Chapter 2 Fabrication of Sub-60-nm Contact Holes in Silicon Dioxide Layers
2.2 Experimental
Table 2.1 lists the experimental flows and parameters used for preparing the sub-60-nm contact holes in the silicon dioxide layer. Electron beam exposure was performed on a Leica Weprint 200 stepper. The silicon dioxide layer was grown by wet oxidation using a mixture of hydrogen (8000 cm3/min) and oxygen (4999 cm3/min) gases at 978 °C in a low-pressure furnace. The electron beam energy was 40 keV, the beam size was 20 nm, and the exposure dose was 14 µC/cm2. The developer for the JSR positive electron beam resist (MES-1EG) was an aqueous 2.38%
tetramethylammonium hydroxide (TMAH) solution. A positive-tone electron beam resist was spin-coated on a silicon wafer (150 mm diameter) and baked at 110 °C for 120 sec. The thickness of the resist film was ca. 650 nm. After exposure and a post-exposure bake (110 °C for 120 sec), the wafer was developed using the TMAH solution. Again, a hard-bake was applied to the wafer (110 °C for 120 sec). A chemical shrinkage procedure was then applied. The chemical shrinking agent (AZ R-200, Clariant) was coated over the resist pattern, and the wafer was again soft-baked. The film thickness was ca. 400 nm. A mixing-bake step at 110 °C for 70 sec was then undertaken. The chemical shrinking agent reacted with the resist acid that diffused from the resist pattern. A cleaning step, using AZ R-2, which can wash away unreacted materials, was applied to the process. A hard-bake step was
performed at 110 °C for 2 min. Finally, we applied a plasma process using a mixture of CHF3 and CF4 gases to etch the underlying silicon dioxide layer. Critical dimensions were evaluated using either an in-line scanning electron microscope (SEM, Hitachi S-6280) or a cross-sectional SEM (Hitachi S-4000).
2.3. Results and Disussion
2.3.1 Chemical shrinkage processes for the electron-beam resist
As has been reported previously [3], the mixing-bake temperature and time are the critical factors for fabricating sub-100-nm contact holes in a resist when using the chemical shrinkage technique. Basically, the residual acid in the resist pattern diffuses into the side-wall regions of the shrinkage agent, which leads to cross-linkage reactions of the shrinkage agent. The extent of hole shrinkage in the positive resist depends upon the intrinsic acid diffusion behavior, such as its diffusion coefficient, and the time. Figure 2.1 illustrates the effect of mixing-bake temperature on the CDs of various contact holes for the positive E-beam resist. Characterization using top-down SEM clearly indicates that the CDs of contact holes at any size gradually narrow as the mixing-bake temperature is increased from 100 to 120 °C. An initial hole size of 140 nm formed by electron-beam patterning is suitable for shrinking down to the sub-100-nm level. We evaluated the profile of the contact holes at various mixing-bake temperatures by using cross-sectional SEM. The SEM pictures in Fig.
2.2 clearly illustrate the profile of the contact holes. Initially (prior to applying the chemical shrinkage agent), the contact holes have vertical sidewalls and a smooth surface. After a 105°C mixing-bake (subsequent to applying the chemical shrinkage agent), the sidewalls and surface covered with shrinkage agent appear to be slightly
distorted. When the mixing-bake temperature was increased to 110 °C, the contact holes remain open, but we observe an overhang effect in the contact holes formed after heating at either 115 or 120 °C. We attribute this finding to the effect of thermal flow of the shrinkage agent on the resist’s sidewalls at higher bake temperatures. This type of defect is not observed when viewing from by SEM from above the surface, but is very apparent from the cross-sectional SEM image. Clearly, the upper limit of the mixing-bake temperature for the chemical shrinkage technique is 110 °C.
Figure 2.3 demonstrates the effect of mixing-bake times on the CDs of various contact holes for the positive resist at 110 °C. We observe that the hole dimensions decrease rapidly during the first 60 sec, but then gradually reach a plateau. Therefore, we believe that the optimal time for the mixing-bake process in this study is 70 sec.
