Absorbance-modulated lithography is a relatively new optical patterning method where a thin layer of photochromic molecules is placed between the far-field optics and photoresist. These molecules can be made transparent or opaque by illuminating with wavelengths λ1 or λ2, respectively. By simultaneously illuminating this layer with patterns of both wavelengths it is possible to create an absorption mask capable of subwavelength resolution. This resolution comes at the price of limited contrast and depth-of-focus resulting in poor process latitude. Here it is shown that by using TM polarization for λ1 and integrating a plasmonic reflector process latitude is increased by up to 66%.
© 2011 OSA
The resolution of far-field optical lithography is typically limited by diffraction, requiring the use of shorter wavelengths and higher numerical apertures to improve performance . This has resulted in the exponential increase in complexity and cost of state-of-the-art optical lithography systems . In contrast, near-field optical techniques such as evanescent-field contact lithography, can achieve sub-diffraction limited resolution by utilizing the evanescent high spatial frequencies to form the image. Although impressive results have been achieved , fundamental problems such as expensive mask fabrication, the need for intimate contact between a rigid photomask and photoresist, and mask wear have prevented this method from being widely adopted.
To overcome these issues while maintaining sub-diffraction limited resolution a relatively new lithography technique, absorbance-modulated optical lithography, has been developed that uses far-field optics to control near-field waves . This technique uses a spin-on absorbance-modulation layer (AML), located directly above the photoresist, as an optically-activated photomask [5,6]. The AML is made up of photochromic molecules that can be optically switched from opaque to transparent via exposure to short-wavelength ultraviolet light (λ1) and switched back, transparent to opaque, via exposure to longer-wavelength visible light (λ2). It is possible to optically control the localized level of transparency of the AML by locally controlling the intensity ratio, λ1:λ2. One simple way to do this is absorbance-modulated interference lithography (AMIL) where the AML is simultaneously exposed to a standing-wave interference pattern of λ2 and uniform illumination of λ1 (Fig. 1 ) .
At the optical nulls of the standing wave the λ1:λ2 intensity ratio is high so the AML is transparent; elsewhere the ratio is low and the AML remains opaque. This effectively creates an optically-activated absorption-grating photomask on the surface of the photoresist with the period of the standing wave. The photoresist below the AML is exposed, via λ1, through the activated transparent regions. Previous work has shown that the width of the transparent region is dependent on the λ1:λ2 intensity ratio and quality of the optical nulls [4,5]. At large intensity ratios it is possible to reduce the width of the transparent regions to deep subwavelength dimensions enabling near-field-lithography resolution. Since the AML is spun on to the recoding stack intimate contact is certain, eliminating the mask related challenges that are problematic in near-field contact lithography. However, AMIL is still plagued by many other problems associated with near-field lithography, such as limited contrast and depth of focus, resulting in poor process latitude. We address these issues here.
Using finite element method (FEM) modeling we have found that it is possible to improve the contrast and depth of focus in AMIL by using the optimal polarizations for λ1 and λ2 and integrating a plasmonic reflector beneath the photoresist. For λ2, transverse-electric (TE) polarization is required to form deep optical nulls and achieve maximum fringe contrast. However, for λ1, FEM simulations have shown that these absorption gratings perform similarly to comparable-sized metal and dielectric gratings; increased transmission and contrast is predicted for transverse-magnetic (TM) polarization (H field is parallel with the grating) , compared with the TE illumination of λ1 that has been used in previous absorbance-modulation lithography experiments . We test this prediction here.
Further improvements are seen in the FEM simulations by integrating a plasmonic reflector beneath the photoresist to improve contrast and depth of focus. This is counterintuitive because typically in optical lithography bottom-layer antireflection coatings (ARC) are used to minimize substrate reflections and suppress the unwanted formation of a standing wave in the photoresist. However, plasmonic reflectors have the ability to regenerate evanescent waves in the near field with the forward and reflected fields interfering to create a symmetric intensity profile through the depth of the photoresist . FEM simulations show that this effect increases the contrast and depth of focus of the evanescent fields of λ1 in the photoresist with minimal effects on resolution.
