`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. ED-28, NO. 11, NOVEMBER 1981
`
`1405
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`[24] D. E. Widmann and H. Binder, “Linewidth variations in photo-
`
`
`
`
`
`
`
`
`resist patterns on profiled surfaces," IEEE Trans. Electron
`
`
`
`
`
`
`
`
`
`Devices, vol. ED-22, p. 467, 1975.
`
`
`
`
`
`
`
`[25] C. H. Ting, private communication.
`
`
`
`
`
`
`
`
`
`
`[26] L. B. Rothman, “Properties of thin polymide films,” J. Electro-
`
`
`
`
`
`
`
`
`chem. Soc. , vol. 127, p. 2216, 1980.
`‘
`
`
`
`
`
`
`
`
`
`[27] W. T. Scott, “Correlated probabilities in multiple scattering,”
`
`
`
`
`
`
`
`
`
`
`.
`Phys. Rev., vol. 76, p. 212, 1949.
`
`
`
`
`
`
`
`
`
`
`[28] J. J. Goldstein, J. L. Costley, G. W. Lorimer, and S.J.B. Reed,
`
`
`
`
`
`
`
`“Quantitative X-ray analysis in the electron microscope,” SEM,
`
`
`
`
`
`vol. 1, p. 315, 1977.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`[29] R. K. Watts, W. Fichtner, E. N. Fuls, L. R. Thibault, and R. L.
`
`
`
`
`
`
`
`
`
`Johnson, “Electron beam lithography for small MOSFET’s,” in
`
`
`
`
`
`
`
`
`19801EDM Tech. Dig, p. 772, Dec. 1980.
`
`
`
`
`
`
`
`
`
`
`
`Linewidth Control in Projection Lithography
`
`
`
`
`Using a Multilayer Resist Process
`
`
`
`MICHAEL M. O’TOOLE, MEMBER, IEEE, E. DAVID LIU, AND MARK S. CHANG, MEMBER, IEEE
`
`
`
`
`
`
`
`
`
`
`Abstract—Linewidth control using a tri-layer resist system on wafers
`
`
`
`
`
`
`
`
`
`
`with topography is investigated. An absorbing dye is incorporated in
`
`
`
`
`
`
`
`
`
`
`
`the bottom layer to improve the usable resolution. Resist patterns of
`
`
`
`
`
`
`
`
`
`l-um lines and spaces over aluminized topography are demonstrated
`
`
`
`
`
`
`
`
`
`
`
`using a projection aligner. The advantages of a multilayer system are
`
`
`
`
`
`
`
`
`investigated using an exposure and development simulation program
`
`
`
`
`
`
`
`
`for optical lithography. The relative contributions of planarization and
`
`
`
`
`reflection suppression are discussed.
`
`
`
`
`
`1.
`
`INTRODUCTION
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`sorbing dye in the bottom polymer. The tri-layer structure is
`
`
`
`
`
`
`
`analyzed theoretically using an exposure and development
`
`
`
`
`
`
`
`
`simulation program for optical lithography [7]. The program
`
`
`
`
`
`
`
`assumes diffraction-limited optics and considers the numerical
`
`
`
`
`
`
`
`
`
`
`aperture of the imaging lens, the imaging wavelength, the par-
`
`
`
`
`
`
`
`
`
`
`tial coherence factor of the illumination system, and the focus
`
`
`
`
`
`
`
`
`
`
`
`error to generate an intensity pattern on the surface of the
`
`
`
`
`
`
`
`
`
`
`resist. The resist is then exposed and developed using the model
`
`
`
`
`
`
`
`
`
`
`
`
`described by Dill er a1.
`[8]. The final output is a simulated
`
`
`
`
`
`
`
`
`
`line-edge profile in positive resist. The simulations in this
`
`
`
`
`
`
`
`
`
`
`paper are for Hunt positive resist developed nominally for 15 s
`
`
`
`
`
`
`
`
`
`
`
`
`
`in a mixture of two parts MF312 developer with one part water.
`
`
`
`
`
`
`
`
`
`The exposure and development parameters for the resist [9]
`
`
`
`
`
`
`
`
`
`
`were measured using equipment similar to Dill’s for the expo-
`
`
`
`
`
`
`
`
`
`
`sure parameters and to Meyerhofer’s [10] for the development
`
`parameters.
`
`
`
`
`
`
`
`
`
`By using the exposure and development parameters of the
`
`
`
`
`
`
`
`
`
`resist in conjunction with the simulation program, the relative
`
`
`
`
`
`
`
`
`contributions of planarization and of reflection suppression to
`
`
`
`
`
`
`
`
`
`linewidth control are shown for Him geometries. The simula-
`
`
`
`
`
`
`
`
`
`tion results provide an analytical understanding of the problem
`
`
`
`
`
`
`
`
`
`and aid in process optimization. Experiments with the tri-layer
`
`
`
`
`
`
`
`
`technique are conducted for Him geometries over aluminized
`
`topography.
`
`
`
`
`
`II. LIMITS OF CONVENTIONAL PROCESSING
`
`
`
`
`
`
`
`
`
`
`The usable resolution of a projection system varies with the
`
`
`
`
`
`
`
`substrate topography and material. Current projection aligners
`
`
`
`
`
`
`
`
`
`can resolve submicrometer features with positive resist and con-
`
`
`
`
`
`
`
`
`
`ventional resist processing on a planar and nonreflective sub-
`
`
`
`
`
`
`
`
`
`strate.
