- Email: [email protected]
Spectrophotometric determination of trace water in organic solvents with a near infrared absorbing dye
Talanta Talanta 44 (1997) 1949 1958
ELSEVIER
Spectrophotometric determination of trace water in organic solvents with a near infrared absorbing dye Mingshu
Li, G i l b e r t E. P a c e y
~Twmistry Department, Miami UniversiO,, Os[ord, OH 45056, USA Received 26 April 1996: received in revised form 23 September 1996: accepted 23 September 1996
Abstract
A spectrophotometric method for the determination of trace water in organic solvents using a near infrared absorbing dye has been developed. This method is based on the effect that a minor change in polarity of the solvent caused by trace water content determines the extent of aggregation of a near-infrared dye monomer. This change can be detected spectrophotometrically. The calibration curves for methanol, ethanol, and isopropanol were determined. This method has the highest sensitivity (em= 16.73 unit) for water in isopropanol and the lowest sensitivity (em = 2.806 unit) for water in methanol. The correlation coefficient (R) 2 values for the regression lines ranges from 0.990-0.998. The linear range of the method for ethanol is 0.001 0.5%, for isopropanol is 0.001-0.1%, and for methanol is 0.001 1.0%. The limit of detection for ethanol, isopropanol, and methanol are 0.0001, 0.0001, and 0.005% water, respectively. The developed method is sensitive, simple and easy to operate, and the cost of analysis is low. © 1997 Elsevier Science B.V.
Ke),words: Infrared absorbing dye; Organic solvents; Trace water
1. Introduction
The determination of water in organic solvents has been of interest to chemists for many years [1]. The majority of water determinations in organic solvents are performed by the Karl Fisher titration [2]. The Karl Fisher titration method requires special equipment and expertise to obtain good accuracy and precision [3]. The method exhibits interferences from oxidizing agents, unsaturated compounds, and sulfur compounds [4 6]. The Karl Fisher titration method is not considered a low level method [7]. In addition, there are safety concerns about the highly toxic
reagent. Recent refinements in the Karl Fisher method have concentrated on improving accuracy and extension to other solvent systems [2,4,6,8]. The Karl fisher method has been automated using flow-injection analysis [9] where improved precision and accuracy were observed, but the detection limit was a disappointing 0.03%. Other methods that are used include thermal conductivity detection gas chromatography [10]; electrochemical sensors based on thin-film perfluorosulfonate ionomer (PFSI) coated with cellulose triacetate; polyvinyl alcohol (PVA)-H3PO4; or PVA-PFSI-H3PO4 composite films operated in a pulsed voltammetric mode [1]; an organic phase
0039-9140/97/$17.00 ~' 1997 Elsevier Science B.V. All rights reserved. PII S0039-91 40( 96)0211 1 -X
1950
M. Li, G.E. Pacey/Talanta 44 (1997) 1949-1958
enzyme electrode [11] where the enzyme activity has been shown to be strongly dependent upon the water content; an FIA-spectroscopic method [12] based on the reaction between group IV and V metal halides and water; HPLC methods where the water peak was detected by either electric conductivity [5] or optical absorption [13]; Solvatochromic effect [3,14,15]; and a spectrofluorimetric method [7] based on formation of the exciplex of pyridoxal. More recently, FIA-NIR spectrophotometry [16] was used for the direct determination of water in organic solvents by using the O - H stretch absorbance bands. In most of these methods either the detection limits were too high or the method did not work in highly polar solvents. During investigations of near infrared dyes in this laboratory, it was observed that these dyes appeared to be unstable in alcohol solvents at room temperature. However, this instability was not observed in extremely dry alcohols. Upon further investigation, it was discovered that the dye dimerized or polymerized in the alcohol solutions. The extent of polymerization depends on the solvent polarity, temperature, and the basicity. Trace amount of water in the solvent alters the polarity of the solvent and that the polymerization reaction was directly proportional to the water concentration in the alcohol. This paper discusses a new near infrared dye, spectrophotometric method for the determination of low level water in polar organic solvents.
2. Experimental 2.1. Chemicals
Methanol was purchased from Fisher Scientific (acetone free, absolute methanol with water concentration of 0.02%). This methanol was further purified as suggested by Lund and Bjerrum [17]. A 200 ml volume of methanol reacted with 24 g of magnesium turnings for 6 h (reaction is vigorous under heating). A 3 1 amount of methanol was added and the mixture was refluxed for 5 h and then the methanol was distilled into 500 ml glass bottles containing 5 g of molecular sieves. The
first 50 ml of collected methanol was discarded. This procedure produced about 2 1 of dry methanol. Ethanol was obtained from Quantum Chemical Corporation (200 proof dehydrated alcohol with water concentration of 0.02%) and was further purified with the same method as used for methanol. Isopropanol was purchased from fisher Scientific (suitable for electronic use, water concentration was 0.04%) and was dried with molecular sieves for 2 weeks. 2.2. Synthesis of the near infrared dye
The near infrared dye, 2-[4'-chloro-7-(3"-ethyl2"-benzothiazothiazolinylidene)-3',5'-(l",3"-propanediyl) - 1'3'5' - heptatrien - 1' - yl] - 3 - ethylbenzothiazolium iodide as shown in Fig. 1, was prepared as the bromide analog [18]: ethyl iodide and 2-methylbenzothiazole were refluxed in dimethlyformamide at 153°C overnight. The mixture was refrigerated for 2 h and then slowly added diethyl ether to initiate crystallization of the product. The crystals were filtered, washed with diethyl ether, and dried under reduced pressure. The dried product, 3-ethyl-2-methyl-benzothiazolium iodide, was refluxed with N-[(3-(anilinomethylene)-2chloro-l-cyclohene-l-yl)-methylene]analene momohydrochloride in ethanol under the presence of anhydrous sodium hydrochloride for 1 h and kept in a refrigerator for 1 h. The crystals were filtered off, washed with distilled water, benzene, and ether and dried under a vacuum. The product was recrystallized in ethanol and dried under a vacuum at 50°C for 6 h. The nuclear magnetic resonance spectrometry (NMR) and matrix assisted laser desorption ionization mass spectrometry (MALDI-MS) spectra were presented in Fig. 2 and Fig. 3.
CH~
CH3 Fig. 1. Structure of dye 1.
M. Li, G.E. Pacey / Talanta 44 (1997) 1949-.-1958
1951
a.
4-
s ~ s
Cl 2'
"[
L
r
s---U~
t I
CH~
I
CH~
2"
>HSO
HzO
- CH 3
4,7
III
|tt
'] ....
