ISSN:0003-2654    

DOI:10.1039/AN98914FX037

出版商:RSC    

年代:1989

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98914BX039

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, OCTOBER 19x9, VOL. 114 1173 Piezoelectric Devices for Mass and Chemical Measurements: an Update A Review* John J. McCallum Laboratory of the Government Chemist, Queen's Road, Teddington TWI I OLY, UK Summary of Contents Introduction Theory Acetoin Acrylonitrile Ammonia For ma I de h yde Halothanes Hyd razine Hydrocarbons Hydrogen Hydrogen Cyanide Hydrogen Sulphide Nitro benzene Nitrogen Oxides Organophosphorus Compounds Ozone Phosgene Propylene Glycol Dinitrate Quartz Crysta I Micro ba la nce Species in Solution and Solution Properties Styrene Sulphur and Sulphur Dioxide Toluene To I u e ne Di isocya na te Total Organic Chlorine Vinyl Chloride Water lmmunosensors and Biochemical Uses Crystal Arrays Alternative Crystal Designs Com mercia I Systems Conclusions References Keywords: Review; piezoelectric crystal Introduction Over the last 10 years or so there has been an increased interest in producing chemical, and more recently, biological sensitive devices.Much of the research has involved the characterisation and modification of a number of electronic devices such as field effect transistors (FETs), pellistors, metal oxide semiconductor devices, piezoelectric crystals and sur- face acoustic wave (SAW) devices. Electrochemistry has been applied to both chemical and biological applications and new optical techniques, for example, surface plasmon resonance (SPR) and optical fibre based sensors, have appeared. The use of chemometrics, mathematical modelling, pattern recognition and optimisation techniques, in conjunction with arrays of sensors promises the possibility of an enhanced flexibility with respect to the extraction of useful data and minimisation of background interference effects.Indeed, * Crown copyright. papers incorporating the use of these techniques have started to appear in the chemical and biochemical sensor literature. In 1983, Alder and McCallum1 published a review on the use of piezoelectric crystals as mass and chemical sensors. This review updates and extends the bibliographic cover in this field and notes the use of newer and complementary piezo- electric designs for sensing applications, including the SAW device and a cantilever design. A brief outline of chemical sensors appeared in 1985 by Snook2 in which he described piezoelectric crystal, optical fibre, electrochemical and semiconductor devices.A review (in Japanese) of enzyme based sensors, including enzyme coated piezoelectric detectors, by Osa and Anzai3 contains 11 references. Also in Japanese, Karube and Suzuki4 very briefly summarised 14 papers involving the use of biological materials on FET devices, piezoelectric crystals and SAW devices. Mierzwinski and Witkiewid reviewed the use of piezoelectric devices as air pollution detectors and listed 36 references. Sirokyh surveyed the present state of sensors and incorporated1173 ANALYST, OCTOBER 1989, VOL. 114 the use of thin films including piezoelectric crystals, SAW sensors, thin film oxide resistive and capacitative sensors, diode and transistor based sensors. This work, in Czecho- slovakian, contains 98 references. In a report, Guilbault7 reviewed his group’s work.This consists of 2.5 references for the introductory text and to 17 publications stemming from the group’s activities. Some of the hitherto unpublished work from this report is discussed later. N ylanders reviewed the field of chemical and biochemical sensors i n 198.5 including FETs, catalytic gas sensors, electro- chemical. optical and mass sensitive devices such as piezoelec- tric crystals and SAW dcvices. This work has about 94 references. I n a paper describing the future applications of micro- engineering, Hirschfeld‘) discussed the piezoelectric crystal as a possible basis for a gas vapour sensor. He stated that the ”device in principle is adcquate for this task” but also stated that it did not work because the “selectivity of the quartz piezoelectric microbalance is due to the selectivity of the absorbing chemical layer, but nobody has succeeded in making a layer that will absorb just one chemical vapour out of all possible ones that may be present.” He suggested that an alternative approach would be to use an array of crystals and examine the response pattern.The array could be reduced to a single crystal, driven at one of the higher harmonics, coated with different chemical layers and interrogated. He suggested that even this could be improved by using “a comb-like array of multiple crystals produced using photolithography.’’ Indeed, Hager et al. 10 reported the use of a single coated cry%tal for multi-component analysis. Concentration data were obtained from the crystal using the initial rate measure- ments of binary and tertiary mixtures of toluene, halothane (CF3CHCIBr) and ethanol.Turner et ~1.11 and EdInonds12 have edited works on biosensors and chemical sensors, respectively, both of which contain contributions on piezoelectric and SAW devices. A review article by Guilbault and Jordan1-? has just appeared. This contains 129 references, the majority of which were covered in reference 1. Theory SauerbreyI4.l’ has been given the credit for suggesting the use of piezoelectric crystals as the basis of a sensitive micro- balance. He derived an equation for AT-cut crystals, operat- ing in the thickness shear mode, which expressed the change in frequency (AF) of the crystal as being proportional to the mass deposited on t o the surface of the device.Lostisl6 and Stockbridge17 have covered the derivation in detail. An alternative treatment of the derivation was produced by Horvath and Aranyi,ls which takes into consideration the presence of some foreign material (i.e., a coating) on the surface of the crystal. Lu1‘) and Lu and Lewis20 have investigated mass and film thickness determination using piezoelectric crystals. Some doubt as to the general validity of the Sauerbrey equation has been mooted by Bcitnes and SchrQder.21 They reworked some of the results obtained by Guilbault’s group and produced a different interpretation of these figures. The work of Ho p t ~ 1 . 2 2 indicates support for this equation. Beitnes and SchrGdcr” suggest that the method of applying the coating may play a part in determining the actual arca covered.It is clear from an article by Warner23 that some of the deviations from the normal form of Sauerbrey’s equation can bc explained by closer examination of its derivation and the simplifications made “en route.” Much has been made of the fact that the T-cuts have zero temperature coefficients over a wide range of temperatures, e.g., AT-cut crystals are often quoted to have a zero or almost zero temperature coefficient over the range -20 to +60°C. Experiments have shown that temperature has an effect on the frequency of the crystal. This, usually, has been attributed as resulting from manufac- turing “errors” during the cutting and polishing of the crystal. Other factors, including the design and dimensions of the crystal, influence how a change in temperature affects the frequency of oscillation.For maximum stability it has been suggested that the crystal should be thermostated. It has been established that for contoured 2.5-MHz crystals, Sauerbrey’s results are valid over the mass loading range 700-7000 pg and that extrapolation to lower mass loadings is possible. However at higher mass loadings, where the applied mass becomes an appreciable fraction of the mass in the vibrating system, the calibration becomes curved. This devia- tion may arise from potential energy being stored in the coating, severance of the bond between quartz and the coating or the second-order terms in the derivation of the form of Sauerbrey’s equation, which were thought to have a negligible effect on the crystal response, becoming significant.Thc degree to which any of these factors play a part can be determined by observing whether they are a function of mass, type of material used or whether there is a change in the quality factor ((3) of the crystal. The change in Q is usually observed as a change in the electrical resistance. One other effect that has been observed during experiments employing coated crystals for gas vapour determination is a sudden jump in frequency from, for example, around 10 MHz to about 30 MHz. This sudden change may result from the crystal changing from one mode of oscillation to another. As the mass loading of the crystal increases Q can decrease to such an extent that other, lower Q modes, become allow- able.24 Acetoin Tetrabutylphosphonium chloride was reported by Suleiman et ~ 1 .2 5 to be a good coating for the detection of acetoin. It was reported to be completely reversible and sensitive in the p.p.b. range and linear over the range 8-120 p.p.b. of acetoin. Results were presented over the range 8-147 p.p.b. but no equation for the line was produced. If these data are inserted into a curve-fitting program, it is found, based on all nine points, that the calibration fits the equation d F = 2.01 [acetoin] + 124.84 with a correlation coefficient of 0.997 and where the concentration of acetoin is in p.p.b. The flow-rate was found to have a marked effect on the sensitivity, response time and reversibility. A reversibility of ca. 9.5% was obtained in 10 min at a flow-rate of 270 ml min-* and in 20 min at 130 ml min-1.A study was carried out which showed that a 30-s exposure to the analyte, at a flow-rate of 30 ml min-’ gave a significant response and complete reversibil- ity in 5 min. A number of interferents were examined and these are listed in Table 1. For comparison, the sensitivity to acetoin is 2120 Hz p.p.m.-*, which means that the coating is Table 1. Responses obtained by Suleiman el ~ 1 . 2 5 for a number of potential interferents. Sensitivity values have been included. Sensitivity to acetoin on the same basis is of the order o f 2120 Hz p.p.m.-l Species Acetone . . . . Benzaldehyde . . Benzyl alcohol Ethanol . . . . Ethylacetate . . Ethyl butyrate . . 2-Furaldehyde Limonene . . Methanol . . Pinene . . .. Octanol . . . . Concentration, p.p.m. . . . . 62461 . . . . 2097 . . . . 107 . . . . 19466 . . . . 17061 . . . . 10442 . . . . SO . . . . 81 1 , , . . 30722 . . . . 764 . . I . 389 Response/ Sensitivity/ Hz Hzp.p.m.-1 29 1 0.00s 31 0.015 32 0.299 203 0.010 29 0.002 42 0.004 23 0.46 6 0.007 47 0.002 8 0.010 77 0.198ANALYST, OCTOBER 1989, VOL. 114 1175 between 4610 and 1.06 X 106 times more sensitive to acetoin than the potential interferents. No indication of the interfer- ence effects of atmospheric moisture was given. Hahn et ~ 1 . 2 6 described the use of neutralised semicarbazide coated crystals for the detection of acetoin in air. Twenty-five other potential coatings were mentioned briefly together with an indication of how well they responded. The crystal was driven at its third overtone ( i e ., 27 instead of 9 MHz). The detector sensitivity was 12.4 Hz 1 pl-1 (12.4 Hz p.p.b.-l) and a linear response was obtained over the range 50-80 pl 1 - 1 . Acetone, ethanol, chloroform, acetaldehyde, hexyl acetate, ethyl butyrate and octanol were examined as possible interfer- ents. Relative sensitivities of these species fell in the range 0.000000 (acetaldehyde)4.000029 (hexyl acetate) with the relative response to acetoin being 1. The effect of moisture on the response was determined. A much higher response to acetoin was obtained in dry nitrogen at a flow-rate of the order of 90 ml min-1 than in an air stream of relative humidity (RH) 58%. In the latter instance, optimum sensitivity occurred at a flow-rate of 200 ml min-1.The humidity level was also shown to affect the response to acetoin (0% RH 710 Hz); at 40% RH this was 630 Hz and remained constant up to 90% RH. Between 90 and 95% RH the response decreased by 22% and overloaded at 97% RH. A field test method was also described. Acrylonitrile An indirect method for the determination of acrylonitrile by chemically oxidising it to give hydrogen cyanide was attemp- ted by Guilbault.7 The hydrogen cyanide was then to be detected by a suitably coated piezoelectric crystal. Several materials were tested but none showed promise. It was intended to examine some biochemicals including cytochrome G , methemoglobin, rhodanase and (3-glucosidase. Ammonia Moody et al. 27 incorporated pyridoxine hydrochloride (vitamin B6 hydrochloride) in a high relative molecular mass poly(alkoxy1ate) matrix (Antarox CO-880).The useful life- times of pyridoxine hydrochloride and pyridoxine hydro- chloride - polymer coated crystals were compared. The addition of the polymer improved the lifetime of the coating to 10-53 d. Calibrations were obtained for the coating including the polymer, which were AF = 6.41[NH1] + 20 for the freshly coated crystal and AF = 6.27[NH3] + 20.4 after 53 d. Further work on this coating was reported by Lai et a1.278,29 Both papers list a number of potential interferents (Table 2) and in the second paper'') a frequency change of 211 Hz was Table 2. Frequency changes obtained by Lai et al.29 for interferences with Antarox CO-880 - pyridoxine hydrochloride coated crystals AFfor 5-ml samples passed over the crystal coatings/Hz Conce ntra- Gas tion.p.p.m. NH3 . . . . 0.463 32 S O ? . . . . . . 101 NO? . . . . 75 H C I . . . . . . 109 CO? . . . . 1480 H 2 S . . . . . . 116 R o o m a i r . . . . - Dried room air . . - Triethylamine . . 3.6 36.4 Pyridoxine hydrochloride Pyridoxine on Antarox Antarox hydrochloride CO-880 CO-880 199 320 - 4s 21 44 34 47 35 32 4s 1496 2674 19 32 30 27 43 41 10 16 11 0 0 0 21 1 383 12 41 - - - - noted when the crystal was exposed to 5.3 mg dm-3 of triethylamine. This is contrary to the results of Hlavay and Guilbault30 who claimed that this compound produced no effect on the frequency. Hydrogen chloride was found to be a serious interferent as it reacted with the supporting matrix. It was suggested that Antarox CO-880 could be used as a coating for HCI.29 Capillary tube and brush methods for coating thc crystals were examined28 and it was shown that there was a greater sensitivity obtained from the crystals coated by painting.Mass loadings by both methods were comparable. The difference in sensitivity was attributed to the uneven coating produced by the capillary tube method. A multi-sensor system was developed by Fraser et d . 3 1 for the determination of airborne contaminants such as ammonia. This system used a poly(vinylpyrro1idine) (PVP) coated crystal for the determination of ammonia, a silver chloride coated crystal for water and an RS 590 KH temperature sensor (RS Components, Birmingham, UK). Data from these sensors were passed to a microcomputer based on an Intel 8080 microprocessor.The computer compared these measure- ments with a set of standard conditions to produce a concentration value for ammonia. The crystals responded rapidly to the ammonia in the first few minutes of exposure with an equilibrium being established after 15 min. The silver chloride response could be related to the PVP crystal response over the humidity range MO% RH at different temperatures and with different polymer mass loadings and surface coverage on the crystal. A flow-rate of 50 ml min-1 was found to be the best. Calibrations were produced for the ammonia response of the PVP coated crystal, water on both the PVP and silver chloride crystals at 25, 30 and 35°C and a calibration of the amount of water resident on a PVP coating that affects the amount of ammonia sorbed from a gas stream containing 4 p.p.m.of ammonia. The data were entered into the micro- computer to describe the standard conditions. With this system it was possible to determine ammonia in the range 6 p.p.b.-10 p.p.m. The use of the Langmuir - Blodgett (LB) technique to coat the surface of 18-MHz AT-cut crystals was reported by Ross and Roberts.32 Layers of o-tricosenoic acid (tnTA) and tetra-4-tert-butyl silicon phthalocyanine dichloride (ttbPcSiC12) were applied to the surfaces and produced linear responses of frequency change versus the number of applied monolayers. They suggested that an enhanced acoustic coupling effect may be obtained when using the LB film method. Crystals coated with five monolayers of the coatings were exposed to ammonia and hydrogen sulphide over the concen- tration range 0-100 p.p.m.Linear responses were obtained and the sensitivity for oTA to ammonia was 0.03 Hz p.p.m.-l and for ttbPcSiCI20.14 Hz p.p.m.-l. A pre-concentration technique involving the use of denuder tubes is being studied by Ali and Alder.33 The denuder tube is coated, internally, with tungstic acid to sorb ammonia in the test gas stream. After a period of time, the tube is heated rapidly to desorb the ammonia. This gives rise to a pulse of ammonia that is detected using a piezoelectric crystal. The operational conditions have not yet been fully optimised but the results presented so far show an enhancement of the signal. Formaldehyde Formaldehyde dehydrogenase, together with glutathione and nicotinamide adenine dinucleotide, was coated o n to a quartz crystal by Guilbault34 with the intention of monitoring formaldehyde.Excellent selectivity was obtained with little response to other aldehydes or alcohols. Sensitivity to formaldehyde was of the order of 50 Hz p.p.m.-I. The coated crystal was usable for ca. 10 d.1176 ANALYST. OCTOBER 1989, VOL. 114 Halothanes A piezoelectric crystal based detector for halogenated hydro- carbons was presented at the 1980 IEEE-BME (Biomedical Engineering) conference by Kindlund et d . 3 5 The piezo- electric gas monitor was developed further,36 incorporating a pre-concentration stage, to determine the concentration of halothane (CFiCHCIBr). The crystals were coated with a silicone oil (DC-190). The pre-concentrator consisted of a 200-pm deep basin etched into a 0.75-mm thick glass slide.The basin was then coated with a 6.5-pm thick layer of the same silicone oil. A second glass slide was used as a lid. A Pelticr heat sink was mounted on both sides of the cavity and used as a cooling and heating element. In the application, the test gas was passed continuously through the cooled cavity for some hundreds of seconds and then the heat sink was used to generate a pulse of vapour (ca. 10 s). With DC-190, the absorption energy of water was ca. 0.64 eV molecule-' and for hydrocarbons ca. 0.35 eV molecule- 1 ; this difference was insufficient to allow the determination of hydrocarbons in the presence of water vapour at 1-2% absolute concentration. Two methods for improving the discrimination were then examined.The first involved the use of an 80-cm length of q l o n tubing the inner surface of which was coated with Teflon. This separated the water and, say, the halothane pulses in time. Water was the most delayed. Perfluorinated polymer tubing (20 cm), permeable to water, successfully removcd water. A linear calibration over the range 0.5-100 p.p.m. of halothane was obtained (AF,,, = 6 Hz for 100 p.p.m.). Laughing gas ( N 2 0 ) was examined as a potential interferent. It gave no response by itself and did not change the response of a given halothane concentration. Toluene and ethanol were examined briefly. Tolucnc sensitivity was found to be 0 . 1 Hz p.p.m.