`In device fabrication, however, the image is projected
`
`
`
`
`
`
`
`onto a nonplanar, reflective surface covered unevenly with
`
`
`
`
`
`
`
`resist. The resulting usable resolution degrades to approxi-
`
`
`0018-9383/81/1lOO-l405$00.75 ©19811EEE
`
`
`
`
`
`
`
`
`
`N OPTICAL lithography, the demand for small feature sizes
`
`
`
`
`
`
`
`
`
`has resulted in optical projection printers with higher numer-
`
`
`
`
`
`
`
`ical apertures, closer tolerances,
`lower imaging wavelengths,
`
`
`
`
`
`
`
`and better alignment capabilities. These improvements extend
`
`
`
`
`
`
`
`
`the theoretical resolution limit of projection lithography into
`
`
`
`
`
`
`the submicrometer range. However, the practical resolution
`
`
`
`
`
`
`
`
`
`limit has been considerably larger due to the difficulty in main-
`
`
`
`
`
`
`
`
`taining a constant resist linewidth over substrate topography.
`
`
`
`
`
`
`
`
`In an attempt to improve linewidth control over topography,
`
`
`
`
`
`
`
`several multilayer resist processes have recently been proposed
`
`
`
`
`
`
`
`
`
`and demonstrated [1] -[6] . In the multilayer system, the sub-
`
`
`
`
`
`
`
`
`strate topography is planarized by a bottom polymer layer. In
`
`
`
`
`
`
`
`addition, reflections from the underlying topography can be
`
`
`
`
`
`
`
`
`
`eliminated by choosing an absorptive material for the bottom
`
`polymer.
`
`
`
`
`
`
`
`
`
`The technique outlined here extends the tri-layer scheme of
`
`
`
`
`
`
`
`
`Bell Laboratories [1], [2] by incorporating a selectively ab-
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Manuscript received May 1, 1981 ; revised July 24, 1981.
`
`
`
`
`
`
`
`
`
`
`
`
`M. M. O’Toole and E. D. Liu are with Hewlett Packard Laboratories,
`
`
`
`
`Palo Alto, CA 04304.
`
`
`
`
`
`
`
`
`
`
`M. S. Chang was with Hewlett Packard Laboratories, Palo Alto, CA.
`
`
`
`
`
`
`
`
`
`
`
`
`
`He is now with Seeg Technology, Inc., San Jose, CA 95131.
`
`
`
`Page 1 of 6
`
`TSMC Exhibit 1046
`
`TSMC v. IP Bridge
`IPR2016-01377
`
`
`
`
`
`Page 1 of 6
`
`TSMC Exhibit 1046
`TSMC v. IP Bridge
`IPR2016-01377
`
`

`

`1406
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. ED-28, NO. 11, NOVEMBER 1981
`
`
`1RELATIVE INTENS!TY
`
`
`
`'
`
`.5
`.5
`
`
`
`
`I
`
`V
`
`
`
`1.3
`
`
`THICKNESS (pm‘) 5
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 3. The fractional intensity coupled into I-IPR 204 positive resist
`
`
`
`
`
`
`
`
`
`
`(N = 1.69— 10.12) on a silicon substrate as a function of the resist
`
`thickness.
`
`
` .5
`
`RELATIVE INTENSITY
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`1.2!
`
`
`
`Fig. 1. Resist step. coverage.
`
`fl
`l-pm-thrck resrst pattern over 0.5-um
`step.
`
` '-
`
`
`
`mappSE '(mj/Icmz)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`5m.
`
`1
`
`
`
`
`
`
`
`.5
`
`
`
`
`
`
`
`
`
`
`
`1.5
`1.
`
`
`
`
`RESIST THICKNESS (pm)
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig.4. The exposure energy density (dose) required at 436 nm to
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`achieve l-um lines and spaces in HPR 204 on (a) silicon and (b) alu-
`
`
`
`
`
`
`
`
`
`
`
`
`
`minum for A =436 nm, NA = 0.28, perfect focus,‘and partial co-
`
`
`
`herence factor a = 0.7.
`
`
`
`
`
`
`
`
`DISTANCE (pm)
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 2. The image intensity pattern of a periodic l-um line and space
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`for the case of (a) perfect focus and (b) 3 pm of focus error. A = 436
`
`
`
`
`
`
`
`
`
`
`nm; NA = 0.28; partial coherence factor (a) = 0.7; square aperture.
`
`
`
`
`
`
`
`(Dashed line is the intensity at the reticle.)
`
`
`
`
`
`
`
`
`
`
`
`
`
`Variations in exposure are due to nonuniform illumination
`
`
`
`
`
`
`
`
`
`
`of the mask and to the standing-wave effect. The exposure
`
`
`
`
`
`
`
`
`
`variation due to nonuniform illumination of the mask is gen-
`
`
`
`
`
`
`
`
`
`
`
`
`erally less than 5 percent and much smaller than that due to
`
`
`
`
`
`
`
`
`the standing-wave effect. The standing-wave effect is related
`
`
`
`
`
`
`
`
`
`to multiple reflections of the electromagnetic waves [12] in
`
`
`
`
`
`
`
`
`
`
`
`the resist and in the underlying films. Small variations in the
`
`
`
`
`
`
`
`
`
`resist thickness or in the thin semitransparent layers under the
`
`
`
`
`
`
`
`
`
`
`
`resist can cause large variations in the energy coupled into the
`
`
`
`
`
`
`
`
`
`
`resist.
