Purification of trace amount of metal impurity from ultrapure water using membrane purifier/filter
02/01/2005
The metal removal performance of two IE membranes are evaluated and compared under practical conditions
By Yukio Hashimoto, Mutsuhiro Amari, Makoto Komatsu, Kunio Fujiwara
Manufacturing processes for microelectronics devices and silicon wafers are exposed to high-purity water more frequently than to any other chemicals. The ultrapure water (UPW) produced at the central water system in a fab is of extremely high purity, with each critical metal concentration typically below 1 part per trillion (ppt). However, it is difficult to maintain such high purity out of the central system during distribution to points of use for wafer cleaning.
 Figure 1. SEM of Graft Membranes: HDPE nonwoven and UPE microporous membrane (inset). Source: Mykrolis Corporation |
In the front end of line (FEOL) and back end of line (BEOL) wet cleaning processes, UPW wafer rinsing tools are designed to process many wafers in a short period of time in order to reduce cost of ownership. The processes require a high flow rate of UPW and need to be continuously fed with water of the highest purity level. We have reported the effectiveness of a metal purifier/filter cartridge assembled from an ion exchange membrane prepared from a radiation-induced graft polymerization method.1 The advantage of using the membrane purifier device at point of use was also reported.2 In this paper, practical considerations such as multiple metal contamination and colloidal impurity in UPW, the effect on the performance, and impact on microelectronics device manufacturing are discussed.
Ion exchange membrane
 |
Ion exchange (IE) membranes were prepared by radiation-induced graft polymerization. Two different polyethylene microporous membranes were used for the preparation of the IE membrane as base material: a high density polyethylene (HDPE) non-woven fabric sheet (pore size 5 µm, thickness 0.27 mm) for IE membrane A, and an ultrahigh molecular weight (UPE) membrane (pore size 1 µm, thickness 0.15 mm) for IE membrane B. Polystyrene sulfonic acid groups were introduced as a graft chain. The ion exchange capacity of membrane A was 300 meq./m2, and membrane B was 90 meq./m2. Figure 1 shows SEM photographs of such IE membranes after sulfonation. The maximum loading of ion exchange capacity was varied based upon material crystallinity inherited from the preparation method of the membrane. Depending on the base material, the mechanical strength was decreased due to a side reaction during the irradiation process of the electron beam to create radicals on the surface of the polyethylene membrane. As a result, there was an IE capacity loading limit for the UPE membrane by this method.
The metal removal performance of the grafted membrane has been previously reported.3 Both membrane A and B removed Na and Cu from metal-spiked UPW. Metal removal efficiencies were calculated from metal concentration changes, from feed to filtrate, in a filtration experiment using 47-mm diameter disc membrane samples. Over 99 percent metal removal efficiencies were obtained from both ion exchange membranes A and B when Cu-spiked UPW (100 ppb) was challenged. The kinetics of an ion exchange reaction in a dilute solution is known to be the boundary-layer diffusion controlled-surface reactionbetween a membrane surface and a metal ion. The metal concentration decreases exponentially in membrane thickness direction during filtration by capturing the metal ion with sulfonic acid of the ion exchange group. Smaller pore size IE membrane B has a larger Brunauer, Emmett and Teller (BET) surface and a faster metal adsorption reaction rate than the larger pore size IE membrane A. IE membrane A has a larger membrane thickness, and a longer retention time of fluid as a purifier device, however, both membrane A and B have a similar capability for metal removal in model solutions. The pore size and thickness of the membrane directly affect the metal removal efficiency.
Comparison of IE membrane performance as a practical application
Na and Fe are the two common and critical metal impurities in UPW. Na contamination is ubiquitous and Fe dissolves into water by equipment corrosion. Alkali metals Na and K have smaller complex forming affinity to styrene sulfonic acid than Ca or Cu. Na is two units smaller in its equilibrium constant than an alkali earth metal or transition metal ion. Figure 2 shows metal concentration changes of filtrate from 47-mm diameter discs IE membranes A and B when mixed metal-spiked solutions containing 50 ppb of Cu and Na were challenged. Close to 100 percent of the Cu was removed in both IE membranes during the filtration. With regard to Na, over 99 percent was initially removed from membrane B. However, Na started gradually leaking out with filtration volume. On the other hand, for IE membrane A, Na retention started at 99 percent-similar to IE membrane B-and slowly decreased. After some amount of filtration, Na retention of IE membrane A became stable at around 90 percent.
