2.1.

Switching Device

Our approach to overcome the limitation of area in the SLM panel is to use display technology. Adoption of TFT-based backplane made on glass substrate will make it very easy to enlarge the panel area. But, to achieve ultrasmall pixel pitch (about $1\mu \mathrm{m}$), it is indispensable to use photolithographic tools for semiconductor technology.

A stitching process was developed to make the SLM panel, the area of which was larger than the field size of the photolithographic tool. To reduce the process costs and time, the periphery area of the SLM panel was patterned using a projection aligner tool. Therefore, a special patterning process utilizing the combination of two different photolithographic tools was also developed as shown in Fig. 1.

Fig. 1

Stitching process map for the combination of stepper and aligner tools. Active area composed of pixel array was patterned by stepper and the periphery area was patterned by aligner.

In addition to the scale down of pattern size, the channel length and width of TFTs also should be reduced. It is generally known that, for the case of oxide semiconductor TFTs, the short channel effects do not appear up to very small channel length such as tens of nm. Therefore, oxide semiconductor TFTs can be good candidate for switching device of SLM panel. A short channel oxide TFT with channel length of $1\mu \mathrm{m}$ was developed as shown in Fig. 2. For the case of the SLM on glass with $7\text{-}\mu \mathrm{m}$ pixel pitch, oxide TFTs with $2\text{-}\mu \mathrm{m}$ channel length was used.

Fig. 2

Stitching short channel oxide TFT with channel length of $1\mu \mathrm{m}$. (a) Top view, SEM image and (b) the transfer characteristics of $1\text{-}\mu \mathrm{m}$ channel length of oxide TFT.

Another merit of oxide semiconductor TFTs is extremely low off-currents. According to the scale down of pixel pitch, relative area of pixel capacitor compared with entire pixel area will be decreased. The allowable level of off-currents for switching TFTs in LCD should be lowered according to the decrease in the pixel pitch. Due to the wide band gap of the oxide semiconductor (larger than 3 eV), the off-currents of oxide semiconductor TFTs are fairly low compared with Si-based TFTs, such as a-Si TFTs and LTPS. Therefore, oxide semiconductor TFTs can cope with required off-current level of $1\text{-}\mu \mathrm{m}$ pitch pixel.

To preserve the interference properties of reflected light, the optical flatness of the reflector should be maintained. An organic planarization layer was formed using coating and hard baking. Figure 3 shows the cross-section diagram of SLM on glass with the planarization layer, light block layer, and reflector layer.

Fig. 3

Cross-section diagram of SLM on glass (SLMoG). ECB mode reflective LC cell is driven by oxide TFTs.

2.2.

Liquid Crystal

For an optical modulator, we used a liquid crystal and optimized it to our system with high anisotropic refractive index of $\mathrm{\Delta }n=0.25$ and minimum cell gap of $2.5\mu \mathrm{m}$. It is well known that liquid crystal suffers a crosstalk among neighboring pixels when a pixel pitch approaches the cell gap distance.¹⁴ To verify the optical modulation characteristics when a pixel pitch is small, we designed the test pattern, such as Fig. 4. Pixel pitches are $1.6\mu \mathrm{m}$ in the horizontal direction and $4.8\mu \mathrm{m}$ in the vertical direction and every pixel is connected to the electrodes through contacts. The red contact is connected to the electrode A, which is used as a variable voltage source, whereas the blue contact is connected to the electrode B, which is connected to zero voltage. Therefore, when we apply the voltage to electrode A, liquid crystals on the pixels with red contacts will change their direction appropriate to the voltage while the liquid crystals on the pixels with blue contacts will not move.

Fig. 4

Test pattern to verify the optical modulation characteristics. Pixel pitches are $1.6\mu \mathrm{m}$ in the horizontal direction and $4.8\mu \mathrm{m}$ in the vertical direction.

