All reagents were of analytical grade, and solutions were prepared with deionized water from an Organo Puric- MX system. Non-ionic surfactant standard solution was prepared by dissolving 0.100 g of polyethylene (10) octylphenyl ether (Triton X-100, Tokyo Chemical Industry Co. Ltd.) in 1 L of water. A working standard solution was prepared by accurate dilution of the stock solution with water. The potassium acetate solution was prepared by dissolving 8.33 g of potassium acetate in 250 mL of water. To prepare a 2 × 10⁻⁴ mol·L⁻¹ solution containing the counter ion, 0.024 g of TPPS (>98.0%, C₄₄H₃₀N₄O₁₂S₄·xH₂O; Tokyo Chemical Industry Co. Ltd.) was dissolved in 500 mL of water.

Spectrophotometric measurements were carried out with a Shimadzu UV-1800 spectrophotometer using 1 cm cells. Tetrafluoroethylene test vessels (FEP) (30 mL) were obtained from Nalgene. The vessels were used for more than one experiment. A Taitec model SR-1 horizontal shaker was used to shake the 30 mL FEP test tubes.

A sample solution (5 mL) that contained non-ionic surfactants was divided into 6 aliquots, which were then transferred into 30 mL FEP tubes. 10 mL of potassium acetate solution and 0.20 mL of TPPS solution were added into the sample. The tubes were shaken mechanically for 5 min (150 times per min). The solution was discarded, and the remaining solution was removed by tapping the tube on a paper; the ion-associate adhering to the inner vessel wall was dissolved with 5 mL of water. The absorbance of the solution was measured at 417 nm using a 1 cm glass cell and a spectrophotometer.

The adhesion of counter dyes and ion-associate on the vessel is influenced by the type of vessel employed [12] . Thus, a selection of different vessels was investigated and their effect studied (Table 1). High blank values of 0.373 and 0.451 were obtained with TPPS used an ion pair for polycarbonate (PC) and polysulfone (PSF), respectively; in this case, vessels had hydrophilic functional groups such as carboxy and sulfonyl groups. In contrast, tetrafluoroethylene (FEP) and polypropylene (PP) are characterized by hydrophobic properties. The hydrophobicity of these polymers is typically expressed in terms of critical surface tension and water contact angle (Table 1) [17] . Our data suggested that the water contact angles depend on the adhesion of TPPS and the ion-associate on the wall. Interestingly, we found that when the contact angles were small, the blank values were large and that ion-associate adhesion was low, the opposite trend was observed when the contact angles were large. A PP vessel showed a small blank value when used for the first time; however, repeated use of this vessel caused an increase in blank values. Notably, FEP did not show this behavior and therefore, was chosen as the vessel in our experiments. The adhesion surface hydrophobicity was investigated with glass vessels via silylation with 5% (v/v) dimethyldichlorosilane in toluene for 1 d.

After silylation treatment of the inner surface of borosilicate, the glass tube was rinsed twice with 10 mL of toluene and thrice with 10 mL of methanol. When the silylated glass was used as a vessel, the adhesion of the ion-associate was more than that in non-treated glasses; as a result, the slope of the calibration curve for non-ionic surfactants was found to be three-fold steeper than those in non-treated glasses; the blank value was also reduced (Figure 1). This finding suggested that adhesion of the ion-associate is higher when a hydrophobic surface is used. We would like to remark that the silylated glass could not be used repeatedly because the slope of the calibration curve decreased and the blank value increased.

Potassium ions form complexes with non-ionic surfactants [18] , therefore, the concentration of potassium ion was investigated in our systems. Data in Figure 2 show that the absorbance reached a constant value when 7

^*Absorbance difference between that of blank and of the sample; ^**Data were extracted from Ref. [20] ; the unit of surface is mJ·m⁻².

mL of 4 mol·L⁻¹ potassium acetate solution was added. In addition, the potassium ion forms a 1:1 or 1:2 complex with the non-ionic surfactant, and 1 mole of potassium ion is surrounded with 7 oxyethylene units [20] . However, a large amount of potassium ion was necessary because potassium does not form a very strong complex with the non-ionic surfactant [21] .

