Charting Progress In Semiconductor Fabrication Advances in photoresist resins pave the way for production at 193 and 157 nm Stephen C. Stinson C&EN Northeast News Bureau

From the ACS meeting

E lectronic semiconductor devices are ubiquitous—from computers to cars to industrial process con­

trol. At a symposium on polymers for micropatterning and nanopatterning sponsored by the Division of Polymeric Materials: Science & Engineering, chemists reported progress in the fabri­cation of such devices. Their progress will make possible the data-processing units and memory chips of the future.

A key semiconductor fabrication technology is microlithography—the etching, doping, and plating of devices onto semiconductor surfaces at resolu­tions of micrometers to nanometers. Achieving ever-finer patterns—needed for smaller and faster devices—depends on polymeric resins called photoresists. — -

It became clear during the sym-posium that microlithography might eventually be superseded by advances in new technolo­gies—"drawing" of patterns on re­sist layers by electron beams, for example, or "printing," "rubber-stamping," or molding of images at exquisitely fine resolution onto semiconductor surfaces. But as several sessions made clear, ad­vances are still being made in mi­crolithography at the finer resolu­tions made possible by short-wavelength light sources.

In New Orleans, chemistry professor C. Grant Willson of the University of Texas, Austin, sur­veyed the history of lithography and resists. He put up the familiar equation, F = &(A/NA), in which the minimum feature size (F) is proportional to the wavelength divided by the numerical aper­ture (NA) of the imaging lens. But Willson also pointed out that the resist resin must be transpar­

ent to that wavelength. And although re­sists routinely fend off dopants or plated metals, they must also be impervious to any etching of the substrate.

Fabricators apply photoresists to semiconductor surfaces and expose those surfaces to ultraviolet light through a patterned mask to uncover the areas where the dense network of thin lines and tiny spots that make up a device are to be laid down.

Resists are classed as positive or nega­tive tone. A positive-tone resist is insolu­ble in the developer solvent but is solubi-lized by exposure to UV light. A negative-tone resist is soluble in the developer solvent but is cross-linked and rendered insoluble by exposure to UV light. The specifications for the photoresist depend on the UV wavelength used.

For some years, fabricators have used a positive-tone resist of phenol no-volac formulated with 2-diazonaphtho-

Photoresists yield positive or negative images

i •H I Mask

Photoresist layers

Substrate

Exposed area solubilized

Exposed area v cross-linked

Developer solvent

Etch

Remove photoresist

Positive image Negative image

quinone. Novolac is a thermoplastic, al­kali-soluble, phenol-formaldehyde resin whose polymer chains have stopped short of cross-linking to the three-dimensional, thermosetting, insoluble form. The diazo compound is a dissolu­tion inhibitor.

When exposed to light, the diazo com­pound photolyzes to indene-2-carboxylic acid. When light-exposed areas are devel­oped with aqueous alkali, the indenecar-boxylic acid leaches out, and the novolac resin also dissolves away.

Chemists long thought that the diazo compound somehow physically blocked dissolution of the novolac. But Lewis W. Flanagin, one of Willson's former gradu­ate students, presented the group's analy­sis in New Orleans that dissolution inhib­itors of all kinds interact with phenolic hy-droxyl groups to reduce their acidity and thus delay ionization of phenolic novolac hydroxyls needed for dissolving to occur.

As the industry moved from imaging light sources ranging from 350 to 450 nm down to the 248-nm light from kryp­ton fluoride lasers, it had to find new re­sists. One answer was poly-^-styrene (PS-OH), pioneered by Hitachi. But it took the development of chemical am­plification by Willson, chemistry profes­sor Jean M. J. Frechet of the University of California, Berkeley, and polymer chemist Hiroshi Ito of IBM's research

laboratory in San Jose, Calif., to •f^ bring PS-OH into routine use.

In chemical amplification, the resist is PS-OH protected as the ^^-butoxycarbonyl derivative and blended with a photoacid generator. Photoacid generators such as diphenyliodonium and triphenylsulfonium salts were pi­oneered for curing adhesives by polymer chemistry professor James V. Crivello of Rensselaer Polytechnic Institute, Troy, N.Y.

Photolysis of these salts pro­duces iodine or sulfur radical cat­ions, which abstract hydrogen at­oms from the surroundings to generate hydrogen ions in the light-exposed areas. Hydrogen ions catalyze decomposition of te?f-butoxycarbonyl groups when the semiconductor wafer is baked, which frees up phenolic hydroxyl groups for dissolving by alkali de­veloper and regenerates more protons.

The Semiconductor Industry Association, San Jose, Calif., has established a "road map" into the

SEPTEMBER 13,1999 C&EN 3 7

science/technology

next century, setting mile­stones for adoption of ever-shorter wavelengths of light for imaging of ever-finer fea­tures. For the period 2001-03, the target is 193 nm from ar­gon fluoride lasers, and after 2003, the target is 157 nm from fluorine lasers.

But meeting these mile­stones demands development of still more new resists. Ben­zene derivatives absorb at 193 nm, so novolacs and PS-OH won't work. Carbonyl groups are transparent at 193 nm, so carboxyl groups are among those that can be used to im­part alkali solubility.