-l A detection limit of 100 p.p.m. of ethanol was obtained when the perfluorinated polymer tubing was incorporated into the system.This signal was similar to that obtained from 0.5-1 p.p.m. of halothane. Seven coatings wcre examined by Cooper et al.37 for their suitability to monitor anaesthetic gases such as halothane, enflurane and dinitrogen oxide. Mastic medical grade adhe- sive Type A proved to be the best with respect to stability, speed of response, ease of application and sensitivity to these gases. The following sensitivities wcre quoted (based on the use of six coated crystals): halothane, 2.63 k 0.15; enflurane, 2.61 ? 0.23; and dinitrogen oxide, 1.86 -t 0.25 Hz per MAC per pg of coating where MAC is the minimum alveolar concentration for the gas (in vol% at 760 mmHg; 0.75 for halothane, 1.65 for enflurane and 101 for dinitrogen oxide). All the coatings tested showed significant responses to water.For simplicity, a standard gas stream was used (37 "C and 40% RH). The maximum response obtained when the crystals were exposed to 100% carbon dioxide was 4 Hz. Hy drazine Trimcthylamine has been reported by Guitbault7 to be a suitable coating for the determination of hydrazine. A linear dynamic range of 20-220 p.p.m. was obtained when samples were injected into a carrier stream. Hydrocarbons Nine chromatographic stationary phases and seven analyte species were studied by Edmonds.38 The selectivity of the coatings depended on the solubility of the analyte species in these materials. Table 3 summarises the responses. Theoret- ical calculations were also reported indicating that for the optimum response the sample cell containing the crystal should be as small as possible.He determined that 97.5% of the equilibrium value would be achieved after 300 gas changes in the cell and 99.75% after 500. The effect of coating volume and position on the response was examined. Edmonds38 also suggested the use of multiplex- ing and solving simultaneous equations to obtain concentra- tion data from multiple crystal systems. Mierzwinski and Witkiewiczj'" examined the response characteristics of four coatings to 12 organic analytes. The effects of different mass loadings of coating and the position and surface coverage of the coating on the crystal on the sensitivity of the device were also examined. The results on coating position agreed with the findings of Ullevig ct I Z ~ . , ~ ~ who demonstrated that the area under the electrode was the most sensitive to mass.It was demonstrated that the mass of the coating plays a part in determining the sensitivity of the coating to the analyte. A summary of the sensitivity data39 is shown in Table 4. A further paper appeared in 1987 (in English) that extended this work?' The use of hexa- and tetra-epoxy derivatives of octa- cosahydro[l2]cyclacene as coatings for a number of aromatic species was reported by Elmosalamy et a1.32 They suggested that selectivity could be obtained by way of the cavity size present in these crown-type compounds. Hydrogen The detection of hydrogen in ambient air using a 6-MHz AT-cut crystal coated with palladium on both sides was reported by Hosoya et al.43 The response of the crystal was followed when a sample of hydrogen was introduced into the carrier stream, either air or nitrogen. An increase in frequency was obtained when the hydrogen was present suggesting that some species were being removed from the crystal surface.They suggested that oxygen pre-absorbs on the palladium surface when exposed to air and that, when exposed to hydrogen, it reacts forming desorbable water. The response time was of the order of 1 min and complete reversibility occurred in ca. 2 min. The sensitivity decreased as the temperature increased from 10 to 35 "C. No interference was observed from 1000 p.p.m. V/V of methane, ethane or propane, from 100 p.p.m. of dinitrogen oxide, from 1000 or 5000 p.p.m. of carbon dioxide or from 10000 p.p.m. of sulphur dioxide.Calibrations for hydrogen in air and in nitrogen were produced over the range 0-1%. Grcater linearity was obtained in the presence of air. Table 3. Relative response data obtained by Edmondsjx Ethyl- Coating Chloroform benzene Tricresylphosphate . . . . 1.00 Carbowax 20-M . . . . . . 1 .00 Silicone gum rubber . . . . 1.00 (3,[3'-Oxydi(propionitrile) . . 1 .OO Squalane . . . . . . . . 1 .OO Dinonyl phthalate . . . . . . 1 .OO Pluronic L-64 . . . . . . 1 .OO Rubber solution . . . . . . 1.00 Ethylene glycol succinate . . 1 .00 3.23 3.20 0 2.61 9.18 4.51 3.57 13.84 3.37 o-Xylcnc 3.20 3.53 0 3.10 9.48 5.47 3.85 9.34 2.70 Acetone 0.41 0.42 0 0.71 0.38 0.18 0.15 0 1.30 Hcxanc Cyclohexane Hcptcnc 0.06 0.36 0.4x 0.09 0.21 0.34 0 0 0 0.52 0.26 0.34 0.77 I .55 1.65 0.2 1 0.43 0.45 0.15 0.17 0.26 0.35 1 .so 0.91 0.66 - - Chloroform respondHz 35.0 162.5 12.0 21 .o 30.0 61 .o 158.5 27.0 24.0ANALYST. OCTOBER 1989, VOL.I14 1177 Table 4. Sensitivity data obtained by Mier7winski and Witkiewicz.3" Coatings: PCR, 4-cyano-4'-pentylbiphenyl; PPAB, 4-pentyl-4'- propylazobenzene; PMAOB. 4-propyl-4'-methylazoxybcnzene; and OLWCh. cholesteryl oleylcarbanatc. Sensitivities quoted are for a coating mass equivalent to a 1500-Hz frequency changc Table 5. Response of a number of coatings to 7 p.p.m. of nitrobenzcne45 Amount Response timdmin of coating/ Response/ Coating kHz Hz Adsorption Desorption Sensitivity of the coatingllO-" Hz g-' ml Compound Bcnzenc . . . . . . Tolucne . . . . . . Chlorobcnzenc . . . . rn-Dichlorobenzene , .o-Dichlorobenzene . . Nitrobenzene . . . . Ethanol . . . . . . Butan-1-01 . . . . . . Hexan-1-01 . . . . . . Ethylacetate . . . . Butylacetate . . . . Hexylacetate . . . . PCB 0.9 2.6 8.0 37.5 54.6 i50.0 1.1 6.4 24.5 1 .(I 6.1 34.5 PPAB 0.7 2.2 6. I 33.0 42.5 98.0 0.6 3.0 21.3 0.6 3.5 25.0 PMAOB 0.8 2.4 6.7 35.0 48.0 105.7 0.9 6.0 22.0 0.9 5.4 34.5 OLWCh 0.4 1 .5 3.5 21 .o 25.7 48.6 0.4 2.5 21.5 0.4 2.1 17.7 Hydrogen Cyanide A range of metal salts was examined by Alder et a1.44 as coatings for the detection of hydrogen cyanide in air. To try to minimise the problems of humidity they reasoned that by choosing a metal complex with ligands that were both volatile when free and good leaving groups, they could obtain a mass amplification effect. As mass was lost from the crystal in the presence of cyanide the frequency would increase whereas with moisture a frequency decrease would be expected.They reported some results obtained using crystals coated with bis(pentane-2,4-dionato)nickel with a calibration for cyanide over the range 13-93 p.p.m. (continuous flow stream). The effect of humidity was studied over the range 40-92% RH. Instead of using the usual measure of sensitivity (Hz p.p.m.-I) a rate-dependent sensitivity (Hz min-I p.p.m.-l) was used. They showed that moisture played a part in the reaction between the cyanide and the complex. They also stated that the nickel complex was prone to hydrolysis from the presence of atmospheric moisture. The coated crystal, in this instance, is a "one shot device," i.e., one response per crystal.Hydrogen Sulphide Ross and Roberts32 reported the use of the Langmuir - Blodgett film deposition method to prepare coated crystals for the study of hydrogen sulphide and ammonia. Linear responses to H2S were obtained over the range 0-100 p.p.m. with coTA and ttbPcSiC1, with sensitivities of 0.36 and 0.06 Hz p.p.ni.-I, respectively. Nitrobenzene A series of coatings were examined by Sanchez-Pedreno et ~ 1 . ~ ' for their sensitivity to nitrobenzene. They used 14.9-MHz crystals (HC-25 mounts). Generally the coatings were applied by dissolution in chloroform and then smearing the entire surface of one of the electrodes, as uniformly as possible, to give a coating mass corresponding to 8.9 kHz. Poly(ethy1enimine) (PEI) however, supplied as a 1 + 1 aqueous mixture, was diluted further with ethanol before application.Charcoal was sieved through a 212-pm test sieve to remove the largest particles and then placed in an in-house produced applicator. Air was passed through this applicator at the slowest rate that would deposit a fine black film on to a piece of filter-paper. This ensured that only the finest charcoal particles were applied to the crystal. The crystal ceased oscillation when charcoal corresponding to 0.75 kHz was applied; therefore a loading of 0.57 kHz was used. Charcoal . . . . 0.57 - PEG-400 . . . . 8.8 PEG-750 . . . . 8.9 PEG-1000 . . . . 8.9 PEG4540 . . . . 8.9 Quadrol . . . . 8.9 Tetrabasc . . . . 8.9 Triethanolaminc . . 8.8 Uncoated (blank) 0.0 - 136 - 53 - 86 - 24 - 14 - 48 -31 -18 -13 16 7 10 4.8 2.8 8 6 4 3.5 13.2 3.2 4.4 2.6 2.4 5 2.8 2.8 2.4 Table 5 summarises the responses obtained when the crystals were exposed to 7 p.p.m.of nitrobenzene in dry air. A series of six absorption/desorption cycles were run on five of the test coatings. The reproducibility of the response was 1.3. 2.S, 3.5, 4.4 and 2.4% for charcoal, PEG-400, PEG- 750, Quadrol [ N , N , N ' , N'-tetrakis(2-hydroxypropyl)ethylene- diamine] and Tetrabase [4,4'-methylene-bis(N,N-dimethyl- aniline)], respectively. Calibrations and linear ranges for the five coatings are listed in Table 6. Other potential coatings were examined but were rejected because of their high bleed rate, e.g., dicyclohexylamine, diphenylamine and trioctylaniine. Nitrogen Oxides Hepher-46 and Edmonds rt al.47 described the use of a manganese dioxide coated crystal in adsorption studies of nitrogen dioxide. The crystal was coated by melting man- ganese nitrate tetrahydrate on to the surface of the electrode at 36 "C.The crystal was then taken up to 177 "C and baked at this temperature to produce the manganese dioxide. These workers reported that for the best results with this coating the humidity should be kept at around 1500 p1 1 - 1 and in a practical detector system this would limit its usefulness. Organophosphorus Compounds In 1980, Guilbault P t d.48 submitted a report to the United States Air Force on the use of piezoelectric crystals to detect organophosphorus compounds. Two papers from this report have been published4y.s0 and their content was covered in reference 1.The work involved the use of a crystal coated with a mixture of 1 -dod e c y 1-3 - h y d r ox y im i n om e t h y 1 p y ri d i n i u m iodide (3-PAD), Triton X-100 and NaOH. 'The use of copper complexes as coatings suitable for the detection of chemical warfare agents was reported by Guil- bault et a/." Included in this study were copper butyrate - ethylenediamine, copper butyrate - diethylenediamine and XAD-4 - Cu2+ - amines. Diisopropyl methylphosphonate (DIMP) was used as the model compound. Bidentate copper complexes gave the best results and the XAD-4 - C U ~ + - diamine complex gave a greater affinity than the corresponding copper butyrate complex. The greater response was associated with the larger adsorption surface area of the polymer. Further enhancement o f the response was achieved by the addition of sodium hydroxide in glycerol to make the coating more basic.Although the XAD-4 - Cu2+ - diamine complex gave the best sensitivity, it had a short lifetime. To improve this, the crystal was coated with poly(hexadecy1 methacrylate) and then sprayed with the polymer. The excess of material was removed using soft paper. Table 7 lists the sensitivities of the coatings and Table 8 the results of an interference study. Uncoated 9-MHz crystals with different electrode materials were examined for their sensitivity to DIMP by Kristoff and1178 ANALYST, OCTOBER 1989, VOL. 114 Table 6. Performance data for coatings sensitive to nitrobenzene4s Linear Least squares Amount/ range, No. of Coating kHz p.p.m. points SlopelHL p.p.m.-l Intercept/Hz Charcoal .. . . . . 0.57 0.73-7.6 5 -13.1 + 0.3 44*1 PEG-400 . . . . . . 8.8 2.1-9.7 4 -7.3 f 0.0 2 + 2 PEG-750 . . . . . . 8.9 2.2-9.7 4 -11.4 f 0.5 6 + 4 Quadrol . . . . . . . . 8.9 2.2- 10.7 4 -6.4 f 0.4 3 f 3 Tetrabase . . . . . . 8.9 2.2-7 .0 3 -4.1 k 0.0 2 f 0 Correlation coefficient 0.999 0.999 0.998 0.996 1 .000 Table 7. Sensitivities reported by Guilbault et al.” for coated crystals exposed to DIMP Sensitivity (AF)l Hz per p.p.b. of DIMP per pg Coating of coating XAD-4-Cu2+ . diamine . . . . . . 2.62 3-PAD . . . . . . . . . . . . 1.9* SE-30 . . . . . . . . . . . . 0.4 Poly(hcxadecy1 methacrylate) . . . . 0.5 * If 3-PAD is incorporated with a base and a surfactant then a higher response is expected (see references 43-45). Table 8.Interference study using a copper - tetraamine chelate coated crystal. The response to 15 p.p.m. of DIMP was 591 Hz51 Interferent Carbonmonoxide . . Ammonia., . . . . Hydrogen sulphidc . . Sulphurdioxide . . . . Hydrogen chloride . . Benzene . . . . . . Toluene . . . . . . Chloroform . . . . . . . . . . . . . . . . . . . . Concentration of interferent, p.p.m. . . 1250 . . 1250 250 . . 1250 250 . . 1250 250 . . 1250 250 . . I000 . . 1000 . . 500 100 * An irreversible response was obtained. AFl HZ 0 26 2 18 0 12 4 40” 15” 9 13 49 11 Table 9. Response of PVP - TMEDA and PVBC - TMEDA coatings to interferents. All responses were obtaincd in the presence of 10 p.p.b. of DIMP54 AFIHz Concentration, PVP - Interferent p.p.m. TMEDA None (DIMP, 10p.p.b.) . . 1000 40 Nitrogendioxide .. . . 1000 40 Hydrogenchloride . . . . 1000 41 Ammonia . . . . . . 1000 48 Sulphurdioxide , . , . 1000 41 Carbonmonoxide . . . . 1000 40 Hydrogensulphide . . . . 1000 41 Benzene . . . . . . . . 100 43 Toluene . . . . . . . . 100 40 Chloroform . . . . . . 100 42 Car exhaust (diluted 1 + 10) - 42 PVBC - TMEDA 10 11 14 14 10 10 11 10 10 17 I 1 Guilbault .-52 The sensitivity increased and the selectivity decreased in the order of gold, silver and then the nickel electrode. Linear calibrations were produced over the concen- tration range 0-50 pg 1- 1 for all three electrode materials and also at 20, 30, 40 and 50°C for the crystals with gold electrodes. Interferences were few. Nitrogen dioxide was shown to be a serious interferent with the silver electrode crystals.The use of piezoelectric crystal devices and chemFETs (chemically sensitive field-effect transistors) as chemical warfare agent detectors was described by Cassidy and Thomas-53 in the introduction to their report on the synthesis of resin polymers with the ability to remove these species. It is noteworthy that no indication as to whether these polymers were deposited on crystals or chemFETs was made. Guilbault and Kristoff-54 mixed (tetramethylethylenedi- amine)copper(II) chloride (TMEDA) with either of two polymers, poly(vinylpyrro1idine) (PVP) or poly(vinylbenzy1) chloride (PVBC), to sensitise it to DIMP. A linear response was obtained over the range 0-30 p.p.b. for the PVP - TMEDA coating and 0-20 p.p.b. for PVBC - TMEDA. Above 30 p.p.b. the PVP - TMEDA coating was saturated and, for the PVBC - TMEDA coating there was an increase in sensitivity above 20 p.p.b.without the presence of saturation. A response time of 1-5 min and recovery time of 1-7 min were rcported for PVP - TMEDA and 4-30 and 20-120 min, respectively, for the PVBC - TMEDA coating. Table 9 lists the results on a number of possible interferents. Water caused a serious interference with the PVP-based material but not with the PVRC-based one. With the PVBC - TMEDA coating, up to 80% RH can be tolerated. The use of protein coatings (antibodies and enzymes), in particular cholinesterase and acetylcholinesterase, for the detection of organophosphorus compounds was described by Guilbault et al.5-5 Equal amounts of this material were applied to both sides of the crystal using a microlitre syringe and immobilised using the appropriate ratios of glutaraldehyde and bovine serum albumin applied in the same manner.The excess of glutaraldehyde was removed by washing with a phosphate buffer solution. The crystals were then placed in a desiccator for at least 1 h before use. Calibrations were produced for Malathion { 0,O-dimethyl S-[ 1,2-di(ethoxycar- bonyl)ethyl]phosphorodithioate} ( A F = 280 Hz at 1 p.p.m.) and DIMP (AF = 180 Hz at 10 p.p.b.). An antibody coating for parathion was produced and tested. An example of the response to 35 p.p.b. of parathion was produced indicating a frequency change of the order of 900 Hz in 4 min for this concentration. Complete reversibility appeared to occur in ca. 6 min. Reproducibility for all concentrations tested was better than 4%.A linear calibration in the p.p.b.-p.p.m. range was claimed. It was noted that “No interferences from humidity were observed as long as it remained constant. A larger response was recorded at higher humidities than at low humidities, hence calibration curves at different humidities are first required.” The enzyme coatings were found to have a lifetime of ca. 40 d stored in a desiccator at room temperature whereas the antibody coatings had a lifetime of ca. 18 d stored similarly. Mention was made of a portable system, the PZ101, marketed by Universal Sensors (see later). Ozone Fog and Reitz-56 coated 9-MHz crystals, AT-cut with gold electrodes, with poly(butadiene) in toluene to determine ozone. After coating, the crystals were stored over activated charcoal to eliminate trace amounts of toluene and ozone from the surrounding air.They suggested that the coating reacts with ozone to form ozonides.ANALYST, OCTOBER 1989, VOL. 114 1179 An uncoated crystal was used as a reference and, by way of a frequency mixer, compensated for background fluctuations. A detection limit of less than 10 p.p.b. was obtained and frequencychangesof 3.2Hzfor70p.p.b.,25 Hzfor500p.p.b. and 28 Hz for 550 p.p.b. were obtained. It was noted that the poly(butadiene) coating was weakly hygroscopic but over the range 28-75% RH there was no effect on the response from 94 p.p.b. of ozone provided that the water concentration remained constant. Air samples were taken from the breathing zone of a welder via Teflon tubes mounted around his helmet and the results obtained with the crystal device were compared with those generated from an AID ozone analyser and Draeger tubes.Reasonable agreement was reported. Phosgene Suleiman and Guilbault57 coated 9-MHz crystals with methyl- trioctylphosphonium dimethyl phosphate dissolved in chloro- form with a view to producing a phosgene detector. It was shown that by increasing the mass of the applied coating the sensitivity of the crystal to phosgene could be increased and for the best reproducibility, the amount of coating applied to the crystal should lie in the range 15-20 pg. The flow-rate of the gas stream was shown to have an effect on the sensitivity. The maximum sensitivity was obtained between 5 and 10 ml min-1.To give a rapid removal of the phosgene after exposure the flow-rate was increased to 70 ml min-1. A number of interferents were examined at the 320 mg 1-l level, which are listed in Table 10. At very high concentra- tions, ammonia was found to interact irreversibly with the coating and increase its sensitivity to phosgene but was shown to decrease the lifetime of the detector. A linear calibration over the range 5-140 pg 1-1 was reported and a loss in sensitivity of ca. 12.5% after 6 weeks was claimed. Propylene Glycol Dinitrate A number of coatings were tested by Turnham et al.58 for their suitability to detect and monitor propylene glycol dinitrate (PGDN), a torpedo fuel. Table 11 lists the responses obtained Table 10. Detector responses to 320 mg I-' of interferent (flow-rate. 10 ml min-'; mass of coating, 17.95 ~ g ) ~ ' Interferent SO* .. . . . . . . CZH3CI . . . . . . NO* . . . . . . . . NH3 . . . . . . . . HCl . . . . . . . . HZS . . . . . . . . H*NNH* . . . . . . co* . . . . . . . . Air . . . . . . . . Response/ Hz 4 2 3 21,17,11 16 18 0 0 14 Table 11. Response of different coatings to propylene glycol dinitrate at the 1 p.p.m. leve1.58 Crystals used were operated at 10 MHz Load- Response ing/ factor/ Coating kHz Hzp.p.m.