Tools and devices

Making things small: lithography and nanofabrication

Every computer chip is a nanoscale object made by printing. The techniques that pattern silicon, and the research methods that push beyond them, are collectively called lithography. They are the top-down half of nanofabrication: start with a flat surface and define structures on it by exposure, etching and deposition.

Photolithography

The Wikipedia article on photolithography describes the basic process. A light-sensitive material called a photoresist is spread on a substrate, usually a silicon wafer. A photomask carrying the desired pattern is placed over it, and light shone through the mask changes the resist chemically where it lands, making the exposed areas either more or less soluble in a developer. After development, the pattern in the resist is transferred into the material beneath by etching, by depositing new material, by plating or by implanting ions. Repeated many times with different masks, this builds up the layers of an integrated circuit.

The limit is the light itself. As the same article explains, the wavelength used determines the smallest feature that can be formed in the resist, which is why the industry has moved from ultraviolet to deep ultraviolet and on to extreme ultraviolet light. The long effort to print ever-smaller features with light is one of the great engineering stories of the past half-century, and it is the reason ordinary transistors now have critical dimensions in the nanometre range.

Writing with electrons

For research, where only a few copies of a pattern are needed, electrons are often more useful than light. The Wikipedia article on electron-beam lithography describes scanning a focused electron beam to draw custom shapes directly in an electron-sensitive resist, which is then developed like a photoresist. Its great advantage, the article notes, is that it can draw arbitrary patterns without a mask at resolutions below 10 nm; its great drawback is low throughput, because the beam has to visit every point of the pattern in turn. That trade-off confines it mainly to making photomasks, to low-volume production and to research and development.

In practice, electron-beam lithography is the workhorse behind a large share of the nanoscale device literature. The metal contacts on a single nanotube or nanowire, the tiny gaps used to measure molecules, and many of the test structures in spintronics and quantum-device papers are drawn this way. When a paper says that electrodes were patterned by electron-beam lithography, it is describing a slow, precise, one-at-a-time process well suited to a laboratory and poorly suited to a factory.

The wider toolbox

Beyond light and electrons, the Wikipedia article on nanolithography sorts the field's methods into four broad groups: photolithography, scanning lithography, soft lithography and a set of miscellaneous techniques. It notes that nanolithography grew from the need to fit ever more sub-micrometre features into integrated circuits, and that lithographic techniques in general are far older than their nanoscale versions.

Scanning-probe lithography uses the tips of probe microscopes to pattern surfaces directly, by removing material or by writing new material onto a substrate. The nanolithography article names dip-pen nanolithography, in which a tip deposits molecules much as a pen deposits ink, as the most widely used of these methods. Probe tips can write very small features but, like electron beams, they work one point at a time. The instruments themselves are described on seeing and touching atoms.

Soft lithography uses stamps, moulds or masks made of elastomers such as polydimethylsiloxane. A stamp can be inked with molecules and pressed onto a surface, transferring a pattern of self-assembled monolayer in a single step; the article lists microcontact printing among these methods. Soft lithography is cheap and works on curved surfaces, which made it popular in chemistry laboratories without access to cleanroom equipment.

Nanoimprint lithography, listed among the miscellaneous techniques, creates patterns by mechanically deforming a resist with a mould, much like pressing a seal into wax. Once a master mould has been made, often by electron-beam lithography, copies can be imprinted quickly, combining high resolution with reasonable throughput.

Top-down meets bottom-up

Lithography is the classic top-down approach. Its opposite is bottom-up fabrication, in which molecules or particles arrange themselves into structures, the subject of the self-assembly page. Each approach has the other's weakness. Lithography is precise and can make arbitrary patterns, but it becomes slow and expensive at the smallest scales. Self-assembly works naturally at molecular scales and in parallel, but it produces only the structures its components are designed to form, and it tolerates defects.

Much current research combines them: a lithographic pattern guides where self-assembling molecules, block copolymers or particles settle, so that coarse features come from the top and fine features from the bottom. The paper that introduced DNA origami, discussed on DNA as a building material, framed its goal in exactly these terms, as a way for bottom-up methods to reach the complexity of top-down ones.

Reading fabrication sections

Methods sections in device papers often compress weeks of cleanroom work into a paragraph. Useful questions are how many devices were made and how many worked, whether the reported feature sizes were measured or merely designed, and which steps limit yield. A striking result from one device out of many is interesting physics but a weak basis for claims about technology. For how such devices are then used, see molecular electronics and nanowires.