' ....
] "J'''
....
8.0
I .... 7,0
' ....
I .... 5.0
' ....
I ....
' ' ' ' ' l
5.0
~L.0
....
' ....
I ......... 3.0
I ....
' ....
2.0
I .... l,O
' ....
I''" 0.0
PPH
Fig. 2. N M R spectrum of dye 1. 200 MHZ NMR, DMSO as solvent.
2.3. Safety consideration Methanol, ethanol, and flammable and toxic by with skin, and swallowed. at room temperature and suspect agent.
isopropanol are highly inhalation, in contact Chloroform is volatile is highly toxic cancer
2.4. Apparatus A Hewlett Packard 8452A Diode Array Spec-
trophotometer was used to scan the spectra and absorbance measurements. Tekmar utility bath was used for heating and temperature control. All glassware was cleaned with detergent, rinsed with distilled water, rinsed with acetone, dried in the oven at 100°C for 1 h and cooled in a dedicator before use. M A L D I - M S spectra were obtained with the Bruker Reflex II T O F - M S mass spectrometer equipped with a N d : Y A G laser model Minilase-10. N M R spectra were acquired with Bruker AC-200 M H Z N M R spectrometer.
M. Li, G.E. Pacey /'Talanta 44 (1997)1949-1958
1952
a.i.
~
3500
3000
S
CI
S ----~'~'l
CH 3 [C2sH2aSzN2CI] +
[M-C2H~+2H]2~ 2500
CH~ FW = 491. I
233. 1 I
2000
1500
49
.l
i000
500
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
m/z
Fig. 3. M A L D I - M S spectrum of dye 1. N d : Y A G laser (355 nm), matrix was :~-cyano-4-hydroxy cinnamic acid, and CH~CN as the solvent. N u m b e r of shots was 40.
2.5. Reagents
2.6. Procedure
1.125 × l0 --3 M dye solution was made in m e t h a n o l - c h l o r o f o r m (50:50) solution and was kept in the refrigerator. U p o n storage, aggregates of the dye were formed and the solution was no longer useful for the analysis. The solution can be dept up to 2 months in a refrigerator and 1 week at room temperature. A 1.0 × 10-3 M potassium hydroxide solution was prepared as follows: dissolve a certain amount of potassium hydroxide in dried ethanol, filter the solution, and dilute to the volume with ethanol. The standard stock solution of 1% water was made with dried solvents and deionized water.
A 10 ml volumetric flask was filled about half full with the sample solution, add 0.2 ml of K O H solution and 0.2 ml of dye solution, and then dilute with the sample to the volume and mix well. The solution mixture was placed into a hot water bath with the cap on at 60-70°C for 10 min and cooled at r o o m temperature for 15 min. The reduction of absorbance was measured at about 800 nm depending on the solvents. The water concentration in the solvents was derived from the calibration curve. The calibration curve was obtained by measuring the absorbance of 0, 50, 100, 200, 300, and 500 p p m (v/v) standard solutions.
M. Li, G.E. Pac~:v ,, Talanta 44 (1997) 1949 1958
These standard solutions were prepared in 10 ml volumetric flasks adding 0, 0.05, 0.10, 0.20, 0.30 and 0.50 ml water working solution (10 000 ppm) and following the same procedure of sample analysis described above.
3. Results and discussion
Near-infrared dyes absorb light between 600 1200 nm and exhibit large molar absorptivities ( ~ 200000 Abs cm-1 M 1), thereby enhancing their potential for low level detection. In addition, most NIR dyes are strongly fluorescent in solution [18-20]. When NIR dyes are used for analytical applications, detection can be carried out in the longer wavelength regions creating a reduced background and interferences from coexisting species. Despite these attributes and the large number of NIR dyes available, the use of NIR dyes has not gained much attention in analytical applications. Dye I has been used as a hydrophobicity probe for aliphatic alcohols and other water-miscible organic solvents [18] with water/organic solvent ratio ranged from 50-100 (pure water). This dye has a molar absorptivity of 1.83 x 105 [18] and absorbance maximum wavelength of 800 nm in ethanol. Absorption spectra of the dye changed significantlydepending on the solvent hydrophobicity. With higher concentration (over 50%) methanol, the maximum absorption band of the dye solution was at 811 nm. When the methanol concentration was decreased from 50% to 0 (pure water), the absorption peak at 811 nm disappeared and a new peak at 698 nm appeared. It was suggested that the observed change in the NIR absorption spectra is a result of dye dimerization [18]. It is known that aqueous NIR dye solutions, especially at higher dye concentrations, exhibit an absorption band different from the absorption band that can be observed in less concentrated solutions or in less polar organic solvents. These spectral changes have long been attributed to dye molecule aggregation. The NIR dyes also display this type of behavior in water, where they tend to form dimmers (2 = 698 nm) or higher aggregates (,i = 4 5 0 - 5 0 0 nm) because of the strong dispersion forces associated with the high polarizability of the
1953
polymethane chain [21]. Experimental results in this lab revealed that dimerization conditions for the cyanine dyes depend on the following factors: dye concentration, water concentration, temperature, reaction time, exposer to light, and the basicity of the solution. The dependence of the dimer formation on the dye concentration and water content in the solution has been proven by West and Pearce [21] and Patonay et al. [18]. Dimmers usually do not form in pure organic solvents. The dimmersband and the higher-aggregate band were observed at 698 and 450 nm, respectively, in 1 x 10 5 M aqueous solutions. Fig. 4 presents the spectra obtained in this lab for the dye in 60% water/ethanol solution with increasing basicity and reaction time. Five peaks are observable in the wavelength range of 320 820 nm. The peak at 800 nm is the dye monomer, 692 nm is the dimer peak, and the rest are the peaks for higher aggregates of the dye. As the solution basicity increased, the 656 nm and 380 nm aggregate peaks were decreased and the 496 nm peak increased, the 692 nm dimer peak initially persisted and then started to decrease when the 656 nm peak disappeared. As the 656, 698, 380 nm aggregate peaks decreased, the 800 nm monomer peak increased until the dimer peak became a shoulder of the monomer peak and then decreased with increasing basicity. These changes in the spectra indicate that dimers form in neutral solutions at high concentrations of water in aliphatic alcohols. When a small amount of base is added, the polymer at 656 nm and the dimer decompose to their monomer. When the basicity is further increased, polymerization to higher aggregates happens. Using the optimized conditions for the method, the 656 and 698 nm peaks did not exist (Fig. 5), instead, higher aggregates represented by the peak at 496 nm were formed. There are at least three isosebestical points at 575, 405 and 363 nm. These points indicate that there are three equilibrium systems in the solution. In the solution without added basicity (spectrum 1 in Fig. 4(a)), dimers and the aggregates at 656 nm were formed and there is equilibrium between the aggregate of 656 nm and the monomer. With increasing basicity (spectra 2 5 in Fig. 4(a)), the dye is distributed between the dimers and the aggregates at 496 nm. When the
M. Li, G.E. Pacey /Talanta 44 (1997)1949-1958
1954
a 1.91G"~.