The binary hologram is calculated as follows. The object is letters of “NANO SLM,” which are located 20 cm behind the SLM. The object’s light wave, which is propagated to the SLM plane, is calculated using the angular spectrum method.¹⁵ If the object wave’s amplitude is $|O(x,y)|\exp [j{\phi }_{O(x,y)}]$ and the reference wave’s amplitude is $|R(x,y)|\exp [j{\phi }_{R(x,y)}]$ at the SLM plane, then the binary hologram value for each pixel is designated as 1 when $\mathrm{Re}[O(x,y)R*(x,y)]=|O(x,y)|$ $|R(x,y)|\cos [{\phi }_{O(x,y)}-{\phi }_{R(x,y)}]\ge 0$, otherwise it is zero. The pixels with the binary hologram value of 1 is connected to the variable voltage through red contact while those with the binary hologram value of zero are connected to constant zero voltage. Even though a liquid crystal works well according to the voltage, there are some concerns with respect to the observed reconstructed hologram image. When zero voltage is applied to the electrode A, we expect no hologram image because there will be no motion of a liquid crystal. However, the test pattern of Fig. 4 has the red and blue contact structures, which may induce some reconstructed hologram images.

Figure 5 shows a simple model for the contact structure of Fig. 4. When light shines on the contact structure of Fig. 5(a), there will be a diffracted beam from each contact. Red contacts are located exactly same to the pixel position of the binary hologram value of 1 so that the interference among diffraction lights from red contacts will reconstruct the hologram image, “NANO SLM,” at 20 cm distance from the SLM. Also, blue contacts are positioned at the complementary to the red contact position along the horizontal direction but shifted by $d_{\text{cont}}$ along the vertical direction. Babinet’s principle¹ states that the electric field from the complementary points will have the same magnitude but 180 deg out of phase compared with that of the original points. Therefore, the diffracted lights from the blue contacts will make the same hologram, “NANO SLM,” with the same amplitude but 180 deg out of phase and shifted along the vertical direction by $d_{\text{cont}}$. Graphically, the diffracted electric field from the blue contacts in Fig. 5(a) can be considered as a summation of the diffracted electric fields from every blue contact in Fig. 5(b) and the diffracted electric fields from the green contacts in Fig. 5(c), which has the same amplitude but is 180 deg out of phase and shifted along the vertical direction by $d_{\text{cont}}$ as compared with the diffracted electric field from the red contacts.

Fig. 5

Diffracted electric field from blue contacts of (a) can be considered as a summation of the diffracted electric field from every blue contact of (b) and the diffracted electric field from green contacts of (c), which has the same amplitude but 180 deg out of phase compared with that from the blue contacts.

Figure 6 shows the phase difference, $\mathrm{\Delta }\phi$, at the reconstructed point between the diffracted electric fields from the red contacts and the green contacts. At horizontal plane where the zeroth-order hologram image along the vertical direction will appear, the phase difference, $\mathrm{\Delta }\phi$, will be just $\pi$ because there is no path difference from the contacts to the hologram reconstruction plane if we assume that the distance between the SLM and the reconstruction point is larger than the SLM size such that the electric field from the red contacts will interfere destructively with that from the green contacts, so that there will be no hologram due to the contact structure. The $m$’th-order hologram image along the vertical direction occurs at the angle, $\sin {\theta }_m=m\lambda /d_{\text{panel}}$. Because there is a path difference between the red contacts and the green contacts to reconstruction point, $d_{\text{cont}}$ $\sin {\theta }_m$, the phase difference, $\mathrm{\Delta }\phi$, between the electric field from the red contacts and that from the green contacts will be as follows:

Fig. 6

Phase difference, $\mathrm{\Delta }\phi$, at the reconstructed point between the diffracted electric field from the red contacts and that from the green contacts’ hologram image estimation.

Therefore, the phase difference depends on the ratio between the shifted contact distance and the panel pitch along the vertical direction, $d_{\text{cont}}/d_{\text{panel}}$. The reconstructed hologram image will be bright due to the constructive interference when $\mathrm{\Delta }\phi =2k\pi$ but will disappear due to the destructive interference when $\mathrm{\Delta }\phi =(2k-1)\pi$, where $k$ is the integer number.

