• 沒有找到結果。

Chapter 1 Introduction

1.4 Organization of this thesis

The thesis is organized as following: In chapter 2, the working principle of a switchable grating and the theories of micro elements, including the micro-hinges, the binary phase grating and the stress-induced actuator, are presented. The topic of chapter 3 includes a summary of micro-fabrication technologies and the instruments to characterize the performance of the switchable grating. Chapter 4 is devoted to”thin film optics”, which was used to predict the performance of a micro-PBS with stacked films. The development of low stress silicon nitride films are also described here. In chapter 5, we focus on the structure design and the measurement of the micro-PBS. In chapter 6, we propose and demonstrate the micro-reflective system. In chapter 7, the

main contributions of this dissertation and a proposal of future work are given.

Table 1-1 Pros and cons of the four micro optic pickup heads.

Pros Cons Waveguide with focused gratingP[1]P

1. High integrity 1. Low coupling efficiency

2. Difficult realization of high numerical aperture lens by the grating coupler Diffractive optical element (DOE) planar approachP[2]P

1. Low coupling loss

2. Simple fabrication approach

1. Requirement of precise alignment on both sides of the substrate

2. Low integrity with dynamic devices 3. High off-axis aberration

Stacked optical pickup schemeP[3]P 1. Simple design

2. Simple fabrication process

1. Lower integrity with dynamic devices, such as with micro actuators.

Free-space optical bench using the surface-micromachiningP[4]P 1. High possibility of dynamic

calibration in assembly.

2. High possibility to generate a diffraction-limited focused spot.

1. Cumbersome fabrication and assembly process

2. Application only in the infrared spectrum.

Chapter 2

Principles of Microoptics and Microactuators

In conventional CD and DVD drives, an optical pickup (OPU) with a single laser beam is used for sequential data retrieving. The data rate is proportional to the rotation speed of the spindle motor. However, the maximum rotation speed is partly limited by the dynamic characteristics of the objective lens actuator. Using multiple beams in parallel is a straightforward solution to improve the data rate. Several methods have been demonstrated to achieve simultaneous reading of multiple tracks, such as generating multiple beams by using a diffractive optical elementP[5-6]P, a diode laser arrayP[7]P, or a combination of laser diodes and a beam combinerP[8]P. Alon et al. proposed a pickup using diffractive optical elementsP[5]P. A grating was used to split an illumination beam into multiple beams to read several tracks of the optical storage medium. The reading beams can be generated uniformly by a grating; however, they are not suitable for multi-beam recording due to the insufficient power of each beam. To overcome the limitation, Shih proposed an optical pickup employing a liquid crystal grating to switch alternatively between single-beam recording and multi-beam readingP[9]P. The liquid crystal grating, however, is of large size and high cost.

In this section, we present the working principle of a micro-silicon-based free-space switchable grating to switch between single-beam recording and multi-beam reading in the micro optical pickup. The switchable grating is composed of a binary phase micro-grating mounted on a bimorph actuator by micro-hinges. Low stress silicon nitride is used as the optical material of the binary phase grating for its high transparency in the visible spectrum and its superior chemical and mechanical

properties. The switching function is achieved using an electrostatic driven stress-induced curved polysilicon based micro actuator.

2.1 Overview of the switchable grating for a micro optical pickup

The free-space integrated micro-optical pickup is based on the surface micromachining technology. As shown in Fig. 2-1, the optical bench, including a collimator, a switchable grating, a polarization beam splitter, a folded mirror and an elliptical diffractive lens are monolithically integrated on a silicon substrate. The vertical binary phase grating is fabricated on a stress-induced actuator and can be raised above the path of the incident beam for data recording (Fig. 1(a)) and lowered to create multiple beams for data retrieving under variable electro-static force. (Fig. 1(b)).

The optical axis of the system is designed to be 400 µm above the silicon substrate.

The laser can be mounted on its side with the aid of three-dimensional alignment structuresP[10]P or attached on a 400-µm-thick micro-chip so that its emitting spot is aligned with the optical axis. First, the laser light emitted from the laser diode (LD) is collimated by a micro-fresnel lens or a micro-refractive lens. When the grating is on the optic axis, the forward light is further divided into three beams. Part of the beams passes through the beam splitter. The beam splitter is positioned 45° from the optic axis.

The light reflected from the splitter can be used for monitoring the intensity of the LD.

The transmitted light is further bent upward by an integrated 45° mirror, which enables the optical bench to be mounted in parallel to the optical disc. After being focused on the disc by the objective lens, the reflected beams carrying the information of the disc are converged by the same objective lens, reflected by the intensity beam splitter, and focused again by an elliptical diffractive lens onto a planar quadrant photodetector attached on the Si substrate.

(a) Single-beam state

(b) Multi-beam state

Fig. 2-1. Schematic of the micro-optical pickup with a switchable grating:

(a) single-beam recording and (b) multi-beam reading.

