Small optics with large opportunities
Tiny but mighty, micro-optics play a key role in photonic devices for applications such as communications, imaging, and data storage. Micro-optics are typically diffractive or refractive elements with apertures ranging from 100 µm to 2 mm and are fabricated as discrete or array-based components. To date, measuring micro-optics has largely been a tedious and manual task. Conventional techniques such as contact profilometry or scanning white-light microscopy are slow, labor intensive, and lack sufficient resolution. Automated interferometers offer a viable alternative to more traditional lab- oriented platforms.
Small-aperture laser-based Twyman-Green interferometers provide a good base for automated metrology (see figure 1). The Twyman-Green is a phase-shifting interferometer with separate test and reference legs that can be operated with either a plano wavefront or a spherical wavefront (for testing flats or nominal spheres respectively). By including an automated x, y, and z motion system with advanced software capabilities, various surface parameters such as radius of curvature (Rc ) can be automatically measured for each element of an entire lens array.
No matter what level of automation an interferometer has, the basics of measurement are the same. Consider a convex microlens. In the Twyman-Green system, a collimated wavefront is passed through an objective lens to create a converging spherical wavefront. If the microlens is placed at the axial location where the spherical wavefront converges to a point, the light retro-reflects into the interferometer (see figure 2). This location is known as the catseye position (Lcat). If the microlens is moved closer to the objective lens such that the wavefront from the interferometer has nominally the same shape as the microlens surface, it lies at the confocal position (Lconf). This is the position at which surface figure data is obtained. The distance between the catseye and confocal positions is the Rc of the lens.
In order to measure Rc , the system must determine how far away the lens lies at each point from the 'correct' position (the point of minimal defocus). An automated system can find this point by observing the defocus in the interferograms at each position. Very high accuracy Rc measurements can be made by correcting the distance between measurements with the defocus using:
where 
is the catseye focus offset and Δcat is the catseye focus offset.
Surface figure data show how the spherical surface of the microlens deviates from the perfect spherical wavefront generated by the objective lens. The data are acquired through a phase measurement in which a piezoelectric stack moves the reference surface a known amount and a computer tracks the resultant change in fringes. This technique can differentiate between bumps and holes and provides higher accuracy measurements than static fringe analysis.
To ensure best quality data, optics should be aligned prior to data acquisition using an automatic nulling routine. This routine takes a phase measurement and determines the lateral and vertical offset by analyzing the tilt and focus of the resultant interferogram. In an automated system, the lens is adjusted in x, y, and z to minimize aberrations caused by misalignment, minimizing fringes and optimizing data quality. With hundreds of lenses in a single array, performing this task manually becomes time prohibitive.
Commercially available interferometers enable automated, high-volume production testing by leveraging the advancements in computers, software analysis, and motion control. Modern metrology is no longer viewed as a production bottleneck, but as a value-added tool for rapid process feedback to improve manufacturing. oe
Eric Felkel and John Roth are with Zygo Corp., Middlefield, CT.
