18 Apr 2006
Micro-optics are enabling new generations of compact optical devices thanks to their functionality and cost-efficiency. Thomas Asshauer and Matthias Merschdorf explain.
Micron-sized optical elements have been available commercially for more than a decade, but the technology is only just starting to be exploited in new types of optical systems. Micro-optics clearly has the potential to design devices that are smaller - and often cheaper - than is possible with conventional macro-optics, but that is far from the end of the story. Micro-optics should be seen as an enabling technology that will play a crucial role in the "century of the photon".
The term micro-optics refers to optical elements that have feature sizes in the micrometre range. In particular, it comprises refractive lenses with aperture sizes ranging from about a millimetre down to a few microns (figure 1) as well as diffractive optical elements (DOEs) that can essentially be considered as complex grating structures (figure 2). Micro-optics is also often associated with the use of microtechnology for fabrication, which includes the wafer production methods that were originally developed for the electronics industry.
To understand the potential of micro-optics, first consider a single laser diode. Light produced by the laser can be collimated by macroscopic optics, but micro-optics offers a much smaller and cheaper solution (figure 3). For laser-diode bars, where emitters are arranged in a row, microlenses offer an additional benefit, since microtechnology makes it easy to fabricate an array of optical elements. In this case, an array of cylindrical lenses with the same pitch as the emitters in the laser bar can be used to collimate the output from the individual emitters, in combination with a cylindrical lens in the transverse direction.
Even further integration can be achieved by stacking the collimated diode bars to create extremely compact laser sources that can deliver up to a few kilowatts of output power. Subsequent micro-optics can be used to homogenize the stack's beam profile, and to enable efficient coupling into optical fibres. In a similar way, matching microlens arrays can be used to couple light in and out of telecommunication fibre arrays, which avoids the need to match each fibre individually with a collimation lens.
Micro-optics also serves as a key enabling technology in new image sensor designs with "smart pixels". These pixels integrate amplification and image-processing electronics onto the same chip, but suffer from low fill factors because the light-sensitive area is reduced relative to the size of the pixel. Arrays containing up to millions of microlenses, each one measuring just 5-25 mm, compensate for the light loss and focus the incoming light onto the active area.
Further applications arise from the possibility of splitting a macroscopic beam into a number of sub-beams. For example, in the so-called Shack-Hartmann sensor, each microlens in an array forms its own focus spot, and the distortion of the spot pattern enables the curvature of an incoming wavefront to be determined with great accuracy. First used in astronomy, these sensors are now being used in applications ranging from aberration testing of optical components to laser correction of human eyes.
Refractive beam homogenizers rely on the same principle. In this case, a micro-lens array divides the incoming beam into sub-beams that are then overlayed by subsequent optics, which acts to average out a non-uniform light-distribution across the beam. Beam shaping can also be achieved by diffractive optical elements, such as fanning out a beam into arrays of many sub-beams with equal intensity, or even more complex light patterns.
Other types of micro-optics are involved in homogenous illumination and brightness enhancement of microdisplays, ranging from mobile phone screens to beam- projection systems, while shaping the light output from LEDs is another rapidly growing field of application. New possibilities also arise when the packaging concept becomes an integral part of the micro-optics design.
Properties of micro-optics
Despite the astonishing functionality enabled by micro-optics, the basic laws and limits of optics still apply - but not necessarily the laws of classical ray optics. It is therefore important to be aware of some of the peculiarities that arise when using micron-sized optical elements.
For a start, it is well known that light beams cannot travel forever without diverging. In the case of micro-optics, however, this restriction can limit transfer distances to the centimetre range, or even less. The diffraction effects that arise at small apertures must also sometimes be considered to determine the actual intensity distribution in the focus region.
There are also important practical limitations to the ways in which DOEs can be used to modify waveforms. For example, the diffraction angle and efficiency depend strongly on wavelength, which make DOEs a good choice for use with a monochromatic laser beam but usually not for a broadband light source. Stray light behaviour and efficiency is generally not as good as with refractive optics, particularly when large deflection angles are desired. And for large angles the required structure sizes of the elements become very small, making them difficult and expensive to manufacture.
Another key issue is that micro-optics cannot be considered as off-the-shelf components, despite a steadily growing range of standard elements. One reason for this is that the product specification must include the size and geometry of the microlens array, as well as the normal parameters of macroscopic optics such as radius of curvature, aperture diameter and material. Further choices include the manufacturing technology, and therefore the cost structure.