I.S331-
=_
1. I~99
0 Clm
14.5
O. 7865"7
'3 '2 O.
38329-
O.0000. 400
500
600 ~IAVELEMGTH
700
800
b i •8 7 4 0 -
I • #9~9.
==
I . 1158-
9
0
O.7366S.
5 6 7
O. 3 5 7 5 3-
8 9
-0. 021Sg 400
SO0
600 UAVELEMGTH
700
800
Fig. 4. Absorption spectra of the dye in 60% water/ethanol solution: (a) with different K O H concentration: 1-neutral solution; 2-2.8 x 10 4 M; 3 - 8 . 4 x 10 . 4 M; 4 - 1 . 4 x 10 - 3 M; 5-2.8 x 10 3 M; (b) under different reaction time: 5 - 5 min; 6 10 rain; 7 - 2 0 rain; 8 - 3 5 rain. Spectrum 9 was obtained after heating the solution of 2.8 x 1 0 - 3 M in K O H 10 rain at 70°C.
basicity of the solution was further increased or the basic solution sat for long time (spectra 6 - 8 in Fig. 4(b)), the equilibrium shifted toward the monomer and the polymer at 496 nm. The polymerization rate of the monomer to the aggregate at 496 nm is slow. It took 30 min to convert spectrum 5 - 8 in Fig. 4(b) at room temperature. But when the solution was heated for 10 min at 70°C, the conversion was completed almost immediately (spectrum 9 in Fig. 4(b)).
When the system was going through higher-aggregation and a reduction of the monomer peak was observed, the sensitivity was higher than the dimerized system. This was one of the reasons that the determination of trace amount of water in organic solvents is possible in this method. Representative absorption spectra of the dye in dry ethanol (spectrum 1) and in 0.05% water in ethanol solution (spectrum 2) are presented in Fig. 5. The peak reduction at 800 nm was used for
M. Li, G.E. Pacey /Talanta 44 (1997) 1949-1958
the quantitative analysis of trace water concentrations in organic solvents. As it was mentioned earlier, besides the dye concentration and water content in the system, basicity of the solution is an important factor for the process being dimerization or higher-aggregate formation. It can be seen in Fig. 4(a) that dimers were formed in neutral ethanol solutions and with increasing basicity the dimer peaks would disappear. When the solution was basic and heated at a certain temperature for a period of time, even a little change in polarity by trace amounts of water in the solvent caused aggregation of the dye in the solution. The extent of aggregation was proportional to the water concentration in the solvent. Both monomer and H-aggregate peaks can be used for the measurement, but he monomer peak at 800 nm is the most sensitive. Basicity of the solution was adjusted by adding different amount of KOH/ethanol solution. Reagent grade K O H was kept in a desiccator and the carbonate impurities were insoluble in alcohols and were removed by filtration. The result of the basicity experiment is shown in Fig. 6. With increasing K O H concentration from 0-8.0 × 10 4 M, the net absorbance of 0.05% water/ethanol solution increased, reaching a maximum at 4 x 10 4 M of KOH. Therefore, the K O H concentration of 4 x 10 -4 M was chosen. It was pointed out [18] that formation of H-aggregates of the dye was observable in 1 x 10 4 M aqueous solution. If the concentration is lower
1
6 70555
t.)
0 56235
l--
0 41916
2
o 1/1
.13
<
0 27593
0 13273
0 01049 4OO
5O0
600
7OO
8O0
Wavelength (nm) Fig. 5. Absorption spectra of the dye in dry ethanol [1] and in 0.05% water/ethanol solution [2].
1955
04,
o3i "'Q 0 2
< 01 B
0
.
0
2
.
.
.
4
.
.
5
.
.
.
8
10
KOH concentration, M X 10000 Fig. 6. Effect of KOH concentration on the absorbance of the dye in water:ethanol solution.
than 1.25 x 10 5 M, then the higher aggregate band diminished. In the range of dye concentration tested, no dimer peaks were seen. When the dye concentration increased from 0-1.3 x 10 4 M, the absorbance for both blank and 0.2% water/ethanol solutions increased (Fig. 7(a)) and reached the maximum absorbance. The rates of absorbance increases were different. Blank absorbance was increasing faster than the water solution. The absorbance difference between blank and water solution was biggest at the dye concentration of 6.8 × 10 5 (Fig. 7(b)). Since at this concentration the absorbance of dry solvent was about three absorbance unit, accurate absorbance measurement is questionable. Therefore, a dye concentration of 4.5 x 10--5 M was selected. Heating prevented dimer formation and catalyzed the higher aggregation process. Heating was also the main factor for the dye responding to trace water concentration. At room temperature, the lowest water concentration measurable was 1% in ethanol after 45 rain. When the solution was heated, there were some changes in the spectral properties of the dye (Fig. 8). Both in dry ethanol and 0.05% water/ethanol solution, the peak at 800 nm shifted to 816 nm and the peak at 460 nm move to 495 nm when the temperature increased from 43.2-62.3°C. The peak reduction at 800 nm was much bigger in the water/ethanol solution than in dry ethanol. When the temperature of the solution was increased, the absorbance for both blank and 0.05% water/ethanol solution decreased (Fig. 9). After about 50°C, the blank
M. Li, G.E. Pacey /Talanta 44 (1997) 1949-1958
1956
absorbance persisted and remained constant up to 85°C. The absorbance of 0.05% water solution was further decreased and kept constant between 60-85°C. The maximum absorbance difference occurred between 60-70°C. Therefore, heating at 60-70°C was utilized. The aggregation reaction under elevated temperature was a reversible process. Heating provided the energy for shifting the reaction equilibrium toward the formation of the aggregates. When the solution was cooled at room temperature, absorption peak of the monomer increased with decreasing temperature and was stable after 15 min at room temperature. Formation of the aggregates at 496 nm from their monomers was a relatively slow process. 4
-
O ¢0
~2 O
< • blank
e 0.05
%
1
,, 0
o ±
1
005
01
0.15
Thousandths
Dye solution, M 2,
0.7500
I
2~
0.5623 0.4191
4
0.2759
4
0.1327 0.0(X)0
500
400
700
600
800
Wavelength (nm) Fig. 8. Absorbance spectra of the dye at different temperature and water contents. Spectra 1 and 2 are for dye in dry ethanol at 43.2 and 62.3°C, respectively. Spectra 3 and 4 are for the dye in 0.05% water in ethanol solution at 43.2 and 62.3°C, respectively.