-1 Comments OV-275 (dieyanoallylsilicone) . . 47.5 146 Stable Tricresylphosphate . . . . . . 22 20 Volatilc 1,2,3-Tris(2-cyanoethoxy)propane 20 70 Volatile OV-105 . . . . . . . . . . 16 9 Volatile OV-225 . . . . . . . . . . 53 30 Slowly Nitrile silicone gum (X6-60) . . . . 31.5 75 Volatile Tetracyanoethylated pentaerythritol 31.5 80 Volatile volatile for these materials.Dicyanoallylsilicone (OV-275) gave the best results with a sensitivity of 146 Hz p.p.m.-'. It was noted that OV-275 did not bleed from the crystal whereas the other materials tested were too volatile. This coating was found to come to equilibrium in 2-3 min (4 p.p.m. level) with a concentration indication in 20 s. A recovery time of less than 1 min was reported. At high coating levels there was an increase i n sensitivity to other contaminants, e.g., propan-2-01; therefore it was endeavoured to keep the level of coating applied to <40 kHz. Turnham et a1.58 produced a prototype instrument with a built-in microprocessor similar to that described by Fielden et al.s'" Their device used two crystals and a trap in front of one of the crystals to remove the PGDN.The prototype ran satisfactorily in the laboratory for up to 3 months with no obvious reduction in sensitivity, selcctivity or stability. Quartz Crystal Microbalance The quartz crystal microbalance has been shown to be very sensitive to small increments of mass. McKeownhO reported their use in measuring the sputtering of various materials from their surface when placed in a molecular beam. 10-MHz AT-cut crystals were used and it was determined that ," 1 Hz change corresponded to a thickness change of 0.166 A . The sputtering rate for gold in argon at normal incidence was then measured. Plots of p, the sputtering rate in atoms per ion, as a function of the beam energy were produced, which were in reasonable agreement with those obtained by other workers.McKeown noted a sputtering threshold considerably lower than that reported previously (10 cc 50 eV) but suggested that further work was rcquired with neutral bcams to confirm this observation. Andersen and Bay61 measured the heavy-ion sputtering yield for 45-keV Arf irradiation of amorphous silicon and reported absolute yields for 16 different ions (7 d Z d 83). The sputtering yield, S , was calculated from where ko is the calibration constant of the quartz (in this instance 5.97 X 10-6 g Hz-I), A F is the frequency change caused by N , impinging ions of mass M I , M2 is the target-atom mass and No is Avogadros number. The term MIIM2 was included as it was assumed that all the projectiles stuck to the target.The experimental and theoretical sputtering yields differed by <15% (Table 12). The energy dependence of the lead sputtering yield on silicon over the range 25-500 keV was determined and again good agreement between experiment and theory was obtained. The sputtering yield of silicon was studied further by Blank and Wittmaack62 and measured when bombardcd with 10-149 keV argon and 10-540 keV xenon. Ullevig and Evans63 used these devices to measure sputter- ing yields and ion beam damage to thin organic films. They reported significant structural and chemical changes in some of the polymers examined and, with polystyrene the layer formed after bombardment was found to be highly cross- linked and, possibly, even graphitic. Visual examination of this revealed a dark greyish film that did not dissolve in chloroform, the solvent used to disperse the coating originally.Species in Solution and Solution Properties It has becn shown by Nomura and co-workers64,hS that the change in the oscillation frequency of quartz crystals in organic solutions depends on the specific gravity and viscosity of the solvcnts used. The frequency was shown to decrease with increasing electrolyte concentration added, depending on the increase in the specific conductivity. Frequency shifts (AF)1180 ANALYST, OCTOBER 1989, VOL. 114 were reported that agreed well with calculated values derived from A F = a& + - c where a , h and c are constants determined by the crystal, d is the density and '1 is the viscosity of the solvent.Bel'kov and Malinovskya66 used 5-MHz AT-cut crystals to determine the mass of the solid residue remaining after evaporation of the solvent. A thermostated system was used to give the greatest frequency stability. The effect of metal ions on a piezoelectric crystal in aqueous solutions was studied by Nomura and Maruyamah7 with particular reference to the adsorptive determination of iron(II1) as its phosphate. One side of the crystal was covered with a glass plate and the leads were protected with epoxy resin to prevent electrolysis. Profiles of frequency versus immersion time of the crystal in an aqueous solution contain- ing 0.5 mM of metal ions and acetate buffer (1 mM, pH 4.7) were obtained for magnesium(II), aluminium(IIT), iron(II1) and lead(TT). For comparison purposes, the profile of a reagent blank was also presented. Some responses for magnesium concentrations over the range 0-0.6 mM were also obtained.A linear calibration was reported for Fell1 over the range 1 x 10-5- 1 x 1 0 - 4 M and described by the equation [FeIII] = (AF/15.8) x 10-SM where AF is measured in Hz. The standard deviation was 7.81 Hz (9.6%) for five determinations of 5 x 10-5 M iron. Table 12. Comparison of theoretical and experimental sputtcring yields obtained by Andersen and Bayb1 S Ion N . . . . . . . . . . . . Nc . . . . . . . . . . . . Al . . . . . . . . . . . . CI . . . . . . . . . . . . Ar . . . . . . . . . . . . v . . . . . . . . . . . . Zn . . . . . . . . . . . . Se . . . . . . . . . . . . Kr . . . . . . . . . . . . Ag . . . .. . . . . . . . Sn . . . . . . . . . . . . Te . . . . . . . . . . . . Xe . . . . . . . . . . . . Tm . . . . . . . . . . . . Pb . . . . . . . . . . . . Bi . . . . . . . . . . . . Experiment 0.43 0.85 1.12 1.35 1.75 2.16 2.56 3.1 3.3 4.0 4.4 3.9 3.8 5.0 4.8 4.8 Theory 0.47 0.77 1.06 I .ss 1.67 2.12 2.78 3.1 3.3 3.7 3.9 4.0 4.0 4.4 4.6 4.6 Table 13. Tolcrance limits for anions in the determination of 5 X 10-6 M of silver(I)70 Tolerancc limit [anion: silver(1) molar ratio] Anions 100 Sulphate, perchlorate, nitrate, thiocyanate 10 Carbonate, thiosulphate 1 Chloride 0.1 Cyanide. bromidc <o. 1 Sulphide, iodide Table 14. Tolerance limits for ions in the determination of 1 X of silver(I)71 M Tolcrancc limit [ion: silver(1) molar ratio] Ions Interference effects of a number of species (up to 0.5 mM) were investigated.Ten-fold molar amounts of lead, alumin- ium, bismuth, sulphide and thiosulphate interfered whereas similar amounts of manganese(TI), cobalt(II), nickel(Il), copper(II), zinc(II), cadmium(ll), halides, cyanide and thio- cyanate did not. Nomura68 incorporated a 9-MHz crystal into a flow injec- tion (FI) system for the determination of sulphate. Barium sulphate was prepared in the FI system and deposited on to the crystal. The effects of flow-rate and the length of the reaction tubing were studied. A linear calibration was produced at a flow-rate of 1.2 ml rnin-l with a reaction tube length of SO cm and a reaction time of 5 min over the range 0.5-100 UM of sulphate. The response for 100 VM of sulphate was of the order of 275 Hz. Deposited barium sulphate could be removed from the crystal by the use of ethylenediaminetetraacetic acid (EDTA).Hager et af.6') used piezoelectric crystals to determine the dissolution and film formation rates o f iron. The crystal was used as one electrode in a three-electrode system. The reference electrode was Ag - AgCl and the counter electrode was platinum. The iron was plated on to either 5-MHz (silver electrodes) or 10-MHz (silver or gold electrodes) crystals at four different current densities in the range 20-400 mA cm-2 from a solution of 2 . 4 ~ ( 1 . 2 ~ ) iron(I1) sulphate at room temperature. The frequency of the crystal was monitored during the plating process. Film thicknesses i n the range 2000-12 000 8, were obtained. Linear potential corrosion scans were run in deoxygenated 1 N (0.5 M) Na2S04 at room temperature (22 -t 2°C) over the range -1 to + 1 V with respect to the Ag - AgCl electrode.The cell current and potential were recorded together with the crystal frequency. Plots of current and AFldt against the crystal electrode potential gave a close correlation. It was also noted that in the passivation range (0.4-0.6 V versus the reference electrode) the current versus potential plot showed more structure than the AFldt versus potential plot. The current plot showed many local maxima and minima whereas the latter showed only two spikes (sharp peaks). The current in the passive region was not uniquely related to the crystal frequency and, by inference, to the passive film thickness. Three papcrs dealing with the determination of silver(1) in solution have been produced by Nomura and co-~orkers.7(~72 In the first, Nomura and Iijima70 described methods for determining silver(1) by passing silver(T) solutions in an EDTA - ammonia - ammonium chloride buffer through a sample cell fitted with a platinum counter electrode, an Ag - AgCl reference electrode and the platinum electrode of a 9-MHz AT-cut crystal. Calibrations were obtained for A F against the concentration of silver(1). A linear calibration was obtained over the range 5 X 10-7-1 x 10-5 M of silver(1) with the equation [Ag+] = (AH19.1) x 10-6 M where AF is in Hz.The standard deviation was 2.66 Hz (2.8%) for five determi- nations at a concentration of 5 x 10-6 M . A linear calibration was also obtained over the range 5 x 10-8-5 x l o - 7 ~ .A calibration was produced without the presence of EDTA in the buffer; this also produced a linear calibration over the range 5 x 10-7-1 x 10-'M of silver(1) with the equation [Ag+] = (AH24.5) x 1 0 - 6 ~ . After the run, the silver(1) was electrolysed from the electrode of the crystal into the reagent blank at +0.S V with rcspect to the Ag - AgCl electrode. Flow-rate and pH optimisation studies were carried out together with a study of tolerance to other species. The tolerance data are shown in Table 13. A simplified cell design was reported two years later.71 A zinc rod was used as the counter electrode. which was connected to the platinum electrode of the crystal by a copper wire. 'The experiments were carried out in an EDTA - tartrate buffer.The concentration measurements were made in aANALYST. OCTOBER 1989, VOL. 114 1181 similar manner to the previous paper.70 After sevcral measurements, the crystal was removed from the cell for cleaning with nitric acid, water and acetone. Again flow-rate and pH dependency studies were carried out together with tolerance limits for ions present while determining 1 x 10-SM of silver(1). The tolerance data are given in Table 14. A linear calibration over the range 1 x 10W-1 X 1 0 - ' ~ was obtained with a standard deviation of 11.9 Hz (5.0%) for five determinations of 1 x 10-SM of silver(1) with the equation [Ag+] = (AH23.8) x 10-6 M where AF is in Hz. I n a similar paper, Nomura and T ~ u g e ~ ~ produced another flow cell design for determining silver(1) in solution.An EDTA - acetate buffer (pH 4.6) was used to mask interfer- ences. The sampling procedure was similar to the previous papers. 'Two different oscillators were used, one transistor based and the other integrated circuit (i.c.) based. From experiments without EDTA present, responses could be obtained from copper(1I) and manganese(I1) when the transistor oscillator was used but not with the i.c. oscillator. The EDTA removes these interferences. Calibrations were obtained and found to follow the equations [Agf] = (AFl8.36) X 10-6 and [Agf] = (AH10.8) x 1 0 - 6 ~ over the concentration range 1 x 10-6-3 x 1 0 - ' ; ~ of silver(1) for the transistor and i.c. based oscillators, respectively. Standard deviations were 1.42 (1.7) and 1.60 Hz (3.0%) based on five determinations of 10-5 and 5 x 1 0 - 6 ~ of silver(I), respectively.A similar method to that described in reference 68 for the determination of copper was reported in the Japanese l i t e r a t ~ r e . ' ~ In this instance copper was electrodeposited on to the crystal. Copper(I1) solutions in the concentration range 1 x 10-6-1 x 10-JM were analysed. The standard deviation was 4.97 Hz (1.74%) for six determinations of a 2 X I ~ - ' M copper(I1) solution. Piezoelectric crystals have been uscd to determine iodide in ~olution.7"7~ Reference 74 takes an electrochemical approach whereas reference 75 is based on the change in the solution characteristics which, in turn, affect the frequency of the crystal. In both instances the iodide standard was produced using potassium iodide.The former study74 used the crystal as part of an electro- chemical cell. The gold clectrodes of the crystal were first plated uith platinum and then with silver and, in operation, were held at -0.05 V with respect to Ag - AgCI. The best buffer was sodium borate - sodium hydroxide (pH 9). A linear calibration over the range 1 x 10-6-1 x 10-5 M was obtained with the equation [I-] = (AH14.1) x 10 6 M where AF is in Hz. The standard deviation was 2.51 Hz (3.3%) for six determinations of 5 x 20-6 M of iodide. Iodide plated on to the crystal during each determination was removed by electrolysis at -0.4 V each time. The latter study75 incorporated the crystal and platinum plated gold electrodes into the apparatus described in refer- ence 70.An acetate buffer was used (1 mM, pH 4.1). The effect of temperature, pH, flow-rate and viscosity of the solution on the frequency was studied. Removal of iodide from the crystal was achieved using a 0 . 0 1 ~ ammoniacal buffer (pH 9.4) containing 2 mM of thiosulphate. A linear calibration over the range 0.5-7 PM was obtained and described by [I-] = (AFl20.0) X 1 0 F M where AFis in Hz. The standard deviation was 4.02 H7 (4.0%) based on five determinations of a 5 VM iodide standard. A quartz crystal microbalance method was used by Bruck- enstein and Swathirajan76 to determine the surface coverage of under-potentially deposited species at immersed elec- trodes. The leads to a 10-MHz crystal were insulated and the real surface area of the electrode was determined by an oxygen chemisorption method by potentiostating at + 1.2 V I .~ ~ ' Y S U S SCE in 0.2 M sulphuric acid and then determining the charge for the reduction of the oxide monolayer. The crystal microbalance was calibrated by calculating the mass deposited (using Faraday's law) from the charge passed during the galvanostatic deposition of silver. Bruckenstein and Shay77 used cyclic voltammetry and a quartz crystal microbalance to elucidate the mechanism for the formation of the first monolayer of electrosorbed oxygen at a gold electrode. The mechanism is a three-step process: A U - ( H ~ O ) ~ ~ ~ + Au-OH + H+ + e- Au-OH -+ Au=O + H+ + e- Au=O + H20 + O=Au-(H20),dS The crystal microbalance and the electrochemical methods gave good agreement with respect to mass changes on the surface of the gold electrode.The underpotential deposition (UPD) coverages obtained for silver and lead in acetonitrile agreed with those measured by rotating disc electrode and rotating ring - disc electrode methods. Kanazawa and Gordon78 derived a simple relationship for the change in oscillation frequency of an AT-cut crystal in contact with a fluid in terms of material parameters of both the quartz and the fluid. The relationship is expressed by f = -So (TI PI /WQPQ)' wheref, is the frequency of oscillation of a dry crystal, qL and pL are the absolute viscosity and density of the liquid, respectively, and pQ and pu are the elastic modulus and density of the quartz, respectively. The relationship was tested for aqueous solutions contain- ing various amounts of either glucose or ethanol.The results obtained agreed well with theory a5 did those of Nomura and Minerr~ura~~ for sucrose. This agreement with theory was also noted by Kanazawa and Gordon.80 The properties of piezoelectric crystals in various salt solutions were studied by Shou-Zhuo and Zhi-HongX' with a view to developing the total salt content of natural waters; 9- MHz AT-cut crystals with silver electrodes were used. The method involved immersing the crystal in water (at 25 t 0.5"C) and noting the frequency. The crystal was then transferred into the electrolyte or test solution and the new frequency determined. For determination of the total salt in water, calibrations were obtained over the range 0.5-4.5 mmol 1-1. The effect of temperature on the response was examined together with the effect of changing the salt/electrolyte being determined.The sensitivity of the crystal to the analyte depended on the capacitance in the oscillator circuit. For example, for the salts tested a sensitivity of 232.1 k 9.6 Hz (mmol I - I ) - l was obtained with the presence of a 12-pF capacitance and -285 k 10.3 Hz (mmol I-')-' with 24 pF. Yao et d . 8 2 used an electrogravimetric method to determine the concentration of mercury in solution. A series of mercury standards was studied over the concentration range lo-7-10-5~ in a carrier electrolyte solution of sodium chloride, the concentration of this being optimised at M and at a pH between 6 and 7. The crystals uscd had either gold or silver electrodes but the gold gave the better results.For the electrodeposition, the crystal electrodes were made the cathode in the electrochemical cell with aluminium electrodes as the anodes. The separation between these electrodes and the crystals was determined and a separation of 2 mm was used in subsequent experiments. The crystal was immersed in the test solution (at 30°C) and left to stabilise (usually achieved in 30 s) and the frequency noted. Electro- deposition was then carried out for 5-10 min depending on the concentration of mercury. The electrochemical cell was then switched off and the frequency of the crystal re-determined. A plot of the frequency change against concentration was linear and the concentration of mercury was related to the frequency change by the equation [Hg2+] = 6.41 x AF x M.Good agreement with the Sauerbrey equation was obtained for the frequency change due to the deposition of the mercury.1182 ANALYST, OCTOBER 1989, VOL. 114 Table 15. Interference study carried out by Yao et af.82 The concentration of HgCI? in each instance was 5 x 20-6 M Species does not interfere at x times the molar concentration of mercury Species 200 K+, Na+. Caz+, Mg2+ 40 Znz+ 1 0 1 Cu2+ Fe3+, Ni2+, Co2+, Cd2+, Br- Seriousinterferents . . . . . . S'- , I -, co+, s2032- Table 16. Response obtained from various coated crystals to styrene85 Coating Sensitivity/ Hzp.p.m.-l Uncoated . . . . . . . . . . . . . . . . 0.023 4-rert-Butylcatechol . . . . . . . . . . . . 0.014 p-Hydroxyazobenzene . . . . . . . . . . 0.019 o-Mercaptobenzoic acid .. . . . . . . . . 0.023 2-Amino-4-chlorobenzcnethiot hydrochloride . . 0.015 Poll st) rcnc . . . . . . . . . . . . . . 0.083 PEG-400 . . . . . . . . . . . . . . . . 0.117 Table 17. Data obtained by Hartigansh on the amount of SOz sorbed by amines Average Standard slope/ deviation/ Amine pK, Hz nmol-1 Hz nmol-1 Diethanolamine . . . . . . Triethanolamine . . . . . . Isopropanolamine . . . . . . Diisopropanolaminc . . . . . . Triisopropanolamine . . . . . . Dodecylamine . . . . . . . . Didodecylamine . . . . . . Tridodecylaminc . . . . . . 