Articles Anal. Chem. 1996, 68, 1089-1094
Holographic Sensor for Water in Solvents Jeff Blyth, Roger B. Millington, Andrew G. Mayes, Emma R. Frears,† and Christopher R. Lowe*
Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, U.K.
The diffraction color of a gelatin holographic diffraction grating changed as a function of the water activity when immersed in a “wet” hydrophobic liquid. Quantification of the absorption maximum of the diffracted light showed that it was related, after calibration, to either the water content or the water activity of the solvent. The holographic diffraction grating measured water contents of hydrocarbon solvents at sensitivities comparable to that of the Karl Fischer coulometric titrator and over a wide range of water contents. A grating immersed in xylene revealed a visible color change when the water content was increased from 47 to 120 ppm. Conversely, the holographic grating responded to ethanol in water in the range 0-1% (w/w). The inexpensiveness and simplicity of silver halide holographic reflection gratings, combined with their relatively high sensitivity, suggests that these devices might find widespread application as immersible water activity sensors for hydrophobic liquids. The measurement of water content/activity is of great significance in the food, textile, electronics, and pharmaceutical industries. Water activity (aw) is the fractional form for what is most familiarly expressed as a percentage in the term relative humidity (RH) for a gas phase, i.e., the ratio of the partial vapor pressure of water in the gas to that above pure water at the same temperature. It is especially important in biotechnology to be able to monitor and control water activity (aw) rather than water content;1 for example, when using lipases in organic solvents to perform esterification reactions in the food and fine chemicals industries. Different materials in equilibrium at a given temperature may have the same aw values but widely differing water contents, and it has been shown that the rates of lipase-catalyzed reactions in various solvents depend on aw values rather than water contents.2 Traditional methods for the measurement of RH, such as the hair sensor, psychrometry, and Peltier dew point methods,3 and the more recently introduced piezoacoustic, metal oxide ceramic, field effect transistor, and optical fiber devices4,5 tend to be used † Current address: London Hospital Medical College, Arthritis and Rheumatism Council Building, 25-29 Ashfield Street, London E1 1AD, U.K. (1) Halling, P. J. Tetrahedron 1993, 48, 2793-2802. (2) Robb, D. A. Biocatalysis 1994, 9, 277-283. (3) Brundrett, G. W. Criteria for moisture control; Butterworths: London, 1990. (4) Blakemore, C.; Baker, W. In Humidity Sensors and their Contribution; McGivern, W. H., Ed.; NPL Teddington: London, 1986; pp 54-64. (5) Russell, A. P.; Fletcher, K. S. Anal. Chim. Acta 1995, 170, 1209-1216.
0003-2700/96/0368-1089$12.00/0
© 1996 American Chemical Society
exclusively in the gas phase and are inappropriate for use in liquids. The RH values are usually monitored in an enclosed space above the target liquid, and the assumption is made that the measured partial water vapor pressure in the gas phase is in equilibrium with that of the liquid below. Direct measurement of water content, as opposed to water activity, in liquids is usually achieved by exploitation of the Karl Fischer reagent system,6 although for many systems of interest in biotechnology, this chemical approach is cumbersome, insensitive to aw changes, and subject to interferences. Consequently, there is a requirement for a simple and sensitive sensor which can monitor water activity directly within the liquid phase. Holographic diffraction gratings, when illuminated by white light, are known to act as sensitive wavelength filters. The gratings comprise gelatin-silver halide photographic emulsion coated onto glass plates and are fabricated by passing a single diverged laser beam through a holographic plate backed by a mirror. Interference creates holographic fringes lying in planes parallel with the gelatin surface and approximately half a wavelength apart within the 6 µm thickness of the gelatin film. Under white light illumination, the finished grating acts as a specular reflector of diffracted light for a specific narrow band of wavelengths and holographically recreates the monochromatic image of the original mirror used in its construction.7 If a gelatin-based holographic grating is immersed in the test liquid, absorption of water causes the grating to swell perpendicular to the glass substrate, thereby increasing the fringe separation and causing longer wavelengths to be selected for reflection from the holographic mirror.8 This report investigates the utility of holographic gratings for the measurement of water activity in liquids, follows wavelength changes as a function of the water content of the gratings, and identifies causes of observed nonideal behavior. EXPERIMENTAL SECTION Reagents and Materials. All solvents were AR grade and dried over molecular sieves, except xylene, which was dried over sodium wire for a month prior to use. The head space in sealed solvent bottles was flushed with dry nitrogen. Controlled moisture doping of hydrophobic solvents over the experimental range was performed by adding increments of saturated stock solvent to the dried solvent. Direct addition of water to hydrophilic solvents was used. (6) Oradd, C.; Cedergren, A. Anal. Chem. 1994, 66, 2603-2607. (7) Denisyuk, Y. N. Opt. Spectrosc. 1965, 18, 152-157. (8) Spooncer, R. C.; Al-Ramadhan, F. A.; Jones, B. E. Int. J. Optoelectron. 1992, 7, 449-452.