`Fig. 3 shows the fractional intensity coupled into a
`
`
`
`
`
`
`
`
`
`
`
`
`
`film of resist on a silicon substrate as a function of the resist
`
`
`
`
`
`
`
`
`
`
`
`thickness. For an exposure wavelength of 436 nm and a resist
`
`
`
`
`
`
`
`
`
`
`
`indexN = 1.69 — 10.012, a 64-nm change in the resist thickness
`
`
`
`
`
`
`
`
`
`
`
`can cause a 50-percent change in the energy coupled into the
`
`
`
`
`
`
`
`
`
`
`
`resist. The energy coupled into the resist is periodic with pe—
`
`
`
`
`
`
`
`riod 7\/2n. Essentially, random variations in resist thickness
`
`
`
`
`
`
`
`
`
`
`occur as the resist covers the substrate topography.
`In addi-
`
`
`
`
`
`
`
`
`
`
`
`tion, changes in the reflectivity of the features under the resist
`
`
`
`
`
`
`
`
`
`cause additional variations in the amount of energy available
`
`
`
`
`
`
`
`
`
`for resist exposure. The exposure variations are most evident
`
`
`
`
`
`
`
`
`
`as the resist lines traverse steps. Resist features approaching
`
`
`
`
`
`
`
`
`
`
`the resolution limit of the projection lens show increased line-‘
`
`
`
`
`
`
`
`
`width instability because of the nonzero intensity discussed
`
`previously.
`
`
`
`
`
`
`
`
`The standing-wave and bulk effects may be simulated using
`
`
`
`
`
`
`
`
`
`a computer program for the simulation of optical projection
`
`
`
`
`
`
`
`
`printing. Fig. 4(a) simulates the nominal exposure required
`
`
`
`
`
`
`
`
`
`
`
`
`
`for a periodic 1-,um line and space pattern as a function of the
`
`
`
`
`
`
`
`
`
`
`resist thickness of positive resist on (a) a silicon substrate, and
`
`
`
`
`
`
`
`
`
`
`
`(b) an aluminum substrate. The nominal dose is defined as the
`
`
`
`
`
`
`
`
`
`exposure energy density required to obtain the mask linewidth
`
`
`
`
`
`
`
`
`
`
`
`
`
`in the resist. The bulk effect is evident by the gradual rise of
`
`
`
`
`
`
`
`
`
`
`
`the curve; and the standing-wave effect is evident by the peri-
`
`
`
`
`
`
`
`
`
`
`
`odic variation h/Zn, or 128 nm for an exposure wavelength of
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`mater a 1.5-um feature size for an aligner with a numerical
`
`
`
`
`
`
`
`
`
`
`
`aperture of 0.3 and an imaging wavelength of 436 nm. The
`
`
`
`
`
`
`
`
`
`nonplanar, reflective surface gives rise to two effects which
`
`
`
`
`
`
`
`
`
`
`
`limit the usable resolution of the aligner. The first effect is
`
`
`
`
`
`
`
`
`
`
`
`related to large thickness variations of the resist near steps, or
`
`
`
`
`
`
`
`
`
`
`the “bulk effect.” The second effect is related to multiple
`
`
`
`
`
`
`
`
`reflections from the substrate, or the “standing-wave effect.”
`
`
`
`
`
`
`
`
`
`
`
`The bulk variation in the resist thickness as it covers a step
`
`
`
`
`
`
`
`
`
`
`
`is demonstrated in the micrograph of Fig. 1.
`If the resist on
`
`
`
`
`
`
`
`
`
`
`
`
`top of the step and the resist next to the step receive equal
`
`
`
`
`
`
`
`
`
`
`
`
`exposure, the resist on top of the step will clear first. The re-
`
`
`
`
`
`
`
`
`
`
`
`sist over the step may continue to develop while the thick
`
`
`
`
`
`
`
`
`
`
`
`resist next to the step clears, resulting in a narrowing of the
`
`
`
`
`
`
`
`
`
`
`resist line over the step. The narrowing is more pronounced
`
`
`
`
`
`
`
`
`for linewidths approaching the resolution limit of the aligner’s
`
`
`
`
`
`
`
`
`
`
`objective lens and for areas slightly out of focus or influenced
`
`
`
`
`
`
`
`
`
`
`
`by scattered light. Fig. 2(a) is the calculated intensity of the
`
`
`
`
`
`
`
`
`
`
`
`image of a periodic l-Mm line and space pattern produced by a
`
`
`
`
`
`
`
`
`
`
`
`lens with a numerical aperture of 0.28 at a wavelength of 436
`
`
`
`
`
`
`
`
`
`
`
`
`nm and a partial coherence factor [11] of 0.7. The curve is
`
`
`
`
`
`
`
`
`
`
`
`normalized so that large clear areas have an intensity of 1.0.