Figure 3 shows that the absorbance of the sample and of the blank decreases when temperature is increased. In addition, we found that the difference in the absorbance between the sample and the blank is constant between 10˚C and 30˚C. Thus, the sensitivity of our method is independent of temperature changes close to room temperature. Interestingly, when TBPE was used, the blank value hardly changed with temperature; in the case of the sample solution, we observed the maximum absorbance at temperatures around 35˚C [17] .

The nature of the non-ionic surfactant was not influenced by pH changes, because it does not contain functional groups that can be affected by pH changes. In contrast, the counter ions may be influenced by pH changes. Thus, the effect of pH was investigated. Data in Figure 4 show that, while the absorbance difference is virtually constant at pH ranges between 8 and 12, the absorbance of the sample and the blank increases at values lower than 8. This may be due to the fact that at these pH values, the association of TPPS and potassium complexes of non-ionic surfactant decreases. This may also explain why the blank value hardly changed at pH around 8 [16] .

Adhesion phenomena occurred when the solution was vigorously shaken. We suggest that the ion-associate gathers around the bubbles, which promote adhesion on the inner walls. Therefore, the effect of shaking time was also investigated, and the result is shown in Figure 5. Our data clearly show that absorbance reached a constant value when the shaking time was greater than 4 min.

The effect of shaking ratio (from 0 to 300 time·min⁻¹) was also investigated by changing the shaking speed of the shaker. Results displayed in Figure 6 show that the absorbance reached a maximum value when the shaking ratio was set to be 150 time·min⁻¹. Increasing the shaking speed further decreased the adhesion of the ion-asso- ciate.

Bubble condition was influenced by the vacant-volume-ratio of the vessel, because bubble formation is affected by this parameter. The volume of the sample and air ratio of the inner vessel was thus investigated (Figure 7). The adhesion of the ion-associate decreased, when the vacant-volume ratio-decreased by more than 60%. Thus, according to our data, the most suitable vacant-volume-ratios range between 40% and 60%.

The ion-associate adhering on the inner walls was dissolved with selected solvents and spectrophotometrically monitored (Table 2). In particular, alcohol could not dissolve the ion-associate; moreover, solubility decreased when the carbon content in the alcohol increased. Notably, high dissolution was achieved only with water. In particular, when water is added, both the complex between potassium and the non-ionic surfactant and that with TPPS dissociated. The proposed method, therefore, does not require the use of solvents such as ethanol or

^*Absorbance difference between that of blank and of the sample.

methyl cellosolve [10] [15] - [18] .

The calibration curve, prepared using a standard procedure, exhibits an excellent linear correlation coefficient (0.984), and the sensitivity for Triton X-100 (expressed as absorbance of 1 mg·L⁻¹ solution) was found to be 0.146. The molar coefficient for Triton X-100 was 1.04 × 10⁵ l·mol⁻¹·cm⁻¹ and the detection limit is 0.2 μg/L.

Potassium ion forms a helical coil with non-ionic surfactants; one turn of the helix consists of 6 - 7 EO units [19] . Commercially available non-ionic surfactants usually contain about ten POE units. Therefore, we investigated the effect of the number of POE units in the non-ionic surfactant, and the results are shown in Figure 8. When

the number (n) of POE units ranged between 2 and 5, the sensitivity was low. In contrast, when n ranged between 7.5 and 10, the sensitivity increased. Therefore, it can be concluded that more than 7.5 units of POE are necessary to generate a complex of potassium ion with Triton X-100.

The influence of 1.0 mg·L⁻¹ of ferric, aluminium ion, and zinc ions was investigated using calibration curves for Triton X-100. A mixture of these cations showed a decrease in the slope of the calibration curve. However, addition of EDTA (as a masking agent for cations) successfully eliminated their influence. This approach was thus included in the standard procedure.