For assembly, a technology packaging process at the wafer level was reviewed for manufacturing ferroelectric LCOS devices^[10]; it can be applied to nematic LCOS devices as well. At the research level, phase-only LCOS devices are being assembled largely at the die level at present^[11] because of its flexibility and practicality^[12]. Performance characterization of LCOS devices and their applications are mainly focused on the utilization and optimization of phase gratings^[6,13,14], the improvement of spatially varying phase responses^[15,16], multi-phase spatial compensation^[17], and even optical phase compensation^[18,19], as well as crosstalk mitigation in LCOS-based optical switches by constructing asymmetrical output positions in space^[13,20].

One of the most important features of a phase-only LCOS device is its use of optically non-linear liquid crystal (LC) materials, which are sensitive to the working temperature. Some of the theoretical and experimental research works have been carried out to analyze the optical properties of LCs when the temperature changes. Wu et al. derived a four-parameter model^[21] for describing the temperature effect on the refractive indices of LCs based on the Vuks equation^[22]. Lin et al. also examined the optical properties of twisted nematic LCs in different temperature ranges^[23].

All the studies show that temperature is an essential parameter in determining the optical properties of LCs. However, there is a lack of systematic research of the LCs’ temperature dependence on the overall performance of phase-only LCOS devices. This is partly because the main application interest of LCOS devices in the past was on optical intensity modulation rather than phase modulation, and the former is less or little affected by temperature variations, particularly when the binary mode is used. As a result, temperature sensing and performance compensation are not always considered for those devices in practice. For phase-only LCOS devices, optical phase modulation of the incident light is an essential performance parameter, and it can easily be affected by a small change of the working temperature, resulting in a significant and sometimes dramatic change in the outcome of corresponding optical diffractions. For instance, Fig. 1(a) shows the simulated spatial distribution of a diffracted optical intensity at a replay plane in the far field when a phase-only blazed grating pattern is displayed on an LCOS device at a constant temperature. The pixel pitch is assumed to be 8 μm, and the grating period is 160 μm. In this situation, the optical intensity of the background is less than −70dB in comparison to the normalized intended diffraction peak at position zero. Figure 1(b) shows the changed spatial optical intensity when there is only 1% phase error caused by temperature variation. In this case, the background optical intensity at some positions can increase by 20 or even 40 dB, leading to an unacceptable level of the crosstalk in telecommunication switches or wavelength-selective switches.

In this Letter, the temperature effect on the performance of phase-only LCOS devices using the nematic LC BLO37 is investigated systematically, including the material birefringence and dielectric anisotropy, as well as the device dynamic phase modulation depth and threshold voltage.

A phase-only LCOS device was assembled in-house using a die-level process^[12] with a 5 μm thick LC layer of BLO37. It has 1280×720 pixels with a pixel pitch of 15 μm, and it can achieve a 2π phase modulation at the light wavelength of 1550 nm for telecommunication applications. In this study, the wavelength used was 633 nm. This was for a better understanding of the effect of an unbalanced optical response over a wide phase modulation range of 2π and more. This would also allow us to explore the effect of lower phase modulation depths (still>2π) at higher temperatures, since the device was assembled to achieve a minimum 2π phase modulation depth for infrared (IR), which could be 5π–6π or more in the visible range. In addition, using visible light can be a convenient way to characterize phase-only LCOS devices assembled for IR applications if it is done properly^[15]. Figure 2 shows a basic setup for characterizing the temperature effect of phase-only LCOS devices. A light beam emits from an He–Ne laser into the system through a collimated beam expander. It then passes through polarizer 1, which is aligned to ensure that the light is polarized in parallel to the alignment direction of the LC materials in the LCOS device. After passing through a beam splitter (BS) and encountering the blazed grating uploaded on the phase-only LCOS device, the light beam is diffracted back and its spatial intensity profile is measured by a photodiode for measurement (PDM). In this setup, a heating stage is put next to the LCOS device for the adjustment of the device working temperatures in a range of 15°C–55°C, as the birefringence of most of the LC materials will reduce significantly when the temperature goes higher. The temporal changes of the diffracted optical intensity are recorded using a data acquisition card (DAQ). The voltage applied across the LC layer is up to ±5 V, with a DC-balanced square waveform generated by the CMOS circuitry on the silicon backplane. The actual voltage can be set at any of the 256 individual levels between zero and the maximum voltage.

Birefringence is probably the most important property of LC materials for phase modulation, as it determines the speed of light propagation depending on the direction and polarization of the light wave traveling through LC materials. For phase-only nematic LCOS devices with a positive electrically controlled birefringence (ECB), the director of the LC molecules normally switches from almost “planar” to “homeotropic” when a voltage is applied. In the experimental setup shown in Fig. 2, the incident polarization direction is aligned with the initial direction of the LC molecules, and the electro-optic switching has access to almost the complete range of refractive index changes, from the extraordinary refractive index (ne) to ordinary refractive index (no), i.e., the magnitude of birefringence. For the validity of studying the temperature effect in LCOS devices, it is only meaningful to operate in the temperature range before the LC material reaches its clearing point^[24].

Birefringence depends not only on the working temperature but also on the wavelength of the light. A quantitative equation was proposed^[25] to describe the birefringence dispersion as a function of wavelength in the visible and IR spectral regions. It showed good agreement with the measured results for the LC material BLO37 at several wavelengths, demonstrating its relatively high value of birefringence, especially in the near-IR region (0.224 at 1550 nm)^[26].

