Anal. Chem. 2000, 72, 465-474
Visual and Colorimetric Lithium Ion Sensing Based on Digital Color Analysis Etsuko Hirayama,† Tsunemi Sugiyama,† Hideaki Hisamoto,† and Koji Suzuki*,†,‡
Department of Applied Chemistry, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan, and Kanagawa Academy of Science and Technology, 3-2-1 Sakado, Takatsu-ku, Kawasaki 213-0012, Japan
A new optical analytical method, “Digital Color Analysis (DCA)”, is proposed based on a digital color analyzer instead of the conventional optical methodology, “Spectrophotometry”. The digital color analyzer is a hand-heldsize instrument for measuring “colors”, and it can transform the color information into numerical values, color library data, etc., that can be treated as analytical information. DCA gives us a more informative analytical method than spectrophotometry by treating colors as digital information. In addition, DCA can also simulate the optimum color variations for optimization of the visual sensor with computer assistance. By utilizing colors as digital information, colorimetric analysis that has been used for only semiquantitative analysis can serve as an accurate determination method. On the basis of DCA, we developed a plasticized PVC film optode and a paper optode for Li+ determination in saliva. After the optimization of color variation and the detection range for the Li+ measurements, the optode membrane gives colorless gray in the Li+ therapeutic range (at 10-3 M) in saliva. Consequently, whether or not the optimum therapeutic Li+ concentration is maintained can be easily evaluated with these optodes. Especially, the sensing paper optode can be easily handled within a short measurement time (∼80 s) which is suitable for home use. Using the digital color analyzer with QxQy coordinates, a linear relation calibration curve can be obtained over the range from 10-5 to 10-1 M Li+, in which the analyzer can detect a concentration difference of ∼0.1 mM Li+. For the near future, an accurate and simple analysis is needed for a health check at home that does not require going to a hospital. The optode based on DCA has great potential for this analytical purpose. With the conventional optical methods, highly accurate quantitative analysis has generally been performed by measuring an optical signal or spectra by spectrophotometers. In these methods, spectrophotometric values such as absorbance or fluorescence are normally measured. Thus, when the obtained spectrum is complicated, determination often becomes difficult. However, even if the spectrum changes are quite complicated, our eyes recognize * Corresponding author: (phone) 81 45 563-1141 ext. 3444; (fax) 81 45 5630446. † Keio University. ‡ Kanagawa Academy of Science and Technology. 10.1021/ac990588w CCC: $19.00 Published on Web 12/23/1999
© 2000 American Chemical Society
them simply as color changes. Determination utilizing the colors themselves is a perceptual method and is not similar to spectrophotometry. In analytical chemistry, Reilley et al.1,2 utilized quantitative color information instead of spectrum data for the first time in 1961 and evaluated the color changes of indicators. Since then, the color-change quality of various acid-base indicators3-7 and metal indicators8-16 has been analyzed. In these studies, the capability of indicators was evaluated by numerically representing the range and rapidity of color change at the end point of titration. However, these analytical methods which introduce color information are semiquantitative. We proposed a new optical analytical method, Digital Color Analysis (DCA), based on a variety of color information for applications with an analytical purpose. In this method, the observed colors are treated as digital information using a digital color analyzer. The concept of DCA consists of the analysis and utilization of “color” information, where a digital color analyzer is used as a means of utilizing “colors”. The digital color analyzer used in this study (COLORTRON) is hand-held-size and is capable of measuring colors quite easily. It can rapidly transform color information into numerical values by processing the data with a personal computer which is connected to the color analyzer. This analyzer can easily determine colors, calculate the tristimulus values (X,Y,Z) with the obtained spectrum data, and the XYZ values can be converted into a variety of numerical color data and (1) Reilly, C. N.; Flaschka, H. A.; Laurent, S.; Laurent, B. Anal. Chem. 1960, 32, 1218-1232. (2) Reilly, C. N.; Smith, E. M. Anal. Chem. 1960, 32, 1233-1240. (3) Bhuchar, V. M.; Kukreja, V. P.; Das, S. R. Anal. Chem. 1971, 43, 18471853. (4) Bhuchar, V. M.; Agrawal, A. K. Analyst (Cambridge, U.K.) 1982, 107, 14391450. (5) Cacho, J.; Nerin, C.; Ruberte, L. Anal. Chem. 1982, 54, 1446-1449. (6) Fernandez, A. M. C.; Chozas, M. G. Talanta 1987, 34, 673-676. (7) Kotrly, S.; Vytras, K. Essays on Analytical Chemistry; Pergamon Press: Oxford, 1977; pp 259-280. (8) Calatayud, J. M.; Marti, M. C. P.; Vives, S. S. Analusis 1985, 13, 87-89. (9) Cacho, J.; Garnica, A.; Nerin, C. Anal. Chim. Acta 1984, 162, 113-122. (10) Vytrasova, K. V. J.; Kotrly, S. Talanta 1975, 22, 529-534. (11) Cacho, J.; Lopez-Molinero, A.; Castells, J. E. Analyst (Cambridge, U.K.) 1987, 112, 1723-1729. (12) Prasad, K. M. M. K.; Rahhem, S. Talanta 1991, 38, 793-799. (13) Prasad, K. M. M. K.; Rahhem, S. Anal. Chim. Acta 1992, 264, 137-140. (14) Prasad, K. M. M. K.; Rahhem, S. Analusis 1992, 20, 401-406. (15) Rahhem, S.; Prasad, K. M. M. K. Talanta 1993, 40, 1809-1814. (16) Prasad, K. M. M. K.; Vijayalekshmi, P.; Sastri, C. K. Analyst (Cambridge, U.K.) 1994, 119, 2817-2821.