To confirm the FEM results, various AMIL exposure are performed using TE or TM polarization for λ1 and with or without an integrated plasmonic reflector. The process latitude, the percent increase in dose required to increase the linewidth by 50%, is measured for each exposure condition. An increase in process latitude signals an improvement in contrast and depth of focus for the exposing wavelength. Similar to the FEM results the experiments showed that the process latitude increases when switching from TE to TM polarized light for λ1. A further increase is seen for the samples exposed with TM polarized light when a plasmonic reflector is placed below the photoresist.
For these experiments the AML layer is comprised of a polymerized azobenzene-based molecule chosen for its optical and thermal characteristics. However, this class of photochromic molecules exhibits the formation of photoinduced surface-relief gratings (SRG) in the presence of a standing wave interference pattern. In the last section of this paper, 4.2, we will analyze the formation of SRG in the AML and discuss its consequences on the contrast of AMIL.
2. Finite element method modeling
For modeling AMIL a two-dimensional (2D) full FEM modeling system using Comsol Multiphysics software suite , has been developed. The accuracy and validity of this model had been tested previously , by comparing the results for idealized 2D absorption gratings to an analytical calculation method developed by Rytov . For AMIL, full FEM modeling is needed to model the fields beneath the AML (i.e. in the photoresist and at the plasmonic reflector) since analytical methods such as Rytov analysis, assume that the gratings are infinite.
In the model the inputs are defined as a TE-polarized standing wave (800nm period) at λ2 with an imperfect null of 0.25% of the peak intensity and uniform illumination of λ1 with either TE or TM polarization. The material stack defined in the model was given parameters to mimic the recording stack used in the experiments: a 200-nm thick AML, a 10-nm thick polyvinyl alcohol (PVA) layer, a 40-nm thick photoresist layer above either an anti-reflective coating (ARC) or plasmonic reflector (see Fig. 6 ). The results for λ1 intensity in the photoresist for the TE, TM and TM with a plasmonic reflector case are shown in Fig. 2 . Also plotted in Fig. 2 are the line-spread function for each case at depths of 0 nm, 20 nm, and 40 nm into the resist layer. From these plots it is possible to see that contrast and depth of focus is increased by using TM polarization and adding a plasmonic reflector
From the line-spread function it is possible to qualitatively determine the relative process latitude for the three cases. This is done by looking at the slope of the line-spread function at the bottom of the photoresist, where the higher the slope results in higher process latitude. Using this metric it is clear the TM polarization should result in higher process latitude than TE polarization. The highest process latitude is achieved by using TM polarization with a planar plasmonic reflector below the photoresist. Also noticed in these results is that when using a plasmonic reflector the intensity is higher due to the reflected power traveling back through the photoresist. This should result in a lower exposure dose when using a plasmonic reflector.
3. Absorbance-modulated interference lithography
A schematic of the Lloyd’s mirror based AMIL setup used is shown in Fig. 3 . The source used for λ1 was a mercury lamp fiber light. The broadband light from the lamp travels through uniform illumination optics, a 405 nm filter, and a linear polarizer before reaching the sample. At the surface of the sample the 405 nm light is either TE or TM polarized with an intensity of 1.5 mW/cm2 uniform over a 25 mm × 25 mm area.
For λ2 a 50 mW, 532 nm wavelength solid-state diode pumped laser with a coherence length >1 m was used. This laser travels through a spatial filter with a non-collimated output, resulting in a Gaussian shaped intensity profile, centered at the intersection of the mirror and the sample, with peak intensity of 9 mW/cm2. A silver mirror is used for the Lloyd’s mirror, instead of the conventional aluminum mirror, due to its higher reflectivity at 532 nm. Also the sample chuck has an integrated Peltier cooler for controlling the sample’s temperature. For the process latitude experiments the Lloyds mirror is set to θ=19.4° for a period of 800 nm. Using the combination of uniform illumination for λ1 and a Gaussian shaped intensity profile for λ2 allows for the testing of many different intensity ratios on a single sample for a given λ1 exposure dose (Fig. 4 ).