2,2'-(m-Tolylimino)diethanol . . N-Phenyldicthanolamine . . . . 8.87 7.82 9.3s 9.5 6.4 10.63 11.0 9.93 4.07 4.2 17.3 30.8 50.55 16.37 65.7 34.7 74.77 100.3 21.2 23.3 2.42 2.91 0.206 1.78 6.41 1.82 10.46 13.68 0.98 1.43 Table 18.Retention data produced by Hartigang6 so2 Diethanolamine . . . . . . 91.2 Triethanolamine . . . . . . 79.8 Isopropanolamine . . . . . . 88.6 Amine retention, % Diisopropanolamine . . . . . . 80.5 Triisopropanolamine . . . . . . 34.9 Didodecylamine . . . . . . 81.2 Tridodecylamine . . . . . . 11.3 N-Phenyldiethanolamine . . . . 1.4 2.2 '-( rn-Toly1imino)diethanol . . 3.4 Dodecylamine . . . . . . . . 30.6 Standard deviation, % 3.11 4.21 5.86 2.08 2.06 3.21 9.83 0.578 0.216 0.265 Table 19. Amount of SOz absorbed by PEA coatings8x Mass of coating Mean frequency AFlm (total)/ (each side)/pg drop/Hz Hz pg- 8 9 10 1s 20 30 5 0 11 000 12 000 13 000 19 000 2s OOO 38 000 62 (MI0 Mean: 680 670 650 630 625 630 620 645 Two different crystal cleaning strategies were tested, heating at 195 "C and cleaning with different concentrations of HI.A 10% HI solution gave a cleaning efficiency of 99.7%. A study of interferences was carried out, which is sum- marked in Table 15. The adsorption of tetrabutylammonium (TBA) ions at the gold electrode - electrolyte interface was studied by Kusu et al.83 The gold electrode was glued to a quartz disc using epoxy resin. The signal amplitude and the phase were measured at the modulation frequency using a lock-in amplifier. In the presence of TBA ions there was an increase in the amplitude of the signal. Similarly, Oda and Sawadas4 incorporated a piezoelectric crystal device in a flow cell as a photoacoustic detector to monitor chromatographic effluents. It should be noted that this technique is essentially electroacoustic rather than strictly piezoelectric in the manner described in the rest of this review.Styrene Potential coatings for styrene were examined briefly by McCallum.85 The sensitivities obtained (Table 16) were of the order of those normally obtained for moisture. This indicates that more research is required to achieve a better discrimina- tion between styrene and water. Sulphur and Sulphur Dioxide An autospray system was constructed and evaluated by Hartigan86 in his investigation of amine coatings for the determination of sulphur dioxide. Initially 25 amines were examined with respect to response and drift. Of these ten were chosen for further study and are detailed in Table 17 together with their pK, values. The sensitivity was dependent on the mass of coating applied to the crystal.A linear response relating the amount of SO2 sorbed to the mass of coating was obtained (see Table 17). Within a series of amines the response increased as the basicity decreased. Didodecylamine gave a response higher than expected. This was felt to be due to this material being a solid and that solids did not hinder the ability of the crystal to oscillate. With liquids some of the energy of the crystal is dissipated in the liquid and hence its ability to oscillate is impaired. Hartigan suggested that with strong amines, interaction with the sulphur dioxide occurred at the surface of the coating whereas with the weaker amines the SO2 permeated further into the coating and, hence, gave a greater response. A dependency of the peak shape on the basicity was noted and traces were obtained for the dodecylamines tested.The stronger amines held the SO2 more tightly than the weaker ones. Table 18 lists a summary of the retention data. To overcome problems of coating loss, it was suggested that it could be possible to design a suitable polymeric coating with properties similar to the better amines examined. As a first choice he suggested that Amberlite may match the charac- teristics of N-phenyldiethanolamine (PEA) and would war- rant study. No interference studies were reported. Cheney et al.87 described the use of a 1-mm thick dimethyl silicone membrane in the gas line to form a permeation cell to monitor a sulphur dioxide source. The use of the membrane is two-fold; it allows the dilution of the SO2 from the range 100-1000 p.p.m.down to 1-10 p.p.m. (linear range reported for an ethylenedinitrotetraethanol coated crystal) and it does not allow the passage of NO2 across it. The membrane yielded no measurable NO2 at up to the 500 p.p.m. level. Two the~es46~88 extended this work on amines as coatings for sulphur dioxide. The first, by Cooke,s8 repeated the work of Hartigan86 on PEA and triethanolamine (TEA). He found that PEA was a poorer coating than suggested by Hartigan and examined whether the method of coating the crystal could have played a part in the sensitivity difference. No significant difference in sensitivity was found when the coating was applied by painting, spraying or by use of a syringe. The effect of the mass of PEA applied to the crystal on the amount of SO:! sorbed was examined and this is reported in Table 19.Triethanolamine was more sensitive to sulphur dioxide and was examined in some detail with respect to flow-rate, temperature effects, moisture and the use of a porousANALYST. OCTOBER 1989, VOL. 114 1183 membrane to minimise bleed (loss) of coating from the crystal. Higher sensitivities could be obtained with greater flow-rates but bleed and creep of the coating became a problem. A flow-rate of 50 ml min- I was found to be the best compromise. Calibrations over the ranges 0-4 p.p.m. and 0-140 p.p.m. for SO? on TEA were produced in "dry" air (700 p.p.m. of water) at a flow-rate of 50 ml min-1 and at 20 and 30°C. A higher sensitivity to the analyte was obtained at 20 than at 30 "C.At 20 "C the calibrations were linear over the range 0-5 p.p.m. and at 30°C up to 3 p.p.m. Table 20 lists the effect of temperature on the equilibrium frequency response to 1.2 p.p.m. of SO?. When the humidity of the carrier gas was changed from 560 p.p.m. of water to 1170 p.p.m. , there was a frequency change of ca. 200 Hz (estimated sensitivity of 0.33 Hz p.p.m.-I). When the coated crystal was exposed to a change in the moisture content from 750 to 1500 p.p.m. in the presence of 0.32 p.p.m. of SO2 in the carrier gas stream, the frequency change was 33 Hz (0.04 Hz p.p.m.-l, 25 "C) and at 1.2 p.p.m. of SO2 this was 75 Hz (0.1 Hz p.p.m.-l, 25°C). The use of gas permeable membranes (microporous poly- propylene; Celgard 2500,3400 and 5511) was examined to try to minimise the rate of bleed of the coating.These were placed in the cell in the form of a small tube lining the crystal compartment. They proved successful in reducing the bleed rate of TEA but at the expense of sensitivity, response time and recovery time. The response of TEA to NO2 was examined briefly by injecting known amounts of NO2 into the carrier gas stream and the results are summarised in Table 21. The idea of varying the detector cell temperature to improve the response and recovery has been suggested. Cookc88 stated that his system had a detection limit of 0.01 p.p.m. and that a concentration of 0.08 p.p.m. was well within its capabilities. For this concentration of SO2 at 20"C, the response should be of the order of 80 Hz and at 30°C this would be about 12 Hz.He also noted that TEA was not a practical coating for the routine detection of sulphur dioxide because of its high bleed rate and suggested that a polymeric amine would be better. The second thesis, by Hepher,46 developed the work of Cookex8 to the stage where field trials of coated crystals could be undertaken. N-Phenyldiethanolamine, TEA and a 1,3- diene system bound to a divinylbenzene - styrene copolymer were tested for their sensitivity to SO2. Two methods to evaluate the response were used: initial reaction rate (IKR) mode of operation and equilibrium shift (ES) mode of operation. The results from the field trials gave reasonable agreement. Table 20. Temperature effect on the equilibrium response to 1.2 p.p.m. of S0288 Experiment 1 2 3 4 5 6 7 8 Temperaturd "C 20 25 40 30 30 25 20 18.5 Equilibrium response/ Hz 7.50 480 110 320 300 465 730 900 Table 21.TEA response data88 Concentration, Exposure time/ Response/ Gas p.p.m. min Hz N O ? . . . . . . 3.33 6 300 1.66 6 165 1.66 12 270 S O ? . . . . . . 3.33 6 730 A mercury displacement reaction was utilised by Suleiman and GuilbaultXg to determine sulphur dioxide concentrations. The sample containing sulphur dioxide was bubbled through a mercury(1) nitrate solution. This liberated free mercury which was carried by the gas stream to the detecting crystal. Varying the temperature of the reaction vessel had an effect on the response obtained and on the purge time. At higher tempera- tures it was also necessary to re-condition the drying tube more often.At a fixed purge time of 10 min, the best response was obtained at 27 "C. The only major sources of interference with this coating, at the occupational safety and hygiene association (OSHA) threshold limiting value (TLV) levels, arose from sulphide gases. To reduce this interference, the test samples were also passed through a silver nitrate solution. This gave an 18% reduction in the sensitivity but, as the interferences were removed it was tolerable. Calibrations over the range 20-450 p.p.b. were produced from 2.5-ml samples and 2.5-85 p.p.m. from l-ml samples. The detection limit was of the order of 0.08 ng of SO2. The above procedure was modified to allow the determina- tion o f sulphur in liquids.')() Standard sulphur solutions were prepared by dissolving butyl sulphide in either ethanol or isooctane.Either 5- or 10-pl aliquots were transferred into a sample boat and this was inserted into the vaporisation stage of the rig. The vaporised sample was then combusted in an oxygen - nitrogen (30 + 70) gas stream. As above, the sulphur dioxide produced was bubbled through mercury(1) nitrate to liberate mercury. The mercury was then detected using a crystal with gold electrodes. Linear calibrations were obtained over the range 1-400 p.p.m. if a 5-PI sample was used and over 1-150 p.p.m. with 10 PI. Sensitivities for the two sample sizes were 1 and 2 Hz p.p.m.-l, respectively, at the 100 p.p.m. sulphur level. It was noted that the detector "is highly sensitive and is not seriously affected by the chlorine and nitrogen content of the materials.The detector will probably perform as well for the determination of sulphur in different matrices of hydrocar- bons. " It would be interesting to discover the source and level of the nitrogen and chloride content in the above mentioned reagents . Snook and Zaftgl described the use of a Quadrol coated crystal for the determination of sulphur dioxide. They claimed that this coating was hydrophobic and that it selectively absorbed sulphur dioxide. On examination of the structure of this material some sensitivity to water would be expected. Samples (1 ml; of known concentration) were injected into a carrier gas stream flowing at 60 ml min-1. The sensor is claimed to detect 0.06 p.p.m. of sulphur dioxide. Sensitivities were of the order of 20 Hz p.p.m.-l at the 1 p.p.m.SO? level and ca. 6 Hz p.p.m.-l at 10 p.p.m. Triethanolamine coated crystals were used by Edmonds et al.92 in a study of the effect of temperature on the sorption and desorption of sulphur dioxide. The crystals were placed in a stainless-steel cell, which in turn was connected to a test rig. This cell was then warmed or cooled to the required temperature (between 0 and 60 "C) for the exposure to sulphur dioxide. A number of Peltier cooling devices were tested for their suitability. Two methods of extracting the data from the continuous exposure of sulphur dioxide were used: equilibrium shift (ES), where the crystal reaches equilibrium with the ambient atmosphere, and an initial rate of reaction (IRR) method (see also references 46 and 88).With the IKR method pseudo-first- order kinetics were assumed and the sorption was allowed to continue only while it was pseudo-first-order. The SO2 - TEA followed a Langmuir-type adsorption isotherm. Plots of the detector response and coating bleed rate versus temperature were obtained. As the temperature increased, the bleed rate of the coating increased and the detector response decreased.1184 ANALYST, OCTOBER 1989, VOL. 114 Their results"' suggest the use of temperature cycling for the determination of species such as sulphur dioxide; a low temperature is required for sorption and a higher temperature to aid the removal of the analyte. The ES method is the more sensitive technique, yielding lower detection limits, but it does not give real time results below ambient conditions whereas with IRR a rapid result is offset with a higher detection limit.Toluene Bazhenov et ~ 1 . ~ 3 applied a solution containing (EtO),SiPh, methyl acctate and 1 M hydrochloric acid to the surface of a crystal at 3 - 4 5 "C and heated the resulting film at 115-125 "C for 55-65 min. 'The sensor produced had a high sensitivity to toluene vapour, stable response characteristics and a long service lifetime. Crystals coated with DC-190 silicone oil were reported by Kindlund et ~ 1 . ~ 6 to have a sensitivity of 0.1 Hz p.p.rn.-I. Karmarker and Guilbault')4 were noted in reference I to have used a Nujol mixture of tl-uns-carbonylchlorobis(tri- phenylphosphine)iridium( I), rrans-[ IrC1(CO)(PPh3)~], to detect aromatics. The effect of moisture was nil and aliphatics could be detected only at high concentrations. Pohling')' examined the response of trans-[ IrCl- (CO)(PPh3).] to a number of organics.The response of the coated crystal was dependent on the flow-rate if it was below 200 ml min-1 . 'This coating had a moisture sensitivity of 0.01 1 Hz p.p.m. 1 and a linear range of 100&20000 p.p.m. When exposed to benzene, the calibrations produced consisted of two linear regions; one region indicated a sensitivity cxl' between 5 arid 7 x 10-3 Hz p.p.m.-I and the other between 8 and 12 X lo-' Hz p.p.ni.-1. The break point varied. Acrylonitrile gavc responses similar to benzene. The sensitiv- ity for this coated crystal increased in the order benzene, toluene, ethylbenzene, xylenes and cyclohexane. Most compounds examined gave a reversible response.Aniline was irreversible. The sensitivity was of the order of 20% higher if the carrier gas used was argon rather than air and was at least two orders o f magnitude less than that claimed in reference 94. Pohling9S reported a practical detection limit for benzene of m. 600 p.p.m. A portable detector was constructed and tested by Ho et ~ 1 . 9 " to monitor toluene in the environment of a Danish printing works. This work added to a previous paper97 in which they used a Carbowax 550 coating on the crystals and Nafion tubing to remove water from the gas stream. In this instance, the device was a two-crystal system with the sensor crystal coated with Pluronic F68 and an uncoated reference crystal. Clean air was generated by passing the air through silica gel, activated charcoal and a filter.During atmospheric sampling this clean-up line was switched o u t and the line containing the Nafion tubing switched in. Dry air or nitrogen was passed across this drier to remove watcr from this sample stream. The optimum mass loading of the coating was 60 yg and the optimum flow-rate was 100 ml min-'. The response and recovery times were 40 and 30 s, respectively. Inorganic vapours such as CO, NH3, SO2 and HCI were not expected to interfere at the 100 p.p.m. level. Organic vapours did give some interfcrence cspecially benzene and alkylbenzenes. The frequency response to 100 p.p.m. of toluene was almost the same as that of 100 p.p.m. of toluene plus 5 p.p.m. of benzene, p-xylene, ethylbenzene or mesitylene. Water interfered but was removed by the Nafion drier.A linear calibration was obtained over the range 1-200 p.p.m. with an average standard deviation of 2%. The portable instrument gave results that agreed well with two reference methods: a photoionisation detector and Drager tubes. The instrument described in this work is available commercially as the PZl0l instrument (see later). Toluene Diisocyanate The initial work on the use of piezoelectric crystals for monitoring toluene diisocyanate (TDI) (a mixture of 2,4- and 2,6-diisocyanatotoluene) was carried out by Isaac.98 He examined a number of mcthods of coating the crystals and found that the use of a cotton swab gave the best results. A number of different relative molecular mass poly(ethy1ene glyco1)s (PEGs) were tested for their sensitivity to the diisocyanate and to water.The greatest sensitivity was obtained using lower relative molecular mass PEGs. This improvement in sensitivity was related to the number of OH groups present. A portable instrument, with very restricted power-pack lifetime, was described. McCallum and co-workersXS.99 published results on a range of materials with respect to their sensitivity to TDI and water. These are listed in Table 22. Of those tested. trioctylphos- phine oxide gave the greatest sensitivity. A portable instrument capable of being programmed to correct for changes in relative humidity was described.S('.85 This instrument used two crystals coated with different materials whose response characteristics to both TDI and water were known. By solving a pair of equations it was possible to carry out the correction for humidity changes over the range 4&70% RH.This two crystal approach was u5ed instead of pre-drying the gas stream because when TDI and moisture come into contact on a surface a deposit of toluenediamine is obtained in addition to the TI11 self polymerising. Therefore any pre-treatment by some drying method would result in the removal of the TDI also. In 1085, Morrison and Guilbault1()() used various silicone oils as coatings for TDI; 9-MHz AT-cut crystals were coated with silicone fluid FS-1265, DC high vacuum grease or Mastic LS420. They suggested that the interaction of the TDI with the silicone oil involved the insertion of the CO of the isocyanate group into a silicon - oxygen bond in the oil.Frequency changes of 98,64 and 57 Hz, respectively, were given for TDI at the 100 p.p.b. level. Response times to 90% of the maximum frequency change were 2.83, 2.70 and 5.04 min, respectively. No interferences were noted from 100 p.p.m. of C 0 2 , NH3, SO? or HCI. Of the common organic solvents tested, only ketones and alkylbenzenes interfered. They also noted "although water was not found to interfere directly with the TDT analysis, it was observed that the amount of water that adhered to the crystal face (coated or not) was a function of the relative humidity and not of the coating mat eri a1 . " The sensitivities were one order of magnitude greater than that produced by McCallum,Xs Table 20, for an SE-30 coated crystal (43 Hz p.p.m.-l) but the response times were much shorter.~ Table 22. Sensitivitiu of various coatings to TDI and waters'"85 Scnsitivity/Hz p.p.m. - 1 Coat i n g - Water U ncoa t ed . . . . . . . . Gelatrne . . . . . . . Repelcote . . . . . . . . PIX-400 . . . . . . . . PEG- IS40 . . . . . . . . Cobaltchloride . . . . . . Calciumchloride . . . . . . Trioctylphosphine oxide . . Dithizone , . . . . . . . SE-30 . . . . . , . . . . 