After 30 min at room temperature, the conversion process was still active (Fig. 4(b), spectrum 8). At higher temperature, aggregate formation immediately complete in higher concentration of water/ ethanol solution. When water concentration is very low (0.1-0.001%), heating was needed to catalyze the aggregation process. Fig. 9 is the result of the experiment to determine the reaction time needed at 70°C. With increasing heating time from 0 - 3 0 min, the absorbance of the blank solution was decreased at almost a constant rate. But 0.05% water solution exhibited rapid reduction in absorbance between 0 - 1 0 min heating at 70°C was utilized, Fig. 10. Calibration curves (Fig. 11) were obtained under the optimized conditions for methanol,
• blank
,>
,~ 0.05 % H 2 0
08
cD
O ¢fO
o 06 < ¢'~ 04 < O
ii
o
01
005
O
02
O
_
0.15
Thousandths
Concentration of Dye, M*1000 Fig. 7. Effect of the dye concentration. (a) Absorbance of the dye in dry ethanol and 0.05% water/ethanol solution. (b) Net absorbance of the 0.05% water solution.
0 20
•
L
•
40
60
80
100
Temperature, C Fig. 9. Temperature effect of the dye in dry ethanol and 0.05% water/ethanol solution.
M. Li, G.E. Pace)'/Talanta 44 (1997) 1949 1958
• Blank
1957
1 ,
,~ 0.05% water
e 08
08
• methanol • isopropanol
0.6--
e~
,<
ethanol
06
~o4
04 @
¢,
A
02--
i
_
_
-sl
0
0
10
20
30
40
Reaction Time, rain
0
002
004
006
Water Concentration, v/v %
Fig. 10. Effect of heating time on the dye absorbance dry ethanol and 0.05% water/ethanol solution.
Fig. 11. Calibration curves.
ethanol, and isopropanol and the results of the curves are summarized in Table 1. This method has the highest sensitivity (era= 16.73 unit) for water in isopropanol and the lowest sensitivity (em = 2.806 unit) for water in methanol. The correlation coefficient (R) 2 values for the regression lines ranged from 0.990-0.998. The linear range of the method for ethanol is 0.001 0.5%, for isopropanol is 0.001-0.1%, and for methanol is 0.001-1.0%. The limit of detection for ethanol, isopropanol, and methanol are 0.0001, 0.0001 and 0.005% water with relative standard deviations of 2.62, 2.25 and 2.59% at 0.05% water concentration, respectively. The limits of detection depend mostly on the preparation of dry solvents.
time that trace amount of water in polar organic solvents can be determined by the fact that a minor change in polarity of the solvent caused by trace water content determines the extent of aggregation of a near-infrared dye monomer, which can be detected easily by measuring the absorbance reduction at 800 nm. By using a NIR dye, the detection was moved to longer wavelength where background interferences were minimized. Sensitivity of the method was very high, owing to the high molar absorptivity of the dye solution and formation of a higher aggregate in the system. The limits of detection were lower than that of most reported methods. The developed method is simple, easy to operate, and the cost of analysis is low. Current work in our laboratory concerns the possibility of applying this method to different organic solvents.
4. Conclusions
In this study we demonstrated for the first
Table l Results of calibration curves Solvent
Equation A = a + b [C]
R -~
Linear range v/v, %
LOD Wv, %
Methanol
a b a b a b
0.991
0.001
1.00
0.005
0.998
0.001-0.10
0.001
0.995
0.001
0.001
Ethanol Isopropanol
= = = = = =
0.002649 ± 0.004509 2.806 ± 0.117 0.006375 ± 0.006375 7.825 4- 0.153 0.03783 4- 0.03462 16.73 + 0.82
0.50
1958
M. Li, G.E, Pacey / Talanta 44 (1997) 1949-1958
References [1] T. Mitchell Jr. and D.M. Smith, Aquametry, part Ili (The Karl Fisher Reagent), A Treatise on Methods for the Determination of Watter, A Wiley-Interscience Publication, Wiley, 1977. [2] I. Nordin-Andersson and A. Cedergren, Anal. Chem., 57 (1985) 2571-2575. [3] H. Langhals, Anal. Lett., 23(12) (1990) 2243 2258. [4] A. Cedergren and C. Oradd, Anal. Chem., 66 (1994) 2010-2016. [5] T.S. Stevens and K.M. Chritz, Anal. Chem., 59 (1987) 1716-1720. [6] C. Oradd and A. Cedergren, Anal. Chem., 67 (1995) 999 1004. [7] Y. Ci and X. Jia, Talanta, 31(7) (1984) 556-558. [8] C. Oradd and A. Cedergren, Anal. Chem., 66 (1994) 2603-2607. [9] C. Liang, P. Vacha and W.E. Van Der Linden, Talanta, 35(1) (1988) 59-61. [10] T. Mitchell Jr. and D.M. Smith, Aquametry, part I, A
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
Treatise on Methods for the Determination of Water, Wiley-Interscience Publication, Wiley, 1977. J. Wang and A. Reviejo, J. Anal. Chem. 65 (1993) 845-847. J. Rhee, P.K. Dasgupta and D, Olson, Anal. Chim. Acta, 220 (1989) 55 63. J.S. Fritz and J. Chert, Am. Lab., July (1991) 24J-24Q. S. Kumoi, H. Kobayashi and K. Ueno, Talanta, 19 (1972) 505-513. S. Kumoi, K. Oyama, T. Yano and H. Kobayashi, Talanta, 17 (1970) 319 327. S. Garrigues, M. Gallignani and M. Guardia, Anal. Chim. Acta, 281 (1993) 259 264. H. Lund and J. Bjerrum, Ber. Stsch. Chem. Ges., 64 (1931) 210. G. Patonay, M.D. Antoine, S. Devanathan and L. Strekowski, Appl. Spectro., 45(3) (1991) 457 461. G. Patony, Advances in near-infrared measurements, 1 (1993) 113 138. M. Matsuoka, Infrared Absorbing Dyes, Plenum Press, 1990. W. West and S. Pearce, J. Phys. Chem., 69(6) 1965 1894-1903.