3.1 Absorbance- modulation layer
The AML used in these experiments was comprised of polymerized 4’-[[(2-methacryloyoxy)ethyl]ethylamino]azobenzene (pMAEA). The synthesis of this photochromic material is described in detail in [11,12]. pMAEA is a polymerized azobenzene based molecule that has two distinct absorption spectrums for its trans and cis isomers. The thermally stable trans isomer undergoes a photoisomerization reaction when exposed to λ1 = 405 nm, forming the cis isomer. The reverse reaction can take place quickly by exposing the cis isomer to λ2 = 532 nm or slowly at room temperature via thermal decay. The absorbance spectrum of a 200 nm thick film of pMAEA is shown in Fig. 5 after exposure to 405 nm light, to maximize the cis concentration, and 532 nm light, to maximize the trans concentration. The important parameters for this material are the absorption contrast at the exposing wavelength and the thermal decay constant, measured to be 2.3 and 2 × 10−3 at 15°C, respectively.
3.2 Experimental procedure
The recording stack for the polarization experiments consisted of four spin-coated layers; 200 nm of pMAEA, 10 nm PVA, 40 nm of AZ 1518 photoresist, and 230 nm of ARC on a silicon substrate (Fig. 6a). For the plasmonic reflector experiment the ARC was replaced with 60 nm of Ag deposited via sputtering (Fig. 6b). The PVA interlayer was necessary to prevent the solvent, chlorobenzene, used for spin-coating the pMAEA from attacking the underlying photoresist.
A Peltier cooler was used to cool the sample to 15°C during exposure to minimize the cis-trans thermal decay rate. First the samples were exposed to only λ2 for 15 min to assure maximum trans concentration. Next λ1 was turned on and the samples were exposed for various times to achieve λ1 doses ranging from 4.5 to 13.0 J/cm2. After exposure the AML and PVA were stripped using a 60 s dip in trichloroethylene and deionized (DI) water, respectively. The photoresist was then developed for 20 s using MIF 326 developer diluted 3:2 with DI water and then thoroughly rinsed with DI water. The developed samples were then characterized using atomic-force microscopy (AFM).
4. Results and discussion
AFM images of the gratings are taken at various distances from the sample’s edge, corresponding to different λ2:λ1 intensity ratios. Figure 7 shows AFM images of samples exposed with TE, TM and TM polarization with a plasmonic reflector for various λ1 exposure doses at an incident intensity ratio of 4. The full-width half-maximum (FWHM) of the exposed regions (trenches) were measured and averaged over 36 μm of grating length.
4.1 Process latitude
By plotting the FWHM versus exposure dose (Fig. 8 ) it is possible to determine the minimum linewidth and process latitude for the various exposure conditions, as is summarized in Table 1 . The minimum linewidth is approximately the same for samples exposed with TE and TM polarization without a plasmonic reflector. However, for TM polarization the process latitude is improved by 32% and 22%, for intensity ratios of 4 and 5, respectively. It is also noticed that the process latitude decreases as the power ratio increases. This is consistent with previously reported FEM simulation results where both the contrast and depth of focus are reduced as the intensity ratio increases past a relatively low threshold value, with TM polarization out performing TE .
By replacing the ARC with a plasmonic reflector it is seen that the process latitude when using TM polarization is increased even further, 34% and 82% for power ratios of 4 and 5 respectively. This increase is due to the refocusing upon reflection of high spatial frequencies resulting in an increased depth of focus and the creation of a symmetric intensity profile of the evanescent waves throughout the depth of the photoresist. As expected, when we go to higher power ratios where high spatial frequencies begin to dominate the exposure, the benefit of the plasmonic reflector becomes greater. However, this increase process latitude does not come with out a price. As is evident in the experimental data the minimum linewidth achievable at a given power ratio increases when the plasmonic reflector is used. This increase can be explained by a broadening of the transparent aperture in the AML by the reflected λ1 intensity.
4.2 Surface-relief gratings
In the FEM simulations the AML is modeled as a static layer where only its absorption characteristics can be changed. In reality this is not true for pMAEA, which, like most polymerized azobenzene molecules, exhibits the formation of photoinduced SRG when illuminated with a standing-wave interference pattern. This well known phenomenon is not trivial and can have undesirable effects on the performance of AMIL.
