Introduction

Distributed feedback lasers (DFB) are widely used as a light source in optical communications. DFB’s use a diffraction grating as part of the optical resonator for precise definition of the wavelength of emitted light. The pitch of such gratings is typically in the range of 200 nm or 240 nm, which makes patterning such gratings by means of traditional, contact mask lithography difficult and expensive. An alternative method of manufacturing such gratings is use of interference photolithography, where the periodic pattern of the grating is created by interference of two coherent laser beams. A typical manufacturing system of this kind looks more like an optics laboratory setup, uses little or no automation at all, and presents an optical hazard because of free laser beam propagation. Controlling such a system requires an engineer or a highly skilled technician, its throughput is extremely low, and it often becomes a bottleneck in the manufacturing process.

The first system to overcome the difficulties described above was developed and built by Holographic Lithography Systems, Inc. (Bedford, Massachusetts, USA), now a division of Optical Switch Corporation (Richardson, Texas, USA). The tool, known as PC2, is shown in Fig. 1. The tool operates as follows: 

• A laser-generated beam is split and coupled into two fibers;
• Ends of fibers are mounted on two carets that can move along their corresponding rails as well as change their tip and tilt angles;
• The two beams exit the fibers as wide cones of light, that can be centered on the exposure area using beam alignment and optical beam conditioning, where they form an interference pattern with precise pitch, defined by the relative position of the carets;
• A wafer covered with photoresist is mounted on a chuck and positioned in the exposure area;
• The wafer is exposed to a precise dose (exposure time calculated during calibration of the system);

Later in the process, the wafers with the periodic pattern developed in the photoresist go through wet or dry etch step, which transfers the photoresist pattern onto the wafer surface.

Fig. 1.  PC2 Lithography Tool

Hardware Architecture

The hardware architecture of the tool is shown in Fig. 2.

Fig. 2. Hardware Architecture of the Tool

The choice for rail and chuck motion hardware was mostly dictated by the requirements for stability during exposure. Both rails and chuck are controlled using servomotors and standalone motion controllers that communicate with the PC over RS232 interface. Once in position, brakes are engaged and the motors are killed. The chuck controller is also responsible for dispensing and evacuation of the fluid used for coupling the laser beams between the beam conditioning optics and the wafer. Optionally, the wafer can be rotationally aligned with the system axis using processing of the image of the wafer flat, acquired by one of the auxiliary analog cameras.

Fiber input coupling, fiber output tip and tilt, and beam polarization are controlled using piezo motors and a drive that communicates with the PC over RS232. The system periodically undergoes an automatic alignment process whose goal is to maximize the laser beam power on the wafer, align the beams on the center of the chuck and equalize the power of the beams. A digital camera is used as the beam sensor: the camera looks down at a reflective wafer and detects the amount and distribution of the reflected light. The same camera is used for calibration of the beam intensity and calculation of the exposure time. Polarization is optimized using additional fixed polarizers (“analyzers”) and photometric diodes whose readout is digitized by a PCI-6025E board.

Use of high power laser gives rise to optical safety issues. In order to prevent exposure of the operator to the laser beam, the system is completely enclosed and all panels are equipped with interlocked switches. When any panel is disturbed, its switch is opened and the main laser shutter is immediately closed. After the panel is returned to its position, the user (engineer) has to access a special screen where he can reset the interlock chain and re-enable control over the main shutter.

Software Architecture

The software for the system was written in LabVIEW. This programming environment was chosen for its flexibility, high programming speed and ease of maintenance: because of high degree of customization and specific needs of each customer, tools may not be identical, either in software or in hardware. Motion controller calls were written using the controller command set and saved as command files, which allowed the applications engineers to tune system motion parameters in the field without modifying the main code.

The software architecture of the PC2 control system is shown in Fig. 3. The program consists of two parallel loops – a control loop and a status loop. The control loop is implemented as a state machine with various functional states accessible to the user depending on his privileges. The status loop continuously polls the subsystems and displays their context-sensitive status.

A noteworthy feature of the control program is “global optimization” – automatic alignment and calibration of all subsystems of the tool. This feature is implemented in a separate virtual instrument (VI) that uses LabVIEW VI Server functionality to exchange parameters with each subsystem VI. This VI is called from the main program. Essentially, it serves as a macro that automates user actions. When the user selects this option, the following sequence of events is invoked consecutively for each subsystem VI:

• A reference to the VI is created;
• Buttons, that would normally be pressed by the user, are “pressed” by the program;
• Subsystem state is analyzed, errors and timeouts checked;
• Upon successful completion of the step, the next subsystem is invoked

Fig. 3. Software Architecture of the Control System

Fig. 4 shows one of the user interface screens. On the top of the screen, a status indicator bar is shown. It is displayed permanently. The main part of the screen shows an interface for beam centering subsystem. An insert shows global optimization status display. There are a total of 19 screens broken by functionality, including 11 functional subsystem screens (See Fig. 3).

Fig. 4. Beam centering subsystem screen with status bar and optimization progress dialog window

Acknowledgements

The author would like to express his gratitude to Adam Kelsey and Mark LeClerc of Optical Switch Corporation for their help in developing the PC2 control system.