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000 465
Figure 1. Concept and data-processing diagrams for Digital Color Analysis (DCA).
color library data.17,18 In the case of calculating the tristimulus values, the characteristics of a standard observer’s eyes and the spectrum data of a light source are included in the calculation process as shown in Figure 1. The basic tristimulus values can be further transferred into several-figured color data such as HSB, RGB, L*a*b*, etc., or QxQy as a complimentary color data. These color coordinates have their own unique chromaticity diagrams. By transforming color information into numerical values, colorimetric analysis that has been used only as semiquantitative analysis can serve as an accurate method. Moreover, even if the spectrum is quite complicated to analyze, it is possible to make an accurate determination without the spectrum data by the color changes, on the basis of color difference or digital color library data for the obtained color. DCA can produce calibration curves by converting the spectrum into tristimulus values and plotting the color on a chromaticity diagram. DCA can also create a color library for direct visual calibration and simulate the optimum color changes based on limited experimental color data using a personal computer with suitable software. Thus, DCA has several major advantages for its utilization in the calibration/determination of an analyte using color data as follows:19 (1) calibration/determination based on numerical color data, (2) calibration/determination based on color library data, (3) (17) COLORTRON User manual; Light Source Computer Images, Inc., Larkspur, CA, 1994. (18) Prasad, K. M. M. K.; Rahhem, S.; Vijayalekshmi, P.; Sastri, C. K. Talanta 1996, 43, 1187-1206. (19) Sugiyama, T.; Kamae, Y.; Yamamoto, N.; Hisamoto, H.; Suzuki, K. Proposal of Color-based Chemical Sensings. Presented at the 45th Annual Meeting of the Japan Society for Analytical Chemistry, Sendai, Japan, September, 1996; Abstr. no. 2A06.
466
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
calibration/determination based on chromaticity diagrams, (4) suitable visual calibration/determination based on color simulating calibration which can be used for the design of an optimum colorbased optode. As one of the effective applications of DCA, we here demonstrate the determination of lithium ion in the millimolar concentration range which fits the manic-depressive therapeutic Li+ concentration level using Li+ film or paper optodes. Some lithium salts such as Li2CO3 are used to treat manicdepressive illness and hyperthyroidism. The manic-depressive therapeutic Li+ concentration range in blood is 0.75-1.25 mM, and it is known that more than 1.5 mM Li+ is toxic and 3.5-4.5 mM Li+ is lethal. Because of the toxicity of excess lithium salt and the requirement to maintain the concentration of Li+ within the narrow effective therapeutic range, it is necessary to control the Li+ concentration in the human body.20,21 Although Li+ can be determined by blood analysis, it is preferable to determine the Li+ concentration in saliva because it is more easily extracted and allows the patients to monitor Li+ concentration for themselves. The Li+ normally exists in saliva at a concentration twice that in the blood; the effective therapeutic Li+ range in saliva is 1.5-2.5 mM. Monitoring the concentration of a specific clinically relevant substance is important for controlling overprescription and for confirming that a patient actually took the medicine. However, in most cases, clinical analysis is difficult outside the hospital. Eastman Kodak Co. has developed a clinical colorimetric slide sensor for a serum lithium sample,22 but it is still in hospital use. Therefore, one of the aims of the present study is to construct an optode device that can be used at home to determine Li+ for treatment of manic-depressive illness. For such a device, simplified operation and the possibility of visual determination are desired. For simplified operation, we introduced a mechanism of dry chemistry23,24 and attempted to construct a film-type sensor (membrane optode) which needs only the dropping of the sample onto the optode film device which includes all the reagents necessary for analysis. For this purpose, a PVC membrane film optode containing a lithium ionophore, an indicator dye, and a lipophilic additive was prepared, and its response characteristics were investigated. In addition, to construct an optode device with shorter response time and easier handling compared with that of the PVC film optode, a test-paper-type optode for disposable use was prepared. The sample was dropped onto the paper optode, and the color-change characteristics were measured by a digital color analyzer. For visual determination, an attempt was made to produce the color change passing through a colorless gray point, which can be performed by screening the dye1,25-27 in the optode membrane (20) Frezzotti, A.; Gambini, A. M. M.; Coppa, G.; Sio, G. D. Scand. J. Clin. Lab. Invest. 1996, 56, 591-596. (21) Gorham, J. D.; Walton, K. G.; McClellan, A. C.; Scott, M. G. Ther. Drug Monit. 1994, 16, 277-280. (22) Bodman, V.; Arter, T.; Maseiewicz, F.; Dychko, D.; Schaeffer, J.; Winterkorn, R. Clin. Chem. 1992, 38, 1049. (23) Walter, B. Anal. Chem. 1983, 55, 499A-514A. (24) Curme, H. G.; Columbus, R. L.; Dappen, G. M.; Eder, T. W.; Fellows, W. D.; Figueras, J.; Glover, C. P.; Goffe, C. A.; Hill, D. E.; Lawton, W. H.; Muka, E. D.; Pinney, J. E.; Rand, R. N.; Sanford, K. J.; Wu, T. W. Clin. Chem. 1978, 24 (8), 1335-1342. (25) Flaschka, H. Talanta 1961, 8, 342-354.
that represent complementary color data are quite useful. When (QxQy) coordinates are used, a linear calibration curve can be obtained with the optode, and the calibration curve covers an extensive dynamic concentration range of typically 4-5 orders of magnitude. The combination of the color analyzer or membrane optode for visual analysis, which involves simplified accurate hardware, and the digital color analysis (DCA) as software is quite useful for chemical sensing devices for accurate analyte determination with ease. Here we demonstrate the lithium-ion determination with this concept that can provide a simple health check and examination at home. THEORY Response Mechanism and Theoretical Equations for the Ion Optode. The response mechanism of the Li+ optode constructed in this study is based on the ion extraction/exchange from the aqueous phase (sample solution) to the organic phase (optode membrane) as shown in Figure 2. The ion-pair extraction/ exchange can be described by the following, eq 1: + + + i+ w + So + DH R o a Si R o + Do + Hw
Figure 2. Ion-extraction/-exchange model of the optode based on a neutral ionophore and a lipophilic cationic dye in relation to the ionoptode response mechanism (i+, cation to be extracted (analyte, Li+ in the present case); H+, proton; S, neutral ionophore (TTD14C4); R, lipophilic anionic additive (TFPB); D, the color-changeable dye (KDM11); D′, screening dye (KD-S1). The subscripts o and w represent the organic phase and water phase, respectively.) For factors and coefficients, see the Theory section.