With polymer chains limit­ed to largely aliphatic units, their etch resistance is low. Among the thinkers who have contributed to understanding of etch resistance are Robert R. Kunz, senior staff member at Massachusetts Institute of Technology's Lincoln Labo­ratory, and polymer chemist Yoshitake Ohnishi of NEC Corp., Tsukuba, Japan.

The Kunz number is the sum of the atomic weights of the carbon atoms in the polymer repeating unit divided by the formula weight of the repeating unit. High Kunz numbers indicate good etch resistance. The Ohnishi number is the total number of atoms in the repeating unit divided by the number of carbon at­oms minus the number of other atoms. Ohnishi numbers close to 1.00 imply good etch resistance.

Some chemists have made resins with polycyclic chain units such as nor-bornyl or adamantyl because their high carbon-hydrogen ratios confer favor­able Kunz and Ohnishi numbers. Other workers have incorporated silicon, which also imparts etch resistance.

For example, Ralph R. Dammel, group leader for technology in the AZ Electronic Materials business at Clari-ant Corp., Somerville, N.J., described the company's AX-1000P 1:1 resist

AX-1000P is a copolymer of methac-rylate esters of 2-methyl-2-adamantanol and mevalonic lactone (six-membered lactone of 25-dihydroxy-2-methylpen-tanoic acid) compounded with an "oni-um" salt photoacid generator. Photoly­sis of the salt generates protons, which open the pendant lactone rings, freeing up the alkali-soluble acid.

In other work, Clariant investigators studied copolymers of maleic anhy-

Photoacid generators liberate protons . . .

(C6H5)3S C4F9SO3 /TV, (C6H5)2S + C6H6 + C4F9S03-H+

. . . which free phenolic hydroxyl groups . - .

H+,

C(CH3)3

+ CH2==C(CH3)2f

+ co2t + H+

. . . rendering exposed resin alkali-soluble

dride, 5-norbornene-2-carboxylic acid (NCA), and isobornyl, tert-butyl, and 2-hydroxyethyl esters of NCA. They compounded a photoacid generator with this copolymer also.

In addition to the general alkali-solu-bilizing function of the esters on hydrol­ysis, the 2-hydroxyethyl ester improved adhesion to the substrate, and the isobornyl ester lowered the glass transi­tion temperature of the resin for anneal­ing. After spin-coating the semiconduc­tor wafer with resist solution, fabrica­tors bake the wafer to anneal the resin film and ease the stresses in dried film.

Such subtle modulation of the ester content of the resist resin copolymer is an example of several fine-tuning strate­gies among interacting components of resist systems that were described in New Orleans. In addition to adhesion to substrate and ease of annealing, acid mi­gration and T-topping were two other problems of photoacid generators dis­cussed in the sessions.

T-topping takes its name from the shape of test patterns of parallel trench­es that researchers create in resist films to gauge performance. Ideally, the trench is composed of perfectly parallel sides, with absolute 90° angles defining the walls' junction with the bottom.

When T-topping occurs, the shape of the trench in cross section resembles two capital letter Ts (TT), with the top of the trench partially to completely closed over. T-topping can happen when there is a delay between exposure to UV light and baking of a wafer to consummate

the reaction between protons and protected ester groups.

In that time, small amounts of amines that are always present in clean-room air can neutralize the protons at the very top of the resist film. The result is that the developer dis­solves the resist down to the bottom of the trench, but leaves "eaves" and "roofs" at the top.

One solution offered by Kyle W. Patterson, a graduate student in the Willson group, is a copolymer containing 10% of a free carboxylic acid. This acid content raises the background alkali solubility, allowing disso­lution of any neutralized sur­face. A resist so formulated shows no T-topping after a 100-minute delay.

Another effect of delay is mi­gration of the acid. Photolysis of the oni-um salt liberates protons within the ex­posed area, but delay results in slow diffu­sion through the resin out of that area, eroding the resolution. Senior scientist Allen H. Gabor of Arch Chemicals, North Kingston, RL, proposed in New Orleans that a photoimageable base be used to trap migrating protons. Gabor worked with investigators at Lucent Technolo­gies' Bell Labs, which has an agreement with Arch to commercialize new semi­conductor fabrication technology.

Instead of the usual diphenyliodonium heptafluorobutanesulfonate, which pro­duces protons free of the sulfonate coun-terion, the Arch investigators used N-cy-clohexylsulfamate as counterion. Photol­ysis of that acid generator leads to a zwitterion. In the unexposed areas, the sulfamate remains intact If hydrogen ions migrate into unexposed areas, they are neutralized by the sulfamate nitrogen.

As progress was presented to meet the new 193-nm milestone in microli-thography, Willson gave a glimpse of what the world of 157 nm might look like. At that wavelength, carbonyl groups are opaque, so researchers seek new acid sources for photolysis and al­kali development.

One thrust is into fluorine chemistry. Willson cited preliminary work by poly­mer chemistry professor Patrick E. Cassidy of Southwest Texas State Uni­versity, San Marcos, on polymer chains with units of polyvinyl alcohol and polynorbornene or of norbornenol and polytetrafluoroethylene.^

3 8 SEPTEMBER 13,1999 C&EN