Membrane compatibility for nanofiltration applications in DUV lithography
02/01/2008
EXECUTIVE OVERVIEW
Typical photochemical solutions contain a variety of additives, often in trace amounts. The slightest unintended alteration of a photochemical formulation’s chemical composition could have a detrimental effect on the photolithography process yield. The added complexity of photochemical formulations leads suppliers of liquid microcontamination control technologies to develop filtration solutions that effectively remove the destructive particles and gels from photochemicals without disrupting photochemical integrity during the filtration process.
During photochemical filtration, the chemical compatibility of the filtration materials (polymeric membrane) with the photochemicals being filtered is very important. Photochemicals may cause degradation, swelling, and loss of the membrane material’s mechanical strength, all of which could compromise the integrity of delicate photochemistries and the effectiveness of filtration.
Essentially, properly chosen membrane-based filters remove damaging particulate and gels from photochemicals while maintaining membrane integrity, so as to prevent alteration of the delicate chemical formulation of photochemical materials. This provides new opportunities for fabs and photochemical suppliers to enhance the performance, repeatability and reliability of semiconductor lithography processes.
Here we discuss the results of several studies of chemical compatibility between membrane materials and photochemicals, including an extensive chemical compatibility study of ultra-high molecular weight polyethylene (UPE) and polyamide membranes with a variety of photochemicals, including photoresists, edge-beads-remover, and several organic solvents commonly used in photochemical processes. This study was enhanced by a complete filter and photochemical compatibility study based upon post-usage evaluation results of a filter that had been used for filtration of a BARC (bottom anti-reflective coating) photochemical in actual photolithography processes in a wafer manufacturing fab. Finally, this paper includes a joint study between Brewer Science Inc. and Entegris, Inc., with respect to the effects of nanofiltration on the integrity of various BARC materials, which details the nature of the polymer membranes used in the filters and the interactions of the membrane surface with the photochemical filtrates.
Photochemical materials
Photoresist and anti-reflective coating (ARC) materials are two major categories of photochemicals used in the lithography process of wafer-fabrication in the semiconductor industry. While photoresist is the core material in this process, ARCs are essential in helping the photoresist achieve even better performance by eliminating or reducing reflections [1].
In the lithography process, photoresist plays two major functions. First, upon responding to the incoming radiation, it can precisely produce pattern information on the substrate. Second, it must have the ability to protect the underlying substrate during the subsequent etch or ion-implantation process.
When IC technology reached the 0.25µm generation, traditional i-line lithography, not capable of printing features of that size, was supplanted by deep-UV-illumination (DUV). In turn, a new resist chemistry was developed because of strong UV absorption of diazonaphthoquinone (DNQ) at approximately 250nm.
The resists for DUV technology are known as chemically-amplified DUV resists. In this chemically-amplified resist system, a small amount of organic acid will be generated by the decomposition of a photoacid-generator (PAG) during the exposure step. The acid acts as a catalyst and will induce a cascade of subsequent chemical transformations in the resist-film. These transformations change the solubility of the resist in the developer.
DUV resists consist of three main components: a polymer (resin), a PAG, and a solvent. The composition differences between DUV resists and conventional i-line resists are: 1) the molecular weights of the polymers in DUV resists (25,000 to 50,000+) are much higher than that of i-line resists (~1,000) [2]; 2) DUV resists use PAG, instead of PAC, used in i-line resists; and 3) one or more polymers are used in DUV resists.
Because of the catalytic nature, DUV resists have higher sensitivity than do i-line resists. While this is a big advantage, it also causes problems for DUV technology. An interruption of the catalytic-chain reaction by any mechanism may lead to catastrophic resist-failure [3]. The most common example of this is basic airborne contamination, which may reduce the proton concentration at the film surface, resulting in T-top defects [2]. One way to solve this problem is to add additives to the resists [4-5], which make DUV resists more complex than i-line resists.
The complexity of both modern DUV resists and the delicate BARC (see details in reference [6]) formulations place challenges on handling and delivery of these photochemicals in the IC manufacturing process. Filtration solutions developed for these chemicals must have the ability to effectively remove destructive particles and gels from the liquid without disrupting the integrity of photochemicals during the filtration process.