`
`
`
`
`
`
`
`
`
`
`The dashed line represents the ideal intensity profile, or that
`
`
`
`
`
`
`
`
`
`
`
`which exists at the reticle for a perfect chromium line. The
`
`
`
`
`
`
`
`
`
`
`nonzero intensity of the imaged line due to diffraction allows
`
`
`
`
`
`
`
`
`
`
`
`
`some exposure of the resist in an area where the resist should
`
`
`
`
`
`
`
`
`
`
`remain unexposed. Focus error, shown in Fig. 2(b), and scat-
`
`
`
`
`
`
`
`
`
`tered light further contribute to the undesirable exposure of
`
`
`
`
`
`
`
`
`
`
`the resist line. The undesirable exposure allows the resist lines
`
`
`
`
`
`
`
`
`
`
`
`in some areas to continue to develop and narrow, While areas
`
`
`
`
`
`
`
`
`
`
`having thicker resist or receiving less exposure have not yet
`
`cleared.
`
`
`
`
`Page 2 of 6
`
`Page 2 of 6
`
`

`

`O’TOOLE et al.: LINEWIDTH CONTROL IN PROJECTION LITHOGRAPHY
`
`
`
`
`
`
`
`
`
`
`1407
`
`
`
`
`
`0.5nm Hunt HPR
`
`0.13pm Si W
`l—3um Bottom
`
`Polymer Layer
`
`
`
`
`
`
`
`
`
`
`
`
`Substrate
`
`
`
`
`
`
`
`Fig. 6. Tri—layer resist system.
`
`TRANSMlTTAVCE
`LO
`
`
`
`‘ J l l
`
` GO
`
` l i l
`
`
`400
`5OQ
`600.
`7OQ
`800
`WAVELENGTd (NM)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 7. Transmittance spectra of (A) 1.5-um HPR 204 and (B) 1.7—um
`
`
`
`
`
`
`
`
`
`
`HPR 204 with 1.5-percent concentragtion of dye by weight in solution
`
`
`
`
`
`
`
`
`
`
`
`after hard bake in a box oven at 160 C for 30 min.
`
`
`
`
`.
`2
`anROSE (mJ/cm )—|—l_'v—P‘I_l_ll
`
` 1B.
`
`.5
`
`
`
`
`
`2. fl
`1. 5
`2. 5
`1.5
`
`
`
`
`
`BOTTOM POLYMER THICKNESS (pm)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 8. Nominal dose versus bottom polymer thickness for the tri—layer
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`system of Fig. 6 on an aluminum substrate for (a) HPR 204 for the
`
`
`
`
`
`
`
`
`
`
`bottom polymer and (b) HPR 204 with 1.5-percent dye. Simulations
`
`
`
`
`
`
`
`
`
`are for l-pm lines and spaces under perfect focus.
`
`
`
`
`
`
`
`
`of several polymers measured with a spectrophotometer. Curve
`
`
`
`
`
`
`
`
`
`
`(A) shows the transmission spectrum of a 1.5-um cOating of
`
`
`
`
`
`
`
`
`
`
`
`
`Hunt 204 positive resist baked at 160°C for 30 min in a box
`
`
`
`
`
`
`
`
`
`
`oven. Curve (B) shows the transmission of a 1.7-um coating of
`
`
`
`
`
`
`
`
`
`Hunt positive resist with the addition of a 1.5-percent concen-
`
`
`
`
`
`
`
`
`
`
`
`tration by weight of dye. The Hunt film absorbs 20-25 percent
`
`
`
`
`
`
`
`
`
`
`
`
`of the exposure light at 436 nm in a single pass, while the Hunt
`
`
`
`
`
`
`
`
`
`
`204 with dye absorbs nearly 92 percent of the exposure light.
`
`
`
`
`
`
`
`
`
`
`Positive resist without dye may be made more absorbing by
`
`
`
`
`
`
`
`
`hardbaking at a higher temperature or for a longer time [14] .
`
`
`
`
`
`
`
`
`
`
`Fig. 8 shows the simulated nominal dose required to print 1-
`
`
`
`
`
`
`
`
`
`
`
`
`14m lines and Spaces as a function of the thickness of the bottom
`
`
`
`
`
`
`
`
`
`
`
`polymer for the tri-layer system of Fig. 6. Curve ((1) assumes
`
`
`
`
`
`
`
`
`
`
`
`
`the absorption given for the Hunt resist of Fig. 7(A), and curve
`
`
`
`
`
`
`
`
`
`
`
`(b) assumes the absorption for Hunt resist with dye in Fig.
`
`
`
`
`
`
`
`
`
`7(3). The addition of dye significantly reduces the variation
`
`
`
`
`
`
`
`
`
`
`in the nominal exposure dose due to the standing-wave effect.
`
`
`
`
`
`
`
`
`
`
`
`From Fig. 8(b), a 1.1 urn of positive resist with dye‘should
`
`
`
`
`
`
`
`
`suppress reflections from the underlying topography. In other
`
`
`
`
`
`
`
`
`
`words, the highest point on the substrate topography must be
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 5.