The reason for this trend with respect to mixing-bake time is that the process is dependent on the abundance of diffusion acid in the resist pattern, which is limited. If the amount of acid in the resist is increased, the CD will become narrower. The initial 140-nm hole formed by electron-beam patterning is the only one of the holes that we have studied that is suitable for being shrunk down to a sub-100-nm hole. Next, we compare the effects of the soft-bake, mixing-bake, and hard-bake processes after applying the chemical shrinkage agent. The control conditions for soft-baking, mixing-baking, and hard-baking are 85 °C for 70 sec, 110 °C for 70 sec, and 110 °C for 2 min, respectively. Figure 2.4 clearly indicates that the hard-bake temperature has no significant effect on the hole shrinkage ratio. The soft-bake temperature does have a significant effect above 110 °C, which we ascribe to the thermal flow effect discussed earlier. Among these baking processes, the mixing-bake temperature exhibits the most significant effect on the hole shrinkage ratio.
The hole shrinkage mechanism is closely dependent on the abundance of residual acid in the resist pattern. We exposed the wafer in the cleanroom environment (class
10) after initial contact hole definition for various delay hours. Figure 2.5 indicates that the shrinkage ratio fluctuates between 33 and 34% after various delay times.
Although not illustrated here, the SEM images for these holes taken from above are very similar. It has been reported in the literature that a chemically amplified resist is sensitive to the molecular base, and leads to T-top and footing problems for the positive resist when not immediately developed (post-exposure delay)[4]. The molecular base in the cleanroom might have reacted to some extent with the surface acid. Most acids under the resist film, however, are not influenced by the molecular base from the air in the cleanroom. Therefore, the diffusion of the acids out of the resist still occurs during the mixing-bake process and the delay time has no effect.
2.3.2 Fabrication of 53-nm contact holes
The fabrication of sub-60-nm contact holes in a silicon dioxide layer by the chemical shrinkage technique has not been reported previously. To ensure the applicability to nano-fabrication techniques, the resist should tolerate the etching process. Figure 2.6 depicts the etch selectivity that we estimate from the ratio of the plasma etch rates of silicon dioxide and the positive resist under various mixtures of gases. The selectivity gradually increases upon increasing the ratio of CHF3 from 0 to 0.75, and then increases abruptly upon a further increase in the gas ratio to 1. It has been suggested in the literature [5] that oxygen byproducts formed during silicon dioxide etching can react with carbon residues, especially at fluorine/carbon ratio < 2.
Hence, the polymer formation blocks any further etching process.
The positive resist (650 nm thick) was coated onto a wafer upon which a silicon dioxide film had been grown. A variably shaped electron beam was used to pattern a 140-nm hole in the positive resist (Fig. 2.7a). The chemical shrinkage process was
then undertaken by spin-coating the shrinkage agent onto the resist pattern, followed by a soft-bake at 85 °C for 70 sec. The wafer was then subjected to a mixing-bake at 110°C for 70 sec. The residual acid diffuses out from the resist pattern into the shrinkage agent, which leads to acid-induced cross-linkage reactions taking place.
After washing off the unreacted shrinkage agent and baking again at 110 °C for 2 min, the contact hole in the resist layer now has a 93-nm diameter (Fig. 2.7b). At the stage, the shrinkage ratio is ca. 33.6%. After the resist pattern had shrunk, the wafer was sent for plasma etching to fabricate a contact hole. Interestingly, the dimension of the contact hole in the silicon dioxide is not 93 nm: Figure 2.8 indicates that the hole size is 53 nm. The total shrinkage ratio of the hole diameter after the chemical shrinkage and plasma etch processes is 62.1%. What happens to the contact holes in the silicon dioxide layer during the plasma etch process? The sidewall deposition of residual polymers during the plasma etch process plays a significant role in narrowing the contact hole dimensions.