The formation of photoinduced SRG in azobenzene films has been the subject of many papers, yet the exact mechanism of their formation is debated and most likely multifaceted . However, two things that are agreed upon are that the material moves from areas of high intensity of light to areas of low intensity and the magnitude of the grating depends of the optical fluence and electric field gradient [13,14]. Figure 9 shows the progression of the formation of SRGs at different intensity ratios during AMIL exposures, which can be as high as 40 nm for large exposure doses and high λ2:λ1 intensity ratios (Fig. 9a). Plotting the height data normalized by the λ2:λ1 intensity ratios (Fig. 9b) allows trends to be observed: at first, when the exposure dose is small, the formation of the grating is linear and is dominated by λ2; at larger doses the SRG goes through a transition region before following a new linear formation curve, with the flood exposure at λ1 acting to smooth out the SRG. With further intensity increase the magnitude of the SRG saturates at a height that depends on the intensity of λ2.
Since the material flows from areas of high intensity light to low intensity the thickness of the AML at the optical null grows during the exposure. This increases the absorption in the transparent regions. Likewise, the thickness of the AML is decreased at the optical peaks resulting in decreases absorption in the opaque regions. This effectively reduces the absorption contrast, calculated as the optical density ratio between the opaque regions at the optical peaks and the transparent apertures at the nulls as seen in Fig. 10 . An absorption contrast reduction of up to 25% is observed due to this effect, which may be responsible for some of the linewidth broadening observed in our experiments, and points to the need for AML materials with reduced susceptibility for forming SRGs in future.
In this paper we have confirmed experimentally the results from 2D FEM simulations that show increase contrast and depth of focus in AMIL when using TM polarization for λ1 and integrating plasmonic reflector below the photoresist. This was done by measuring the process latitude for various exposure conditions. At an intensity ratio of 4, a 32% increase in process latitude was gained by using TM polarization for λ1 and a further 34% increase was achieved by integrating a plasmonic reflector into the resist stack, for a total increase in process latitude of 66%. We also discussed the formation of photoinduced SRG in the AML during exposure and its detrimental effects on the contrast of AMIL.
The authors would like to thank Rajesh Menon, Hsin-Yu Tsai, and Samuel W. Thomas III for synthesizing and donating the pMAEA material. This research is supported by the Marsden Fund Council contract No. UOC0806.
References and links
1. S. Okazaki, “Resolution limits of optical lithography,” J. Vac. Sci. Technol. B 9(6), 2829–2833 (1991). [CrossRef]
2. E. Muzio, “Optical lithography cost of ownership (COO) – final report for LITG501,” SEMATECH, (2000) http://www.sematech.org/docubase/document/4014atr.pdf.
3. T. Ito, T. Yamada, Y. Inao, T. Yamaguchi, N. Mizutani, and R. Kuroda, “Fabrication of half-pitch 32 nm resist pattern using near-field lithography with a-Si mask,” Appl. Phys. Lett. 89(3), 033113 (2006). [CrossRef]
7. J. Foulkes and R. J. Blaikie, “Influence of polarization on absorbance modulated subwavelength grating structures,” J. Vac. Sci. Technol. B 27(6), 2941–2946 (2009). [CrossRef]
9. COMSOL Inc, 744 Cowper Steet, Palo Alto, CA 94301, www.comsol.com.
10. S. M. Rytov, “Electromagnetic properties of a finely stratified medium,” Sov. Phys. JETP 2, 466–475 (1956).
11. H. S. Ho, A. Natansohn, C. Barrett, and P. Rochon, “Azo polymers for reversible optical storage. 8. The effect of polarity of the azobenzene groups,” Can. J. Chem. 72(11), 1773–1778 (1995). [CrossRef]
12. C. Barrett, A. Natansohn, and P. Rochon, “Cis-trans thermal isomerization rates of bound and doped azobenzenes in a series of polymers,” Chem. Mater. 7(5), 899–903 (1995). [CrossRef]
13. N. K. Viswanathan, D. Y. Kim, S. Bian, J. Williams, W. Liu, L. Li, L. Samuelson, J. Kumar, and S. K. Tripathy, “Surface relief structures on azo polymer films,” J. Mater. Chem. 9(9), 1941–1955 (1999). [CrossRef]
14. C. J. Barrett, A. L. Natansohn, and P. L. Rochon, “Mechanism of optically inscribed high-efficiency diffraction gratings in azo polymer films,” J. Phys. Chem. 100(21), 8836–8842 (1996). [CrossRef]