Button: Contact CNFButton: MultimediaButton: About CNF
Button: Getting StartedButton: PublicationButton: REU ProgramButton: Events & SeminarsButton: Education OutreachButton: TechnologiesButton: Lab Equipment

Button: Lab User


SPIE Handbook of Microlithography, Micromachining and Microfabrication, Volume 1: Microlithography

Section 2.7 Resists: 2.7.3 Negative Resists


2.7 Resists
2.7.1 Charge Dissipation
2.7.2 Positive Resists
2.7.3 Negative Resists
2.7.4 Multilayer Systems
2.7.5 Inorganic and Contamination Resists
2.7.6 Other Research: Scanning Probes and Thin Imaging Layers
Table of Contents


2.7.3 Negative Resists

Negative resists work by cross-linking the polymer chains together, rendering them less soluble in the developer. Negative resists tend to have less bias (often zero) than positive resists. However, they tend to have problems with scum (insoluble residue in exposed areas), swelling during development, and bridging between features.

A reasonable starting point for developing a negative resist process is to choose a development time twice as long as the time needed to clear the unexposed resist and an exposure dose just sufficient to ensure acceptable resist thickness loss on all features (e.g., no more than 10%). From there, fine tuning of development time, dose, and postexposure bake conditions may be needed to optimize feature sizes, improve critical dimension control, and minimize resist scum.

2.7.3.1 COP

COP is an epoxy copolymer of glycidyl methacrylate and ethyl acrylate, P(GMA-co-EA), commonly used for negative exposure of mask plates. [147] [122] This is a very high speed resist, 0.3 C/cm2 at 10 kV, with relatively poor resolution (1 um). [158] COP also has relatively poor plasma etch resistance and requires spray development to avoid swelling. Because cross-linking occurs by cationic initiation and chain reaction, the cross-linking continues after exposure. Therefore, the size of features depends on the time between exposure and development. Unless speed is very critical, COP is probably not a good choice for a negative resist.

    EXAMPLE PROCESS: COP NEGATIVE MASK PLATE

    1. Soak mask plate in acetone > 10 min to remove photoresist. Rinse in isopropanol, blow dry.

    2. Clean the plate with RIE in oxygen. Do not use a barrel etcher. RIE conditions: 30 sccm O2, 30 mTorr total pressure, 90 W (0.25 W/cm2), 2 min.

    3. Immediately spin COP, 3 krpm

    4. Expose, 10 kV, 0.3 C/cm2 (Other accelerating voltages may be used. The dose will be different.)

    5. Spray develop, MEK (methyl ethyl ketone) : ethanol 7:3 for ~30 s.

    6. Rinse in MIBK (methyl isobutyl ketone) : isopropanol 1:3 for ~30 s (using spray or spinner).

    7. Rinse in isopropanol for ~30 s. (spray or spinner). Blow dry with nitrogen.

    8. Inspect pattern, repeat steps 5-7 as necessary.

    9. Descum in a barrel etcher, 150 W, 0.6 Torr O2, 0.5 min.

    10. Etch chrome in wet etch from Transene or Cyantek (acetic acid and ceric ammonium nitrate) ~1 min. Rinse in water. Blow or spin dry.

    11. Strip with RIE in O2 or by soaking in acetone. (rinse in IPA, blow dry).

2.7.3.2 Shipley SAL

Shipley Inc. [156] produces the popular SAL resist, which comes in a variety of versions and viscocities. SAL has three components: a base polymer, an acid generator, and a crosslinking agent. After exposure, a baking cycle enhances reaction and diffusion of the acid catalyst, leading to resist hardening by cross-linking. Common alkaline photoresist developers will dissolve the unexposed regions. The acid reaction and diffusion processes are important factors in determining the resolution, [159] and a tightly controlled postexposure baking process is required. The postexposure bake is usually on a feedback-controlled hotplate with a suction holder to ensure good thermal contact. The extent of the cross-linking reaction is therefore affected by the thermal conductivity of the sample and by the cooling rate after the bake. Resolution of 30 nm has been demonstrated at very low voltage, [160] and 50 nm wide lines have been fabricated using high voltage. [161] SAL-606 has 0.1 um resolution in 0.4 um thick films, exposed with 40 keV electrons at 8.4 uC/cm2.