The etch mechanism for contact hole fabrication in a silicon dioxide layer is very complicated, with the dimensions of the pattern formed during the etch process being controlled by a balance between the amount of polymer deposited and the etch conditions. Explanations have been proposed in the literature [6-9] regarding micro-loading and aspect ratio-dependent etching (ARDE) to explain the observations made during etch processes. Micro-loading describes the variations of the etch rate between areas having different pattern densities, with features in low-pattern-density areas etching faster than features in high-pattern-density areas. Effects that are due to the pattern dimensions, which includes effects related to transport of etchant species into the pattern, or transport of etch products out of the pattern, are generally referred to as ARDE. Table II lists a series of data for the contact holes obtained after the chemical shrinkage and dry etch processes. The smaller holes exhibit a higher
shrinkage percentage for the etch process than the larger holes. We attribute this observation to the pattern dimension effect. As the pattern size is reduced, the probability for the flux of incoming polymer species to interact with the sidewall of the contact hole increases. Figure 2.9 depicts the relationship between the ratio of the hole dimension before and after etching and the ratio of the hole perimeter to the hole area of these nano-scale contact holes. We find that a linear dependence exists for the series of holes studied. This finding suggests the pattern reduction arising from sidewall polymer deposition during the etch process has an inverse relationship to the pattern diameter. The probability of transportation of a polymer species onto the sidewall of a contact hole is related linearly to the inverse of the contact hole size (after chemical shrinking). As a consequence, the contact hole in the silicon dioxide layer becomes smaller than expected after the etching process. We estimate that the uniformities (1 sigma) of the proposed method for contact hole formation after the processes of lithography, chemical shrinkage, and plasma etching are 3.21, 3.16, and 2.76 nm, respectively.
2.4 Summary
We have established a successful fabrication technique for preparing sub-60-nm contact holes in a silicon dioxide layer by electron-beam lithography. We have discussed in detail the many factors that influence the performance of the shrinkage process, such as the mixing-bake temperature, mixing-bake time, and hole dimensions before and after chemical shrinkage. Using this chemical shrinkage technique (mixing-bake of 110 °C for 70 sec) and an etch gas of CHF3/CF4 (1:1), we obtained a minimum hole dimension of 53 nm. This technology meets the requirements2 for contact hole fabrication in the year 2009. We propose that a nano-hole effect occurs
during the etch-assisted shrinkage reaction because smaller holes have a higher percentage of polymer deposition in the resist sidewall than do larger holes.
Table 2.1 Process conditions for the fabrication of sub-60-nm contact holes in a silicon dioxide layer.
Oxidation Processes
Oxidation temperature 978°C, SiO2 thickness 150 nm
soft-bake 110°C, 120 sec
Post-exposure bake 110°C, 120 sec Development TMAH, 60 sec
Hard-bake 110°C, 120 sec
Chemical Shrink Processes
Soft-bake 85°C, 70 sec
Mixing-bake 110°C, 70 sec
AZ remover (R-2) 10% IPA and 90% H2O, two puddles 40s/20s Hard-bake 110°C, 120 sec
Etch and Resist Stripping Processes
Etch time 60 sec
Gas components CF4, 20 sccm/CHF3, 20 sccm Chemicals H2SO4/H2O2 = 3:1, 120°C, 10 min
Table 2.2 The diameters of contact holes formed after the processes of chemical shrinkage and plasma etching in SiO2, and their ratios (R = hole radius).
Hole diameter after chemical shrinkage
(x)
Hole diameter after dry etching in SiO2
(y)
y/x (percentage) Perimeter/area (2πR/πR2)
246 nm 233 nm 94.7% 0.0163
149 nm 128 nm 85.9% 0.0268
133 nm 98 nm 73.6% 0.0301
119 nm 79 nm 66.6% 0.0336
93 nm 53 nm 53.3% 0.0430
Temperature ( o C)
Figure 2.1 Dependence of various mixing-bake temperatures for 70 sec on the critical dimensions (CDs) of contact holes formed by the chemical shrinkage technique. The initial hole sizes were 140, 180, 240, and 360 nm, respectively; top-down SEM was used to measure CD.
120oC 115oC
110oC 105oC
initial
Figure 2.2 Cross-sectional SEM image of contact holes in the resist after various mixing-bake temperatures.
Time (s)
20 40 60 80 100 120 140
C.D. (nm)
80 120 160 200 240 280
320
140 nm180 nm 240 nm 360 nm
Figure 2.3 Dependence of various mixing-bake times on the CDs of contact holes formed by the chemical shrinkage technique. The initial hole sizes were 140, 180, 240, and 360 nm, respectively; top-down SEM was used to measure CD.