0.0 I8 0.110 0.027 0.03x41.osst 0.051 0.053 3 0.0 1841.O24 0.026 0.012 * Very dependent on previous exposure. .i. Dependent on mass of the coating. $ Response was too erratic. TDI 5 * - 2s 6 25 76 43 12 -ANALYST. OCTOBEK lYXY, VOL. 114 1185 Total Organic Chlorine Gebhardt 1 0 1 dcscribed a piezoelectric method for the determi- nation of the total organic chlorine content in water.Thc crystals were coated with Amine 220, a gas chromatography stationary phase with a high affinity for HCI and almost none for sodium chloride which was reported to be the major interfcrcnt in the determination of thc total organic chlorine or halogen. The detection limit was of the order of 100 pg of HCI and 3 p.p.b. of chloroform should be detectable in a 100-pI aqueous sample. Vinyl Chloride Amine 220 [2-(8-heptadecenyl)-2-imidazoline- I-ethanol] coated crystals were reported, by Guilbault,7 to produce a linear response to vinyl chloride over the range 0.6-75 p.p.m. The vinyl chloride underwent a chromatc reaction bcfore detection by the crystals. The response obtained was irrevers- ible but reproducible for coatings in the range of Amine 220 applied (7-17 pg) and 15 samples could be injected into the gas stream before the coating failed. The crystals were exposcd continuously for periods of betwcen 0.5 and 3 min to 2 and 5 p.p.m.levels o f vinyl chloride. Several organochlorine compounds were tcstcd for interfer- cnce effects but only chlorobenzene and tetrachloroethylene were found to respond significantly. Water Tahara et ~ 1 . ~ ~ ) ~ coated an 8.95-MHz crystal with a plasma polynierised layer of styrenc and then exposcd this matcrial to fuming sulphuric acid in a nitrogen gas stream to give a polymer with sulphonic acid groups. The crystal was then mounted in a ccll (volume ca. 0.8 cm?). A stream containing thc sample and a dry gas stream was switched into the cell on the basis of 1 min of each l i e ., thc crystal is exposed alternately to sample ( 1 min) and dry gas ( I min)]. The frequency change occurring between the two streams was noted and was relatcd to the concentration of water. An excellent performance for the system over the range 1 p.p.m.-lO volo/o was reported and littlc or no interference from 07, H2, CO, CO?, C12, NH3 and various hydrocarbons and fluorocarbons. A linear response was obtained over the range 1-10' p.p.m., following the equation log AF = 0.39710g[H70] + 1.142. The system maintained a stabilised performance for more than 6 months. Japanese patents have been applied for by Shimadzu1°7.1()~ based on this coating procedure. It should be noted that the coating and proposed method of use described in these papers and patentsI0~-IO3 are very reminiscent of that described by King in his 1966 patent.Ioi King'\ patents1O5.1Oh on a water analyser were developcd into a series of commercial instruments by Du Pont and are described later.This patent lOi also described other potcntial applications for the device including the determination of aromatics in preference to paraffins. The use of four coatings on miniature 10-MHz crystals for measuring relative humidity was described by Randin and Ziillig.107 Three polymeric coatings: HEM - AMPS (HEM = 2-hydroxyethyl methacrylatc, AMPS = 2-acrylamido-2- methylpropane- I -sulphonic acid), cellulose acetate and a modified epoxy resin [prepared from a solution containing the two components CY22I and HY956 (Ciba-Geigy)], and SiO, were dip coated (polymers) or evaporated ( S O , ) on to the crystal surface.Streams of dry and water-saturated nitrogen were mixed at a flow-rate of 500 cm3 min-1 in a detcction chamber accommodating 24 quartz crystals. The selectivity ot the coatings was assessed either by adding a third gas or by replacing the nitrogen by another gas. A summary of the selectivity results is given in Table 23. Long-term stability measurements were obtaincd at 44% KH. Table 23. Selectivity of coatings expressed as the change in %RH caused by an interfering gac at 50%) RH and 25 "C. The response was measured with respect to n i t r ~ g e n " ' ~ Interfering gas (concentration in vol %) CHj CO CO? MeOH* EtOH* Me,CO* Coating (100) (100) (100) (1.7) (0.83) (3.2) HEM-AMPS .. . . 0 0 4 x 5 5 Celluloce acetate . . 0 0 7 20 14 30 Modifiedepoxy resin . . 0 0 I0 17 12 12 * The concentrations ot MeOH. EtOH and Mc2C0 were obtained by saturating 10% of thc drl nitrogen in the wlvcnt and the relative humidity bj mixing the appropriate aniounts of dry and wet nitrogen. A Flory - Huggins typc response was reported, which was associated with the first molecules sorbed tending to looscn the polymer structure locally and making it easier for subsequent niolcculcs t o enter the matrix in thc same n c i g h bo u r h ood . At RH > 70% crystals coated with HEM - AMPS ceased mcillation. With poly(AMPS) this occurred at RH > 50%. 'This effect was found to be dcpendent on the mass and composition of thc coating. Randin and Zullig concluded that the best system was HEM - AMPS but this was limited to 70% RH.King10x examined the response of a number 0 1 materials to moisture. Thesc included Quadrol, PEG- 1540, PEG'IOM, Versamid 900, Versamid 940 and polystyrene and also an uncoated crystal. A frequency change of the ordcr of 250 Hz could be obtained with air at 75%) RH at 25 "C. The response for the PEGS varicd in proportion to the number of OH groups prescnt in the molecule. The response of the versamid coated crystal increased as the mass loading increased but in cxperimcnts where the moisture content was gradually increased then decreascd, a hystcrcsis effect was noted. The magnitude of this cffcct increased with the thickness of the coating. The non-linear responscs obtained were explained in a similar manner to that dcscribed by Randin and Zullig.1()'7 Plots of Ah'/% RH against coating mass produced a curved response whereas plots of log AF/% KH against coating mass were linear.Alcohols were shown to interfcrc with the humidity response. Butanol produced the greatest interfer- ence. Immunosensors and Biochemical Uses A quartz crystal microbalancc technique was used by Shons ei ul. 10') to detcrrnine the levcl of antibonding activity i n solution; 5-MHz crystals were coated with Nybar C followed by an antigen (bovine serum albumin or horse gamma- globulin). The crystal was Mrashed, then exposed to the anti-sera at 20°C for 4 min at pH 7. The crystal was washed, dried and the frequency re-mea5ured. The technique provided a rapid qualitative and quantitative measurc of antibody activity.Thompson et rtl. 1 1 0 evaluatcd 2.5- and 5-MHz crystals with respect to their potcntial as a liquid phase immunoscnsor and described thc construction of a flow-through cell design. They initially examined how the form of the coating affected the intcrfacc properties by coating the crystals with a film of stearic acid (using Langmuir - Hlodgett technology) or by treating the surfacc with a deactivating species such as dichlorodimethylsilane or trichlorovinylsilarie. The silanes were hydrolyscd rapidly by the surface oxide/hydroxide groups to produce a coating with a hydrophobic character. The surface hydrated slowly o n immersion in watcr. The chosen antigen [goat antihuman immunoglobulin G (IgG)] was then immobilised on to the crystal.Again two approaches to immobilisation were examined. The first1186 ANALYST, OCTOBER 1989, VOL. 114 involved the use of a thin film of a poly(acry1amide) gel to support the antigen. By adjusting the monomer ratios it was possible to optimise the polymer with respect to rigidity and pore size. The antigen was fixed to the gel using a glutaral- dehyde linkage. Alternatively, the crystal was silanised using (3-glycidoxypropyl)trimethoxysilane. The epoxy functions were then oxidised to the active aldehyde groups, which were then coupled with the NH2 groups of the antigen. Tests were carried out in water and a frequency change of the order of 50 Hz was obtained for 40 pg of the antigen. Some experiments were carried out using whole blood samples but the crystals ceased oscillating after a few minutes.This was thought to be due to the cells and proteins seeping into the connection between the platinum foil and the electrode. More success was obtained with blood diluted 1 + 1 with water. In this instance the crystals oscillated for more than 1.5 h. The reactions of IgG and anti IgG on silver were studied by cyclic voltammetry and by electrogravimetry using a quartz microbalance by Grabbe et al. 111 Binding of these species both individually and together in phosphate buffers at pH 7 and 12.7 were examined. Silver electrodes were used. Anti IgG absorbed on to the surface over the potential range 0.2-0.9 V. It appeared to undergo structural rearrange- ments on reduction. The antibody was absorbed more strongly at pH 7 than at 12.7.Simultaneous measurements of current and crystal fre- quency against voltage were produced. A method for the immunoassay of antigens was patented by Oliveira and Silver112 in which quartz crystals were coated with poly(2-hydroxy-3-dimethylaminobutadiene) and then with the antigen human y-globulin. The coated crystal was immersed in a solution containing a pre-determined amount of a specific antibody and an unknown concentration of antigen and the change in frequency was determined. Seiko Instruments”3 are developing a piezoelectric immu- nosensor for the detection of Candida albicans, a yeast-like fungus, which is found in the human body. The group oxidised, anodically, the plated platinum electrodes of some AT-cut crystals. On this surface, the anti-Candida antibody was immobilised.The frequency of the crystal was noted before and after dipping into a suspension of Candida. The frequency change, due to the immunoadsorption of the fungus, was noted and could be correlated with the concentra- tion of Carzdida in the range 106-S X 10s cells cm-3. A short paper appeared in 1986 describing their results. 114 This group went on to examine the possibility of a piezoelectric based sensor that had Protein A immobilised on its surface. 115 The aim was to determine immunoglobulins. The protein was fixed to the surface via (y-aminopropy1)tri- ethoxysilane. The analytical procedure for determining IgG was similar to that described by Nomura (e.g., reference 70) for determining silver in solution. By adjusting the incubation time, the concentration range which could be examined was 10-6-10-2 mg ml-1 of human IgG.The analysis pattern of mouse IgG sub-classes was obtained. Crystal Arrays Carey’s group, at the University of Washington, have applied a chemometric approach to arrays of chemical sensors and have produced a package of pattern recognition techniques by the name of ARTHUR.116 The paper involved the application of the package to an array of 27 coated piezoelectric crystals each of which was coated with one of 27 different gas chromatography stationary phases. Response data were obtained on 14 analytes including aromatics, organophos- phorus pesticides and water. This produced an initial array of dimensions 27 (coatings) x 14 (analytes). By using statistical methods, including varimax rotation and hierarchical cluster analysis, they searched for “a sub-set of coatings that described at least 95% of the information or, in statistical terms variance, contained in the original complete data set.’’ No consideration was made to the sensitivity of the coatings to generate this sub-set but sensitivity and range data were generated once the best selectivity had been obtained.Seven eigen vectors were found and coatings were chosen from each of the vector groups for trial. In theory, the coatings with the highest contribution to the vector would be the best choice but they suggested that other factors may have to be considered, e.g., availability of the coating material. Sensitiv- ity can be considered at this point. In this instance, a coating can be eliminated and the principal component analysis procedure repeated.Their second aim was to determine what types ot chemical interaction governed selectivity. By using the varimax eigen vectors and grouping the materials they found that possible interactions could be grouped into classes such as Lewis acidity, van der Waals interaction and polarity. They also noted that “the usefulness of coating selection by using pattern recognition is dependent on the choice of coatings selected and the analytes used to test them.” Further work on pattern recognition techniques, including expressions describing sensitivity, signal to noise, selectivity and the limit of detection of individual sensors, was carried out by Carey and Kowalski.117 An array of seven sensors (either piezoelectric crystals or SAW sensors) and either seven or three analytes was used.A direct calibration matrix for the sensor array was generated by examining the response of each coated device to four concentrations of each analyte in duplicate. This matrix, C, is given by where C is an n x rn matrix of concentration of rn components in n standard samples, R is an n X p matrix of p sensors and n samples and K is the calibration matrix calculated by using standards and is a p x rn matrix of rn analytes and p sensors. Their studies indicated that better results could be obtained using an array of sensors greater than the number of analytes. Nine 9-MHz crystals in an array were each coated with a different partially selective material by Carey et al.118 to evaluate their capability for the multi-component analysis of organic vapours. The coatings, chosen from an initial group of 31 by applying a principal component analysis method, were bis(2-ethylhexyl) sebacate, ethylene glycol phthalate, Quad- rol, octahexyl vinyl ether, 1 ,2,3-tris(2-cyanoethoxy)propane, silicone SE-54, silicone DC-710, dioctyl phthalate and silicone OV-225. The sensitivities and selectivities obtained for the array are listed in Table 24. The evaluation of the system was carried out using two calibration techniques, multiple linear regression (MLR) and partial least squares (PLS), to quantify samples containing either two or three components of known composition. These test mixtures were: m-dichlorobenzene - 1,1,2-trichloro- ethane, 2-methylpentan-2-01- water and m-dichlorobenzene - 1,1,2-trichloroethane - 1,2-dichloropropane. Plots of the sum of the crystal responses were presented for the components of both binary mixtures together with C = RK Table 24.Sensitivity and selectivity for a crystal array when exposed to test mixturcs118 Sensitivity1 Coating Hz p.p.m. ~ Sclcctivity rn-Dichlorobcnzcne - 1,1,2-trichlorocthane . . . . 0.141 0.170 0.023 0.170 0.010 0.620 2-Methylpentan-2-01 -water . . 0.058 0.620 rn-Dichlorobenzene - 1,2-dichloropropanc - 1,1,2-trichlorocthanc . . . . 0.093 0.12s 0.010 0.186 0.018 0.145ANALYST, OCTOBER 1989, VOL. 114 1187 response patterns of the array to all three mixtures. Deviation from linearity occurred at the 2000 p.p.m. level for m-di- chlorobenzene and at the 12000 p.p.m. level for water.The response patterns obtained show some degree of collinearity. As the collinearity increased the error in the analysis increased. It was shown that the PLS method was better than the MLR method, which was attributed to the ability of the former to filter noise. Alternative Crystal Designs Two new piezoelectric designs have appeared in the sensing literature, a cantilever device and SAW devices. Although they are not of the standard bulk wave design they are worth mentioning briefly. O’Connor and Patton119 have patented a cantilever beam device that can be coated in the same manner as conventional crystals. The device is based upon a design for stress/pressure measurement. It can be made of silicon or quartz and can be contained, together with all of the supporting electronics on a semiconductor chip.The device is reported to follow the Sauerbrey equation with respect to mass loading. The design could be produced with a built-in heating element. There is increasing interest in the use of SAW devices for chemical sensing. A brief introduction to these devices, together with piezoelectric crystal devices, can be found in the chapter contributed by Bastiaans in reference 12. A more theoretical approach to surface acoustic waves was produced by Farnell120 who considered the single crystal case. He discussed how the solutions to the wave equation could be applied to piezoelectric and non-piezoelectric materials. Also considered was the perturbation caused by the presence of a thin layer of material to the substrate on the surface wave.Temperature sensitivity of SAW devices was discussed. Wohltjenl21 discussed further the operation of SAW devices and reported a simplified equation describing the change in frequency obtained when the device is mass loaded. The equation, where hp’ is mass per unit area of the coating and kl and k2 are material constants dependent on the substrate used in the fabrication of the device, is the SAW equivalent of the Sauerbrey equation. A study into oscillator design for use with 158-MHz twin SAW devices was undertaken by Rezgui.Iz2 Wohltjen and Dessy123 examined three alternative methods of extracting data from these devices. Circuitry was produced for measuring changes in amplitude of the wave, the change in the phase angle or the change in frequency when the device was mass loaded.Signal to noise measurements using a quartz SAW were reported (amplitude, 566; phase, 153; and frequency, 115). With a lithium niobate device the ratio using the frequency response was 696. Surface acoustic wave devices have found application for a similar range of analytes as the bulk wave piezoelectric crystals. Similar coatings and coating techniques have been employed. Examples of the usage of these devices include a gas chromatographic detector, 124 a thermomechanical poly- mer analyser, 125 pattern recognition techniques,126,127 gas or vapour sensing,128-131 sensing in fluids132.133 and microgra- vimetryhmmunoassay .134,135 AF = (kl + k2)Php’ Commercial Systems Piezoelectric crystal devices for gadvapour monitoring can be obtained from commercial sources.As noted previously, Du Pont have employed the patents of King (including reference 105) to produce a series of water monitoring systems. The specifications and capabilities of the 530 system were described by Williamson and Janzen.136 It could monitor water over two calibration ranges (0-250 and 0-10000 p.p.m.) and had response and recovery times to the final value of 4 min. Few gases were reported to cause an interference (carbon dioxide was one). Du Pont market a number of piezoelectric based systems for monitoring water, viz., the 560 series, 5600 and the 5700 Moisture Analyzers (see product data; Du Pont, Wedgewood Way, Stevenage SG1 4QN, UK). The instruments switch the analyte gas stream through a filter to remove the water and then switch to an unfiltered stream and monitor the difference in frequency which, in turn, is converted to a concentration.The instruments are reported to be able to detect moisture changes of 0.02 p.p.m. The 560 and 5700 systems are reported to monitor levels up to 9999 p.p.m. of water and with the 5600 package up to 99 999 p.p.m. The crystals used by these systems are in an environmental chamber which is kept at 60 “C. There is also an internal standard generator built into each system. A forerunner of the Du Pont system was produced by Gilbert and Barker and became commercially available in January 1964. The design and performance of the instrument were described by Crawford et ul. 137 Universal Sensors is an offshoot of Guilbault’s group.They market a piezoelectric crystal system under the designation PZ101 (see product data; Universal Sensors, 5258 Veterans