2.2 Microhinges

In the micro-optical bench, all optical elements are three dimensional and fabricated by the micro-hinge technology with the two layer polysilicon process. Hinged structure was first proposed by Pister in 1991P[11]P. The structure allows the surface micromachined polysilicon plates to be patterned by photolithography and then folded into three-dimensional structure. There are four types of hinges, as illustrated in Fig. 2-2.

The simplest setup is the ‘substrate hinge’, as shown in Fig 2-2(a), which consists of a polysilicon (poly-1) plate and a hinge pin constrained by another polysilicon (poly-2) stable. The stable is connected to the substrate through two contact areas, and the plate is free to rotate any degree off the substrate by applying a stop. The width of the pin has to be smaller than the sum of the thickness of the first polysilicon (poly-1) and its sacrifical oxide layers, by which the pin will be able to rotate a full 90 degrees without contacting the substrate and staple. The substrate hinge is used to hinge poly-1 plates to the substrate. In our micro-optical pickup bench, the substrate hinges are applied in the building of the three-dimensional structures of the collimator, the intensity beam splitter, the folded mirror, and the elliptical diffractive lens.

To hinge plates to each other requires the other two types of hinges. Two poly-1 plates can be hinged together using an ‘interdigitated hinge’, as shown in Fig 2-2(b).

Poly-2 strips are connected between interdigitated poly-1 fingers, allowing the plates a relatively ‘concave-down’ rotation of 180 degrees and preventing the two poly-1 plates from pulling apart.

The third type of hinge is a ‘T-style hinge’, which hinges two poly-2 plates together and folds ‘concave up’, illustrated in Fig. 2-2(c). The fourth type of hinge is a ‘scissor hinge’P[12]P, which hinges two poly-1 plates together and folds ‘concave up’, illustrated in Fig 2-2(d). Scissor hinges are the most complicated and also applied in our

micro-optical pickup bench. It is applied to connect the binary phase grating and the stress-induced actuator. The grating is realized using microhinges and spring-latches similar to the substrate-hinge based collimator. However, the hinges of the grating are fixed on the stress-induced actuator instead of Si substrate.

Fig. 2-2. Four types of hinges. (a) Substrate hinge, which is used to hinge released structures to substrate. (b) Interdigitated hinge, which is used to hinges two

‘concave down’ structures. (c) T-style hinge, which is used to hinges two

‘concave up’ structures. (d) Scissor hinge, which is used to hinge two poly-1 plates together and folds ‘concave up’.

Unlike the substrate hinges, there is no ‘pin’ in the other three types of hinges.

Given the typical film thicknesses, a substrate hinge designed to rotate 90 degrees must have a pin no more than the total thicknesses of the poly-1, the first and second

geometries wider than the above total thicknesses, making them much stronger than substrate hinges.

2.3 Binary phase micrograting

Typical macro-optical components, such as mirror, lens and prism are described by ray optics: treating light as geometrical rays which will be refracted and reflected at optical interface according to the Snell’s law. Rather than being refracted at continuous surface profiles, in diffractive optical elements (DOEs) light is diffracted at the periodic microstructure of the element. Specific continuous surface profiles are not easy to be realized in a system sized of a few microns. Therefore, DOEs are more attracted for integrating within micro optical systems. DOEs have several advantages: 1) their diameters as a few hundreds of micrometers can be defined using photolithography; 2) their thickness is on the order of an optical wavelength and suitable for micromachining processP[13-15]P. To determine the optical path and efficiency of light transmitting in DOEs, scalar diffraction theory and vector diffraction theory are two main theories used to design. Vector diffraction theory is a rigorous electromagnetic theory based on Maxwell equation. Its numerical computing is time-consuming. In contrast, scalar diffraction based on fourier optics is relatively efficient in calculation. If the variation of the surface-relief is larger than the wavelength of light λ, the optical performance except the phenomenon of polarization of DOEs can be dealt with using the scalar diffraction theoryP[16-18]P. Therefore, the scalar diffraction theory is used on the designs in this thesis work.

Binary phase gratings have the stepped approximation with respect to an ideal continuous profile. Diffraction efficiency of binary phase gratings depends on the number of steps in this staircase-like binary profile. In a typical micromachining

process, a sequence of binary processing steps can be used to generate multilevel profiles. Since each step has a specific height, a smart approach is to make 2PNP discrete levels by processing N mask lithography and etching. For example, a blazed grating is to deflect the incident light through a certain angle with high efficiency. Fig. 2-3 shows an ideal phase grating profile with a perfect sawtooth profile, and a quantized version of that grating with 2PNP level. In this blazed phase grating, if there are 2PNPsteps of equal spaces and thickness existed in one period Λ with height d, the width and height of each step is Λ/2PNP and d/2PNP, respectively.