according to the concept of DCA. Such a color change is easy to recognize visually at a certain point or narrow level of analyte concentration. When the color of the sensor changes from one color to its complementary color, passing through the gray point at nearly 10-3 M of Li+ concentration, it is possible to efficiently evaluate whether optimum therapeutic Li+ concentration is maintained or Li+ is in excess over the toxic limit. To realize this, a Li+ optode based on DCA, TTD14C428 and a functional dye, KDM11,29 were synthesized and used as the lithium ionophore and color-changeable dye, respectively, which were previously developed by our group. The membrane optode is capable of not only visual determination but also accurate determination using a hand-held digital color analyzer. There are various calibration methods that are based on the conversion of colors into numerical values and coordinates (see Figure 1), but especially the (QxQy) coordinates1,30 (26) Bosch, E.; Casassas, E.; Izquierdo, A.; Roses, M. Anal. Chem. 1984, 56, 1422-1428. (27) Calatayud, J. M.; Marti, M. C. P.; Vives, S. S. Analyst (Cambridge, U.K.) 1985, 110, 837-839. (28) Suzuki, K.; Yamada, H.; Sato, K.; Watanabe, K.; Hisamoto, H.; Tobe, Y.; Kobiro, K. Anal. Chem. 1993, 65, 3404-3410. (29) Hisamoto, H.; Tohma, H.; Yamada, T.; Yamauchi, K.; Siswanta, D.; Suzuki, K. Anal. Chim. Acta 1998, 373, 271-289.
(1)
where i+ represents the cation to be extracted (analyte, Li+ in the present case) and H+ represents a proton. The abbreviations of S, DH+R-, Si+R-, and D represent the neutral ionophore, the ion-pair of a protonated dye and a lipophilic anionic additive, the ion-pair of an ion-ionphore complex and a lipophilic anionic additive, and the deprotonated dye, respectively. Subscripts o and w represent the organic phase and water phase, respectively. When the neutral ionophore S complexes with the analyte cation, the cationic dye DH+ deprotonates and becomes neutral. Subsequently, the ion-ionophore complex and the lipophilic anion Rform an ion pair in the organic phase. The total extraction constant, K, is written as the following, eq 2:
K)
aSiRo aDo aH+w aiw+aSoaDHRo
(2)
where atot denotes the activity of each chemical species. Based on all of the equilibrium constants and the reactions in the model shown in Figure 2, the analyte activity, ai+ , is given w by eq 3 (see refs 31 and 32).
a i+ ) w
aSiRo aDo aH+w aSo aDHRo K
) (aRtot - aDtot + aDo)aDo aH+w
(aStot - aRtot + aDtot - aDo)(aDtot - aDo)K
(3)
where atot denotes the total activity of each chemical species. (30) Flaschka, H. Talanta 1960, 7, 90-106.
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
467
Because the absorbance, A, is expressed as eq 4 according to Bouget-Beer’s law, eq 3 is rewritten as eq 5.
A ) baDo ai+ ) w
(baRtot - baDtot + A)AaH+w (baStot - baRtot + baDtot - A)(baDtot - A)K
(4) (5)
Vind,r Jindmind + Vs,r Jsms ) 0
where and b represent the molar absorption coefficient and the optical path length, respectively. Equation 5 represents the general relationship between the ion activity for the analyte (i+) in the water phase and the absorbance value (A) of the optode. For the evaluation of the optode response, introduction of the normalized absorbance, R, is useful, which is defined by the ratio of the activity of the deprotonated dye to that of the total amount of dye as expressed by the following, eq 6
A - A0 R ) tot ) A 1 - A0 a aDo
mixture will always lie on the line joining these two colors. The location of the color point depends on the mixing ratio. In order for the mixture of the color-changeable dye (indicator) and inert dye (screening dye) to exhibit grayness during the color change of the indicator, an experimental condition satisfying the following equation is required1,25
(13)
where “m” represents the volume of solution where the concentration of the solution has to be fixed, “r” represents x or y, and the subscripts “ind” and “s” denote the indicator and screening dye, respectively. V and J values are given by
Vr ) Q r - G r J)
(6)
X+Y+Z Xn + Yn + Zn
(14) (15)
D
where A0 and A1 represent the absorbance values of the dye in the protonated form and completely deprotonated form, respectively. Theoretical Equations for Color Mixing on the Basis of QxQy Coordinates. QxQy coordinates are complementary chromaticity coordinates.1,18 These are analogues of xy coordinates which are calculated from tristimulus XYZ values in order to plot color points in a two-dimensional diagram.30 In this case, QxQy values are calculated by the same procedure adopted in the case of xy, but the transmission spectra T(λ) are replaced with the absorption spectra A(λ). The equations for QxQy are given as follows
Qx )
U U+V+W
(7)
Qx )
V U+V+W
(8)
U)k V)k
∫
780
A(λ)xj(λ)P(λ)dλ
(9)
A(λ)yj(λ)P(λ)dλ
(10)
380
∫
780
380
in which Gr indicates the coordinates of the gray point and J is the optical concentration. Qind,r represents a certain color point during the color transition of the indicator. Assuming that mind ) 1 in eq 13, the volume of the screening dye ms to be added to the indicator can be calculated by solving the simultaneous eqs 16 and 17.