Boulevard, Suite D , Metairie, LA 70003, USA). They offer 9- and 14-MHz crystals uncoated and with a series of coatings for sulphur dioxide, ammonia, aromatics, hydrogen chloride, mercury, pesticides, explosives and phosgene. They also offer to supply crystals with coatings designed to customer require- ments. Edwards produce a film thickness monitor based on piezoelectric crystal devices. One such model is the FTMS (see product data; Edwards High Vacuum, Manor Royal, Crawley RH10 2LW, UK) which can monitor two crystals (6 MHz). The instrument, which is used in the basic crystal micro- balance manner, is capable of displaying both increases and decreases in frequency and, therefore, it can be used to monitor deposition and etching of films.To use, the density (0.1-99.9 g cm-3) and acoustic impedance (1-99.9 X 105 g cm-2 s-1) values of the film material to be uscd are entered. Thicknesses in the range 0.0 nm-999.9 pm can be displayed with a resolution of 0.1 nm. Data for up to five layers can be stored. Wohltjen has set up a company, Microsensor Systems, which markets a detector/monitor system based on SAW devices (see product data; Microsensor Systems, P.O. Box 90, Fairfax, VA 22030, USA). This company also supplies a number of standard SAW sensors, coated and uncoated. Conclusions In general, the conclusions made in the previous review’ are still valid. The limitations of piezoelectric crystals are well known and some work has appeared indicating ways to maximise the recovery of information from any sensor device, e.g., the use of sensor arrays together with chemometrics.The need for high selectivities using this approach appears to be a hindrance instead of a benefit as one is examining the pattern of sorption over the array and, hence, in addition to concentration data it may be possible to identify (certainly to class of compound and in some instances to a few materials) some of the species present. Most of the papers continue to report work carried out around ambient laboratory conditions (20-25 “C) with some optimised at a fixed temperature for the chemical system. It would be interesting to find a report of a sensor that has reasonable response characteristics over a range of, say, 20 “C around ambient but it is likely that at the lower temperatures problems could arise from the kinetics of the interaction between the coating and the “atmospheric” contaminant. Certainly there are papers where a temperature programme is1188 ANALYST, OCTOBER 1989, VOL.114 used to maximise the response and minimise the time required for recovery. Indications of the reproducibility of both the coating method and the results obtained from the coated crystal(s) are now appearing in the literature. There has been an increase in the number of papers covering species in liquids , especially from Japanese workers. The majority of these are concerned with heavy metals. Others have applied enzymes and antigens to the crystals with interesting results.A few new patents have appeared. These mainly describe coating procedures (two from Japan1()3.104 and one from Canadall2) but one of note is that by O’Connor and Pattonlly u7ho have used a totally different design of crystal as a mass sensor. A new commercial piezoelectric system has appeared in the market place from Universal Sensors, the PZlOl. The company offer a range of coated crystals for a number of applications. Du Pont still offer their range of piezoelectric crystal based humidity monitors. What of the future of piezoelectric devices (crystals and SAW sensors)? Certainly the potential is there. However, a practical device requires the careful specification of the sensor and the environment in which it will be used to allow it to be tailored to a particular need.The standard pH electrode is not expected to perform well at very high or very low pH or i n the presence of high concentrations of organics or certain ionic species. It is obvious that a sensor suitable for the determina- tion of, say, sulphur dioxide in flue gases would not be totally capable of determining the same material in the headspace above a vat of an alcoholic beverage as the interferent species present in each instance would be totally different. With the tlue gas application, interference from the other acid gases present (NO,, CO, CO?, HCI, etc) would be the problem whereas with the headspace application ketones, aldehydes and higher alcohols present could be the main source of interference.This consideration is not restricted to piezo- electric devices. AH sensing devices have their strengths and limitations and therefore it is necessary to consider these when applying any sensor system to a new situation. 1 . 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. References Alder, J . F., and McCallum, J . J., Analyst, 1983, 108, 1169. Snook, R. D., Lab. Pract., 1985, July, 7. Osa, T., and Anzai. J . , Kagaku (Kyoro), 1984, 39, 138 (in Japanese); Chem. ilbstr., 1984, 100, 188141g. Karube. 1.. and Suzuki, M., Kaguku (Kyoto), 1986,41.472 (in Japanese). Mierzwinski. A., and Witkiewicz, Z., Uchr. Powietrza, 1984, 18, 73 (in Polish); Chern. Abstr., 1985, 102, 66570s. Sisoky. K.. Sb. Vys. Sk. Chem.-Technol. Praze, Fyz. Muter., Mericirech., 1985, PS, 145 (in Czechoslovakian). Guilbault, G .G.. Report for the US Army Research Office, Grant No. DAAG29-81K-0094, January lst, 1985. Nylander, C., I. fhys. E., 1985, 18, 736. Hirschfeld, T., Chem. Technol., 1986, 16, 118. Hager, H. E., Verge, P. D., and Danielson, J . L). S . , “Proceedings ot the Second International Meeting on Chemical Sensors,” Bordeaux, 1986. Turner, A. P. F., Karubc, I., and Wilson, G . S., Editors, “Biosensors,” Oxford University Press, 1987. Edmonds, T. E., Editor, “Chemical Sensors,” Blackie, Glas- gow, 1987. Guilbault, G. G., and Jordan, .I. M., CRC Crit. Rev. Anal. Chem., 1988. 19, 1. Sauerbrcy, G . Z., f’hys. Verh., 1957, 8, 113. Sauerbrcy. G . Z., 2. Phys., 1959, 155, 206. Lostis, M., Thesis, University of Paris, 1958.Stockbridge, C. D., Vac. Microbalance Technol., 1963, 3, 55. Horvdth, G., and Aranyi, L., Analyst, 1987, 112, 1451. Lu, C.. J . Vac. Sri. Technol.. 1975, 12, 578. Lu, C. and Lewis, 0.. J . Appl. Phys., 1972, 43, 4385. Beitnes, H . , and Schrgder, K., Anal. Chim. Acta, 1984, 158, 57. 22. 23. 24. 25. 26. 27. 28, 29. 30. 31. 32. 33. 34. 35, 36. 37. 38, 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. Ho, M. H., Guilbault, G . G . , and Scheide, E. P., Anal. Chim. Acta, 1981, 130, 141. Warner, A. W., in Wolsky, S. P., and Zdanuk, E. J., Editors, “Ultra Micro Wcight Determination in Controlled Environ- ments,” Interscicncc, New York, 1969. Chapter 5 . Warner, A. W., in Wolsky, S. P., and Zdanuk, E. J . , Editors, “Ultra Micro Weight Determination in Controlled Environ- ments,” Interscience, New York.1969, Chapter 7. Sulciman, A. A.. Hahn, E. C., and Guilbault. G. G., Anal. Lett., 1984, 17, 2205. Hahn, E,. C., Suleiman, A. A., Guilbault, G. G., and Cavanaugh, J . R., Anal. Chim. Acta, 1987, 197, 195. Moody. G. J., Thomas, J . D. R., and Yarmo, M. A , , Anal. Chim. Acta, 1983, 155, 225. Lai, C. S . I . , Moody. G. J., and Thomas, J . D. R., Analyst, 1986, 111, 511. Lai, C. S. I., Moody. G. J., andThomas, J . D. R., Anal. Proc., 1985,22, 11. Hlavay, J., and Guilbault. G . G.. Anal. Chon., 1978,50,1044. Fracer, S. M., Edmonds, T. E., and West, T. S . , Analyst, 1986, 111,1183. Ross, J . F., and Roberts, G . G . , “Procccdings ot the Second International Meeting on Chemical Sensors,” Bordeaux, 1986.Ali, Z . , and Alder, J . F . , poster presented at “Kcsearch and Development ‘Topics in Analytical Chemistry,” Plymouth, Guilbault, G. G . , And. Chem., 1983, 55, 1682. Kindlund, A.. Lundstrom, I., Gcodeon, A., and Hamilton, K . , IEEE Trans. Biomed. Eng., 1980, 27, 544. Kindlund. A., Sundgren, H., and Lundstriim. 1.. Senwr,s AcfuutorJ. 1984, 6, 1. Cooper, J. B., Edmondson, J. H., Joseph, D . M., and Newbower, R. S . , IEEE Trans. Biomed. Eng., 1981, 28, 459. Edmond5, T. E., PhD Thesis, University o f London, 1975. Mierzwinski, A., and Witkicwicz, Z.. Chern. Anal. (Warsaw), 1985, 30, 429 (in Polish with English abstract). Ullevig. D. M., Evans, J. F., and Albrecht, M. G., Anal. Chem., 1982. 54. 2341. Mierzwinski, A , , and Witkiewicz, Z . , Talanta, 1987, 34, 865.Elmosalamy, M. A. F.. Kohnke, F. H., Moody, G . J. Stoddart, J. F., and Thomas, J . D. R., Anal. Proc., 1989, 26, 12. Hosoya, T., Abe, S . , and Yoshida, T., Chem. Lett.. 1984,385. Alder, J . F., Bentley, A. E., and Drew, P. K. P., Anal. Chim. Sanchez-Pedreno, J. A. 0.. Drew, P. K . P., and Alder, J . F., Anal. Chim. Acta, 1986, 182, 285. Hepher, M. J . , PhD Thesis, University of Aberdeen, 1984. Edmonds, T. E., Hepher. M. J . , and West, T. S . , Anal. Chim. Acta, 1988, 207, 67. Guilbault, G . G., Tomita, Y., and Kolcsar, E. S . , Jr.. “A Coated Piezoelectric Crystal To Detect Chemical Warfare Agents in Aircraft,” United States Air Force Technical Report Guilbault, G. G., Affolter, J., Tomita. Y., and Kolesar, E. S . , Jr., Anal. Chem., 1981, 53, 2057.Tomita, Y., and Guilbault, G. G., Anal. Chem., 1980, 52, 1484. Guilbault, G . G . , Tomita, Y., and Kolesar, E. S., Jr.. United States Air Force Report SAM-TR-80-21, July, 1980 (Order No. NTIS AD A09175). Kristoff, J.. andGuilbault, G. G.,Anal. Chim. Actu, 1983, 149, 337. Cassidy, P. E . , and Thomas, R. W., “Synthesis of Resin Polymers for Chemical Agent Defence,” United States Air Force Report SAM-TR-83-48, December, 1983. Guilbault, G. G., and Kristotf, J.. Anal. Chem., 1985,57. 1754. Guilbault, G . G., Ngeh-Ngwainbi, J . , Fowley, P., and Jordan, J . , in Smyth, M. R., and Vos, J . G . . Editors, “Elcctro- chcmistry, Sensors and Analysis,” Analytical Chemistry Symposium Series, Volume 25, Elsevier. Amsterdam, 1986, Fog, H. M., and Reitz, B., Anal. Chem., 1985, 57, 2634.Sulciman, A . , and Guilbault, G. G . , Anal. Chim. Acta, 1984, 162, 97. Turnham, B. D., Lee, L. K . , and Luoma, G. A., Anal. Chem., 1985,57,2120. Fielden, P. K., McCallum. J . J.. Stanios, T., and Alder, J . F., Anal. Chrm. Acta. 1984, 162, 85. 18-19 July, 1988. Acta, 1986, 182, 123. TR80-21, 1980. pp. 335-341.ANALYST, OCTOBER 1989, VOL. 114 1189 60. 61. 62. 63, 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 1“. McKeown, D . , Rev. Sci. Instrum.. 1961, 32, 133. Andersen, H. H . , and Bay, H. L., J. Appl. Phys., 1975, 46, 1919. Blank. P . , and Wittmaack, K., J. Appl. Phys., 1979,50, 1519. Ullevig. D. M., and Evans, J. F.. Anal. Chem., 1980,52,1467.Nomura. T., and Okuhara, M.. Anal. Chim. Actu, 1982, 142, 281. Nomura, T., Okuhara, M., Murata, K., and Hattori, O., Bunseki Kagaku, 1981, 30, 417. Bel’kov, V. M., and Malinovskaya. L. M., Zh. Anal. Khim., 1984, 39, 1614 (English translation). Nomura. T., and Maruyama, M., Anal. Chim. Acra, 1983,147, 365. Nomura, T., Runseki Kagaku. 1987, 36, 93 (in Japanese with English abstract). Hager, H. E., Ruedisueli, R. D . , and Buehler, M. H., Corrosion (Houston), 1986, 42, 345. Nomura, T.. and lijima, M., Anal. Chim. Acta, 1981,131, 97. Nomura, T., and Nagamune, T., Anal. Chim. Actu, 1983, 155, 231. Nomura, T.. and Tsuge, K.. Anal. Chim. Acta. 1985, 169,257. Nomura, T., Nagamune, T., Izutsu, K., and West, T. S . , Runseki Kagaku. 1981. 30. 494 (in Japanese with English abstract).Nomura. T., and Mimatsu, T.. Anal. Chim. Acta, 1982, 143, 237. Nomura, T., Watanabc, M., and West, T. S., Anal. Chim. A m , 1985, 175, 107. Bruckenstein, S . , and Swathirajan, S . , Electrochim. Acta, 1985, 30, 85 1 . Bruckenstein, S . , and Shay, M., J . Elrctroanal. Chem., 1985, 188. 131. KanaLawa, K. K., and Gordon 11, J . G . , Anal. Chim. Acta, 1985, 175, 99. Nomura, T., and Minemura, A., Nippon Kuguku Kuishi, 1980, 1980, 1621. Kanazawa, K. K., and Gordon 11, J . G . , Anal. Chem., 1985, 57, 1770. Shou-Zhuo, Y., and Zhi-Hong, M., Anal. Chim. Acta, 1987, 193, 97. Yao, S . , Tan, S . , and Nic, L.. Fenxi Huaxue, 1986, 14, 729 (in Chinese with English abstract). Kusu, F., Hoshino, H., Takamura, K., and Sawada, T., J . I’hotoacoust., 1984, 1.463. Oda, S . , and Sawada, T., Anal. Chem.. 1981, 53, 471. McCallum. J . J . , PhD Thesis, University of Manchester Institute of Science and Technology, 1984. Hartigan, M. J.. Jr., PhD TheJiJ, Univercity of Rhode Island, 1970. Chcncy, J., Nonvood, T., and Homolya, J . , Anal. Lett., 1976, 9, 557. Cookc, S . , PhD Thesis, University of Aberdeen, 1979. Suleiman, A.. and Guilbault, G . G., Anal. Chem., 1984, 56, 2964. Suleiman, A . , and Guilbault, G . G., Anal. Left., 1984. 17, 1927. Snook. R. D., and Zaft, P. E.. Anal. Proc., 1985, 22, 300. Edmonds, T. E.. Hepher. M. J . , and West, T. S . , Anal. Chim. Acta, 1986, 187, 293. Bazhenov. V. M., Ivashchenko, V. E., Rudykh, T. A., Samodumova. I . M., Slinyakova, I. B., Cherenko, Zh. V., and Popov, K. V., USSR, SUI, 086,000. 15 April, 1984 (Russian Patent, Appl. No. 3 489 998. 8 June 1982), from Otkrytiya, Iiohret., Prom. Obraztsy, Tovarnye Znaki, 1984. 14, 84; C’hem. Ahstr. 1984, 101, 39905~. Karmarkcr, K. H., and Guilbault. G. G . , Environ. Left., 1975, 10, 237. Pohling, D . , MSc Thesis, University of ManChester Institute of Science and Technology, 1983. Ho, M. H.. Guilbault, G . G . , and Reitz, B., Anal. Chem., 1983, 55. 1830. Ho, M. H., Guilbault, G . G., and Reitz, B . , Anal. Chem., 1980. 52, 1489. Isaac. C., PhD Thesis, University of London, 1980. McCallum, J . J . . Fielden, P. R., Volkan, M., and Alder. J . F., Anal. Chim. Acta, 1984, 162, 75. Morrison, R. C.. and Guilbault. G . G . , Anal. Chem., 1985,57, 2342. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. Gebhardt, J . A., Report, EPA 600/4-84-003, Order No. PB84-129121, The Office of Research and Development, US Environmental Protection Agency, Gulf South Research Insti- tute, New Orleans, LA, 1983. Tahara, S . , Kobayashi, J . , and Oka, S . . Anal. Chem. Symp. Ser., 1983, 17, 405. Shimadzu Corporation, Jpn. Kokai Tokkyo Koho, JP 24 234 (84 24 234), 1984, February 7; Chem. Abstr., 1984, 100, 212207f. Shimadzu Corporation, Jpn. Kokai Tokkyo Koho, JP 59 182 334 (84 182 334), 1984, October 17; Chem. Absrr., 1985, 102, 105377f. King, W. H., Jr., US Pat. 3 260 104, 1966. King, W. H . , Jr., US Pat. 3 164 004, 1965. Randin, J.-P., and Ziillig, F., Sensors Actuators, 1987,11,319. King, B., MSc Thesis, University of Manchester Institute of Science and Technology, 1984. Shons, A . , Dorman, F . , and Najarian, L., J. Biomed. Muter. Res., 1972, 6, 565. Thompson, M., Dhaliwal, G. K., Arthur, C. L., and Calab- rese, G. S . , IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 1987, UFFC 34, 127. Grabbc, E . S . , Buck, R. P., andMclroy, 0. W., J . Electroanal. Chem., 1987, 223, 67. Oliveira, R. J., and Silver, S. F., Can. Par., CA 1 163 460,1984, Application 351 302, 1980; Chem. Abstr., 1984, 101, 21940a. Sensor Technology, 1987,3,5, Technical Insights, Englewood, NJ (see also Eur. Pat., Application 0 215 669, 1986, September 16). Muramatsu, H., Kajiwara, K., Tamiya, E., and Karube, I . , Anal. Chim. Acta, 1986, 188, 257. Muramatsu, H., Dicks, J . M., Tamiya, E., and Karube, I., Anal. Chem.. 1987, 59, 2760. Carey, W. P . , Beebe, K. R., Kowalski, B. R., Illman, D . L., and Hirschfeld, T., Anal. Chem., 1986, 58, 149. Carey. W. P., and Kowalski, B. R., Anal. Chem., 1986, 58, 3077. Carey, W. P., Becbe, K. R., and Kowalski, B. R., Anal. Chem., 1987, 59, 1529. O’Connor, J . M., and Patton, J . C., Eur. Pat., 0 072 744, 23 March, 1983. Farnell, G. W., Wave Electron., 1976, 2, 1. Wohltjen, H., Sensors. Actuators, 1984, 5, 307. Rezgui, N.-D., MSc Thesis, University of Manchester Institute of Science and Technology, 1988. Wohltjcn, H., and Dcssy, R., Anal. Chem., 1979, 51, 1458. Wohltjen, H., and Dcssy, R., Anal. Chem.. 1979, 51, 1465. Wohltjen, H . , and Dessy, R., Anal. Chem., 1979, 51, 1470. Ballantine, D. S . , Jr., Rose, S. L., Grate, J . W., and Wohltjen, H., Anal. Chem., 1986, 58, 3058. Setter, J. R., “Fundamentals and Applications of Chemical Sensors,” American Chemical Society, Washington, DC. 1986, Chapter 17. Jarvis, N . L . , Lint, J . , Snow, A. W., and Wohltjcn, H., “Fundamentals and Applications of Chemical Sensors,” American Chemical Society, Washington, DC, 1986, Chapter 18. Birger, W. R., Wohltjen, H., Snow, A. W., Lint, J . , and Jarvis, N. L., “Fundamentals and Applications of Chemical Sensors,” American Chemical Society, Washington, DC, 1986, Chapter 8. Nicuwcnhuizen, M. S . . Nederlof, A. J., and Coomans, A . , Fresenius Z. Anal. Chem., 1988, 330, 123. D’Amico, A., Palma, A., and Verona, E., Sensors Actuators. 1982, 3, 31. Thompson, M., Arthur, C. L., and Dhaliwal, G. K., Anal. Chem., 1986, 58, 1206. Calabrese, G . S., Wohltjen, H., and Roy, M. K., Anal. Chem., 1987, 59, 833. Roederer, J. E . , and Bastiaans, G. J., Anal. Chem., 1983, 55, 2333. Sensor Technology, 1987,3,4, Technical Insights, Englewood, NJ. Williamson, J . A., and Janzen, D. W., Anal. Instrum.. 1972, 10, 175. Crawford, H. M., Heigl, J. J . , King, W. H., Jr., and Mesh, Paper 9100846 B T. J . , Anal. Instrum., 1964, 2, 10.5. Received February 24th, 1989 Accepted April 19th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401173