ELSEVIER
Spectrophotometric determination of trace water in organic solvents with a near infrared absorbing dye Mingshu
Li, G i l b e r t E. P a c e y
~Twmistry Department, Miami UniversiO,, Os[ord, OH 45056, USA Received 26 April 1996: received in revised form 23 September 1996: accepted 23 September 1996
Abstract
A spectrophotometric method for the determination of trace water in organic solvents using a near infrared absorbing dye has been developed. This method is based on the effect that a minor change in polarity of the solvent caused by trace water content determines the extent of aggregation of a near-infrared dye monomer. This change can be detected spectrophotometrically. The calibration curves for methanol, ethanol, and isopropanol were determined. This method has the highest sensitivity (em= 16.73 unit) for water in isopropanol and the lowest sensitivity (em = 2.806 unit) for water in methanol. The correlation coefficient (R) 2 values for the regression lines ranges from 0.990-0.998. The linear range of the method for ethanol is 0.001 0.5%, for isopropanol is 0.001-0.1%, and for methanol is 0.001 1.0%. The limit of detection for ethanol, isopropanol, and methanol are 0.0001, 0.0001, and 0.005% water, respectively. The developed method is sensitive, simple and easy to operate, and the cost of analysis is low. © 1997 Elsevier Science B.V.
Ke),words: Infrared absorbing dye; Organic solvents; Trace water
1. Introduction
The determination of water in organic solvents has been of interest to chemists for many years [1]. The majority of water determinations in organic solvents are performed by the Karl Fisher titration [2]. The Karl Fisher titration method requires special equipment and expertise to obtain good accuracy and precision [3]. The method exhibits interferences from oxidizing agents, unsaturated compounds, and sulfur compounds [4 6]. The Karl Fisher titration method is not considered a low level method [7]. In addition, there are safety concerns about the highly toxic
reagent. Recent refinements in the Karl Fisher method have concentrated on improving accuracy and extension to other solvent systems [2,4,6,8]. The Karl fisher method has been automated using flow-injection analysis [9] where improved precision and accuracy were observed, but the detection limit was a disappointing 0.03%. Other methods that are used include thermal conductivity detection gas chromatography [10]; electrochemical sensors based on thin-film perfluorosulfonate ionomer (PFSI) coated with cellulose triacetate; polyvinyl alcohol (PVA)-H3PO4; or PVA-PFSI-H3PO4 composite films operated in a pulsed voltammetric mode [1]; an organic phase
0039-9140/97/$17.00 ~' 1997 Elsevier Science B.V. All rights reserved. PII S0039-91 40( 96)0211 1 -X
1950
M. Li, G.E. Pacey/Talanta 44 (1997) 1949-1958
enzyme electrode [11] where the enzyme activity has been shown to be strongly dependent upon the water content; an FIA-spectroscopic method [12] based on the reaction between group IV and V metal halides and water; HPLC methods where the water peak was detected by either electric conductivity [5] or optical absorption [13]; Solvatochromic effect [3,14,15]; and a spectrofluorimetric method [7] based on formation of the exciplex of pyridoxal. More recently, FIA-NIR spectrophotometry [16] was used for the direct determination of water in organic solvents by using the O - H stretch absorbance bands. In most of these methods either the detection limits were too high or the method did not work in highly polar solvents. During investigations of near infrared dyes in this laboratory, it was observed that these dyes appeared to be unstable in alcohol solvents at room temperature. However, this instability was not observed in extremely dry alcohols. Upon further investigation, it was discovered that the dye dimerized or polymerized in the alcohol solutions. The extent of polymerization depends on the solvent polarity, temperature, and the basicity. Trace amount of water in the solvent alters the polarity of the solvent and that the polymerization reaction was directly proportional to the water concentration in the alcohol. This paper discusses a new near infrared dye, spectrophotometric method for the determination of low level water in polar organic solvents.
2. Experimental 2.1. Chemicals
Methanol was purchased from Fisher Scientific (acetone free, absolute methanol with water concentration of 0.02%). This methanol was further purified as suggested by Lund and Bjerrum [17]. A 200 ml volume of methanol reacted with 24 g of magnesium turnings for 6 h (reaction is vigorous under heating). A 3 1 amount of methanol was added and the mixture was refluxed for 5 h and then the methanol was distilled into 500 ml glass bottles containing 5 g of molecular sieves. The
first 50 ml of collected methanol was discarded. This procedure produced about 2 1 of dry methanol. Ethanol was obtained from Quantum Chemical Corporation (200 proof dehydrated alcohol with water concentration of 0.02%) and was further purified with the same method as used for methanol. Isopropanol was purchased from fisher Scientific (suitable for electronic use, water concentration was 0.04%) and was dried with molecular sieves for 2 weeks. 2.2. Synthesis of the near infrared dye
The near infrared dye, 2-[4'-chloro-7-(3"-ethyl2"-benzothiazothiazolinylidene)-3',5'-(l",3"-propanediyl) - 1'3'5' - heptatrien - 1' - yl] - 3 - ethylbenzothiazolium iodide as shown in Fig. 1, was prepared as the bromide analog [18]: ethyl iodide and 2-methylbenzothiazole were refluxed in dimethlyformamide at 153°C overnight. The mixture was refrigerated for 2 h and then slowly added diethyl ether to initiate crystallization of the product. The crystals were filtered, washed with diethyl ether, and dried under reduced pressure. The dried product, 3-ethyl-2-methyl-benzothiazolium iodide, was refluxed with N-[(3-(anilinomethylene)-2chloro-l-cyclohene-l-yl)-methylene]analene momohydrochloride in ethanol under the presence of anhydrous sodium hydrochloride for 1 h and kept in a refrigerator for 1 h. The crystals were filtered off, washed with distilled water, benzene, and ether and dried under a vacuum. The product was recrystallized in ethanol and dried under a vacuum at 50°C for 6 h. The nuclear magnetic resonance spectrometry (NMR) and matrix assisted laser desorption ionization mass spectrometry (MALDI-MS) spectra were presented in Fig. 2 and Fig. 3.
CH~
CH3 Fig. 1. Structure of dye 1.
M. Li, G.E. Pacey / Talanta 44 (1997) 1949-.-1958
1951
a.
4-
s ~ s
Cl 2'
"[
L
r
s---U~
t I
CH~
I
CH~
2"
>HSO
HzO
- CH 3
4,7
III
|tt
'] ....
' ....
] "J'''
....