What's more, in many cases micro-optics are designed for customized products. Design and technical support are therefore extremely important, and products are often developed in close co-operation between customer and manufacturer.
Micro-optics manufacturing technologies fall into two distinct categories: those that fabricate the final optical element directly - which include serial processes such as ultraprecise machining (single-point diamond turning), or parallel processes such as wafer-based manufacturing - and those that exploit a "master form" that is used in later replication steps.
Wafer-scale fabrication is an established technology that benefits from the advances in semiconductor process technologies. In this technique, a layer of photoresist is first applied to substrates with diameters ranging in size between 100 and 200 mm. Photolithographic masks are used to control the areas of the photoresist that are exposed to light, and the high precision with which these masks are made enables micro-optical structures to be defined with high lateral accuracy - for example, the array pitch can be patterned with submicron precision.
Greyscale masks provide some flexibility in creating arbitrary shapes in a single exposure, but tend to be difficult and expensive to manufacture at high resolution. An alternative technique is reflow technology, in which the photoresist layer is structured into lens preforms by exposure with simple binary masks. Melting the photoresist then causes surface tension to produce spherical surface profiles of extremely high quality and smoothness, resulting in optics with diffraction-limited imaging quality and very low scattering losses.
In both cases, the photoresist structures can be transferred to the wafer substrate by dry etching methods, resulting in lenses made purely from the wafer-substrate material. For example, fused silica is popular for making high-quality lenses for use with wavelengths in the ultraviolet to infrared range because of its optical, physical and chemical properties. For infrared light with wavelengths ranging from 1.2 to about 8 μm, silicon offers excellent transparency and a high refractive index of about 3.5.
A single wafer produced using these techniques can contain 10,000 microlenses or more. Metallic and dielectric coatings, which can be used for antireflective layers or filters, are readily available for such wafers, and are also much more cost-effective to apply than for the same number of individual lenses.
The other main fabrication technique involves creating a master form, either using photolithography (as described above) or by exposing the photoresist with laser- or electron-beam writing. The resulting photoresist structure is then converted into a metal "shim" by electroplating. The shim can then be used as a mould for replication into plastic optical elements by hot-embossing or injection moulding.
Compared to glass and crystalline materials, the usable temperature range of plastics is more limited and the coefficients of thermal expansion are larger. Also, the absorption of plastics rapidly increases towards UV wavelengths. Therefore, plastic optics are less suited to applications requiring high precision, thermal stability, or high-power light sources.
The choice of manufacturing technology determines how the production cost scales with the quantity of elements produced. Direct manufacturing using a serial technique can provide small numbers of prototypes with the lowest non-recurring engineering and tooling (NRE) cost, but unit prices decrease only moderately as the volumes increase.
In contrast, wafer-based direct manufacturing requires higher NRE investments, but unit costs fall significantly when large numbers of wafers are manufactured. Unit prices range from fractions of a dollar to hundreds of dollars, depending on the element size and the quantity produced.
Injection moulding in plastics becomes an option when large quantities of elements costing just a few cents are needed, provided it fits the application and the packaging concept. However, this method also involves the highest NRE cost and allows for little flexibility for design changes after volume production has begun.
Optical assembly traditionally requires one element after the other to be mounted, aligned and permanently fixed in separately machined mounts, which is a lengthy and labour-intensive process. But microtechnology opens the way to new assembly methods. For example, established processes can yield optical or mechanical structures to align and position micro-optical elements with submicron precision. Soldering pads can also be directly processed on the optical elements, such as has been used in figure 3 for attaching the collimation lens to the laser submount. MEMS devices and optical elements may even be integrated as they use similar technologies.
Arrays of lasers or fibres can also be equipped with matching arrays of micro-optics in a single alignment step. Taking this concept even further, whole wafers of optical and optoelectronic elements can be bonded together using anodic bonding or flip-chip bonding before the final elements are diced.
To sum up, micro-optics offers clear benefits in terms of functionality, cost efficiency and ease of industrial assembly. But they are only just starting to exploit their full potential as an enabling technology for ever smaller and higher performance optoelectronic devices. As more optical system designers learn about the advantages of micro-optics, many new or redesigned optical devices are sure to emerge.