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, OCTOBER 1989, VOL. 114 1191 Oxygen-sensitive Optical Fibre Transducer Philip Y. F. Li and Ramaier Narayanaswamy" Department of instrumentation and Analytical Science, University of Manchester institute of Science and Technology, P.O. Box 88, Manchester M60 IQD, UK An optical fibre oxygen transducer was developed for the measurement of gaseous oxygen. The oxygen transducer is fabricated from a polymer optical fibre, an oxygen-sensitive reagent phase and a gas-permeable membrane. Oxygen is determined by the efficient fluorescence quenching of an immobilised fluorescent indicator. The transducer fabrication, its oxygen response and the performance characteristics of this oxygen transducer a re discussed. Keywords: Optical fibre oxygen transducer; fluorescence quenching; transducer characteristics; optical fibre accessory; ox ygen-sensitive reagent phase The potential applications of optical fibres in analytical chemistry have long been recognised and were reported as early as 1969,' but it was not until recently that interest in optical fibre chemical sensors began to grow steadily, as indicated by the increasing number of publications, which have been reviewed recently.2 This renewed interest might be due to the substantial improvements in the quality, reduced cost and wider commercial availability of optical fibres.In the past, oxygen gas measurements have been performed mainly with solid-state gas sensors, which detect oxygen by utilising its paramagnetic property. Oxygen has also been determined electrochemically with oxygen electrode^.^ The development of an oxygen-sensitive optical fibre transducer is based on the use of a polymer optical fibre as the optical waveguide and on the oxygen quenching of an immobilised fluorescent indicator.The transducer is fabri- cated from a length of single-core polymer optical fibre. The oxygen-sensitive reagent phase is encapsulated at the tip of the optical fibre using a gas-permeable membrane. The oxygen transducer is operated by guiding the excitation radiation from the modulated light source through the optical fibre to the oxygen-sensitive reagent phase. The stimulated fluorescence is collected and guided by the optical fibre to the detector system, where the attenuation in fluorescence intensity is related to the level of oxygen present.The dynamic fluorescence quenching process involves the collisional deacti- vation of the immobilised fluorescent indicator by oxygen. The collisional process is extremely rapid and, as no chemical reaction actually occurs , fluorescence quenching is therefore a reversible process. These are some of the useful attributes for the development of the optical fibre oxygen transducer. The oxygen-sensitive reagent phase employed consists of the fluorescent indicator coumarin 102 adsorbed on an XAD-4 support matrix. This reagent phase was selected on the basis of its optimum sensitivity, emission intensity and stability.4 In this paper, we present the results of the performance characteristics limited to a single transducer. The potential applications of optical fibre oxygen trans- ducers are in the monitoring of industrial processes and in hazardous environments where remote sensing is desirable.Biomedical applications is another potential area because of the size of the sub-millimetre range of the transducer. It presents no electrical hazard and it is immune to electrical interference when operated in an electrically noisy environ- ment. * To whom correspondence should be addressed. Theory of Fluorescence Quenching Collisional quenching of fluorescence is generally described by the linear Stern - Volmer relationship: where I and I, are the fluorescence intensities with and without quenching, respectively, Kq is the bimolecular quenching constant, R , is the lifetime of the fluorescent molecule in the absence of the quencher, Kd is the Stern - Volmer quenching constant, which is the product of K , and R,, and [Q] is the concentration of the quencher.A linear Stern - Volmer relationship normally suggests the existence of a single class of fluorescent compound, the members of which are all equally accessible to the quencher. However, the experimental results from this study and those reported by other workers often showed a non-linear relation- ship when the data were plotted according to equation (l).5-7 The non-linear Stern - Volmer plot is indicative of the presence of a second fluorophore population that is inacces- sible to the quencher.8 An example of the non-linear Stern - Volmer relationship is observed in the quenching of tryptophan fluorescence in proteins by a polar or charged quencher.Iodide is one such quencher that cannot readily penetrate the hydrophobic interior of the protein molecule and only those exposed tryptophan groups located on the surface of the protein are quenched .8 The characteristic downward curvature shows that at higher quencher concentra- tion most of the accessible fluorophores are quenched. The non-linear Stern - Volmer relationship can be analysed by a modified form of the Stern - Volmer equations: wherefis the fraction of the total fluorophore that is accessible to quencher and Kf is the Stern - Volmer quenching constant of the accessible fraction. Experimental Reagents Coumarin 102 was purchased from A. G. Electro-optics. The solid support matrix Amberlite XAD-4 and HPLC-grade methanol were supplied by BDH.A styrene - divinylbenzene copolymer support matrix was employed. Instrumentation The instrumentation of the analytical system employed is shown in Fig. 1. The Perkin-Elmer Model LS-5 luminescence1192 _.) 04 ANALYST, OCTOBER 1989, VOL. 114 vessel a na lyser Needle valve Flow meter spectrometer is equipped with an 8.3-W xenon discharge lamp pulsed at the line frequency (50 Hz) as the excitation source. The source produces a band of energy with a width at half-peak intensity of less than 10 ps. The fluorescence signal is measured at the emission wavelength by the detector system.4 Operating in the fluores- cence mode, a second gating period occurs shortly before the next flash. The second emission signal is subtracted from the first to correct for any contribution from dark current and free from any phosphorescence of lifetime greater than 20 ms.Fluorescence spectra were recorded on an x - y recorder (Perkin-Elmer Model 057). Gaseous oxygen standard I * - Needle Gas Blending System The gas blending system for generating oxygen gas standards has been described previously.4 The oxygen composition of the gas stream was controlled by a gas blender as shown in Fig. 2. Oxygen and the inert diluent gas nitrogen were supplied from cylinders. These gases were mixed by controlling the ratio of the flow-rates of the two gases entering the mixing chamber. The gas mixture was divided into two streams. One stream was passed into a reference Munday cell oxygen analyser (Servomex Model OASOO), where oxygen can be measured quantitatively using the paramagnetic property of oxygen. The remaining gas stream was introduced to the transducer placed inside the flow cell in a direction perpen- dicular to that of the gas flow.rn t - - - Reference oxygen analyser r-l Nitrogen LI Gas blender '7' Flow cell * t Optical fibre oxygen transducer Luminescence spectrometer Fig. 1. Schematic diagram of the analytical system Transducer Fabrication The oxygen transducer was fabricated by a method similar to that described for the optical fibre pH sensor.9 The oxygen- sensitive reagent phase employed was the fluorescent indica- tor coumarin 102 immobilised on an Amberlite XAD-4 support matrix. The oxygen transducer was fabricated from a single-core polymer optical fibre of i.d. 1.0 mm and o.d. 2.2 mm with PVC sleeving (Optronics, Cambridge, UK).The reagent phase was placed against the tip of the polished optical fibre and then encapsulated by a PTFE membrane of pore size 0.5 pm (Millipore U.K., Middlesex, UK). The membrane was retained in position on the optical fibre by a length of heat-shrink polymer sleeving. Optical Fibre Accessory An optical fibre accessory was fabricated to interface a bifurcated optical fibre to the luminescence spectrometer. The accessory is fitted inside the sample compartment. The bifurcated optical fibre divides the functions of light trans- mission and reception between the oxygen transducer and the spectrometer. Procedure The response of the optical fibre oxygen transducer was investigated by the controlled introduction of oxygen to the transducer and the changes in the steady-state fluorescence signals were measured.The experimental arrangement de- signed to determine the transducer response is shown in Fig. 1. The degree of fluorescence quenching was related to the oxygen level introduced to the transduccr. For precision and reproducibility, experiments were carried out by the standard procedures described10 for transducers similar to that described here. For the oxygen transducer, these measurements were determined. from a series of 11 calibration graphs. The sensitivity of the transduccr can be determined from the slope of the modified Stern - Volmer plot of I J ( I , - I ) versus l/[Q]. The flow-rate response of the transducer was studied by passing the gas stream at a rate from 30 to 500 ml min-1 to the transducer and measuring the fluorescence signal. The temperature response of the oxygen transducer was investigated by measuring the fluorescence intensity from the oxygen transducer as a function of temperature. The flowing gas stream (50 ml min-1) was passed into 5 m of coiled stainless-steel tubing before it reached the transducer.The gas temperature was controlled by the environmental chamber. The temperature of the gas was measured independently by a reference thermocouple. The maximum temperature selected was based on temperature specification (80°C) given by the manufacturer. Fig. 2. Schematic diagram of the oxygen gas blenderANALYST, OCTOBER 1989. VOL. 114 1193 The stability of the transducer was determined in the flowing gas stream of oxygen and nitrogen.The stability of the luminescence spectrometer was investigated independently by measuring the changes in reflectance signal with time using an optical fibre probe made with a highly reflecting barium sulphate medium, which is a common reflectance standard. 11 Results and Discussion Transducer Response The oxygen response of the optical fibre transducer is shown by the typical calibration graph in Fig. 3. The dynamic range of this transducer is from 0 to 100% oxygen. The dynamic response of the transducer to an input that approximates to a “step” change in oxygen level from 0 to 100% is shown in Fig. 4. The data obtained showed that the sensor follows closely the first-order response“) and the “step” input response of the transducer is described by the equation qo = Kq,, [ l - exp(- t/T)] .. . . (3) where q,, = [ TD + llq,, qIs is the input which increases from zero to its final value at t = 0, qo is the output, K is the static sensitivity, T is the time constant and D is the differential operator. This equation shows that the speed of response of the scfiwr depends only on the value of the time constant. To check for the conformity to a first-order response and to obtain a more precise value of the time constant, the data extracted from a step-function test were re-plotted semi- logarithmically (Fig. 5 ) . From equation (3), it can be shown that 1 - (qJKq,,) = exp (-t/T) . . . . (4) 100 I 1 Oxygen, % Fig. 3. transducer together with the 95% confidence limits Calibration graph plotted from the average response of the Time/m i n Fig.4. Step-function response of the transducer If Z is defined as 2 = ln[l - (qo/Kqis)] then 2 = -t/T and dZ/dt = -1/T . . . . ( 5 ) The response of the transducer to a “step” input is shown in Fig. 4. The experimental response was compared with the theoretical response calculated from equation ( 5 ) multiplied by the relative response. The correlation coefficient between the theoretical and experimental response is 0.985. A semi-logarithmic plot of Z versus twill produce a straight line for a first-order transducer the slope of which is - 1/T. The time constant was calculated from the slope of Fig. 5. The time constants of the transducer to a “step” increase and subse- quent decrease in oxygen levels were 1.04 and 2.05 min, respectively.Precision The precision and reproducibility of the oxygen measure- ments performed by the optical fibre oxygen transducer were determined from the data from a series of 11 calibration graphs obtained over a period of 11 d. Statistical analyses of these data showed standard deviations which ranged from 3 to 7% of I (where I is the relative fluorescence intensity in arbitrary units). The average values of I together with the 95% confidence limits on the transducer’s measurements are shown in Fig. 3. The relatively large confidence limits obtained are reasonable considering the length of the experimental period involved. Sensitivity Owing to the non-linear response of the transducer’s signal to oxygen levels (Fig. 3), the sensitivity of the oxygen transducer is not constant over the &loo% oxygen range.The most sensitive region was between 0 and 10% oxygen, where the transducer can resolve 1% changes in oxygen, e . g . , from 1 t o 2% or from 5 to 6%, to give a measurable change in signal output. The sensitivity as determined by the slope of the calibration graph decreases as the level of oxygen increases. The sensitivity of the transducer can be expressed from the slope of the modified Stern - Volmer plot shown in Fig. 6, which is equal to 251.2 Torr. Hysteresis Hysteresis is the numerical difference between the average errors at the corresponding points of measurements when approached from opposite directions. This hysteresis effect was observed from the transducer’s measurements (Figs. 7 and 8) and its value was typically 5% of I . These measure- ments were carried out by going through one experimental -3 ‘ 1 1 I I 0 1 2 3 4 5 Time/min Fig.5. Step-function test of a first-order transducer1194 ' Q Q Q Q 90 ANALYST, OCTOBER 1989, VOL. 114 $. Fig. 6. Modified Stern - Volmer plot loo? 0 50 100 Oxygen, % Fig. 7. showing slight hysteresis Typical calibration graph for the optical fibre transducer IWW I 0 50 100 Oxygen, % Fig. 8. Similar calibration graph to that in Fig. 7 cycle starting with 100% and 0% oxygen concentration, respectively, i.e., 100% to 0% to 100% (Fig. 7) and 0% to 100% to 0% (Fig. 8). The hysteresis observed is likely to be in the form of a closed loop because of the reproducibility of the measurements shown by the high precision achieved by the sensor, and its fluorescence signal returns nearly to the original level after one cycle.Flow-rate Response The response of the oxygen transducer to the gas flow-rate in the range from 30 to 500 ml min-1 was studied and the results are shown in Fig. 9. A flat response was observed over the range studied. The flow-rate independence is in agreement with the theory of collisional quenching of the fluorescent indicator by oxygen. The absence of chemical reaction with oxygen implies that the continuous transfer of oxygen t a the transducer is not required in order to maintain a steady-state I 0 250 500 Flow-rate/ml mi n - 1 Fig. 9. transducer Effect of flow-rate on the response of the optical fibre I 0 ' 300 f Average response time/s 0 Fig. 10. transducer Relationship between flow-rate and the response time of the response and therefore the transducer response is indepen- dent of flow-rate. This independence is a distinct advantage over existing electrochemical sensors that consume oxygen and are therefore sensitive to flow-rate.Flow-rate versus Response Time The equilibrium response time of the oxygen transducer to variations in gas flow-rate at a constant 30°C is shown in Fig. 10. The response times of the transducer to introduction and subsequent removal of oxygen showed negative slopes in both instances, which demonstrated that the response time de- creased as the flow-rate increased. Introduction of oxygen to the transducer showed a steeper slope (0.90 ml min-1 s-1) compared with the rate of oxygen removal (0.49 ml min-1 s-1) in Fig.10. However, the difference in response time was reduced to approximately 40 s at 250 ml min-1. The flow-rate dependence of measured response times of the oxygen transducer can be explained as being due to the differences in the rates of diffusion of oxygen through the membrane and matrix material in the transducer. The measured response times are not related to the consumption of oxygen by the reagent in the transducer. Effect of Temperature on Response Time The response of the oxygen transducer was not only indepen- dent of flow-rate but was also influenced by the operating temperature. The results are shown in Fig. 11, where the temperature was varied from 30 to 70°C at a constant flow-rate of 50 ml min-1. It is interesting that the response time for oxygen introduction decreased only slowly, at a rate of 0.007 s OC-1, whereas the response time of the oxygen removal increased at 1.004 s "C-1.ANALYST, OCTOBER 1989, VOL.114 fn . .- E 4 d a 210- fn 0 P 1195 r 21 2801 A,* 0- X,’ x 30 40 50 60 70 Tern peratu rePC Fig. 11. temperature. A, Oxygen removal; and B, oxygen introduction Changes in the average response time of the transducer with The temperature response of the oxygen transducer was generally less than 5% of I over the temperature range studied. The small changes in transducer response with temperature were found not to be related to the spectral shift in the emission spectra of the adsorbed indicator, nor can the changes observed be attributed to the changes in the trans- mission characteristics of the optical fibre over the specified temperature range. Stability The optical fibre oxygen transducer was stable in the presence of both oxygen and nitrogen.The drift in fluorescence signals recorded under the experimental conditions were approxi- mately 1% and 2% per hour, respectively. The instrumental drift monitored in the reflectance mode was less than 0.5% per hour. The oxygen response of the transducer agreed well with the modified Stern - Volmer theory. This resulted in a linear slope obtained from the modified Stern - Volmer plot (Fig. 6). Such a plot can also be used to predict the fraction f of exposed fluorophores accessible to oxygen quenching. The fraction of accessible fluorophores calculated for the reagent phase is approximately 67%.The Stem - Volmer quenching constant (Kf) can also be calculated from the slope of the graph, which in this instance is calculated to be 5.95 x 10-3 Torr - 1 . The reliability of the transducer has been demonstrated by its stability and therefore frequent calibration is unnecessary. High precision and reproducibility can be achieved and the present design of the transducer has a relatively long response time compared with the response of the paramagnetic oxygen analyser of a few seconds.3 However, the response time can be improved by reducing the thickness of the oxygen-sensitive reagent phase. The operating lifetime of the transducer under continuous laboratory use extends to over 3 months without any noticeable change in performance. The completely passive nature of the optical fibre transducer is the key feature which offers some of the principal advantages over other methods of oxygen gas measurements. We are grateful to the Science and Engineering Research Council, the Department of Trade and Industry and a consortium of industrial members of the Optical Sensors Research Unit (OSRU) for financial support of this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 20. 11. References Crum, J. K., Anal. Chem., 1969,41, 26A. Rudolf Seitz, W., Anal. Chem., 1984, 56, 16A. Hitchman, M. L., “Measurement of Dissolved Oxygen,” Wiley, New York, 1978, p. 124. Li, P. Y. F., and Narayanaswamy, R., Analyst, 1989,114,663. Peterson, J . I., Fitzgerald, R. V., and Buckhold, D. K . , Anal. Chem., 1984, 56, 62. Kroneis, H. W., and Marsoner, H. J . , Sensors Actuators, 1983, 4, 587. Lubbers, D . W., and Opitz, N., Sensors Actuators, 1983, 4, 641. Lehrer, S. S . , Biochemistry, 1971, 10, 3254. Kirkbright, G. F., Narayanaswamy, R., and Welti, N. A., Analyst, 1984, 109, 1025. Doebelin, E . O., “Measurement Systems,” McGraw-Hill, New York. 1975, p. 101. Willard, H. H., Merritt, L. L., Dean, J. A., and Settle, F. A., “Instrumental Mcthods of Analysis,” Van Nostrand. New York, 1981, p. 89. Paper 8104953J Received December 19th, 1988 Accepted May 4th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401191