In Fig. 3(a), the birefringence of the LC material BLO37 in our LCOS device was measured as a function of the temperature for three different wavelengths, 445, 532, and 633 nm. It confirms that the birefringence reduces the temperature^[25] and the wavelength spectrum increases. Figure 3(b) shows the phase modulation depth as a function of the applied voltage at different temperatures. It is well known that the phase modulation depth is proportional to the magnitude of birefringence, as the device thickness is fixed, as shown inδ=2π·2d·Δnλ,(1)where δ is the phase modulation depth, d is the thickness of the LC layer in an LCOS device, Δn is the magnitude of the LC birefringence, and λ is the light wavelength. The empirical fitting for the dependence of the phase modulation depth on the applied voltage is shown in δ=δmax(1−10−k(V−Vth)),(2)where δmax is the maximum capability of the phase modulation for a given wavelength (about 7π for our LCOS device at a wavelength of 633 nm), k is a fitting parameter (ranging from 1.42 to 1.84 at different temperatures for our device), V is the external voltage across the LC layer, and Vth is the threshold voltage. It is clear, as shown in Fig. 3(b), that the fitting using Eq. (2) has a good agreement with the experimental data and that the phase modulation depth decreases as the temperature increases. Note that, due to the series of raw data that has been extracted and processed through the empirical fitting equations, the real-time measurement might be diversified because the operating temperature is not constant most of time. Also, for temperatures above 50°C, the phase modulation depth seems to be higher than that of lower temperatures, which is an artifact due to the enhanced phase flicker, which is an indispensable phenomenon of the temporal phase variation in an LCOS device. As a result, unbalanced phase modulations happen due to the instability of the LC materials under study. This leads to subjectively temporal instability of the diffracted intensity detected by a fast response photodiode. Garcia-Marquez et al.^[27] commented that the phase flicker can be reduced only if the LC material is kept at a low temperature range for a certain level of the material viscosity.

The dielectric anisotropy of a nematic LC material under an external electric field can be described by the Maier theory^[28]. The dielectric anisotropy is defined as the difference of the dielectric constant in directions parallel and perpendicular to the long molecular axis, which is given byΔε=ε∥−ε⊥,(3)where Δε is the dielectric anisotropy, ε∥ is the dielectric constant parallel to the LC director (optical axis), and ε⊥ is the dielectric constant normal to the LC director (optical axis). The measured dielectric permittivity and dielectric anisotropy at 1 kHz as a function of the temperature for BLO37 are shown in Fig. 4. Both the dielectric permittivity normal to the long axis and the dielectric anisotropy decrease as the temperature increases. In addition, the LC material has a reasonably high positive dielectric anisotropy, and its threshold voltage can be described by^[29]Vth=4πKiiΔε,(4)where Vth is the threshold voltage, Kii is the elastic constant of the LC material (it is set to be equal to the splay elastic constant K1 since the ECB mode is used for phase-only LCOS devices in this experiment), and Δε is the dielectric anisotropy. The threshold voltage represents the onset voltage for the movement of the LC director in the middle of the LC layer. Here, the threshold voltage is only related to the dielectric anisotropy and the elastic constant. The magnitude of Δε, which usually follows the magnitude of Δn, should be as high as possible for a low Vth.

Figure 5 shows the change of the threshold voltage as a function of the temperature measured in our BLO37 layer. It shows an increase of Vth with the temperature, which is in good agreement with Eq. (4), where Δε decreases with the temperature. It is good to see that in our temperature range from 15°C to 50°C, Vth only increases by 0.3 V and the magnitude of dielectric anisotropy does not decrease excessively, which allows us to drive the LC molecules effectively with a sufficient phase modulation depth for the desired applications. For instance, the test device with the LC Δn=0.282 at room temperature is designed for achieving 7π of the phase modulation depth at 633 nm, which is equal to 2.4π at 1550 nm. If the working temperature is set to be 50°C, the phase modulation can still reach 2.2π at 1550 nm^[26], which is sufficient for most applications.

Many high dielectric anisotropic LC materials, such as BLO37, have a fairly high viscosity and hence a slow response time. The response time (either the rising time or the decay time) is a function of the birefringence, dielectric anisotropy, and temperature, which is given byt=f(Δn,Δε,T).(5)Both the wavelength-dependent birefringence and dielectric anisotropy are affected by temperature as well: Δn=u(T,λ),(6)Δε=ν(T).(7)Therefore, the response time can be shown as t=f(u(T,λ),ν(T),T).(8)Thus, it is clear that temperature is a vital parameter here and the response time of phase-only LCOS devices can be hugely improved by working at a high temperature.

Figure 6 shows the tremendous reduction of the response times, both the rising time and the decay time, in the test LCOS device with BLO37 when the working temperature increases. It also shows that the rising time can be reduced considerably if a large electric field is applied^[30]. For instance, the rising time is about 280 ms when 2 V is applied and less than 50 ms when 5 V is applied. For the decay time, the increase of the applied electric field can also have some effect of improvement. Considering the impact of the phase flickers when the working temperature is above 50°C, 50°C is considered as the optimized working condition for the improvement of the response time.

In conclusion, the temperature-dependent properties of the non-linear nematic LC material BLO37 are investigated systematically in relation to the optical performance and operational conditions for phase-only LCOS devices. Real-time performance of a test phase-only LCOS device as a function of working temperatures is studied on three aspects: the phase modulation depth as defined by the birefringence, the threshold voltage as affected by the dielectric anisotropy, and the response time. The results in Figs. 3 and 5 show that the working temperature could be practically as high as 50°C for significantly fast response times (both rising time and decay time) without compromising the required phase modulation depth for practical applications and the threshold voltage to be driven by the LCOS device.