Fig. 2-3. Phase profile comparison between an ideal sawtooh phase profile for a blazed grating, and a binary approximation to that profile.

However, the precision of the lithography process is limited. As described above, the fabrication process for multilevel phase elements generally consists of a sequence of binary fabrication steps each of which requires an aligment and an etching stepP[19-21]P. Alignment errors occur if the masks for subsequent lithographic structuring processes are slightly shifted relative to existing structures. Etching errors occur if the phase step introduced during individual etching processes deviates from the ideal step profile with precisely defined etching depth and vertical edges. Since our target is to demonstrate the possibility of a switchable grating, a two-level phase grating with rectangular shape

was selected to reduce the process complexity. Under the specific conditions, this two-level phase grating can produce three beams of equal intensity.

To apply the three beam grating in an optical pickup, the diffraction ratio, η, of the 0Pth

Porder beam intensity IB0B and the ±1Pst Porder beam intensities IB±1B should be controlledU to

Uapproach one. In addition, the power efficiency, ηBuB=(IB-1B+IB0B+IB+1B)/IBinputB, where IBinputB is the intensity of the input beam, should be as high as possible. The diffraction angle between the 0Pth Porder and the ±1Pst Porder beams is determined by the optical system layout such as the working distance, the thickness of the cover layer of the disc, and the spacing between the 0PthP and the ±1PstP order beams on the disc, as shown in Fig. 2-4.

0 th +1st

-1st

Spacing Incident Light

λ=633 nm

Working Distance

Cover Layer

Recording Layer Objective Lens

3-beam Grating

Diffraction Angle θ

Fig. 2-4. Optical path of the three beams for reading.

To determine the diffraction angle of the first-order beams, we assumed 40-µm spacing between individual focused spots on the disc. Because the spot spacing is larger than the track pitch, the array of spot would be positioned at an angle from radial direction in practical applications. The micro-pickup is designed to read a disc, which has a 100-µm thick cover layer with refractive index n= 1.6. If the working distance

between the objective lens and the cover layer is 460 µm, underU the Uthin lens approximation for the objective lens, the diffraction angle, θ, is about 4.35°. For a transmissive grating with θ=4.35°, m=1 and λ=633 nm, the period Λ is about 8.35 µm, derived from the equation Λ×sinθ=m×λ. Λ=8 µm was selected in the design. To reduce the process complexity, a grating with rectangular shape was used.

The power in the diffracted beams IB0B and IB1Bof the grating can be determined by the

where m=1 is the diffracted order, fill factor f=w/Λ is the ratio of the line-width to the grating period and phase shift φ is related to the depth of the grating D and the refractive index n of the grating material by the formula

where n=2.102+0.008i is the measured index of refraction of low-stress silicon nitride at λ=633 nm. According to pre-run simulations, it was found when the fill factor was 0.50; multiple grating depths could be used. For example, 101 nm (ηBuB =81.7%), 404nm (ηBuB =83%), and 657nm (ηBu = B74.7%) satisfied the desired diffraction efficiency ratio.

Other grating depths meeting thUe Urequirement were higher than 1000 nm, which was not suitable in surface micromachining processes. To have sufficient mechanical strength and reasonable fabrication yield, the depth of 404 nm was selected, which also yielded high energy utilization efficiency. The contour plot of the diffraction ratio IB0B/IB±1

Bis shown in Fig. 2-5 as a function of the fill factor f and the grating depth D. As can be seen, a diffraction ratio IB0B/IB±1B of about 1 can be obtained provided that f=0.5 and

D=404nm.

Fig. 2-5. Diffraction ratio (IB0B/IB±1B) contours.

According to pre-run simulations, it was found when the fill factor was 0.50;

multiple grating depths could be used. For example, 101 nm (ηBuB =81.7%), 404nm (ηBuB

=83%), and 657nm (ηBu = B74.7%) satisfied the desired diffraction efficiency ratio. Other grating depths meeting thUe Urequirement were higher than 1000 nm, which was not suitable in surface micromachining processes. To have sufficient mechanical strength and reasonable fabrication yield, the depth of 404 nm was selected, which also yielded high energy utilization efficiency. The contour plot of the diffraction ratio IB0B/IB±1 Bis shown in Fig. 2-5 as a function of the fill factor f and the grating depth D. As can be

seen, a diffraction ratio IB0B/IB±1B of about 1 can be obtained provided that f=0.5 and D=404nm.

2.4 Stress-induced micro-actuator

Many new surface-micromachined microactuators have been proposed and demonstrated to integrate various optical elements since the invention of micromotors.