8.0
I .... 7,0
' ....
I .... 5.0
' ....
I ....
' ' ' ' ' l
5.0
~L.0
....
' ....
I ......... 3.0
I ....
' ....
2.0
I .... l,O
' ....
I''" 0.0
PPH
Fig. 2. N M R spectrum of dye 1. 200 MHZ NMR, DMSO as solvent.
2.3. Safety consideration Methanol, ethanol, and flammable and toxic by with skin, and swallowed. at room temperature and suspect agent.
isopropanol are highly inhalation, in contact Chloroform is volatile is highly toxic cancer
2.4. Apparatus A Hewlett Packard 8452A Diode Array Spec-
trophotometer was used to scan the spectra and absorbance measurements. Tekmar utility bath was used for heating and temperature control. All glassware was cleaned with detergent, rinsed with distilled water, rinsed with acetone, dried in the oven at 100°C for 1 h and cooled in a dedicator before use. M A L D I - M S spectra were obtained with the Bruker Reflex II T O F - M S mass spectrometer equipped with a N d : Y A G laser model Minilase-10. N M R spectra were acquired with Bruker AC-200 M H Z N M R spectrometer.
M. Li, G.E. Pacey /'Talanta 44 (1997)1949-1958
1952
a.i.
~
3500
3000
S
CI
S ----~'~'l
CH 3 [C2sH2aSzN2CI] +
[M-C2H~+2H]2~ 2500
CH~ FW = 491. I
233. 1 I
2000
1500
49
.l
i000
500
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
m/z
Fig. 3. M A L D I - M S spectrum of dye 1. N d : Y A G laser (355 nm), matrix was :~-cyano-4-hydroxy cinnamic acid, and CH~CN as the solvent. N u m b e r of shots was 40.
2.5. Reagents
2.6. Procedure
1.125 × l0 --3 M dye solution was made in m e t h a n o l - c h l o r o f o r m (50:50) solution and was kept in the refrigerator. U p o n storage, aggregates of the dye were formed and the solution was no longer useful for the analysis. The solution can be dept up to 2 months in a refrigerator and 1 week at room temperature. A 1.0 × 10-3 M potassium hydroxide solution was prepared as follows: dissolve a certain amount of potassium hydroxide in dried ethanol, filter the solution, and dilute to the volume with ethanol. The standard stock solution of 1% water was made with dried solvents and deionized water.
A 10 ml volumetric flask was filled about half full with the sample solution, add 0.2 ml of K O H solution and 0.2 ml of dye solution, and then dilute with the sample to the volume and mix well. The solution mixture was placed into a hot water bath with the cap on at 60-70°C for 10 min and cooled at r o o m temperature for 15 min. The reduction of absorbance was measured at about 800 nm depending on the solvents. The water concentration in the solvents was derived from the calibration curve. The calibration curve was obtained by measuring the absorbance of 0, 50, 100, 200, 300, and 500 p p m (v/v) standard solutions.
M. Li, G.E. Pac~:v ,, Talanta 44 (1997) 1949 1958
These standard solutions were prepared in 10 ml volumetric flasks adding 0, 0.05, 0.10, 0.20, 0.30 and 0.50 ml water working solution (10 000 ppm) and following the same procedure of sample analysis described above.
3. Results and discussion
Near-infrared dyes absorb light between 600 1200 nm and exhibit large molar absorptivities ( ~ 200000 Abs cm-1 M 1), thereby enhancing their potential for low level detection. In addition, most NIR dyes are strongly fluorescent in solution [18-20]. When NIR dyes are used for analytical applications, detection can be carried out in the longer wavelength regions creating a reduced background and interferences from coexisting species. Despite these attributes and the large number of NIR dyes available, the use of NIR dyes has not gained much attention in analytical applications. Dye I has been used as a hydrophobicity probe for aliphatic alcohols and other water-miscible organic solvents [18] with water/organic solvent ratio ranged from 50-100 (pure water). This dye has a molar absorptivity of 1.83 x 105 [18] and absorbance maximum wavelength of 800 nm in ethanol. Absorption spectra of the dye changed significantlydepending on the solvent hydrophobicity. With higher concentration (over 50%) methanol, the maximum absorption band of the dye solution was at 811 nm. When the methanol concentration was decreased from 50% to 0 (pure water), the absorption peak at 811 nm disappeared and a new peak at 698 nm appeared. It was suggested that the observed change in the NIR absorption spectra is a result of dye dimerization [18]. It is known that aqueous NIR dye solutions, especially at higher dye concentrations, exhibit an absorption band different from the absorption band that can be observed in less concentrated solutions or in less polar organic solvents. These spectral changes have long been attributed to dye molecule aggregation. The NIR dyes also display this type of behavior in water, where they tend to form dimmers (2 = 698 nm) or higher aggregates (,i = 4 5 0 - 5 0 0 nm) because of the strong dispersion forces associated with the high polarizability of the
1953
polymethane chain [21]. Experimental results in this lab revealed that dimerization conditions for the cyanine dyes depend on the following factors: dye concentration, water concentration, temperature, reaction time, exposer to light, and the basicity of the solution. The dependence of the dimer formation on the dye concentration and water content in the solution has been proven by West and Pearce [21] and Patonay et al. [18]. Dimmers usually do not form in pure organic solvents. The dimmersband and the higher-aggregate band were observed at 698 and 450 nm, respectively, in 1 x 10 5 M aqueous solutions. Fig. 4 presents the spectra obtained in this lab for the dye in 60% water/ethanol solution with increasing basicity and reaction time. Five peaks are observable in the wavelength range of 320 820 nm. The peak at 800 nm is the dye monomer, 692 nm is the dimer peak, and the rest are the peaks for higher aggregates of the dye. As the solution basicity increased, the 656 nm and 380 nm aggregate peaks were decreased and the 496 nm peak increased, the 692 nm dimer peak initially persisted and then started to decrease when the 656 nm peak disappeared. As the 656, 698, 380 nm aggregate peaks decreased, the 800 nm monomer peak increased until the dimer peak became a shoulder of the monomer peak and then decreased with increasing basicity. These changes in the spectra indicate that dimers form in neutral solutions at high concentrations of water in aliphatic alcohols. When a small amount of base is added, the polymer at 656 nm and the dimer decompose to their monomer. When the basicity is further increased, polymerization to higher aggregates happens. Using the optimized conditions for the method, the 656 and 698 nm peaks did not exist (Fig. 5), instead, higher aggregates represented by the peak at 496 nm were formed. There are at least three isosebestical points at 575, 405 and 363 nm. These points indicate that there are three equilibrium systems in the solution. In the solution without added basicity (spectrum 1 in Fig. 4(a)), dimers and the aggregates at 656 nm were formed and there is equilibrium between the aggregate of 656 nm and the monomer. With increasing basicity (spectra 2 5 in Fig. 4(a)), the dye is distributed between the dimers and the aggregates at 496 nm. When the
M. Li, G.E. Pacey /Talanta 44 (1997)1949-1958
1954
a 1.91G"~.