{
Vind,x Jind + VS,x JSmS ) 0
(16)
Vind,y Jind + VS,y JSmS ) 0
(17)
In general, when only one screening dye is used, these simultaneous equations cannot be solved unless the color points of the indicator, the screening dye, and the gray point lie exactly on the same line in the QxQy diagram. However, treating Vind,r and Vs,r as the vectors on the QxQy chromaticity diagram in which Gr is an original point, mS should be found by searching for the mS value where the size of the sum of vectors (Vind,rJind + Vs,rJSmS) expressed by eq 18 becomes minimum.
f(ms) ) x(Vin,x Jin + VS,x JSmS)2 + (Vin,y Jin + VS,y JSmS)2 (18)
(12)
By calculating mS using eq 18, the color change that passes through the gray point can be obtained. QxQy, J, and mS values can be calculated by Mathematica software with a personal computer.
where jx(λ), jy(λ), jz(λ) are the color-matching functions for the Standard Observer. A(λ) and P(λ) are the absorbance values of the sample and the intensities of the illuminant as a function of the wavelength, respectively. The color of the dye is presented as one specific color point in the (QxQy) chromaticity diagram, and the (QxQy) coordinates are independent of the concentration of the dye. When two colors are mixed, the color point corresponding to the subtractive
EXPERIMENTAL SECTION Reagent. Lithium ionophore (TTD14C4), tetrakis[3,5-bis(trisfluoromethyl)-phenyl]borate sodium salt dihydrate (TFPB), and 2-nitrophenyloctyl ether (NPOE) were purchased from Dojindo Laboratory (Kumamoto, Japan), where TTD14C4 was developed by our research group.28 Poly(vinyl chloride) (PVC; high molecular weight-type) was purchased from SIGMA (St. Louis, MO). Lipophilic cationic dye (KD-M11) was synthesized according to our previous report.29 The chemical structures of TTD14C4, KD-M11, and TFPB are shown in Figure 2.
W)k
∫
k)
468
780
380
A(λ)zj(λ)P(λ)dλ 100
∫
780
jy(λ)P(λ)dλ 380
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
(11)
The reagents used for the preparation of standard solutions were all analytical grade or the highest grade commercially available. Synthesis of the Screening Dye. The red screening dye, KDS1 (see Figure 2), was synthesized according to the following procedures: Disperse Red 1 (2 g, 6.4 mmol) was dissolved in 20 mL of N,N-dimethylformamide (DMF), and sodium hydride (NaH: 0.23 g 9.6 mmol) was slowly added at room temperature. To this solution octadecyl bromide (2.5 g, 7.7 mmol) was added, and the mixture was stirred for 2 hours at 60 °C. After the reaction period, a small amount of methanol was added to the solution to quench the excess NaH and DMF was evaporated. The resulting residue was extracted three times with chloroform. After the organic phase was dried using Na2SO4 and evaporation, the obtained residue was purified by silica gel column chromatography with hexane-ethylacetate (9:1) as the eluent to yield the final product, 4-nitro-4′-ethyl(octadecyloxyethyl)amino azobenzene (yield: 1.2 g, 33%). 1H-NMR (CDCl3, 270 MHz): δ 0.88 (t, J ) 6.6 Hz, CH3, 3H), 1.22-1.34 (m, CH2, NCH2CH3, 33H), 1.52-1.59 (m, CH2, 2H), 3.44 (t; J ) 6.6 Hz, OCH2, 2H), 3.55 (q, J ) 7.3 Hz, NCH2CH3, 4H), 3.61-3.64 (s, NCH2CH2O, 4H), 6.77 (d, J ) 9.5 Hz, ArH, 2H), 7.89 (d, J ) 9.2 Hz, ArH, 2H), 7.92 (d, J ) 8.8 Hz, ArH, 2H), 8.32 (d, J ) 9.2 Hz, ArH, 2H). Anal. Calcd for C34H54N4O3 (566.83): C, 72.05; H, 9.60; N, 9.88. Found: C, 72.21; H, 9.61; N, 9.75. Standard Solutions. The standard solutions of Li+ (Li+ test solutions) used were 10-5-1 M LiCI in 0.05 M tris(hydroxymethyl)-aminomethane (Tris)/HCl buffer (pH 7-9). The standard solutions for Na+ and K+ (10-5-10-1 M) were prepared with NaCl and KCl dissolved in 0.05 M Tris/HCl buffer (pH 8). The test acid and base solutions used were HCl (pH 2) and NaOH (pH 11) aqueous solutions. Preparation of the PVC Film Optode. The plasticized PVC membrane used for the lithium-ion film optode was prepared by the standard procedure previously reported.31,32 The membrane cocktail containing TTD14C4 (3.49 µmol), KD-M11 (3.49 µmol), TFPB (3.49 µmol), NPOE (40 mg), and PVC (20 mg), which were dissolved in tetrahydorofuran (THF), was dropped on a transparent plastic sheet that was fixed on a spin coater, which was rotated for 5 seconds at 4000 rpm to prepare the thin film. This film was dried under vacuum and cut into a 10 mm × 40 mm segment to fit into a standard glass vessel (10 mm × 10 mm × 50 mm) for the spectrophotometer. For the PVC membrane containing screening dye, the optode membrane was prepared by the same procedure with the lipophilic inert dye KD-S1 (3.53 µmol), NPOE (40 mg), and PVC (20 mg). For the preparation of the “screened” lithium optode membrane which exhibits colorless gray in the process of the color change, the QxQy and J values of these two membranes were calculated from the spectral data for each sample solution using Mathematica (Wolfram Research, Inc., Champaign, IL) as the software for the personal computer. The volume of the screening dye, mS, added to the lithium optode membrane was calculated from the QxQy and J values according to the theoretical equations described in the Theory section. (This program using Math(31) Hisamoto, H.; Miyashita, N.; Watanabe, K.; Nakagawa, E.; Suzuki, K. Sens. Actuators, B 1995, 29, 378-385. (32) Watanabe, K.; Nakagawa, E.; Yamada, H.; Hisamoto, H.; Suzuki, K. Anal. Chem. 1993, 65, 2704-2710.