Analytical Chemistry, Vol. 68, No. 7, April 1, 1996 1089
Figure 1. Experimental setup for hologram fabrication. A He-Ne laser (5-20 mW) beam is diverged by lens L1. P is a pinhole at the focus of L1. The beam is collimated by lens L2 onto mirror M1. H is a holographic recording plate resting on mirror M2. Light from M1 interferes with light returning from M2 to form standing waves, which are recorded in H.
Holographic recording plates (5 in. × 4 in.; 633 nm) were obtained from Agfa Ltd. (Dunstable, Beds, UK) (product code 8E75HD) or Holographic Recording Technologies GmbH (Steinau a.d. Str., Germany) (product code BB640). The Karl Fischer coulometric titrator was from Aquapal Ltd. (Bournemouth, Dorset, UK). Fabrication of Holographic Gratings. Figure 1 shows schematically the experimental setup for fabricating white light holographic diffraction gratings. A holographic recording plate (H) lies on a mirror (M2) under green safe lighting. A front surface mirror (M1) reflects a collimated beam from a He-Ne laser such that the beam strikes H along the perpendicular. M2 reflects the light back to M1, thus producing standing waves or holographic fringes throughout the depth of the recording layer (∼6 µm) on H. P is a pinhole or spatial filter at the focus of lens L1 to improve beam quality. The “emulsion” side of H was arranged face down on M2 so that it was in intimate contact with a thin layer of organic liquid (∼1 mL) sandwiched between M2 and the 10 cm × 10 cm holographic plate. The contactant liquid (usually xylene with a water content of ∼200 ppm) was employed both to reduce light scatter from the surface and to control the moisture level of the grating layer prior to exposure. The finished dried grating exhibited a yellow reflection after processing. The requirement for thorough vibration isolation is virtually eliminated by the arrangement shown in Figure 1. The only critical aspect was absence of movement between the mirror M2 and the holographic emulsion. A creep movement of 2 nm during exposure can cause blurring of the fringe structure and a very poor or absent diffraction in the finished grating. Consequently, prior to exposure, the plate was usually left in position to equilibrate for 24 h when xylene was used as contactant. When the exposure was to be made with the plate under water, it was necessary to allow it to equilibrate for 72 h. The He-Ne laser (5-10 mW) beam was expanded to give a roughly uniform illuminated area with a diameter of about 10 cm. The plate size was arranged to be 10 cm × 10 cm by first cutting off a strip of 2.5 cm × 10 cm from the standard 12.5 cm × 10 cm plate. The off-cut was later used to make 9 mm wide control strips by exposing the strips to white incoherent light and then using the same processing chemistry as that used for the grating plate. Exposure. Initial photometric tests were required to find the exposure time necessary to achieve a density level of 2-3 after 2 min in developer at 20 °C. Density levels were roughly estimated 1090
Analytical Chemistry, Vol. 68, No. 7, April 1, 1996
under the safe light.9 Proprietary holographic plates required around 10 s of exposure, whereas our own required about 40 s. After exposure, the contactant liquid was removed from the plate by a stream of air or wiping prior to immersion in developer. Development of the Exposed Plate. The exposed plates were developed for ∼2 min at 20 °C in the following developer solution: sodium hydroxide (20 g), anhydrous Na2CO3 (60 g), 4-(methylamino)phenol sulfate (Metol) (4 g), and ascorbic acid (30 g) dissolved in a total volume of 1 L of deionized water. The absence of sulfite in this formula necessitated protection from absorption of oxygen; it was used between two closely fitting plastic photographic dishes such that the top dish floated on the developer and the full surface of the developer was exposed to air only when plates were being immersed or removed. Bleaching. The developed silver-loaded plate was briefly rinsed in deionized water and immersed in a bleaching solution comprising sodium hydrogen sulfate (3 g), EDTA iron(III), sodium salt dihydrate (30 g), and potassium bromide (60 g) made up to 1 L in deionized water. The plate was bleached until the last trace of brown colloidal silver had vanished and rinsed. The silver bromide fringes were fixed by immersion for 1 min in 1% (w/v) potassium dichromate solution acidified with 0.5% (w/v) sodium hydrogen sulfate. The grating was given a 3 h rinse in agitated deionized water, dried in a warm air flow, and visually inspected under ambient lighting. A yellow mirror was apparent. Preparation of Grating Strips. The finished grating was cut into strips to match conventional glass spectrophotometer cuvettes (10 mm × 10 mm; 3 mL volume). “Sister” gratings prepared from the same exposed 10 cm × 10 cm sheet were expected to have identical swelling behaviors but frequently displayed small differences in initial λmax values, since the original 10 cm × 10 cm plate was not perfectly illuminated during exposure and bleached in some subtly different way,9 depending on the exposure level. A significant advantage of the 3 mL cuvette was that by removing gelatin from a wet grating except for a central 3 mm × 10 mm section, it was possible to obtain a larger liquid-to-gelatin ratio and thereby minimize the influence of the grating on liquids with very low water contents. The syringe used to inject sample into the Karl Fischer coulometer was first used to fill and agitate the liquid in the cuvette before each