I.S331-
=_
1. I~99
0 Clm
14.5
O. 7865"7
'3 '2 O.
38329-
O.0000. 400
500
600 ~IAVELEMGTH
700
800
b i •8 7 4 0 -
I • #9~9.
==
I . 1158-
9
0
O.7366S.
5 6 7
O. 3 5 7 5 3-
8 9
-0. 021Sg 400
SO0
600 UAVELEMGTH
700
800
Fig. 4. Absorption spectra of the dye in 60% water/ethanol solution: (a) with different K O H concentration: 1-neutral solution; 2-2.8 x 10 4 M; 3 - 8 . 4 x 10 . 4 M; 4 - 1 . 4 x 10 - 3 M; 5-2.8 x 10 3 M; (b) under different reaction time: 5 - 5 min; 6 10 rain; 7 - 2 0 rain; 8 - 3 5 rain. Spectrum 9 was obtained after heating the solution of 2.8 x 1 0 - 3 M in K O H 10 rain at 70°C.
basicity of the solution was further increased or the basic solution sat for long time (spectra 6 - 8 in Fig. 4(b)), the equilibrium shifted toward the monomer and the polymer at 496 nm. The polymerization rate of the monomer to the aggregate at 496 nm is slow. It took 30 min to convert spectrum 5 - 8 in Fig. 4(b) at room temperature. But when the solution was heated for 10 min at 70°C, the conversion was completed almost immediately (spectrum 9 in Fig. 4(b)).
When the system was going through higher-aggregation and a reduction of the monomer peak was observed, the sensitivity was higher than the dimerized system. This was one of the reasons that the determination of trace amount of water in organic solvents is possible in this method. Representative absorption spectra of the dye in dry ethanol (spectrum 1) and in 0.05% water in ethanol solution (spectrum 2) are presented in Fig. 5. The peak reduction at 800 nm was used for
M. Li, G.E. Pacey /Talanta 44 (1997) 1949-1958
the quantitative analysis of trace water concentrations in organic solvents. As it was mentioned earlier, besides the dye concentration and water content in the system, basicity of the solution is an important factor for the process being dimerization or higher-aggregate formation. It can be seen in Fig. 4(a) that dimers were formed in neutral ethanol solutions and with increasing basicity the dimer peaks would disappear. When the solution was basic and heated at a certain temperature for a period of time, even a little change in polarity by trace amounts of water in the solvent caused aggregation of the dye in the solution. The extent of aggregation was proportional to the water concentration in the solvent. Both monomer and H-aggregate peaks can be used for the measurement, but he monomer peak at 800 nm is the most sensitive. Basicity of the solution was adjusted by adding different amount of KOH/ethanol solution. Reagent grade K O H was kept in a desiccator and the carbonate impurities were insoluble in alcohols and were removed by filtration. The result of the basicity experiment is shown in Fig. 6. With increasing K O H concentration from 0-8.0 × 10 4 M, the net absorbance of 0.05% water/ethanol solution increased, reaching a maximum at 4 x 10 4 M of KOH. Therefore, the K O H concentration of 4 x 10 -4 M was chosen. It was pointed out [18] that formation of H-aggregates of the dye was observable in 1 x 10 4 M aqueous solution. If the concentration is lower
1
6 70555
t.)
0 56235
l--
0 41916
2
o 1/1
.13
<
0 27593
0 13273
0 01049 4OO
5O0
600
7OO
8O0
Wavelength (nm) Fig. 5. Absorption spectra of the dye in dry ethanol [1] and in 0.05% water/ethanol solution [2].
1955
04,
o3i "'Q 0 2
< 01 B
0
.
0
2
.
.
.
4
.
.
5
.
.
.
8
10
KOH concentration, M X 10000 Fig. 6. Effect of KOH concentration on the absorbance of the dye in water:ethanol solution.
than 1.25 x 10 5 M, then the higher aggregate band diminished. In the range of dye concentration tested, no dimer peaks were seen. When the dye concentration increased from 0-1.3 x 10 4 M, the absorbance for both blank and 0.2% water/ethanol solutions increased (Fig. 7(a)) and reached the maximum absorbance. The rates of absorbance increases were different. Blank absorbance was increasing faster than the water solution. The absorbance difference between blank and water solution was biggest at the dye concentration of 6.8 × 10 5 (Fig. 7(b)). Since at this concentration the absorbance of dry solvent was about three absorbance unit, accurate absorbance measurement is questionable. Therefore, a dye concentration of 4.5 x 10--5 M was selected. Heating prevented dimer formation and catalyzed the higher aggregation process. Heating was also the main factor for the dye responding to trace water concentration. At room temperature, the lowest water concentration measurable was 1% in ethanol after 45 rain. When the solution was heated, there were some changes in the spectral properties of the dye (Fig. 8). Both in dry ethanol and 0.05% water/ethanol solution, the peak at 800 nm shifted to 816 nm and the peak at 460 nm move to 495 nm when the temperature increased from 43.2-62.3°C. The peak reduction at 800 nm was much bigger in the water/ethanol solution than in dry ethanol. When the temperature of the solution was increased, the absorbance for both blank and 0.05% water/ethanol solution decreased (Fig. 9). After about 50°C, the blank
M. Li, G.E. Pacey /Talanta 44 (1997) 1949-1958
1956
absorbance persisted and remained constant up to 85°C. The absorbance of 0.05% water solution was further decreased and kept constant between 60-85°C. The maximum absorbance difference occurred between 60-70°C. Therefore, heating at 60-70°C was utilized. The aggregation reaction under elevated temperature was a reversible process. Heating provided the energy for shifting the reaction equilibrium toward the formation of the aggregates. When the solution was cooled at room temperature, absorption peak of the monomer increased with decreasing temperature and was stable after 15 min at room temperature. Formation of the aggregates at 496 nm from their monomers was a relatively slow process. 4
-
O ¢0
~2 O
< • blank
e 0.05
%
1
,, 0
o ±
1
005
01
0.15
Thousandths
Dye solution, M 2,
0.7500
I
2~
0.5623 0.4191
4
0.2759
4
0.1327 0.0(X)0
500
400
700
600
800
Wavelength (nm) Fig. 8. Absorbance spectra of the dye at different temperature and water contents. Spectra 1 and 2 are for dye in dry ethanol at 43.2 and 62.3°C, respectively. Spectra 3 and 4 are for the dye in 0.05% water in ethanol solution at 43.2 and 62.3°C, respectively.