ematica is available as supplementary material.) The lithium optode membrane containing the cationic dye and screening dye was prepared on the basis of the calculation results; a membrane cocktail which consists of TTD14C4 (3.49 µmol), KD-M11 (3.49 µmol), KD-S1 (2.70 µmol), TFPB (3.49 µmol), NPOE (40 mg), and PVC (20 mg) was used. Ion Determination with the PVC Film Optode. The prepared PVC membrane was placed in the glass vessel of the doublebeam spectrophotometer (U-2000; Hitachi Co., Ltd., Japan) and immersed in each ion-standard solution. After constant absorbance was maintained for one minute, its absorbance spectra were recorded. The membrane was then removed from the glass vessel. The water droplets were gently removed from the membrane surface, and the membrane was placed on a white Teflon plate. The color of the membrane was measured by a digital color analyzer (COLORTRON, Light Source Computer Images, Inc., Larkspur, CA) in the reflection mode. The same measurements were carried out three times, consecutively, and the average data were adopted. The absorbance at absorption maxima was transformed into normalized absorbance (relative absorbance) by eq 6, and the response curves were obtained using the data for R, in which the A0 and A1 values in eq 6 were the absorbance values corresponding to HCl (pH 2) and NaOH (pH 11), respectively. Preparation of the Optode Paper. The test paper-type optode containing KD-M11 was prepared from the KD-M11/TFPB ion pair to avoid the preconditioning step which uses HCl solution in the measurement procedures. This preconditioning step often made it difficult to obtain a stable response of the sensing papertype optode. The preparation procedure for the ion-pair of protonated KD-M11 cation and TFPB anion is the following: KDM11 (0.0523 mmol) and TFPB (0.0523 mmol) were dissolved in chloroform and washed with 0.1 M HCl twice to completely protonate KD-M11, following which the solution was washed with water to remove NaCl. The solution was evaporated in vacuo, and the KD-M11/TFPB ion-pair was obtained and used for the paper preparation of the optode. After the KD-M11/TFPB ion-pair (4.38 µmol), TTD14C4 (4.38 µmol) and NPOE (50.3 mg) were dissolved in THF, the mixture was dropped onto a coated paper fixed on a spin coater. After rotating the paper for 5 seconds at 4000 rpm with a spin coater, it was dried to remove THF. For this experiment, a special coated paper which has no calcium carbonate was used which was kindly donated by Ohji Paper Co., Ltd., Japan. The coated layer of this paper consists of only kaolin as the pigment and SBR (8%) as the binder. Using the cocktail for the paper optode including the screening dye which consists of KD-S1 (4.44 µmol) and NPOE (50.3 mg) in THF, the paper optode coated with the screening dye was prepared by the same procedure as that for the normal paper optode. In this case, the amount of screening dye added to the lithium test paper was calculated by the QxQy and J values of these two test papers with the same procedure as that for the plasticized PVC optode membrane. The lithium optode paper containing KDM11 and KD-S1 was then prepared by the membrane cocktail consisting of TTD14C4 (3.55 mmol), KD-M11/TFPB ion pair (3.55 mmol), KD-S1 (0.35 mmol), and NPOE (40.7 mg). Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
469
Ion Determination with the Paper Optode. The optode paper was cut into 10 mm × 10 mm pieces for the ion measurements. Each LiCl standard solution was dropped onto a piece of the optode paper. After standing for 2 min, the water droplet was gently removed, and the color of the paper was measured by the digital color analyzer (COLORTRON) in the reflection mode. The measurements were carried out three times, consecutively, and the average data were adopted. Measurements of Selectivity. Measurements of the paper optode were performed for KCl and NaCl solutions to evaluate the Li+ selectivity to K+ and Na+. This experiment was performed with the paper optode containing no screening dye. In this case, the ratio of the color difference, ∆E*/∆E*total, was plotted as a function of ion concentration to obtain the response curves for Li+, Na+, and K+, in which ∆E* is the color difference between the 0.05 M Tris buffer and each standard solution, and ∆E*total is the color difference between the buffer solution and NaOH. The color difference between two colors given in terms of L*,a*,b*18,33 is defined as eq 19.
∆E* ) x(∆L*) + (∆a*) + (∆b*) 2
2
2
(19)
Optimization of Concentration of the Dye in the Optode Membrane. When the absorption spectra are experimentally obtained at a certain concentration (Cdye,0) of the dye, L*a*b* values at various concentrations of the dye (Cdye) can be calculated. The absorption spectra A(λ) at Cdye are given by the following equation
A(λ) ) (Cdye/Cdye,0)A0(λ)
(21)
Therefore, the tristimulus values, XYZ can be expressed as eqs 22-24 using the absorption spectra data.33
X)k
∫
780
10-A(λ)P(λ)xj(λ)dλ
380
Z)k
∫
780
10-A(λ)P(λ)yj(λ)dλ
(23)
10-A(λ)P(λ)zj(λ)dλ
(24)
380
∫
780
380
k)
100
∫
780
P(λ)yj(λ)dλ 380
(25)
The XYZ values at Cdye are obtained by substituting eq 20 into eqs 22-24. The L*a*b* values at Cdye can be calculated as eqs (33) Pauli, H. J. Opt. Soc. Am. 1976, 66, 866.
470
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
Y Yn
1/3
a* ) 500
X Xn
1/3
b* ) 200
Y Yn
1/3
- 16
(26)
-
Y Yn
1/3
(27)
-
Z Zn
1/3
(28)
By plotting ∆ logCLi+/∆E* against Cdye/Cdye,0 values, the optimum concentration of the dye is determined as the concentration which gives minimum ∆ logCLi+/∆E* where the values represent the sharpness of the color change.3,7 The ∆E* value was calculated as the color difference in the test strips of 10-4 M and 10-2 M Li+ in the present investigation. RESULTS AND DISCUSSION Lithium-Ion Film Optode Based on a Lipophilic Cationic Dye. As shown in the following reaction in Scheme 1 and Figure 2, the membrane optode, using the neutral ionophore (S), forms a complex with lithium ion (Li+). Under the condition where the membrane contains the lipophilic anionic additive (R+), the lipophilic cationic dye (DH+) deprotonates in order to conserve electroneutrality in the membrane which causes a color change. + + + Li+ w + So + DH R o a Li SR o + Do + Hw
(Scheme 1)
On the basis of this mechanism, the color change occurs depending on the concentration change in the cation (Li+ in this case). In addition to this color-change mechanism, we attempted a “screening effect” by adding an inactive dye to this system. This screening dye must not influence the mechanism of ion extraction. The mechanism of Li+ determination in this system is shown in the following Scheme 2: + Li+ w + So + DH R o + D′ a
Li+SR-o + Do + D′ + H+ w (Scheme 2)
(22)
Here, k is given by eq 25.
