Abstract
Dissolved oxygen is one of the major contaminants that have to be removed in the production of ultrapure water. The removal of dissolved oxygen from ultrapure water employing microporous hydrophobic membranes has been studied. The membranes have a negligible resistance to the passage of oxygen and provide a larger surface area per unit volume for gas-liquid contact. A study of the mass transfer in the membrane gas-liquid contactor showed that the resistance in the liquid film adjacent to the membrane phase controlled the rate of oxygen removal. The L&&que equation which describes adequately the observed overall mass transfer coefficients, fails to predict satisfactorily the dissolved oxygen concentrations in the parts per billion range. The experimental results also indicate that the membrane modules were capable of reducing the dissolved oxygen content in water to a level of around 8 ppb.
Introduction
Ultrapure water, a quality chemical, is water which is essentially ‘free’ of contaminants, e.g. dissolved ions, silica, particles, organics, dissolved gases and bacteria. Generally, specifications for ultrapure water, call for only parts per billion to trillion of impurities, depending on the type of contamination and varying with the process applications. Production of ultrapure water is one of the key support services for manufacturing of semiconductors and pharmaceuticals, in biotechnology and power industries. In the semiconductor industry, one of the major problems encountered is the presence of dissolved oxygen (DO) in the ultrapure water which may result in the formation of a native oxide layer during wafer cleaning. In order to suppress the occurrence of this native oxide layer, the dissolved oxygen level in the ultrapure water must be kept very low, less than 0.1 ppb is required in some specific applications.
Experimental
The hollow fibre membranes used in the present work were made of microporous polypropylene, of about 0.04 cm OD, and 0.003 cm wall thickness and they were manufactured by Hoechst Celanese. The pores, about 30 nm diameter, cover about 33% of the hollow fibre surface. The hollow fibre modules designed for the present experiments consisted of a 0.02 m diameter polymethylmethacrylate tube acting as a shell. The desired number of hollow fibres, assembled as a bundle, were potted into the shell and glued on both ends of the shell with a liquid epoxy resin (70% epoxy resin and 30% hardener). A total of seven modules with different length and packing fraction have been employed in this study. The characteristics of these modules are shown in Table 1.
Fig1
The apparatus and operating conditions for the removal of dissolved oxygen from ultrapure water using the modules are shown in Fig. 1. The feed water at 22℃ containing about 8 ppm dissolved oxygen was fed into the hollow fibre lumen and purified nitrogen used as purge gas was introduced at the shell side. The inlet flow rates of water and nitrogen were controlled by rotameters. The concentration of the dissolved oxygen in the outlet water stream was measured by a DO meter (Martek) and was used for calculation of the overall mass transfer coefficient KL according to the following equations.
Results and discussion
The effect of operating conditions such as water and sweep gas flow rates on the performance of hollow fibre modules has been studied. The overall mass transfer coefficients were found to be a function of the water flow rate. In order to examine the dependence of the oxygen transfer coefficient on the process parameters, such as liquid and gas flows, the overall mass transfer coefficient expressed in terms of the Sherwood number (kLd/D) versus the water velocity per module length, expressed in terms of the Graetz number ( VLd2/ZD), has been plotted as a log-log graph shown in Fig. 2. For flow through small tubes, it has been suggested that the mass transfer in the fibre lumen can be described by Leveque’s equation, Sh=1.62Pe0.33 shown by the solid line in Fig. 2. It can be seen from the figure that the overall mass transfer coefficients KL obtained from the modules used in this study fall closely to the solid line indicating that the oxygen mass transfer in the liquid phase is dominating.
Fig2
The effect of sweep gas flow rates on the overall mass transfer coefficient is illustrated in Fig. 3 where the overall mass transfer coefficients KL are plotted against the gas flow rate for three modules with fibre lengths of and 10 cm, respectively. As expected the overall mass transfer coefficients are independent of the gas flow rate indicating a negligible gas film resistance.
Firstly, the gas flow rates in Fig. 4 (O&S- 3.4 x 10m6 m”/sec) are considerably lower than those in Fig. 3 (5-260 x 10m6 m3/sec). Based on eqns. (1) and (2)) at a low gas flow rate, the difference between the cocurrent and countercurrent flows is obvious. Of coume, when the gas flow rate approaches infinity, i.e. QJHQo becomes zero, their difference disappears and eqns. (1) and (2) become identical. When the gas flow rates are plotted against the dissolved oxygen concentration in the product stream, the effects of the flow pattern are significant as shown in Fig. 4. If the data in Fig. 4 are transferred to Fig. 3, the effect of the flow pattern becomes negligible as shown in the inset of Fig. 3. However, at the low gas flow rates operated, the results indicate that the gas flow rate has only a small effect on the overall mass transfer coefficient. Thus, it may be concluded that the flow pattern used with the hollow fibre modules has a significant effect on the overall driving force, but has little effect on the resistance to mass transfer.
Fig3
Conclusions
Hollow fibre modules containing hydrophobic membranes have been assessed for their capability to reduce dissolved oxygen to the parts per billion concentration range in ultrapure water production. The overall mass transfer coefficients ofoxygen have been obtained in theliquid phase flowing through the fibre lumenand were observed to be dominated by the in-dividual mass transfer coefficient in the liquidfilm. The Lévèque equation, which describeswell the observed mass transfer coefficient dataobtained in the liquid phase, does not give asatisfactory prediction at very low gas flow ratesand dissolved oxygen levels in the parts per bil-lion range. Countercurrent flow operation wasfound to yield better results than cocurrent flowin terms of both the degree of dissolved oxygenremoval and the minimum sweep gas require-ments.Experimental results also reveal that thelargest reduction in dissolved oxygen obtainedwith the microporous hollow fibre modules usedin this study gives a value around 8 ppb.
上一篇: 磁控溅射制备光滑SAC305薄膜
下一篇: 工业系统中超纯水细菌污染分析