`
`
`
`
`
`
`
`
`
`
`
`
`l-um lines and spaces in 1 pm of resist over 0.5-um polysilicon
`
`steps.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`436 rim and a resist index of 1.69. A 25-percent exposure dif-
`
`
`
`
`
`
`
`
`ference is required to compensate for the standing-wave effect
`
`
`
`
`
`
`
`
`
`
`
`for a 64-nm thickness variation in l um of resist on a silicon
`
`
`
`
`
`
`
`
`
`substrate. A similar exposure difference due to the bulk effect
`
`
`
`
`
`
`
`requires a 250-nm resist thickness variation. Aluminum sub-
`
`
`
`
`
`
`
`
`strates (Fig. 403)) with their greater reflectivity demonstrate a
`
`
`
`
`
`
`
`
`
`larger standing-wave effect. From Fig. 4(b), a bulk thickness
`
`
`
`
`
`
`
`
`variation of about 420 rim is equal to a standing-wave thickness
`
`
`
`
`
`
`
`
`
`
`variation of 64 nm. Both effects can combine near a step to
`
`
`
`
`
`
`
`
`
`result in a significant variation in the nominal dose required
`
`
`
`
`
`
`
`
`and, therefore, severe linewidth control problems. Fig.5 shows
`
`
`
`
`
`
`
`
`
`
`
`
`a micrograph of l-um lines and spaces patterned in 1 pm of
`
`
`
`
`
`
`
`
`
`
`resist over a 0.5-,um polysilicon step. The linewidth is very
`
`
`
`
`
`
`
`unstable near the edge of the steps.
`
`
`
`
`III. TRI-LAYER RESIST WITH ABSORBING DYE
`
`
`
`
`
`
`
`
`
`
`In order to realize the maximum resolution from an aligner,
`
`
`
`
`
`
`
`
`
`
`
`
`the surface of the wafer must approach that of a flat, non-
`
`
`
`
`
`
`
`
`reflecting substrate. The purpose of multilayer systems is to
`
`
`
`
`
`
`
`approximate the ideal surface conditions for exposure. Fig.
`
`
`
`
`
`
`
`
`6 illustrates the multilayer structure used. An absorbing poly-
`
`
`
`
`
`
`
`
`
`
`
`
`mer,
`l to 3 pm thick, is used to planarize the substrate topog-
`
`
`
`
`
`
`
`raphy. The planarized surface enables the uniform dispense
`
`
`
`
`
`
`
`
`
`
`
`of the top resist layer and thus suppresses the bulk effect. The
`
`
`
`
`
`
`
`absorption of the bottom polymer eliminates reflections from
`
`
`
`
`
`
`
`the substrate topography and reduces the standing-wave effect.
`
`
`
`
`
`
`
`
`
`An intermediate silicon nitride layer serves as a reactive ion
`
`
`
`
`
`
`
`
`
`
`etch shield for the pattern transfer to the bottom layer. The
`
`
`
`
`
`
`
`
`
`silicon nitride has an index of approximately 1.8, which mini-
`
`
`
`
`
`
`
`
`mizes reflections from the nitride—resist interfaces.
`If the dif-
`
`
`
`
`
`
`
`
`
`ferential etch rate between the top and bottom polymers were
`
`
`
`
`
`
`
`sufficient, an intermediate layer would not be required.
`
`
`
`
`
`
`
`
`
`Suitable materials for the bottom layer are polymers that
`
`
`
`
`
`
`have good planarization capabilities. Transparent polymers
`
`
`
`
`
`
`
`
`
`
`
`may be made absorbing with the addition of dye. The dye
`
`
`
`
`
`
`
`
`
`
`must dissolve in the polymer and absorb strongly at the exposing
`
`
`
`
`
`
`
`
`
`wavelength.
`In addition, processing is simplified if the dye is
`
`
`
`
`
`
`
`transparent at the alignment wavelength. Transparency at the
`
`
`
`
`
`
`
`alignment wavelength allows detection of the alignment mark
`
`
`
`
`
`
`
`
`
`through the thick bottom polymer. Many of the laser dyes
`
`
`
`
`
`
`
`
`meet these requirements. Fig. 7 shows the transmission spectra
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Page 3 of 6
`
`Page 3 of 6
`
`

`

`1408
`
`
`
`
`
`
`
`
`
`
`
`
`
`IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. ED-28, NO. 11, NOVEMBER 1981
`
`
`
`1.
`
`
`
`
`
`
`
`
`
`
` . 5
`
`. 5
`
`.__.__ 5.1.
`.__L._L L_4
`.
`- L.-
`2 5
`2. Z
`l. S
`1. B
`
`
`
`
`BOTTOM POLYMER TchKNESS (pm)
`
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 10. Simulated linewidth versus bottom polymer thickness for 1-
`
`
`
`
`
`
`
`
`
`
`
`
`
`am lines and spaces for a bottom polymer of (a) HPR 204 and (b)
`
`
`
`
`
`HPR 204 with 1.5-percent dye.
`
`
`
`
`
`
`
` 7
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`after reactive ion etching.
`
`
`
`
`. u polysilicon steps
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`covered with at least 1 pm of Hunt resist with dye in order to
`
`
`
`
`
`
`
`
`suppress reflections and scattered light from the topography.