Compatibility study of photochemicals and polymer membranes
The semiconductor industry has long recognized the importance of nanofiltration in removing particulate contaminants from photochemicals. Polymer membrane-based nanofiltration is widely used by photochemical manufacturers at various stages of photochemical manufacturing, as well as at the point-of-dispense in the lithography process. Increasingly effective filtration processes have become a necessity in the highly competitive industry, as properly-chosen membrane-based filters remove damaging particulate matter and gels from photochemicals while, at the same time, maintaining membrane integrity.
The chemical compatibility of membrane material and process chemistries is an essential factor in increasing filtration effectiveness. If filtration materials are not compatible with process chemistries, filtrate integrity may be seriously compromised. A filtration process that does not take chemical compatibility into account may lead to contamination of process chemistries, particularly if, as in the case of photochemical nanofiltration, the process chemistries are specific and delicate. If process chemistries are compromised through contamination, gross defects can foul runs, batches, or the entire process itself.
Potential chemical compatibility failures can occur as the result of a variety of aggressive reactions between membrane material and chemicals being filtered, depending on the nature of the chemicals. The fine structure of membranes makes them particularly sensitive to attack [6]. In addition, the membrane formation process leaves the polymers stressed and more susceptible to chemical attack. These chemical reactions may cause membrane degradation or swelling, and loss of the membrane material’s mechanical strength.
The degradation of the membrane surface can result in the release of small particles, called particle shedding, and these shed particles can be released into the filtrate. Additionally, the membrane’s pore size can increase, potentially releasing retained particles back into the process stream. The mechanical properties of the filter can be compromised, with a weakened membrane possibly resulting in defects in the filter structure, which may allow for unfiltered chemical to come in contact with the wafer. In these ways, chemical compatibility between the membrane and the chemical being filtered has a direct effect on the integrity of the filtrate, and, in turn, the entire process’ yield.
Compatibility study details
To provide realistically efficient and constructive filtration solutions for photochemical processes, we must examine the interaction (compatibility) between polymer membranes and photochemicals experimentally. In this study, we selected two of the most common materials used to manufacture particle filters for photochemical applications. The two materials are ultrahigh molecular weight polyethylene (UPE) and polyamide, and we selected the following photochemicals for membrane exposure:
- four common photochemical solvents: propylene glycol monomethyl ether acetate (PGMEA), propylene glycol monomethyl ether (PGME), ethyl lactate, and cyclohexanone;
- a KrF photoresist, UV-26; and
- an aggressive edge-beads-remover, Fujifilm RER 450.
The results of this study can serve as guidance to IC fabs to choose the right filtration solution for any photochemical filtration.
 |
Polyamide and UPE membrane samples were cut to size, then soaked in each photochemical at elevated temperature, if possible (for temperature sensitive chemicals, the soaking was done at ambient temperature). Each UPE and polyamide membrane used had a pore size of 50nm. The membrane coupons were submerged in each photochemical inside a container that had been placed on a heating device, the temperature of which was carefully maintained at 40°C throughout the 60-day soaking period.Chemicals were replenished regularly to compensate for evaporation of the organic solvents. The membrane samples were withdrawn after regular soaking intervals (20, 40, and 60 days) and tested for tensile strength, as compared with the control membranes, which had not been exposed to photochemicals. The experimental soaking matrix is listed in Table 1.
Because membrane material is exposed to chemicals and high pressure during filtration, chemical degradation, and the corresponding reduction of mechanical strength of the membrane material, is a distinct possibility. Thus, measurement of the tensile strength of membrane material post-chemical soak is telling, as the reduction in tensile stress and strain is predominant for membrane materials that have been chemically degraded.
An Instron tensile testing system was used to measure the tensile stress and elongation of membrane samples in both the machine direction (MD) and cross-web direction (CWD), which are orthogonal to one another.
 Figure 1. Effect of photochemical exposure on the tensile strength of a) polyamide and b) UPE membranes. |
All results of this experiment are shown in Fig. 1 (MD and CWD results are similar, and only MD tensile stress data are shown as representative). The average of the maximum tensile strength in the elastic range for three membrane coupons is reported for each data point.
Comparing the UPE membrane with the polyamide membrane, elongation values of the UPE membrane material are higher than those of the polyamide membrane. This suggests that the UPE membrane is physically more elastic than the polyamide membrane.