Figure 7 shows the diffraction simulation result according to the variation of the shifted contact distance. The ratio between the shifted contact distance and the panel pitch along the vertical direction, $d_{\text{cont}}/d_{\text{panel}}$, is one-fifth for Figs. 7(a), 7(d), 7(g), and 7(j), two-fifths for Figs. 7(b), 7(e), 7(h), and 7(k), and three-fifths for Figs. 7(c), 7(f), 7(i), and 7(l), respectively. The zeroth-, first-, and second-order simulated hologram image along the vertical direction is Figs. 7(d)–7(f), 7(g)–7(i), 7(j), and 7(k), respectively. The simulation result shows that there is no zeroth-order hologram image along the vertical direction irrespective to the variation of the ratio between the shifted contact distance and the panel pitch along the vertical direction, $d_{\text{cont}}/d_{\text{panel}}$. In Figs. 7(d), 7(g), and 7(j), the ratio of $d_{\text{cont}}/d_{\text{panel}}$ is one-fifth, so that the phase difference of first- and second-order reconstructed hologram is $7\pi /5$ and $9\pi /5$, respectively, according to Eq. (1), so that second-order hologram image is brighter than that of the first-order hologram. In Figs. 7(b), 7(e), 7(h), and 7(k), the ratio of $d_{\text{cont}}/d_{\text{panel}}$ is two-fifths, so that the phase difference of first- and second-order reconstructed hologram between the electric field from the red contact and that from the green contact is $9\pi /5$ and $13\pi /5$, respectively, so that first-order hologram image is brighter than that of the second-order hologram. In Figs. 7(c), 7(f), 7(i), and 7(l), the ratio of $d_{\text{cont}}/d_{\text{panel}}$ is three-fifths, so that the phase difference of first- and second-order reconstructed hologram between the electric field from the red contact and that from the green contact is $11\pi /5$ and $17\pi /5$, respectively, so that first-order hologram image is brighter than that of the second-order hologram.

Fig. 7

Simulation results according to the variation of the shifted contact distance. The ratio between the shifted contact distance and the panel pitch along the vertical direction, $d_{\text{cont}}/d_{\text{panel}}$, is one-fifth for (a), (d), (g), (j), two-fifths for (b), (e), (h), (k), three-fifths for (c), (f), (i), (l) respectively. The zeroth-, first-, and second-order simulated hologram image along the vertical direction is shown in Figs. 7(d)–7(f), 7(g)–7(i), 7(j), and 7(k), respectively.

Figure 8 shows the experimental result for hologram reconstruction with the variation of the applied voltage at the horizontal direction. At zero voltage, there is no hologram image. When the applied voltage is 2 V, the faint hologram image starts to emerge. When the applied voltage is 4 V, the clear image with the maximum intensity is appeared. This voltage range is coinciding with the measured diffraction peak intensity with the applied voltage variation as shown in Fig. 9.

Fig. 8

Experimental results for hologram reconstruction with the variation of the applied voltage at the horizontal direction.

Fig. 9

Diffraction peak intensity as a function of voltage.

2.3.

Driving and Hologram Reconstruction

Driving a huge number of pixels is a gigantic task for the AMSLM. Owing to the short pixel pitch, the resolution of SLM will be very large compared with typical flat panel display.

Multiple numbers of scan driver chips and data driver chips should be bonded to the pad area of SLM. Therefore, to reduce the pad area, pad pitch should be reduced. The minimum pad pitch is limited by the diameter of conducting ball in anisotropic conductive film and align accuracy during bonding process. Multiple, staggered band pad structure is generally adopted for high-resolution display panel as shown in Fig. 10.

Fig. 10

Staggered-type bump configuration for small pitch bonding process.

The number of channels for each driver chip should be increased to cope with high-resolution display, but the maximum number of channels is limited by the size of the driver chip. Considering the yield and price of the driver chip, the maximum size of the driver chip is limited by the field size of the photolithographic tool.

The allowed line time for each scan line is inversely proportional to the horizontal channel number of panel. An increase in the number of scan lines means the decrease in the line time. Therefore, for driving SLM, the current driving capability of switching TFTs should be enhanced. Data driving using a demux circuit can be used to deliver multiple signal at the cost of extra area for the demux circuit and complex driving signals.¹¹

We fabricated SLM on glass having $7\text{-}\mu \mathrm{m}$ pitch pixel driven by oxide TFTs. The size of SLM was 2 in. in diagonal. The resolution of the SLM was 5.8 by 1.2 K and the SLM was driven by three source drivers and two gate drivers bonded on a glass panel as shown in Fig. 11.

Fig. 11

Fabricated SLM on glass substrate. Three source drivers and two gate drivers were bonded for driving the SLM panel.

Hologram images with depth were successfully reconstructed by SLM on the glass substrate as shown in Fig. 12. The “ET” was focused at 0.9 m and “RI” was focused at 1.1 m.

Fig. 12

Reconstructed hologram images by fabricated SLM on glass with $7\text{-}\mu \mathrm{m}$ pixel pitch.