Y)k
() {( ) ( ) } {( ) ( ) }
L* ) 116
(20)
where A0(λ) is the absorption spectra at Cdye,0. On the other hand, the relationship between transmittance T and absorption A is
T ) 10-A
26-28 from the XYZ values:
where D′ indicates the screening dye. In the present Li+ optode, a 14-crown-4 derivative (TTD14C4) was used as the Li+ ionophore that was developed in our laboratory.28 For the color-changeable lipophilic cationic dye, KD-M11 that was also developed by us was used in the present optode.29 KDM11 shows solvent-dependent spectral changes (solvatochromism) as well as pH dependent spectral changes (halochromism), because KD-M11 has both positive and negative charges in both ends of the π-conjugated system of the dye molecule. However, in the present investigation, KD-M11 was used as a simple cationic dye with only the pH dependent spectral change property which causes the color change in the sensing membrane. KD-M11 became yellow in its protonated form and turned blue in the deprotonated form. In the case where the quantity of protonated form of KD-M11 equals that of the deprotonated form, the dye exhibits green. In order to obtain the color change passing
Figure 3. Li+ response curves obtained with the PVC film optode without the screening dye. For the film optode, see the Experimental Section.
through the gray point using KD-M11, a red dye is adequate as the screening dye because red is the complementary color of green. For this purpose, a novel lipophilic aprotic red dye, 4-nitro4′-ethyl (octadecyloxyethyl) amino azobenzene (KD-S1) was designed and synthesized (see Experimental Section). Lipophilic cationic dye KD-M11 (pKa ) 4.5) used as a fundamental indicator was added with the same amount of the lipophilic anionic additive TFPB (R-) in the PVC optode membrane in which NPOE was used as the membrane solvent. Figure 3 shows the response curve for Li+ obtained with the PVC-based film optode containing TTD14C4, KD-M11, TFPB, and NPOE, where the measurements were performed with the solutions of lithium chloride adjusted to pH 7 or 9. These absorbance characteristics are similar to those of the reported optode membrane.32 Figure 4A shows the color changes in the film optode measured by the digital color analyzer which correspond to the response curve at pH 9 shown in Figure 3. In Figure 4, %∆E* indicates the ratio of the color difference ∆E* of 10-5-1 M Li+ or 10-4-10-2 M Li+ to the total color difference (∆E*tot) which indicates the maximum ∆E* value obtained by dropping HCl and NaOH onto the optode membrane. When the optode membrane responses are based on the ion extraction/exchange mechanism, by plotting the absorbance values at λmax against log C, S-shaped response curves such as the curve shown in Figure 3 can generally be obtained.32 We attempted to make it show a remarkable color change near the center of the S-shaped curve in order to detect small concentration changes sensitively over the manic-depressive therapeutic range (near 10-3 M). For pH 7 standard Li+ solutions, the PVC-based optode membrane showed a definite color change according to Li+ concentration changes. When pH 9 standard Li+ solutions were used, the detection limit was lowered by about 2 orders of magnitude compared with that at pH 7, with the center in the S-shaped response curve (normalized absorbance at 0.5) located at log C ) -3.6 as shown in Figure 3. This color change variation is suitable for a clinical Li+ optode that is intended for use in the therapeutic Li+ concentration range. Screening Effect on the PVC Film Optode. The dye for screening requires no color change in any aqueous solution. Therefore, we confirmed that the membrane including the designed and synthesized screening dye KD-S1 did not show a spectral change for pH 1-13 solutions in which the absorption
Figure 4. Color variations in the Li+ test solutions (10-5-1 M Li+, pH 9) obtained by the Li+ sensitive PVC film optode without (A) and with (B) the screening dye, and the QxQy chromaticity diagram illustrating the screening effect using KD-S1 as the screening dye that was added to the KD-M11 based-PVC film optode (C). For the film optode, see the Experimental Section.
maximum of KD-S1 was 497 nm. The response curve of the PVC membrane including KD-M11 and KD-S1 was identical with that of KD-M11. The color changes of the optode membranes with and without the screened indicator were measured by the digital color analyzer, and the typical resulting color of the membranes is indicated in Figure 4A and B, respectively. The QxQy coordinates of these membranes for 10-5-1 M Li+ solutions (pH 9) are shown in Figure 4C. We attempted to draw calibration curves in HSB, RGB, CMY, and CIE-L*a*b* tristimulus color spaces based on digital color data obtained with the digital color analyzer before we adopted the QxQy complementary chromaticity coordinates. In any tristimulus color spaces except for the QxQy diagram, the obtained calibration curves were given as a complicated curved line. Using a calculating software such as Mathematica, these calibration curves become sufficiently useful even though they were complicated lines. However, only in the case of the QxQy chromaticity diagram, can a straight line be obtained as demonstrated in Figure 4C. Furthermore, the coordinates are independent of the dye concentration. Accurate determination by the straight calibration lines is preferable and quite advantageous because all of the ion-selective optodes developed so far based on absorptiometry have used experimental S-shaped calibration curves which often differ from the theoretical response values. Because the straight line connecting the color point of KD-S1 and the gray point crosses the straight line of KD-M11 at a point near 10-3 M Li+ as shown in Figure 4C, it was expected that the color for 10-3 M Li+ would become colorless gray by the Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
471