This phenomenon was explained as follows: At the beginning of filtration, the ion exchange reaction took place between the metal ion and the H-form of the sulfonic acid group. After a certain amount of filtration volume, the surface of the ion exchange membrane was covered with metal ions bound to sulfonic acid groups. When the metal-spiked UPW was continuously fed to the metal-covered surface, a metal complex exchange reaction occurred between the lower affinity metal complex of sulfonic acid, Na, and the higher affinity metal ion, Cu, in the solution. The difference in the Na elution behavior between the two IE membranes is due to IE capacity and BET area differences. More specifically, IE membrane A has a smaller BET surface and a higher IE capacity. Thus, IE membrane A has a thicker ion exchange layer on the surface, allowing Na ions to diffuse deep into the IE layer and avoid the competitive complex exchange reaction between Cu and Na ions.
 |
Another major concern for metal contamination is iron (Fe) dispersed as colloidal particles in UPW. Fe dissolves into high-purity water by the oxidative corrosion of Fe in the process equipment as divalent iron (Fe2+). This can further oxidize to the trivalent iron (Fe3+) which can form mixed oxides expressed as Fe2O3, Fe(OH)2+ and Fe(OH)3. These mixed oxides coagulate and form polynuclear complexes dispersed as colloidal particles in UPW, with wide-ranging particle size distributions and surface charge densities. Particle capture by the charged IE membrane is influenced by the electrostatic interaction between such charged particles and ion exchange groups and the particle diffusion. Smaller colloidal particles of higher charge density overcome inertial forces and are captured on the surface of the ion exchange membrane by electrostatic forces. Larger and heavier colloidal particles with less charge density are more likely to follow streamlines and flow through the ion exchange membrane before
being captured by the ion exchange surface. Therefore, to capture both ionic and colloidal impurities, metal purifier devices have been designed with a laminated structure of ion exchange membrane and microporous UPE membrane of a sub-micron pore size rating. Figure 3 shows the metal removal performance of devices as a function of flow rate with 200 ppb Fe-spiked water at pH 4.5. The ion exchange membrane removed 80 percent to 90 percent of metals at various flow rates, while the laminated devices removed almost 100 percent of Fe (including both ionic and colloidalparticles) from the solution.
Evaluation with integrated circuit (IC) device manufacturer
Our in-house testing demonstrated the advantage of IE membrane A and the laminated structure with a UPE membrane as a purifier/filter. We conducted an evaluation in conjunction with an IC device manufacturer in order to examine the impact of the metal purifier.
Case 1: Metal concentration before and after metal purifier/filter
 |
The device manufacturer evaluated the metal purifier/filter in a UPW line at a point of use for Quick Dump Rinsing Bath made with Quartz. The metal concentration in the UPW was measured by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) before and after the filter/purifier, and compared with a purifier prepared from IE membrane B, and with a polytetrafluoroethylene (PTFE) membrane filter. The result in Figure 4 shows superior performance of the purifier/filter using membrane A.
Case 2: Minority carrier lifetime (MCLT) comparison with and without purifier
 |
Wafers (300 mm) were dipped and soaked in 75ºC UPW for twenty minutes after a 140ºC sulfuric acid/hydrogen peroxide mixture (SPM) for ten minutes, followed by intermetal dielectric (IMD) dry and thermal oxidation. MCLT of the wafer processed with and without the metal purifier/filter were measured and compared. Figure 5 shows that wafers processed with purifiers resulted in higher values of MCLT.
Conclusion
The metal removal performance of two IE membranes were evaluated and compared under practical conditions. Using model UPW solutions with multiple-metal contamination and colloidal impurity, the filtration experiment showed that IE membrane A exhibited superior metal removal capabilities. The advantage of the metal purifier/filter for a UPW line in microelectronics device manufacturing was demonstrated. III Yukio Hashimoto is a senior engineer of application technology development at Nihon Mykrolis, Tokyo. He can be reached at [email protected]. Mutsuhiro Amari is a manager in application technology development at Nihon Mykrolis, Tokyo. He can be reached at [email protected]. Makoto Komatsu is senior research engineer of the Radiation Graft Polymer Project at Ebara Research Co., LTD. He can be reached at [email protected]. Kunio Fujiwara is project leader of the Radiation Graft Polymer Project at Ebara Research Co., LTD. He can be reached at [email protected]. 1. B. Parekh, K. Fujiwara, M. Komatsu, Y. Hashimoto, M. Amari. CleanRooms, Vol. 17, No. 6, June 2003. 2. B. Parekh, Y. Hashimoto, M. Amari, I. Funahashi. A2C2 Magazine, January 2004. 3. See reference 1.References