measurement run in order to equilibrate all surfaces with the environment around the grating. Spectral Measurements on the Gratings. Spectra were recorded on a Perkin-Elmer Model Lambda 16 spectrophotometer with a 2 nm spectral slit width and a cuvette holder temperature controlled to 20 °C. The liquid in the cuvette was mildly agitated with a minimagnet and motor arrangement; gratings were scanned every 3 min, typically over the wavelength range 500-800 nm. “Equilibration time” is defined as the time taken to reach the point where successive λmax values do not differ by more than 0.2 nm. When a spectrum was expected to shift into the infrared it was necessary to use a Shimadzu Model UV-160A spectrophotometer (response curve shown in Figure 6). The diffracted replay wavelengths of newly prepared dry gratings were to some extent controlled through the procedures used during fabrication.10,11 Thus, not only the degree of preswelling prior to exposure but also the type of chemical processing (9) (a) Saxby, G. Practical Holography; Prentice Hall: New York, 1988. (b) Saxby, G. Practical Holography, 2nd ed.; Prentice Hall: New York, 1994. (c) Saxby, G. Manual of Practical Holography; Focal Press: Oxford, 1991. (10) Blyth, J. U.S. Patent 4,563,024, 1986. (11) Walker, J. L.; Benton, S. A. U.S. Patent 4,986,619, 1991.
Figure 2. Effect of spiking various solvents with known amounts of water on the difference peak absorption. The peak absorption (λmax) of a single holographic grating was subtracted from the λmax for the driest sample to give the ∆λmax (nm) value on the ordinate. Measurements were made at 20 °C. Diethyl ether (O), tetrahydrofuran (0), butan-1-ol (4), propan-2-ol (3), and ethanol (]).
used to develop and bleach the hologram determined the final replay wavelength of the contracted emulsion. Thus, in order to compare results from different gratings, changes in λmax (nm) from the driest sample were used. Alternative Method: Visual Assessment of Holographic Grating Response. A grating was fabricated with the recording plate H at 25° to the horizontal mirror M2 (Figure 1). The finished grating reflected ambient light in a direction which was well removed from the ordinary specular reflection from its surface. The grating was cut in half, and each half was separately glued to a glass bottle as follows. The grating was smeared on its glass side with a hot (80 °C) 25% (w/v) gelatin solution and stuck to the inside wall of an equally hot glass bottle and incubated at 80 °C for 2 h to dry the gelatin glue. Each bottle was filled with dry xylene and a magnetic follower. Viewed under ambient light, the bottles could be positioned to replay the holograms. A small amount of moistened xylene was added to one of the bottles and stirred at 20 °C for 20 min. The replay colors were inspected visually. The water contents of the xylene were confirmed by Karl Fischer coulometry. RESULTS AND DISCUSSION Effect of Water Content. Figure 2 shows the wavelength shifts (nm) observed from the driest sample as a function of water content (ppm) for various solvents. It is clear that the greatest sensitivity of the grating to water content occurs with the most hydrophobic solvent, diethyl ether, while sensitivity varies with polarity with both the ether and the alcohol series, being least sensitive to water in ethanol. Figure 3 shows the very high sensitivity to the low moisture levels in one of the more hydrophobic solvents used in this study, xylene. The water content determined on samples taken from the cuvette after scanning the wavelength change was confirmed by Karl Fischer
Figure 3. Effect of water content (ppm) on the absorption maximum (λmax) for xylene at 20 °C. Measurements were made randomly throughout the range investigated. Table 1. Shifts in Absorbance Maxima per Percentage (w/w) Water (∆λmax/%, nm) for Several Solvents of Differing Polarity (log P)a solvent
∆λmax/%
log P
cyclohexane xylene chloroform diethyl ether tetrahydrofuran butan-1-ol propan-2-ol ethanol
7200 1500 1247 95 16 8 6 3.3
3.20 3.10 2.00 0.85 0.49 0.80 0.28 -0.24
a Log P is defined as the logarithm of the partition coefficient of the substance in the standard octanol-water two-liquid-phase system.12
coulometry. The slight scatter in these results is due to the fact that the samples with different water contents were measured at random and the grating was merely rinsed with the next sample prior to recording, rather than being dried out at 80 °C for 30 min prior to each run. In going from a relatively high water content back to a low water content, some hysteresis is evident (see below), and this is the source of the variability. However, the plots of ∆λmax (nm) versus [water] (ppm) show sufficient linearity to construct a unit for comparing the solvents shown in Figures 2 and 3. Table 1 compares the slopes of the plots for each solvent and expresses them as ∆λmax (nm) per 10 000 ppm increment in water content, or nm/%, since 10 000 ppm equates to 1.00% (w/w) mass content of water. It is clear that the response of the diffraction gratings depends on the relative affinity of water for the solvent and gelatin phase and is relatable to a measure of the hydrophobicity (log P) of the solvent12 (Table 1). The absolute value of the unit nm/% also depends on the swellability characteristics of the grating material. It should be noted that sister gratings behave consistently. These effects can be exploited to monitor the water content of (12) Laane, C.; Boeren, S.; Vos, K.; Veeger, C. Biotechnol. Bioeng. 1987, 30, 81-87.