A change in elongation values is observed in the UPE membrane for some chemicals, such as PGME, PGMEA, ethyl lactate, cyclohexanone, and Fuji RER-450. This change is very likely due to the swelling of UPE membrane material in these solvents. The swelling of polymer materials is normal and does not affect chemical compatibility.
For UPE membrane, the tensile strength in both MD and CWD was constant (variation in the values is within the expected variation of the Instron instrument) over the sixty-day experimental period for all the photochemicals tested. Similar test results were observed for the polyamide membrane exposed to the photochemicals listed in Table 1, with the exception of MD strength after exposure to ethyl lactate, which shows a minor tensile strength loss of roughly 20%. These results suggest no chemical degradation due to extended exposure to the photochemicals selected for this study.
In addition to membrane exposure testing, two Entegris Impact 2 V2 (20nm, UPE membrane-based) filters were also analyzed for chemical compatibility testing. These two filters were used in an IC fab’s BARC chemical filtration (ARC29A) and were removed from the product line for testing. The entire filters were analyzed using a variety of test methods. The pressure drop values for both used filters were within 2psi of a new filter for the entire range of flow rates tested. Bubble point testing revealed that both tested filters were integral. SEM/EDS analysis of a membrane sample excised from the filters confirms that the used filters were lightly loaded with particles upstream, but not significantly plugged (Fig. 2).
 Figure 2. Impact 2 V2 (20nm, UPE membrane-based) SEM/EDS analysis results (DUV2, upstream). EDS shown is for captured particles indicated by arrows. |
The downstream side of the membrane was free of particles, which indicates that the membrane effectively removed particles from the photochemical stream during the photolithography process. The tensile strength testing shows little alteration of the membranes’ mechanical properties and no sign of degradation of the membranes exposed to this ARC29A chemical. Results of the device analysis confirm the aforementioned laboratory test’s conclusion that the UPE membrane-based filters are compatible with the fab’s photochemical applications.
Filtration effect on photochemical integrity
While testing membranes for integrity and degradation after exposure to a variety of photochemicals is one aspect of the chemical compatibility study, testing the integrity of the photochemical after filtration is another. One must determine whether the filter removes or alters the filtrate’s ingredients, or if it leaches undesired chemicals into the filtrate. In addition to these issues, one must also examine whether the nanometer-sized pores in the membrane hinder filtration of a photochemical formulation’s larger molecules, which, in turn, could alter that filtrate’s molecular weight distribution. After all, in modern DUV lithography technology, the molecular weights of photochemical resins are as large as 25,000 to 50,000+ amu, while current nanofiltration technology incorporates pore sizes as small as 10nm.
 |
Each of the above issues must be addressed experimentally in a complete study of photochemical and membrane compatibility. Such a study, a joint project between Brewer Science Inc. and Entegris, Inc., evaluated the suitability of UPE membranes in filtration of a variety of BARC chemical solutions in Entegris’ Analytical Laboratories. Three different BARC formulations, which cover a spectrum of applications, were tested. Table 2 shows the chemicals and their application properties.
A 100mL sample of BARC material was filtered through an Entegris UPE filter, Optimizer D300 (CWAX031S2), overnight by recirculation, in order to enhance BARC exposure to the membrane. This filter is rated at 20nm and has 300cm2 of membrane area. In addition to the UPE membrane, the filter is constructed with high-density polyethylene (HDPE) supports, components, and housing. Following the filtration exposure, the BARC chemical was analyzed using gas chromatography-mass spectrometry (GC-MS), ultraviolet-visible spectrometry (UV-Vis), and gel permeation chromatography (GPC), to detect alterations of the filtrates’ respective chemical compositions.
The GC-MS separates the volatile and semi-volatile ingredients of the BARC and examines them using a mass spectrometer. The mass spectrometer detector for gas chromatography captures and displays the unique mass fragmentation pattern of each component that is separated by GC. It is capable of detecting slight changes in the concentration of those organic constituents of the BARC.
As a compensation test method to composition analysis, UV-Vis examines the absorbance of ultraviolet- and visible-light sensitive components of the BARCs. By comparing relative heights of the peaks, the change in quantity of UV-Vis absorptive components can be quantified before and after filtration.
GPC separates components of the BARC sample by molecular weight when the sample passes through a size exclusion column. This technique can detect shifts in molecular weights of the polymer constituents of the BARC by measuring the molecular weight distribution of the filtrates.