15 20 25 30 35 40 45 50
60 80 100 120 140
Temperature ( o C)
S h ri n k ra ti o (% )
soft bake mixing bake hard bake
Figure 2.4 The effects of the various bake processes and temperatures on the shrinkage ratios of the contact holes.
igure 2.5 The effect of delay time on the shrinkage ratio (delay time = the period of
Delay time (hour)
0 5 10 15 20 25 30
Shrink ratio(%)
32.0 32.5 33.0 33.5 34.0 34.5 35.0
F
time between the formation of the initial hole in the resist and the application of the shrinkage agent).
CHF3/(CHF3+CF4)
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Etching Selectivity
1 2 3 4 5 6 7 8
Figure 2.6 The dry etch selectivity of SiO2 to resist at different CHF3 ratios.
Figure 2.7 (a) The 140-nm contact hole in the resist.
Figure 2.7 (b) The 93-nm contact hole formed in the resist after chemical shrinkage.
(a)
Figure 2.8 (a) 53nm contact hole in SiO2 layer without resist stripping (image from top-down SEM).
Figure 2.8 (b) 53nm contact hole in SiO2 layer after resist stripping (image from p-down SEM).
(b)
to
53 nm
(c)
Figure 2.8 (c) 53nm contact hole in SiO2 layer after resist stripping (image from cross-sectional SEM).
tching and the ratio of the hole perimeter to hole area during the etching of the ano-scale contact holes.
y = -16.215x + 1.234
40%
60%
80%
100%
0.00 0.01 0.02 0.03 0.04 0.05
(hole perimeter) / (hole area)
(hole diameter after etch) / (hole diameter before etch)
Figure 2.9 Linear dependence between the ratio of the hole dimension after and before e
n
Chapter 3
Resist Nano-modification Technology for Enhancing the Lithography and Etching
Performance for Nano Contact Hole and Line
3.1 Introduction
During the past two decades, there has been an extremely rapid growth in both the technology and the application of microelectronics, to the point that it now pervades virtually all aspects of commercial and military business. The size and performance of microelectronic devices has been improved substantially, especially in the past few years [1,2]. In the updated International Technology Roadmap for Semiconductors, the 50 nm contact hole in the resist will be used in year 2011. The electron beam direct writing (EBDW), in comparison with optical lithography, is a promising means for controlling and patterning small features, down to sub-100nm [3]. This technology has a cost advantage for production volumes below 100 lots in the future [3]. In EBDW, the Gaussian beam has better resolution than shaped beam.
But, the shaped beam has an at least 10-fold higher throughput than Gaussian beam due to imposing several pixels per shot [4]. In order to achieve the better resolution and high throughput for shaped beam technology, the utilization of thin resist film is inevitable [5]. However, the thin resist will face the challenge of poor etching resistance and serious line edge roughness.
Nano-scale molecules are the possible means to solve the unaffordable etching
resistance and enhance the lithographic performance for the thin resist film generation.
The molecules can be incorporated into the resist to alter its performance. The use of fullerene molecules possesses the advantage of extremely small and monodisperse.
Ishii et al. [6] have been used fullerene molecule (i.e., 5 wt% C60) to modify positive tone resist, and found the fullerene can enhance the 6% etching resistance and not alter the sensitivity. In addition, they claim the negative tone chemically amplified resist incorporated with 3 wt% C60 exhibits strong environmental stabilization in postexposure delay. In the latter report [7], they find the resist sensitivity is degraded by the C60 due to the dissolution-inhibiting effect. Dentinger and Taylor [8] spike 7.9 wt% C60 into poly(methylmethacrylate) resist, and the etching resistance is promoted 8% and 26% for CF4 and Cl2 plasmas, respectively. However, the use of 3-7.9 wt%
C60 in the resists and the deterioration of resist sensitivity elevate the fabrication cost and restrict the further application of this technology.
In this study, the sensitivity curve of resists after spiking with C60 and C70 molecules are investigated. The film stress, etch resistance and the effect of shaped electron beam dose on the contact hole sizes are carefully studied. In addition, the titanium nitride gap-filling and step coverage on contact holes by 0.02% w/v C70-incorporated resists are also evaluated.