After 30 min at room temperature, the conversion process was still active (Fig. 4(b), spectrum 8). At higher temperature, aggregate formation immediately complete in higher concentration of water/ ethanol solution. When water concentration is very low (0.1-0.001%), heating was needed to catalyze the aggregation process. Fig. 9 is the result of the experiment to determine the reaction time needed at 70°C. With increasing heating time from 0 - 3 0 min, the absorbance of the blank solution was decreased at almost a constant rate. But 0.05% water solution exhibited rapid reduction in absorbance between 0 - 1 0 min heating at 70°C was utilized, Fig. 10. Calibration curves (Fig. 11) were obtained under the optimized conditions for methanol,
• blank
,>
,~ 0.05 % H 2 0
08
cD
O ¢fO
o 06 < ¢'~ 04 < O
ii
o
01
005
O
02
O
_
0.15
Thousandths
Concentration of Dye, M*1000 Fig. 7. Effect of the dye concentration. (a) Absorbance of the dye in dry ethanol and 0.05% water/ethanol solution. (b) Net absorbance of the 0.05% water solution.
0 20
•
L
•
40
60
80
100
Temperature, C Fig. 9. Temperature effect of the dye in dry ethanol and 0.05% water/ethanol solution.
M. Li, G.E. Pace)'/Talanta 44 (1997) 1949 1958
• Blank
1957
1 ,
,~ 0.05% water
e 08
08
• methanol • isopropanol
0.6--
e~
,<
ethanol
06
~o4
04 @
¢,
A
02--
i
_
_
-sl
0
0
10
20
30
40
Reaction Time, rain
0
002
004
006
Water Concentration, v/v %
Fig. 10. Effect of heating time on the dye absorbance dry ethanol and 0.05% water/ethanol solution.
Fig. 11. Calibration curves.
ethanol, and isopropanol and the results of the curves are summarized in Table 1. This method has the highest sensitivity (era= 16.73 unit) for water in isopropanol and the lowest sensitivity (em = 2.806 unit) for water in methanol. The correlation coefficient (R) 2 values for the regression lines ranged from 0.990-0.998. The linear range of the method for ethanol is 0.001 0.5%, for isopropanol is 0.001-0.1%, and for methanol is 0.001-1.0%. The limit of detection for ethanol, isopropanol, and methanol are 0.0001, 0.0001 and 0.005% water with relative standard deviations of 2.62, 2.25 and 2.59% at 0.05% water concentration, respectively. The limits of detection depend mostly on the preparation of dry solvents.
time that trace amount of water in polar organic solvents can be determined by the fact that a minor change in polarity of the solvent caused by trace water content determines the extent of aggregation of a near-infrared dye monomer, which can be detected easily by measuring the absorbance reduction at 800 nm. By using a NIR dye, the detection was moved to longer wavelength where background interferences were minimized. Sensitivity of the method was very high, owing to the high molar absorptivity of the dye solution and formation of a higher aggregate in the system. The limits of detection were lower than that of most reported methods. The developed method is simple, easy to operate, and the cost of analysis is low. Current work in our laboratory concerns the possibility of applying this method to different organic solvents.
4. Conclusions
In this study we demonstrated for the first
Table l Results of calibration curves Solvent
Equation A = a + b [C]
R -~
Linear range v/v, %
LOD Wv, %
Methanol
a b a b a b
0.991
0.001
1.00
0.005
0.998
0.001-0.10
0.001
0.995
0.001
0.001
Ethanol Isopropanol
= = = = = =
0.002649 ± 0.004509 2.806 ± 0.117 0.006375 ± 0.006375 7.825 4- 0.153 0.03783 4- 0.03462 16.73 + 0.82
0.50
1958
M. Li, G.E, Pacey / Talanta 44 (1997) 1949-1958
References [1] T. Mitchell Jr. and D.M. Smith, Aquametry, part Ili (The Karl Fisher Reagent), A Treatise on Methods for the Determination of Watter, A Wiley-Interscience Publication, Wiley, 1977. [2] I. Nordin-Andersson and A. Cedergren, Anal. Chem., 57 (1985) 2571-2575. [3] H. Langhals, Anal. Lett., 23(12) (1990) 2243 2258. [4] A. Cedergren and C. Oradd, Anal. Chem., 66 (1994) 2010-2016. [5] T.S. Stevens and K.M. Chritz, Anal. Chem., 59 (1987) 1716-1720. [6] C. Oradd and A. Cedergren, Anal. Chem., 67 (1995) 999 1004. [7] Y. Ci and X. Jia, Talanta, 31(7) (1984) 556-558. [8] C. Oradd and A. Cedergren, Anal. Chem., 66 (1994) 2603-2607. [9] C. Liang, P. Vacha and W.E. Van Der Linden, Talanta, 35(1) (1988) 59-61. [10] T. Mitchell Jr. and D.M. Smith, Aquametry, part I, A
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
Treatise on Methods for the Determination of Water, Wiley-Interscience Publication, Wiley, 1977. J. Wang and A. Reviejo, J. Anal. Chem. 65 (1993) 845-847. J. Rhee, P.K. Dasgupta and D, Olson, Anal. Chim. Acta, 220 (1989) 55 63. J.S. Fritz and J. Chert, Am. Lab., July (1991) 24J-24Q. S. Kumoi, H. Kobayashi and K. Ueno, Talanta, 19 (1972) 505-513. S. Kumoi, K. Oyama, T. Yano and H. Kobayashi, Talanta, 17 (1970) 319 327. S. Garrigues, M. Gallignani and M. Guardia, Anal. Chim. Acta, 281 (1993) 259 264. H. Lund and J. Bjerrum, Ber. Stsch. Chem. Ges., 64 (1931) 210. G. Patonay, M.D. Antoine, S. Devanathan and L. Strekowski, Appl. Spectro., 45(3) (1991) 457 461. G. Patony, Advances in near-infrared measurements, 1 (1993) 113 138. M. Matsuoka, Infrared Absorbing Dyes, Plenum Press, 1990. W. West and S. Pearce, J. Phys. Chem., 69(6) 1965 1894-1903.