screening. For this reason, the mixing ratio of these two dyes was calculated from QxQy values at the 10-3 M Li+ point and KDS1 color point in the QxQy chromaticity diagram, and the PVCbased optode membrane incorporating these dyes was prepared. The line for the KD-M11+KD-S1 in Figure 4C indicates the resulting color changes in this membrane. This calibration curve passed through the colorless gray point at 10-3 M as expected, and a vivid color change before and after the gray point could be successfully produced as shown in Figure 4B. The color difference %∆E* value between 10-4 M and 10-2 M Li+ was 52.1 in the case of only the optode membrane based on KD-M11 and that of KDM11+KD-S1 was 55.8. Although these values are not so different, the ratio of the color difference (%∆E*) between 10-4-10-2 M Li+ indicates that the membrane optode including the screened indicator was slightly larger than that for the membrane based only on KD-M11. It is difficult to visually discriminate a slight color difference during the color transition “yellowish-green f green f bluish-green”, whereas it is easy to discriminate between chromatic and achromatic color. In the case of the screened indicator, it is easy to decide whether a certain color near the gray points inclines to orange or purple which appears near the gray point. ∆E* reflects the human eye response characteristics. In this optode case, %∆E* of the optode using the screened indicator became larger than that using only KD-M11, although the response curves of both membranes were identical. On the basis of this demonstration with the film optode based on KD-M11 and KD-S1, by the conversion of its colors into numerical values, the simulation of the numerical ratio of color mixing and the screening by adding a screening indicator, an appropriate purpose optode could be obtained. In our case, the optode for clinical use that allows for easy visual detection at concentrations near the 10-3 M Li+ point was successfully developed. Sensing Paper Matrix. In the PVC-based optode membrane system, for which the experimental results are well in accordance with the theoretical response values, the color change can be optimized by mixing a screening dye without influence on the response mechanism. However, the response time of the PVCbased optode membrane is normally long (see later discussion) for application of the sensor for home use. Moreover, using large amounts of PVC as a disposable material is undesirable because of environmental problems. For the measurement system based on the digital color analyzer, easier and more rapid measurement with higher reproducibility is available in the reflection mode than is in the transmission mode. Therefore, an attempt was made to construct a simple optode device using white paper as a suitable reflector for the digital color analyzer. The response of the present optodes is based on the solventextraction mechanism between the sample solution as an aqueous phase and the sensing layer as an organic phase containing the functional molecules. The material of the optode membrane and its properties have a great influence on the sensitivity and characteristics of the sensor response. A normal white paper has high hydrophilicity because its main component is cellulose. Thus, it would normally be hard to incorporate the lipophilic ionophore or dye into the cellulose layer. An attempt was then made to use a paper whose surface is coated with the lipophilic matrix, on 472
Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
which the coating layer was expected to behave as an organic phase and be able to incorporate the sensing lipophilic components. On the surface of a commercially available normal white paper for printing with lipophilic dyes (inks), kaolin and calcium carbonate are generally incorporated in the coating phase to make the paper bright and smooth. In addition, casein, starch, and styrene-butadiene rubber (SBR) are often incorporated for stabilizing the coating phase to the base white paper. Moreover, a fluorescent dye, dispersing agent, antiseptic, and antifoamer are also often contained. The paper containing calcium carbonate is called a neutral paper, for which the surface is pH 7-8. This paper is damaged when it contacts acid. To avoid the influence of Ca2+ on the Li+ determination, we used a special coated paper, whose coating reagent consists of only kaolin as the pigment and SBR as the binder (2-3 µm thickness). The surface of this paper was pH 6. The sensing paper for the optode was prepared by spin coating the THF mixture solution containing the ionophore (TTD14C4), the protonated dye-anionic additive ion pair (KD-M11+/TFPB-), and the membrane solvent (NPOE). After drying the paper under vacuum for one hour, all the components were immobilized uniformly on the paper surface. The condition of the paper surface was kept quite dry even though it was touched directly. When the sample solution was dropped on it, the drop formed a half sphere because of its relatively high surface tension, and the water did not spread inside the paper. At this time, a rapid color change corresponding to the ion concentration was observed, as required, within one minute, even though the sample contains as little as 10-5 M Li+. In this case, as we expected, the coating layer on the paper surface behaves as the organic phase in which the ion extraction occurs based on the mechanism of solvent extraction as shown in Scheme 1 and Figure 2. Screening Effect on the Paper Optode. For the paper optode, it is necessary to protonate the lipophilic cationic dye KDM11 in the thin-coated layer of the paper for the start of the measurement as illustrated in Figure 2. In order to protonate KDM11 and simultaneously drive out Na+ which is the counter cation of TFPB in the sensing layer, preconditioning using an acid solution is effective before the measurement. In the case of the PVC membrane, Na+ in the coated layer can be replaced with H+ by immersing the membrane into an acid solution such as HCl of pH 2. However, in the case when the paper optode was immersed into this acid, the tensile strength of the paper was damaged. Therefore, we attempted to exchange the sodium ion with the hydrogen ion before spin coating for casting the sensing components onto the paper surface so that conditioning by acid might not be needed. This ion-exchange procedure is performed by dissolving an equimolar KD-M11 and TFPB-sodium salt in chloroform, followed by washing with an acid and evaporating the solvent. Consequently, an ion pair formed with protonated KDM11 and TFPB anion could be prepared. Using the paper optode based on this ion pair with the Li+ ionophore (TTD14C4) for the sensing components of the paper optode, the measurements can be performed without acid conditioning. The color change in the paper optode was optimized in the same manner as that for the PVC membrane film optode. QxQy values of the paper optode containing the KD-M11/TFPB ion pair
Figure 6. Curve for the optimization of the dye concentration on the paper optode (A) and the color variations to 10-5-1 M Li+ by simulation based on the optimum dye concentration (C/C0 ) 3.34) (B).
Figure 5. QxQy chromaticity diagram of Li+ test solutions (10-5-1 M Li+, pH 8) obtained by the paper optode with and without KD-S1 screening dye (A) and the color variations in the Li+ test solutions obtained by the paper optode with (B) and without (C) the screening dye. For the paper optode, see the Experimental Section.