`
`
`
`
`
`
`
`
`
`In the experiments that follow, a resolution test mask was
`
`
`
`
`
`
`
`
`
`
`used to print lines and spaces over aluminized substrates with
`
`
`
`
`
`
`
`
`
`
`0.5-um steps. A number of bottom polymers were tried. Posi-
`
`
`
`
`
`
`
`
`
`
`tive resist was used because of its superior planarization prop-
`
`
`
`
`
`
`
`
`
`
`erties [13]. The intermediate layer was 130 nm of silicon
`
`
`
`
`
`
`
`
`nitride deposited by plasma-enhanced CVD at room tempera-
`
`
`
`
`
`
`
`
`
`
`
`ture. The top layer of resist was approximately 0.5 [ml of
`
`
`
`
`
`
`
`
`
`
`
`Hunt MPR. A GCA DSW4800 stepper was used to expose the
`
`
`
`
`
`
`
`
`
`
`
`top layer of positive resist. The wafers were then developed in
`
`
`
`
`
`
`
`
`
`
`
`a spray developer with a 2:1 solution of AZ MF312. Pattern
`
`
`
`
`
`
`
`
`
`
`
`transfer from the top resist to the silicon nitride was achieved
`
`
`
`
`
`
`
`
`
`
`
`by plasma etching with CF4 at 4 mtorr. An oxygen reactive
`
`
`
`
`
`
`
`
`
`
`
`ion etch process was used to transfer the pattern to the bottom
`
`
`
`
`
`
`
`
`
`polymer. A 0.1-W/cm2 RF power density at 4-mtorr pressure
`
`
`
`
`
`
`
`
`
`
`resulted in a 70-nm/min etch rate. Undercut was minimal as
`
`
`
`
`
`
`
`shown in the micrograph of Fig. 9.
`
`
`
`
`IV. RESULTS
`
`
`
`
`
`A. Simulation
`
`
`
`
`
`
`
`
`Computer simulations were used for analysis and process
`
`
`
`
`
`
`
`
`optimization.
`In addition to the optical parameters of the
`
`
`
`
`
`
`
`
`aligner and the exposure—development parameters of the resist,
`
`
`
`
`
`
`
`
`
`
`the simulation of the tri-layer system considers the indices of
`
`
`
`
`
`
`
`
`
`
`the materials and thicknesses of the various layers. Steps on
`
`
`
`
`
`
`
`
`
`
`
`the substrate are simulated as a variation in the bottom poly-
`
`
`
`
`
`
`
`
`
`
`mer thickness. The resist linewidth for Him lines and spaces
`
`
`
`
`
`
`
`
`
`traversing steps and the nominal close required were investi-
`
`
`
`
`
`
`
`
`gated. The simulations demonstrate the bulk and standing-
`
`
`
`
`
`
`
`
`
`wave effects associated with the thickness variation of the
`
`
`
`
`
`
`
`
`layers and with the absorption of the bottom polymer.
`
`
`
`
`
`
`
`
`
`The tri—layer simulation results for a l-[rm line and space
`
`
`
`
`
`
`
`
`pattern traversing an aluminum step patterned on an aluminum
`
`
`
`
`
`
`
`
`
`
`substrate are shown in Fig. 10. A l30-nm nitride intermediate
`
`
`
`
`
`
`
`
`
`
`
`layer and a 0.5-um top resist layer are assumed for simulation
`
`
`
`
`
`
`
`
`
`purposes. The bottom polymer is Hunt 204 positive resist,
`
`
`
`
`
`
`
`
`
`
`
`baked at 160°C for 30 min. Fig. 10(a) shows the simulated
`
`
`
`
`
`
`
`
`
`
`
`linewidth for a l-pm line and space pattern versus the thick-
`
`
`
`
`
`
`
`
`
`
`ness of the bottom polymer for a nominal exposure of the top
`
`
`
`
`
`
`
`
`
`
`
`resist of 21 mJ/cm2 at 436 nm, the average exposure of Fig.
`
`
`
`
`
`
`
`
`
`
`
`8(a). The optical parameters are those of the caption of Fig.4.
`
`
`
`
`
`
`
`
`Page 4 0f 6
`
`. a
`. 5
`. 4
`.
`
`
`
`TOP RESIST THICKNESS (pm)
`
`
`
`
`
`
`
`
`
`
`
`
`Fig. 11. Simulated linewidth versus top resist‘thickness for l-um lines
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`and spaces for (a) 0.95-1am bottom layer of HPR 204 and dose of 16
`
`
`
`
`
`
`
`
`
`
`
`
`
`mJ/cm2 for the top layer; (b) 1.03-um bottom layer of HPR 204 and
`
`
`
`
`
`
`
`
`
`
`
`
`
`dose of 25 mJ/cm2 for the top layer; (c) 1.0-um bottom layer of HPR
`
`
`
`
`
`
`
`
`
`
`
`
`204 + dye and 21-mJ/cm2 dose for the top layer of resist.
`
`
`
`
`
`
`
`
`
`
`
`
`The periodic linewidth variation results from multiple reflec-
`
`
`
`
`
`
`
`
`tions from the substrate topology, or the standing-wave effect.
`
`
`
`
`
`
`
`
`
`Since the bottom resist is somewhat absorbing, a thicker bot-
`
`
`
`
`
`
`
`
`
`tom resist partially absorbs the reflections and reduces the
`
`
`
`
`
`
`
`
`standing-wave effect. Bulk effects are not directly observed
`
`
`
`
`
`
`
`
`
`
`with thickness variation of the bottom polymer, since the top
`
`
`
`
`
`
`
`
`
`
`layer of resist was assumed uniformly thick by the simulation.
`
`
`
`
`
`
`
`
`
`If the bottom polymer does not sufficiently planarize the
`
`
`
`
`
`
`
`
`
`
`surface, the top layer of resist will dispense nonuniformly. Fig.