3.2 Experimental
The fullerene molecules of C60 and C70 were purchased from Alfa Aesar Company. The toluene solvent was obtained from E. Merck (Darmstadt, Germany).
The negative NEB-22 resist used in this study was obtained from SUMITOMO Chemical Co., Ltd. (Japan). The resist samples in this study have four types, named NEB, NEB+Toluene, NEB+C60, and NEB+C70, respectively. The NEB means the
NEB-22 resist without any modification. The “NEB+Toluene” means the mixture of 50mL NEB-22 resist and 50mL toluene solvent. For the “NEB+C60-0.01%” sample, the 0.01g C60 fullerene is first dissolved in 50mL toluene, and then mixes with 50mL NEB-22 resist. The final concentration of C60 molecule in the resist is 0.01% w/v. In the same manner, the “NEB+C70-0.02%” uses 0.02g C70 fullerene to prepare the sample.
And the DSE-1010 positive resist used in this study was obtained from DONGJIN Chemical Co., Ltd. (Korea). There are four types of resist samples in this study, named DSE, DSE + Toluene, DSE + C60, and DSE + C70, respectively. The DSE means the DSE-1010 resist without any modification. The “DSE + Toluene”
means the mixture of 50mL DSE-1010 resist and 50mL toluene solvent. For the “DSE + C60-0.01%” sample, the 0.01g C60 fullerene is first dissolved in 50mL toluene, and then mixes with 50mL DSE-1010 resist. The final concentration of C60 molecule in the resist is 0.01% w/v. In the same manner, the “DSE + C70-0.02%” uses 0.02g C70 fullerene to prepare the sample.
Electron beam exposure was performed on a Leica Weprint 200 stepper. The electron beam energy was 40 keV, and the beam size was 20 nm. The developer for the electron beam resist was an aqueous 2.38% tetramethylammonium hydroxide (TMAH) solution. A electron beam resist was spin-coated on a silicon wafer (150 mm diameter) and baked at 95 °C for 120 sec. After exposure and a post-exposure bake (115 °C for 120 sec), the wafer was developed using the TMAH solution for 60 sec.
Again, a hard-bake was applied to the wafer (115 °C for 120 sec). Critical dimensions were evaluated using either an in-line scanning electron microscope (SEM, Hitachi S-6280) or a cross-sectional SEM (Hitachi S-4000). The stress of resist film was measured by TENCOR FLX-2320 instrument. In the stress measurement, the curvatures of bare silicon wafers, resist-coated wafers were determined.
Silicon dioxide film was etched using a reactive-ion etcher (RIE, Tokyo Electron Limited, Model TE5000, Japan). There are two steps for silicon dioxide etching. The operating conditions for step 1 are- 0.2 Torr pressure, 0 W RF power, 400 cm3min-1 Ar gas, and etching gases of CHF3 and CF4 (CHF3+CF4=40 cm3min-1). The operating conditions for step 2 are- 0.2 Torr pressure, 500 W RF power, 400 cm3min-1 Ar gas, and etching gases of CHF3 and CF4 (CHF3+CF4=40 cm3min-1).
The thermal oxide was grown under dry O2 at 900oC in quartz reactor to a thickness of 100nm. After coating the fullerene-incorporated resist onto the thermal oxide and resist patterning, the plasma process of mixing CHF3 and CF4 gases was used to etch the underlying silicon dioxide layer. Then, the contact holes defined by 0.02% w/v C70-incorporated resist were deposited with the titanium nitride (TiN) plug by physical vapor deposition (PVD) and chemical vapor deposition (CVD). The TiN PVD sputter system (ULVAC SBH-3308 RDE system) was used to deposit 200-nm TiN film, and argon and nitrogen were used as process gases. For the film deposition by CVD method, the tool from Materials Research Corporation (MRC) was used to deposit TiN film by gas mixture of TiCl4 and NH3 at 630 oC. The chemical reaction is as follows: 6TiCl4 + 8NH3 → 6TiN + 24HCl + N2.
The other the polysilicon gates with spacer were fabricated with the 0.01% w/v
The other the polysilicon gates with spacer were fabricated with the 0.01% w/v