and that using KD-M11/TFPB and KD-S1 are shown in Figure 5A. Figure 5B,C shows the resulting colors of these paper optodes measured and reproduced by the digital color analyzer. As shown in Figure 5C, the most remarkable color change with the gray point occurred around 10-3 M Li+, which is suitable for Li+ determination in the manic-depressive therapeutic concentration range. Optimization of Dye Concentration of the Paper Optode. In the previous section, the color change was optimized by emphasizing the hue change. An attempt was then made to optimize the dye concentration in the optode membrane so that the largest color difference could be achieved. As the parameter which indicates the color difference perceived with our eyes, the numerical color difference value of ∆E* is generally used7 which is given by eq 19. The dye concentration in the optode membrane that gives maximum ∆E* can be obtained on the basis of the calculation of ∆E* at several dye concentrations. The value of ∆ log CLi+/∆E* vs Cdye/Cdye,0 for the paper optode without a screening dye is shown in Figure 6A. For calculating the color difference at several dye concentrations, the reflection spectra were transformed into absorption spectra, and the calculation procedure described in the Experimental Section was carried out. The results show that the ∆ log CLi+/∆E* value is expected to become minimum at Cdye/Cdye,0 ) 3.34, which indicates that the adequate concentration is 3.34 times the dye concentration used for the experiment. Figure 6B shows the color change pattern based on this determined concentration of the dye. In this case, the optimization of color variation for the determination of 10-5-1 M Li+ samples can be simulated with a personal computer. Response Characteristics of the Lithium-Ion Paper Optode. For Li+ determination in saliva, Na+ and K+ are the most interfering substances. The ion selectivities of the paper optode relative to these ions were evaluated. The results are shown in
Figure 7. Ion-selectivity study of the Li+ sensitive paper optode for Li+, Na+, and K+(A) and the color variations of Li+, Na+, K+ test solutions (10-5-1 M, pH 8) obtained with the Li+ sensitive paper optode (B). For the paper optode, see the Experimental Section.
Figure 7A, in which the vertical axis represents the ratio of the color difference to the total color difference (∆E*/∆E*tot) in the normalized value. This sensing paper exhibited good selectivity for K+ and Na+ with almost no interference if the sample contains under 10-2 M of these ions. Because the concentrations of both Na+ and K+ in the saliva are about 20 mM, these ions scarcely interfere with Li+ determination using the sensing paper. In fact, no interference was observed when the normal saliva (containing no Li+) included 10-2 M Li+ in the test sample. As shown in Figure 7B, the color variations also indicate that this paper optode did not respond to Na+ and K+ up to 10-2 and 10-1 M, respectively. One of the advantages of the paper optode is its short response time. The paper optode showed a stable color value within 80 seconds (t98: 98% response time) for the same Li+ sample. The Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
473
Figure 8. Resolution study of the paper optode using the digital color analyzer in the determination of the Li+ test solutions (10-510-2 M Li+, pH 7.5). For the paper optode, see the Experimental Section.
PVC film optode requires 5-10 minutes (t98: 98% response time) to show a stable value. After five sequential measurements of 10-3 M Li+ with the ion paper optode and the PVC film optode, the relative standard deviations were 8.5 and 3.8%, respectively. The paper optode can possibly be used for several measurements, but the paper has to be pretreated with 0.1 M HCl, which gradually causes damage to the base cellulose paper. Thus, disposable use is desirable for the paper optode. Resolution of the Digital Color Analyzer. The resolution of the digital color analyzer was investigated with the samples containing 10-5-10-1 M Li+ (pH 7.5). The QxQy values are shown in Figure 8 for solutions with different concentrations of Li+. The color difference could be discriminated visually for a concentration difference of about ten times if the screening method was not used. In contrast, using the digital color analyzer, a concentration difference of ∼0.1 mM can be discriminated at a concentration higher than 10-4 M as shown in Figure 8. The digital color analyzer is suitable for use at home because of its hand-held size and easy operation. Using this handy digital color analyzer with the paper optode, accurate ion measurement would be possible at home. The handy digital color analyzer used in the present investigation is a commercially available product which is not best for analytical chemistry measurements. For more accurate determination, a more precise light source, an optical system, a detector, and suitable software are desired. CONCLUSIONS As we demonstrated in the present report, Digital Color Analysis (DCA) can deal with a greater variety of analytical information than spectrophotometry. In addition, DCA can simulate color changes to develop optimum visual sensors. Utilizing these advantageous characteristics of DCA, we attempted to develop an optical chemical ion sensor (ion optode) device based on DCA and here demonstrated making a simple lithium-ion optode for measuring the manic-depressive treatment concentration (millimolar Li+ concentration) level as a typical example of a DCA-based sensor.
474 Analytical Chemistry, Vol. 72, No. 3, February 1, 2000
For this purpose, the PVC film optode containing a screening dye was developed which produces an effective visual color change between complementary colors orange-gray-purple after the optimization of the color change and the detection range for Li+ based on the theories of color mixing and optode response to the ion. The optode membrane achieves a colorless gray at the therapeutic concentration level of Li+ (around 10-3 M) in saliva. In addition, a paper-based Li+ optode was also developed, for which the response time is (e80 s) faster than that of the PVCbased optode. For more accurate Li+ concentration measurement, the numerical values of the paper color were measured by a handy digital color analyzer, and the digital numerical data were used for Li+ calibration. In this case, the complementary QxQy coordinates are effective for the calibration of its linearity. By transforming color information into numerical values, colorimetric analysis that has been used only as a semiquantitative analysis can serve as an accurate method. Moreover, even if the spectrum is quite complicated to analyze, it is possible to make an accurate determination without the spectrum data by the color changes, on the basis of color difference or digital color library data for the obtained color. DCA can create calibration curves by converting the spectrum into tristimulus values and plotting the color on a chromaticity diagram. DCA can also create a color library for direct calibration and simulate the optimum color changes on the basis of limited experimental color data using a personal computer with suitable software. DCA can optimize sensors based on color changes by utilizing colors as digital and numerical information. Recent electronic technologies for color sensing represented by a color scanner or similar instruments have significant potential as high-resolution instruments for DCA. Thus, the way is opened to develop a new optical analytical method other than spectrophotometry. Especially, the DCA-based sensor has great potential for use as a pointof-care (POC) health check at the bedside in a hospital or at home as an environmental substance check, in situ process check, etc., because of its high conventionality. ACKNOWLEDGMENT The authors thank Mr. Mitsuru Kobayashi of Oji Paper Co. for his help in preparing the special coated paper applied to the paper optode. Partial support of this investigation by the Kanagawa Academy of Science and Technology and the Ministry of Science and Technology is acknowledged. This study was also partially supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture. SUPPORTING INFORMATION AVAILABLE The QxQy calculation data for Mathematica software is available as supplementary material. Received for review June 3, 1999. Accepted October 26, 1999. AC990588W