The novolac base polymer has etching properties similar to those of positive photoresists. Unlike photoresist, the shelf life of SAL is on the order of six months at room temperature. Refrigeration extends the shelf life to several years, but care is required to avoid condensation when the resist is dispensed to smaller containers. SAL is a sensitive resist, 7 to 9 uC/cm2 at either 20 or 40 kV, and so is suitable for mask writing. It is interesting to note that, unlike PMMA, the critical dose of SAL does not scale proportionately with accelerating voltage. Although it is not as sensitive as other negative resists (COP, CMS, or GMC) SAL has far better process latitude and resolution.

    EXAMPLE PROCESS: SAL NEGATIVE MASK PLATE

    1. Soak mask plate in acetone > 10 min to remove photoresist.

    2. Clean the plate with RIE in oxygen. Do not use a barrel etcher. RIE conditions: 30 sccm O2, 30 mTorr total pressure, 90 W (0.25 W/cm2), 5 min.

    3. Immediately spin SAL-601, 4 krpm, 1 min.

    4. Bake in 90 C oven for 10 min. This resist is not sensitive to room light.

    5. Expose at 50 kV, 11 C/cm2. Be sure the plate is grounded.

    6. Post-bake for 1 min on a large hotplate, 115 C.

    7. Cool for > 6 min.

    8. Develop for 6 min in Shipley MF312:water (1:1) Be sure to check for underdevelopment.

    9. Descum 30 s with oxygen RIE: same as step 2, 10 s.

    10. Etch with Transene or Cyantek Cr etchant, ~1.5 min.

    11. Plasma clean to remove resist: Same as step 2, 5 min.

2.7.3.3 Noncommercial negative resists: P(SI-CMS) and EPTR

Although not yet commercialized, a very promising negative resist is P(SI-CMS), which combines the high speed of CMS (chloromethylstyrene) with the etch resistance of SI (trimethylsilylmethyl methacrylate). This resist offers at least 10 times the plasma etch resistance of SAL. [162-164] Its silicon component gives excellent resistance to etching in an oxygen plasma by forming a surface layer of silicon oxide. The sensitivity is similar to that of SAL (~10 uC/cm2 at 40 kV) but the resolution is around 0.2 um. P(SI-CMS) will be a good choice when etch resistance is more important than resolution.

The epoxy type resist [165-167] developed at IBM is a combination of a novolac epoxy resin (o-cresol novolac glycidyl ether) and an onium salt (triphenylsulfonium hexafluoroantimonate) photoinitiator. EPTR is a high-speed resist (6 uC/cm2 at 50 kV) with relatively high contrast ( = 6.4) and high resolution (50 nm). While the resolution of EPTR is comparable to that of Shipley SAL, the epoxy formulation allows EPTR to be extended to layer thicknesses exceeding 200 um. [168] The high aspect ratio and thicknesses accessible with EPTR make it uniquely suited for micromechanical applications.


Table 2.5. Comparison of commercially available electron beam resists.

Tone Resolution (nm)
Sensitivity at 20 kV (uC/cm2)
Developer Contact reference
PMMA positive 10 100 MIBK:IPA [135]
EBR-9 positive 200 10 MIBK:IPA [143]
PBS positive 250 1 MIAK : 2-Pentanone 3:1 [147]
ZEP positive 10 30 xylene : p-dioxane [148]
AZ5206 positive 250 6 KLK PPD 401 [152]
COP negative 1000 0.3 MEK : ethanol 7:3 [147]
SAL-606 negative 100 8.4 MF312 : water [156]


Next Sub-Section: 2.7.4 Multilayer Systems

Table of Contents



Back to Top




Button: Search Button: Search Keywords
Cornell University
NYSTARNNINNSF