`
`
`
`
`
`
`
`
`
`
`
`
`11(a), (b) shows the simulated linewidth as a function of top
`
`
`
`
`
`
`
`
`
`resist
`thickness for the bottom polymer thicknesses of 1.03
`
`
`
`
`
`
`
`
`
`
`
`and 0.95 pm, representing two extremes of Fig. 8(a). In both
`
`
`
`
`
`
`
`
`
`
`
`cases, the exposure dose has been adjusted to produce a l-,um
`
`
`
`
`
`
`
`
`
`
`
`line for a 0.5-um-thick top resist. Both the bulk and standing-
`
`
`
`
`
`
`
`
`
`
`
`wave effects are clearly seen.
`If the top resist were uniformly
`
`
`
`
`
`
`
`
`
`
`0.5 pm thick, the nominal exposure required to maintain l-um
`
`
`
`
`
`
`
`
`
`
`
`features would change from 16 mJ/cm2 for a bottom layer of
`
`
`
`
`
`
`
`
`
`
`
`
`0.95 [am to 25 mJ/cm2 for a 1.03-um bottom layer. The ex-
`
`
`
`
`
`
`
`
`
`
`posure variation is too great to achieve linewidth control over
`
`
`
`the entire wafer.
`
`
`
`
`
`
`
`
`
`
`The addition of 1.5-percent dye to the bottOm polymer of
`
`
`
`
`
`
`
`
`
`the tri-layer system suppresses the standing-wave effect of Fig.
`
`
`
`
`
`
`
`
`
`
`
`
`10(a). Fig. 10(b) shows that a l-um film of Hunt resist with
`
`
`
`
`
`
`
`
`dye essentially eliminates the nominal exposure variation due
`
`
`
`
`
`
`
`to reflections from the aluminum topography.
`Fig. 11(0)
`
`
`
`
`
`
`
`
`shows the linewidth variation versus the top resist thickness
`
`
`
`
`
`
`
`
`
`
`for a l-pm-thick bottom resist with dye. The linewidth vari-
`
`
`
`
`
`
`
`
`
`
`ation is entirely due to the bulk effect. Since positive resist
`
`
`
`
`
`
`
`
`
`
`as a bottom polymer has been shown to planarize well, the
`
`
`
`
`
`
`
`
`
`
`top resist thickness can be held to close tolerance, and good
`
`
`
`linewidth control is expected.
`
`
`B. Experiment
`
`
`
`
`
`
`
`
`
`
`
`A resolution test mask was used to print l-,um lines and spaces
`
`
`
`
`
`
`
`
`over 0.5-um topography using the tri—layer system and a GCA
`
`
`
`
`
`
`
`
`
`
`
`Page 4 of 6
`
`

`

`O’TOOLE et al.: LINEWIDTH CONTROL IN PROJECTION LITHOGRAPHY
`
`
`
`
`
`
`
`
`
`
`1409
`
`
`
`
`
`
`b)
`
`
`
`
`
`
`
`
`
`
`
`
`
`l-ym lines and spaces in tri-layer resist with 2.0-pm HPR 204 +
`Fig. 12.
`
`
`
`
`
`
`
`
`
`
`1.5-percent dye as the bottom polymer over 0.5-um aluminum steps
`
`
`
`
`
`
`
`
`
`
`after reactive ion etching.
`(a) Side view to show planarization.
`(b)
`
`
`
`
`
`
`Top view to show linewidth control.
`
`
`
`
`
`
`
`
`
`
`In order to consider the worst case, the
`DSW4800 stepper.
`
`
`
`
`
`
`
`topography was coated with highly reflecting aluminum before
`
`
`
`
`
`
`
`
`
`laying down the three-layer system. HPR 204 resist with and
`
`
`
`
`
`
`
`
`
`
`without dye was used as the bottom polymer. The dye absorbs
`
`
`
`
`
`
`
`
`
`
`strongly at the exposure wavelength and is transparent at the
`
`
`
`
`
`
`
`alignment wavelength. An exposure—focus matrix was used to
`
`
`
`
`
`
`
`determine optimum exposure conditions for the top resist.
`
`
`
`
`
`
`
`
`
`
`
`The image in the top resist is transferred to the silicon nitride
`
`
`
`
`
`
`
`
`
`
`with a plasma etch. The bottom layer is etched using reactive
`
`
`ion etching.
`
`
`
`
`
`
`
`
`
`Fig. 12(a) and (b) shows SEM micrographs of l-um features
`
`
`
`
`
`
`
`
`
`patterned in the 2.0-um HPR 204 bottom resist containing 1.5-
`
`
`
`
`
`
`
`
`percent dye over 0.5-um aluminum topography. Since the 2.0-
`
`
`
`
`
`
`
`
`}.tm bottom resist planarizes the topography well, the thickness
`
`
`
`
`
`
`
`
`
`
`
`
`of the top resist layer is controlled to about 0.03 ,urn. The bulk
`
`
`
`
`
`
`
`
`effect of Fig. 11(0) is minimized. The standing-wave effect
`
`
`
`
`
`
`
`
`
`
`due to the tOpography is eliminated by the absorbing dye, as
`
`
`
`
`
`
`
`
`
`indicated by Fig. 10(b). Because the wafer surface appears
`
`
`
`
`
`
`
`planar and nonreflective, excellent linewidth control of Him
`
`
`
`
`
`features over steps is achieved.