Some schematic of these microactuators are illustrated in Fig. 2-6. The comb drive is actuated by applying electrostatic force between the movable combs and the fixed combs, as shown in Fig. 2-6(a). Micronewtons of force and a few micro-meters of displacement are produced by typical surface-micromachined polysilicon comb drive actuators, which have been demonstrated to drive optical choppersP[23]P and bar-code scannersP[24]P. The linear microvibromotor is actuated by the impacts from the comb drives. The linear microvibromotor shown in Fig. 2-6(b) consists of a set of resonant comb drives and a slider. By adjusting the number and the frequency of impacts, a sub-micron positioning resolution and a travel range exceeding 100µm has been achievedP[25]P. The linear microvibromotor has been employed to actuate a slider dragged mirror to align optical beams for fiber couplingP[26]P. On the other hand, the thermal actuator is actuated by thermal expansion difference between a “cold” arm and a “hot”

arm in a U-shaped geometry as shown in Fig. 2-6(c). By applying a 10 mW driving power, a deflection up to 16 µm and force of 4.4 µN has been achieved by the polysilicon thermal actuatorsP[27]P, which can be used to assemble three-dimensional structuresP[28]P. The scratch drive actuator (SDA) is actuated by applying pulses of electric bias between the polysilicon plate and a bottom electrode separated by an insulator film (SiB3BNB4B), as shown in Fig. 2-6(d). By adjusting the amplitude and the frequency of pulses, a sub µm step size, 20-30 nm, has been achievedP[29]P. The SDA is

ideal for positioning optical elements that require accuracy below 0.1µmP[30]P and has been applied in the optical switchesP[31]P and the variable optical delay lineP[32]P.

To design a switchable grating for a micro-optical pickup, the actuator has to switch alternatively between the state of single beam and the state of multiple beams. The actuation distance in the free end of the bimorph actuator needs to be larger than the illumination range of the incident beam on the optical axis. In a standard electrostatic parallel-plate actuator, actuation distance is determined by the balance between the electrostatic force and the mechanical restoring force. For a linear restoring force, the controllable actuation distance is only one third of the gap between two parallel electrodesP[33]P. To overcome the limited actuation distance, Rosa et al. proposed an external electrode bi-morph actuator to operate over the entire range of motion by preventing electrostatic pull-in instabilityP[34]P. However, the electrical fringe effect between the moving and the fixed external electrodes is relatively insufficient to obtain enough actuating force. Accordingly, the bimorph actuator used in our research is based on the design demonstrated by Chiou et al.P[35]P, which used comb-shape external electrodes and post heat treatment to achieve higher actuation distance under lower voltages.

In our design, the stress-induced actuator is a composite cantilever, composed of a polysilicon layer and a gold film. The cantilever bends upward since polysilicon has a small compressive residual stress while the gold film has a large tensile residual stress.

Schematic of the polysilicon-gold composite cantilever is shown in Fig. 2-7. The curvature and tip deflection of the cantilever were determined by its geometric dimensions and material properties. The curvature of the beam is expressed by the equationP[36]P.

where σB1B and σB2B are the residual stress of polysilicon and gold. EB1B and EB2B are the Young’s modulus of polysilicon and gold.

Fig. 2-6. Schematic of microactuators commonly used in surface-micromachined optical systems: (a) comb drive actuatorP[23]P, (b) linear microvibromotorP[25]P, (c) thermal actuatorP[27]P, and (d) scratch drive actuatorP[29]P.

The symbol m=EB1B/EB2B, n=hB1B/hB2B and K is related by

K =1+4mn+6mn2 +4mn3 +m2n2 (2-4)

where hB1B and hB2B are the thickness of polysilicon and gold. With the radius of curvature ρ and the length L known, the maximum deflection δ at the tip of the beam, where the grating is positioned, can be expressed as

)]

cos(

1

[ ρ

ρ

δ = − L (2-5)

Fig. 2-7. Schematic representation of dimensions of the polysilicon-gold composite cantilever.

According to Equations (2-3)-(2-5), a 140-nm-thick Cr film and a 0.5-µm-thick Au film will be deposited on a 2-µm-thick polysilicon to curve up the composite cantilever.

The length of the designed cantilever is 2000 µm. As illustrated in Fig. 2-8, the proposed switchable grating consists of a 2000 × 260 µmP2P bimorph beam anchored to a bonding pad and a binary phase grating attached to the other end using microhinges and microspring latches. Movable comb fingers are connected to the bimorph beams on both sides; fixed comb fingers are connected to the nitride isolation layer on the surface. The engaged length of the comb finger is 180 µm. To actuate the switch, a

The length of the designed cantilever is 2000 µm. As illustrated in Fig. 2-8, the proposed switchable grating consists of a 2000 × 260 µmP2P bimorph beam anchored to a bonding pad and a binary phase grating attached to the other end using microhinges and microspring latches. Movable comb fingers are connected to the bimorph beams on both sides; fixed comb fingers are connected to the nitride isolation layer on the surface. The engaged length of the comb finger is 180 µm. To actuate the switch, a

相關文件