The entire battery of tests performed in the photochemical integrity study confirms the BARCs’ integrity. The results for DUV2 are shown as representative of the results for all three BARC chemicals examined. Figure 3 shows the GC-MS results for the DUV2 pre- and post-filtration. It is abundantly clear that no alteration of BARC components has taken place.
 Figure 4. UV-Vis analysis of DUV2 before and after filtration with UPE filter. Wavelength values were removed for protection of a proprietary formulation. |
Figure 4 shows the UV-Vis spectrum pre- and post-filtration for DUV2. Once again, there is no change in BARC chemistry. Figure 5 shows the GPC profile for the BARC chemical before and after filtration. The molecular weight distribution of each polymer system has not changed. This indicates that the large polymer molecules were not altered during filtration.
 Figure 5. GPC analysis of DUV2 before and after filtration with UPE filter. Retention volume values were removed for protection of a proprietary formulation. |
null
Conclusion
The results of both a long-term (60 days at 40°C is equivalent to 240 days at ambient temperature, by means of first-order Arrhenius calculations) in-house chemical compatibility test and an actual manufacturing fab-used filter analysis experimentally proved that a UPE membrane is chemically compatible with most of the common organic solvents and chemistries used in photolithography processes. This is also true for polyamide membrane, except for low pH (acid) photochemical applications [6].
Photochemical integrity testing results showed that no changes were observed for the formulation ingredient(s) of DUV1, DUV 2, and DUV 3 before and after filtration using UPE-based filters. There is no evidence that any compounds either leached into the BARCs, or were removed from them during filtration, and the molecular weight distribution of the polymer constituent did not shift.
Essentially, both UPE membrane-based filters and polyamide membrane-based filters are suitable filtration solutions for the most common photochemicals used in modern DUV-lithography technology. The combination of these two membranes is even more powerful and will meet almost all the needs in filtration of photochemicals to enhance the performance, repeatability and reliability of semiconductor lithography processes.
Acknowledgments
Impact and Optimizer are registered trademarks of Entegris. ARC is a registered trademark of Brewer Science. We wish to thank co-authors Yaowu Xiao for his assistance in tensile strength measurements, Aiwen Wu for his contributions relative to generating filtrated BARC samples, and Jian Wei for technical leadership on the project.
References
- R. Buschjost “Understanding Brewer Science’s Bottom Anti-Reflective Coatings,” RIT Microlithography Materials and Processes EMCR, p. 666.
- R. Mohondro, “Photostabilization: Comparing DUV and i-line,” Solid State Technology, Feb. 2003.
- S. Solf, Microchip Manufacturing, Lattice Press, p. 322, 2004.
- D.J. H. Funhoff, H. Binder, and R. Schwalm, “Deep-UV Resists with Improved Delay Capabilities,” SPIE Proceedings, Vol. 1672, p. 46, 1992.
- H. Roschert, K. J. Przybilla, W. Spiess, H. Wengenroth, G. Pawlowski, “Critical Process Parameters of an Acetal-based Deep UV Photoresist,” SPIE Proc., Vol. 1672, p. 33, 1992.
- H. Zhang, A. Wu, J. Wei, R. Buschjost, “Nanofiltration Maintains Photochemical Integrity,” Semiconductor International, September 1, 2007.
Haizheng Zhang received his PhD in chemistry at Boston College and is a scientist at Entegris.
Mufadal Ayubali received his BS in microelectronics engineering and MS in chemistry at the Rochester Institute of Technology and is a scientist at Entegris.
Jacob Andrews received his BS in chemical engineering at Tufts U. and is a process engineer at Entegris.
Xia Man received her MS in sciences at the U. of Guelph and is a scientist at Entegris.
Patrick Antle received his BS in chemistry at Duke U. and is a scientist at Entegris.
Ryan Buschjost, a registered professional engineer, received his BS in chemical engineering from the U. of Missouri and is a process engineering manager at Brewer Science Inc.
Yaowu Xiao received his PhD in chemistry from Syracuse U. and is a scientist at Entegris.
Aiwen Wu received his PhD in chemical engineering at the U. of New Hampshire and is a process engineer at Entegris.
Jian Wei received her PhD in chemistry at Tufts U. and is the manager of analytical chemistry and product evaluation at Entegris, 129 Concord Road, Billerica, MA 01821 USA; ph 978/436 6822, e-mail [email protected].