`
`
`
`
`
`
`
`
`
`
`Fig. 13 shows a SEM micrograph of Him features in a 2.6-
`
`
`
`
`
`
`
`
`
`pm bottom resist
`layer without dye over 0.5-um aluminum
`
`
`
`
`
`
`
`
`steps. Although the bottom resist is partially absorbing with-
`
`
`
`
`
`
`
`
`
`
`out the addition of dye, the standing-wave effect due to reflec-
`
`
`
`
`
`
`
`
`tions from the topography is not completely eliminated, as
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`l-um lines and spaces in tri—layer resist with 2.0mm HPR 204
`Fig. 13.
`
`
`
`
`
`
`
`
`
`
`as the bottom polymer over 0.5-um aluminum steps after reactive ion
`
`etching.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`l-um lines and spaces in tri-layer res1st With 2.0-um HPR 204 +
`Fig. 14.
`
`
`
`
`
`
`
`
`
`
`1.5-percent dye as the bottom polymer over 1.0mm aluminum steps
`
`
`
`after reactive etching.
`
`
`
`
`
`
`
`
`
`
`
`Increasing the thickness or intensely
`indicated by Fig. 10(a).
`
`
`
`
`
`
`
`hardbaking the bottom resist would reduce the standing-wave
`
`effect.
`
`
`
`
`
`
`
`Thicker topography inhibits planarization. Fig. 14 shows a
`
`
`
`
`
`
`
`
`
`SEM micrograph of Him features of 2.0-}.tm bottom layer of
`
`
`
`
`
`
`
`
`
`HPR 204 resist with dye over 1.0-,um aluminum steps. The
`
`
`
`
`
`
`
`linewidth variation is caused by insufficient planarization of
`
`
`
`
`the bottom layer polymer.
`
`
`
`
`
`
`
`
`
`
`
`
`V. CONCLUSIONS
`
`
`
`
`
`
`
`
`
`The tri-layer resist system incorporating an absorbing dye in
`
`
`
`
`
`
`
`
`
`the bottom polymer improves the usable resolution of projec-
`
`
`
`
`
`
`
`
`tion aligners. One micrometer features over topography are
`
`
`
`
`
`
`
`
`
`achievable. Since the effect of the topography are eliminated,
`
`
`
`
`
`
`
`
`
`
`the exposure for each masking layer is essentially constant. The
`
`
`
`
`
`
`
`
`
`
`dye concept offers flexibility for the material selection of the
`
`
`
`
`
`
`
`
`
`bottom polymer and for the exposure system. Simulations of
`
`
`
`
`
`
`
`
`
`the tri-layer system provide an analytical explanation of the
`
`
`
`
`
`
`
`experimental results and aid in process optimization.
`
`ACKNOWLEDGMENT
`
`
`
`
`
`
`
`
`
`
`
`
`The authors wish to thank D. Ilic and P. Marcoux for their
`
`
`
`
`
`
`
`
`
`
`help with the PECVD nitride and G. Rankin for useful discus-
`
`
`
`
`sions on plasma etching.
`
`
`
`
`
`Page 5 of 6
`
`Page 5 of 6
`
`

`

`1410
`
`
`
`
`
`
`
`
`
`
`
`
`
`IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL.» ED-28, NO. 11, NOVEMBER 1981
`
`
`
`[1]
`
`
`
`
`
`REFERENCES
`
`
`
`
`
`
`
`
`
`
`
`
`J. M. Moran and D. M. Maydan, J. Vac. Sci. Technol., vol. 16,
`
`
`
`p. 1620, 1979.
`
`
`
`
`
`
`
`
`
`
`
`
`
`]
`J. H. Bruning, J. Vac. Sci. Technol., VOL 17, p. 1147, 1980.
`
`
`
`
`
`
`
`
`
`
`
`
`
`] K. L. Tai, R. G. Vadimsky, C. T. Kemmerer, J. S. Wagner, V. E.
`
`
`
`
`
`
`
`
`
`
`
`Lamberti, and A. G. Timko, J. Vac. Sci. chhnol., vol. 17, p.
`
`
`1169, 1980.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`[4] K. L. Tai, W. R. Sinclair, R. G. Vadimsky, J. M. Moran, and M.
`
`
`
`
`
`
`
`
`
`
`J. Rand,J. Vac. Sci. Technol., vol. 16, p. 1977, 1979.
`
`
`
`
`
`
`
`
`
`
`
`[5] B. J. Lin, J. Electrochem. Soc, vol. 127, p. 202, 1980.
`
`
`
`
`
`
`
`
`
`[6] B. Carlson and J. Arnold, presented at the Kodak Microelectronics
`
`
`
`
`
`
`Seminar, San Diego, CA, Oct. 1980.
`
`
`
`
`
`
`
`
`
`
`
`[7] W. G. Oldham, S. M. Nandgaonkar, A. R. Neureuther, and M. M.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`O’Toole, IEEE Trans. Electron Devices, vol. ED-26, p. 717,
`1979.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`[8] F. J. Dill, W. P. Hornberger, P. S. Hauge, and J. M. Shaw, IEEE
`
`
`
`
`
`
`
`
`Trans. Electron Devices, vol. ED-22, p. 445, 1975.
`
`
`
`
`
`
`
`[9] M. M. O’Toole, to be published.
`
`
`
`
`
`
`
`
`
`
`
`[10] D. Meyerhofer, IEEE Trans. Electron Devices, vol. ED-27, p. 921,
`1980.
`
`
`
`
`
`
`
`
`
`
`
`
`[11] M. M. O’Toole, “Simulation of optically formed image profiles in
`
`
`
`
`
`
`positive photoresist,” Ph.D. dissertation