Analytical Chemistry, Vol. 68, No. 7, April 1, 1996
1091
Table 2. Demonstration of the Consistency of λmax Values Achieved for Measuring Constant Water Activitya in Solvents with Widely Varying Water Contents
ethanol tetrahydrofuran acetonitrile chloroformc xylene toluene cyclohexane
water contentb (ppm)
λmax (nm)
15138 2147 1160 248 225 183 1500-fold range of water content, the λmax of the diffracted beam was remarkably constant at 590-605 nm, suggesting that water activity rather than water content is the factor determining grating response. This proposal is confirmed by the data given in Figure 4, which shows the λmax values of a gelatin grating immersed in xylene at 20 °C in the presence of various salt hydrate buffers producing nominal water activities in the range 0.02-0.80. It is clear that there is a strong correlation between the peak wavelength of the holographic diffraction grating and the nominal water activity. Hysteresis Effect. It is well established that gelatin13 and other humidity sensors based on biomaterials3 exhibit a sigmoidal relationship between equilibrium water content and the RH or aw of their environment and that the curve follows a different path depending on whether the plot was constructed from sequential measurements running from low to high values or vice versa.13 The phenomenon affects the potential reproducibility of the grating device under circumstances where it may be required to monitor on-line variations in water activity or RH without intermediate drying. Holographic gratings have been used previously (13) Gelfax; Croda Colloids Ltd., Widnes, Cheshire, UK.
1092 Analytical Chemistry, Vol. 68, No. 7, April 1, 1996
Figure 4. Effect of nominal water activity (aw) on the absorption maximum of a gelatin grating in xylene at 25 °C using pairs of salt hydrates1 as buffers to control aw. Hydrate stoichiometry shown in parentheses: [1] lithium chloride (1/0), aw ) 0.02; [2] calcium chloride (2/1), aw ) 0.04; [3] sodium carbonate (1/0), aw ) 0.24; [4] sodium bromide (2/0), aw ) 0.35; [5] sodium phosphate (10/0), aw ) 0.49; [6] zinc sulfate (7/6), aw ) 0.63; and [7] sodium sulfate (10/0), aw ) 0.80.
as gas humidity sensors in conjunction with a fiber-optic assembly.8 This work concluded that holographic sensors suffered from hysteresis and irreproducibility. However, the gratings had been subjected to large differences in chemical processing: for example, the developers and bleaches used led to wide variations in their tanning and hardening properties. In particular, the acid dichromate bleach, when reduced by silver Ag(O), forms Cr(III) and cross-links the carboxylic groups on the gelatin.14 The degree of tanning or hardening affects the elasticity of the gelatin and hence its swelling behavior. In our processing regime, a nontanning ascorbate developer and a ferric-EDTA bleach have been used. The Fe(III) in this bleach acts only as oxidant, and it was too tightly complexed to EDTA to cross-link gelatin. The phenomenon of hysteresis is not apparent on predried gratings made using our procedure until a threshold humidity level is reached, whence the gratings become stretched beyond their elastic limit and do not return to their original λmax when reexposed to drier solvents. Once this has occurred, the gratings need to be “reconditioned” by heating at 80 °C for 30 min. In the case of tetrahydrofuran, this threshold value lies around 2% (w/w) water, and when exposed to this level, the grating would only return to its original λmax (∼590 nm) when heated to recondition it. At water contents 2-3% (w/w)) the Karl Fischer procedure can take over an hour, while the grating approach yields values in