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, OCTOBER 1089, VOL. 114 1197 Ion Chromatographic Separation of Cobalt Cyanide Complexes James C. Thompsen and Alfred B. Carel Research and Development, Conoco Inc., P.O. Box 1267-248RDE, Ponca City, OK 74603, USA An ion chromatographic method for separating various cobalt cyanide complexes is described. The method was developed to study the synthesis of cobalt(ll1) hexacyanide and to determine the purity of cobalt(ll1) hexacyanide solutions. Separation of the cobalt cyanide complexes is achieved on an ion-pairing column, followed by chemical suppression and conductivity detection. An application of the method for the determination of radiolabelled cobalt(ll1) hexacyanide at levels 1000-fold below the detection limit with conductivity detection is described. In this instance, fractions of the effluent are collected and analysed by liquid scintillation counting.Keywords: Cobalt cyanide complex; ion chromatography; liquid scintillation counting; cobalt-60 Cobalt( 111) hexacyanide labelled with XO, "Co or WZo can be used as an aqueous tracer under a wide range of conditions. The hexacyanide i\ known to be very stable ( K = LOh4).' A method was required to check the purity of cobalt(II1) hcxacyanide. Ion chromatographic separation was applied to a study of the synthesis of cobalt(II1) hexacyanide and to the analysis of low-concentration radiolabelled cobalt(1II) hexa- cyanide solutions. Experimental Conditions for Ion Chromatographic Separation A Dionex (Sunnyvale, CA, USA) 24201 ion chromatograph equipped with an ion-pair guard (MPIC-NGI), a separator column (MPIC-NS 1) and an anion micromembrane suppres- sor (AMMS-MPIC) (all from Dionex) was used.The eluent was a mixture of 40 mM potassium cyanide, 2 mM tetrabutyl ammonium hydroxide and 28.5% methanol. The percentage of methanol is critical because a change of just I % will vary the retention times by several minutes. The eluent should be used for only about 3 d , as the stability of the system subsequently deteriorates, and fresh eluent must be prepared. The eluent is pumped at a flow-rate of 1 ml min-l. Regenerant consisting of 12.5 mM sulphuric acid flows through the suppressor at about 4 ml min- 1. The background conductivity varies with eluent age from 20 to 30 $3. The column will overload if more than 25 p.p.m. of cobalt is injected using a 5O-pl injection loop.Cobalt Cyanide Chemistry Interpretation of the ion chromatographs requires an under- standing of the chemistry of the reaction between cobalt and cyanide. The formation of cobalt(II1) hexacyanide from cobalt(I1) chloride and potassium cyanide proceeds in several steps. The first step involves the formation of a cobalt(I1) cyanide precipitate: COCI~ + 2KCN -+ CO(CN)~ + 2KCI . . (1) The next step involves the dissolution of the cobalt(I1) cyanide as three more cyanide moieties are added to the cobalt: CO(CN)~ + 3KCN -+ K,Co(CN), . . . . (2) The species CO(CN)~~- is relatively stable at low concentra- tions (<0.1 M) and only in the absence of oxygen.2 The CO(CN)~~- complex will react by two paths, one with and the other without oxygen.In the presence of oxygen, CO(CN)~~- forms an oxygen bridge complex according to the equation3 ~ C O ( C N ) ~ ~ - + 0 2 -+ [(CN)5CoOOCo(CN)5]'- . . . . (3) This oxygen bridge complex reacts with more Co(CN)+ to form cobalt(II1) pentacyanide3: [ (CN)SCoOOCo(CN)5]6- + 2Co(CN)+ + 2H20 + 4Co(CN)52- + 40H- . . . . (4) The cobalt(II1) pentacyanide can be formed quantitatively when air is added slowly to solutions of cobalt(T1) penta- ~ y a n i d e . ~ Rapid bubbling of oxygen in solutions of CO(CN)~~- increases thc yield of the oxygen bridge complex because it is formed faster than it is de5troyed by reaction with Co(CN)+-. The yield of the oxygen bridge complex is about 88% at 0 "C and 65% at 25 "C.3 'The final product of equation (4), CO(CN)~~-, might also be formed by the photodissociation of cobalt(1II) hexacyanide in bright sunlight.4 Cobalt(II1) penta- cyanide obtained by photodissociation does not show a tendency to reform hexacyanide.4 The reaction of CO(CN)~~- with cyanide [equation (S)] is slow at room temperature in the presence of base.5 Reaction ( 5 ) is catalysed by Co(CN)+.6 Co(CN)52- + CN- + Co(CN)& .. . . ( 5 ) The CO(CN)~~- complex reacts in the absence of oxygen by disproportionation with water according to the equation 2Co(CN)S'- + H20 + Co(CN)SH'- + Co(CN)SOH'- (6) This reaction is fastest at high concentrations of the complex (M.1 M) and in the presence of cations.2.7 The products of equation (6) will react further. The hydride, C0(CN)jH3-, reacts to form hydrogen and CO(CN)~~ - when heated2: ~CO(CN)~H'- -+ 2Co(CN)5'- + H2 .. (7) The product of equation (7) might react with water again according to equation (6). The hydroxide in equation (6) can react with cyanide to form the hexacyanide [equation (8)].2 This reaction is slow unless the mixture is heated. Reaction (8) is also catalysed by CO(CN)~~- .o Co(CN)SOH'- + CN- -+ CO(CN)& + OH- . . (8) The synthesis of cobalt(II1) hexacyanide was described by Bigelow.8 The over-all stoicheiometry is proposed as 2CoC12 + 12KCN + 2H2O -+ 2K,Co(CN), + 2KOH +4KC1 +H, . . . . . . (9) This is the stoicheiometry expected if equations (l), (2), (6), (7) and (8) were balanced and assumes little oxygen is present. Bigelow8 reported that the yields of hexacyanide might be as high as 90%. We modified his procedure by using a single-reaction mixture and reacting the materials for up to 2 h compared with 15 min in his work.Our yields of cobalt(II1) hexacyanide exceeded 99%.1198 ANALYST, OCTOBER 1989, VOL. 114 Table 1. Preparation of cobalt cyanide complexes KCN/g per 5 ml No. H20/ml CoCI2.6H20/g 1st addition 2nd addition Conditions 1 15 0.0216 0.0296 Mix, shake 2 15 0.0216 0.0296 Bubble air 3 15 0.0216 0.0296 Bubble N2 4 15 2.16 1.41 2.82 Boil 130 min 5 18 0.11 mg ml-1 0.87 mg per 4 ml 1.43 mg per 7 ml Boil 90 min 6 15 2.16 1.41 2.82 Boil, bubble air Table 2. Samples analysed by ion chromatography Pcak 1, YO oxygen Pcak 2, Peak 3, Fig. Solution analysed bridge complex % pentacyanide YO hexacyanide 1 2 3 4 5 6 7 8 9 1 0 Analytical-reagent grade K,CO(CN)~ Reaction 1, fresh Reaction 1, aged Reaction 2 Reaction 3 Synthesis (17000p.p.m.Co) Synthesis (10 p.p.m. Co) Synthesis (17 000 p.p.m. Co; 0,) Radiotracer, conductivity Radiotracer, scintillation 0.5 13.2 74.2 - 91.3 26.5 62.7 81.3 __ - - 1.1, o* - 100 7.3 44.9 - - - - * Two values are shown for peaks 2 and 3 to give an indication of the reproducibility 99.5 12.6 8.7 10.8 18.7 0.0 92.7 55.1 98.9,100* - 0 2 4 6 8 10 12 14 16 18 20 Time/mi n Fig. 1. cobalt(II1) hexacyanide Ion chromatogram of analytical-reagent grade potassium 0 2 4 6 8 10 12 14 16 18 20 Time/min Fig. 2. Ion chromatogram of reaction 1 products (fresh) 0 2 4 6 8 10 12 14 16 18 20 Time/m in Fig. 3. Ion chromatogram of reaction 1 products (aged) T ). > 3 U C 0 0 Y .- .- Y I I 1 1 I I I I 4 6 8 10 12 14 16 18 20 Time/m i n Fig.4. Ion chromatogram of reaction 2 products (bubbled 0,) t II I I I I I I I I 0 2 4 6 8 10 12 14 16 18 Time/m i n 0 Fig. 5. Ion chromatogram of reaction 3 products (bubbled N,) I PI 3 0 2 4 6 8 10 12 14 16 18 20 Ti me/mi n Fig. 6. tration) Ion chromatogram of reaction 4 products (high Co concen-ANALYST, OCTOBER 1989, VOL. 114 - L I 1199 t 0 2 4 6 8 10 12 14 16 18 20 Time/mi n Fig. 7. tion) Ion chromatogram of reaction 5 products (low Co concentra- t I I I 1 I I I I I 0 2 4 6 8 10 12 14 16 18 20 Ti me/m in Fig. 8. tion. bubbled 0,) Ion chromatogram of reaction 6 products (high Co concentra- I I 0 Ti me/m i n Fig. 9. dctection) Ion chromatogram of radiolabelled tracer (conductivity 600 1 E d 5001 u 400 200 1 loot 0 5 10 15 20 Ti me/m in Fig.10. tion counting) Ion chromatogram of radiolabelled tracer (liquid scintilla- Preparation of Cobalt Cyanide Complexes Reactions of cobalt and cyanide were carried out under different conditions to optimise the yield for various com- plexes. The preparation of all the complexes was performed in batches without purification. The conditions of the prepara- tions are summarised in Table 1. The batches corresponding to reactions forming complexes other than the hexacyanide are Nos. 1-3 in Table 1. In each of these reactions, cobalt(I1) chloride was added to 15 ml of de-ionised water. In the second and third reactions, air or nitrogen, respectively, was bubbled into the solutions. A total of 5 ml of water containing potassium cyanide was then added. The solutions were subsequently diluted to 25 p.p.m.of cobalt and analysed by ion chromatography. The synthesis of cobalt(II1) hexacyanide was performed using three conditions of reactant and oxygen concentration; these are run Nos. 4-45 in Table 1. In these experiments, cobalt(I1) chloride was added to 15 ml of water in a SO-ml round-bottomed flask containing a stirring bar and equipped with a side-arm and rubber septum. A water-cooled condenser was attached. The CoC12 solution was brought to the boil and the first two equivalents of cyanide were added dropwise through the septum. Solid CO(CN)~ formed during this addition. In reaction 4, air was bubbled through the mixture for the remainder of the synthesis. The second addition of cyanide was started within a few minutes of the first.At this time, CO(CN)~ dissolved and further reaction took place. The product obtained was diluted to about 25 p.p.m. of cobalt and analysed by ion chromatography. Results The ion chromatographic results are summarised in Table 2. Some simple potassium salts are by-products of the reactions. Analytical-reagent grade potassium cyanide, hydroxide and chloride were analysed by ion chromatography to ensure that they would not interfere with the cobalt cyanide peaks. The initial negative response (Fig. 1) in conductivity at about 2 min is a result of the sample solvent diluting the carrier stream and lowering its conductivity. The initial positive peak at about 3 min is assigned to K+ and the second positive peak to chloride. These peaks are prevalent in Figs.2-8. All of these ions eluted very shortly after the sample solvent front and did not interfere with subsequent analyses. Fig. 1 represents the analysis of analytical-reagent grade potassium cobalt(II1) hexacyanide obtained from Aldrich. A small impurity was detected. The chromatogram was obtained using 25 pg 8-1 of cobalt. The sensitivity of the detector is dependent on the age of the eluent, as it deteriorates with time. Therefore, peak areas should not be compared unless they were obtained from analyses run at about the same time. It is possible to compare peak areas within the same chromatogram or with others obtained within 2 h. The absolute retention time of a compound also varies with age of the eluent. A shift of several minutes can occur if the composition of the eluent differs by only a few per cent.In this study, a standard solution was run daily to verify the retention of compounds before new samples were analysed. Standardisation for detector sensitivity was not performed because only the relative chemical composition of the samples was needed. Because standard compounds are not commercially available and are stable for only short periods of time, we assumed that the detector has the same response to each ion for purposes of comparison only. Figs. 2-5 represent analyses of reactions in which the cyanide to cobalt ratio was 5. Figs. 2 and 3 are for the reaction in which the components were simply mixed together and shaken. Fig. 2 is for analysis within a few days after reaction and Fig. 3 is for analysis over a week after reaction. Three components are present in the fresh solution, but only two in aged solution.The first peak was assigned to the oxygen bridge complex, [(CN)SCoOOCo(CN)5]6-, the second to [Co(CN),]2- and the third was identifed as the hexacyanide. It is expected that the oxygen bridge complex would form CO(CN)~~- on ageing,5 which would account for the disap- pearance of the first peak in the aged sample. Repeating the same reaction but with vigorous bubbling of air through the solution gives an increase in the first peak (Fig. 4). It was expected that some CO(CN)~~- would also be obtained as the yield of the oxygen bridge complex is only about 65% at 25 "C31200 ANALYST, OCTOBER 1989, VOL. I14 (Fig. 4). The same reaction was repeated but with nitrogen bubbled through the solution.In this instance, none of the oxygen bridge complex is found (Fig. 5 ) . As oxygen was not rigorously excluded before the ion chromatographic analysis, the first peak in Fig. 5 may be a mixture of CO(CN)~~- and Co(CN)s2-. The hexacyanide complex is formed in small amounts in each of these reactions. Fig. 6 represents analyses of products from the hexacyanide synthesis performed with 17000 p.p.m. of cobalt and a cyanide to cobalt ratio of 7. The product was diluted to 25 p.p.m. of cobalt before ion chromatographic analysis. Co- balt(II1) hexacyanide from two separate reactions was obtained in high yield (98.9 and 100%). The higher yield for the second synthesis was probably a result of the longer boiling time. Fig. 7 represents the same hexacyanide synthesis but at a lower reactant concentration.The cyanide to cobalt ratio was still 7 but only 9.5 p.p.m. of cobalt was used. Pentacyanide was formed exclusively. Two factors might have influenced the low yield of hexacyanide. The first is the low ionic strength o f the dilute reaction mixture. The literature refers to several of the synthesis steps being accelerated in the presence of excess cations,’ which permits the large negatively charged species to come close together during reaction. Secondly, with 10 p.p.m. of cobalt and at 100°C, there are four times as many oxygen molecules as cobalt present in solution. An excess of oxygen during reaction can affect the hexacyanide yield. Fig. 8 represents the ion chromatographic analysis of a cobalt(II1) hexacyanide synthesis with 17 000 p.p.m.of cobalt, a cyanide to cobalt ratio of 7 and continuous bubbling of air through the reaction mixture. The hexacyanide yield decreased to 92.7%. It might be that the reaction path in the presence of oxygen is slower than in the absence of oxygen. Also, as reaction of Co(CN)s2- with CN- is catalysed by CO(CN)~~-,J it is possible that this catalyst is completely oxidised by dissolved oxygen and thus destroyed. This would result in the slow formation of CO(CN)~~-. An important result of this work was the analysis of radiolabelled tracer material. The chemical composition of this material was unknown. The main impediment to chemical analysis of the labelled tracer was the radiation level from a sample that was concentrated enough to cause a response with the conductivity detector.A novel approach was taken in this study. The radiolabelled tracer was diluted to a safe level of radiation before ion chromatographic analysis. At this dilution (less than 1 p.p.b. of cobalt), no response by conductivity detection was measured (Fig. 9). Instead, l-ml fractions were collected at the outlet of the instrument and analysed for radioactive content by liquid scintillation counting. This technique com- bined the separation power of ion chromatography with the sensitivity of liquid scintillation radioassay. The sensitivity of the conductivity detector is about 1 p.p.m. of cobalt whereas that of liquid scintillation detection is less than 1 p.p.b., which represents greater than a 1000-fold increase in sensitivity. A plot of disintegrations per minute versus time for a radio- labelled solution of cobalt-60 is shown in Fig. 10. Two forms of radiolabelled compound were present, CO(CN)~’- and Co(CN),3- The amount of tracer present as cobalt(II1) hexacyanide was 55% in this sample. Conclusion The separation of cobalt cyanide complexes by ion chromato- graphy has been described. The assignment of peaks other than the hexacyanide has been proposed, based on the chemistry of the cobalt - cyanide reaction. The sensitivity of the ion chromatograph was increased over 1000-fold by the use of cobalt-60-labelled cobalt complexes and liquid scintilla- tion detection. 1. 2. 3 . 4. 5. 6. 7. 8. References Dilts, R. V., “Analytical Chemistry,” Van Nostrand, New York, 1974, p. 561. King, N. K., and Winfield, M. E.,J. Am. Chem. Soc.. 1961,83, 3366. Haim. A . . and Wilniart, W. K., J . Am. Chem. SOC., 1961, 83, 509. MacDiarmid, A. G., and Hall, N. F., J . Am. Chem. SOC., 1953, 75, 5204. Bayston, J . H., Aust. J . Chem., 1963, 16, 954. Birk, J . P., and Halpern, J . , J . Am. Chem. SOC., 1968,90,305. King, N. K., and Winfield, M. E . , J . Am. Chm. SOC., 1958,80, 2060. Bigelow. J . H., Inorg. Synth., 1946, 2, 225. Paper 9100437H Received January 26th, 1989 Accepted May 4th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401197

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, OCTOBER 1989, VOL. 114 1201 Determination of Nitrite, Sulphate, Bromide and Nitrate in Human Serum by Ion Chromatography Yoshimasa Michigami, Yoshikazu Yamamoto and Kazumasa Ueda Department of Chemistry and Chemical Engineering, Faculty of Technology, Kanazawa University, 2-40 Kodatsuno, Kanazawa 920, Japan An ion chromatographic method has been developed for the determination of trace amounts of nitrite, sulphate, bromide and nitrate in human serum, using an ODS column dynamically coated with cetylpyridinium chloride. The anions studied were eluted with 1 mM citrate - 2.5% methanol (pH 6.5) as the mobile phase and detected by an ultraviolet detector. The interfering proteins in human serum were removed by an initial filtration through an ultrafilter-paper. The many inorganic and organic anions commonly found in serum had little effect on the determination of the four anions.Recoveries of nitrite, sulphate, bromide and nitrate in serum were 107-1 10,94-106, 106-1 10 and 92-loo%, respectively. The proposed method was also applied to human saliva and urine. Keywords: Serum nitrite, sulphate, bromide and nitrate determination; ion chromatography; cetylpyridinium chloride coated column Nitrite, sulphate, bromide and nitrate are usually present in biological fluids at the p.p.m. level. It has been reported that inorganic sulphate plays an important physiological role in the human body. 1 For nitrate, little information has been pub- lished concerning its metabolism and no conclusive evidence has been found for its essential function in the human body.' In contrast, bromide has been shown to be essential in the human body from toxicological data The concentration of serum bromide is reported to be dependent on eating habits and geographical or environmental factors; drugs containing bromide ( e .g , sedatives and nerve tonics) can also raise the serum bromide level. 1.2 A number of papers have reported the determination of inorganic anions in serum by ion chromatography; these have included the simultaneous determination of inorganic phos- phate, bromide, nitrate and sulphate in human serum1 and the determination of thiocyanate,3 bromide4 and sulphate.5 The first1 requires a 10-fold higher sensitivity for bromide and nitrate, compared with phosphate and sulphate. Few papers have been published on the determination of serum nitrite.This paper describes the simultaneous determination of nitrite, sulphate, bromide and nitrate in human serum by ion chromatography in which the four anions can be recorded at the same sensitivity, employing an ODS column dynamically coated with cetylpyridinium chloride, eluting with a citrate - methanol solution after deproteinisation of the serum using an ultrafilter-paper and measuring the absorbance at 210 nm. The proposed method was also applied to the determination of the four anions in human saliva and urine. Experimental Apparatus and Reagents The ion chromatographic equipment consisted of a pump (CCPD , Tosoh) , a variable-wavelength ultraviolet detector (UV-8000 or UV-8011, Tosoh), an injector (Rheodyne), a column oven (CO-8000, Tosoh) and a pen recorder (YEW Type 3066, Yokogawa).The operating conditions are given in Table 1. All chemicals were of analytical-reagent grade, and the solutions used were prepared with de-ionised, distilled water. Standard solutions of the anions were prepared from the corresponding sodium salts. The eluent of 1 mM citrate - 2.5% methanol was adjusted to pH 6.5 with dilute sodium hydroxide solution and filtered through a 0.45-pm membrane filter before use. Table 1. Operating conditions Column . . . . . . . . TSK gel ODS-80TM column dynamically coated with cetyl- pyridinium chloride Detection wavelength . . 210 nm Columntemperature . . 35°C Flow-rate . . . . . . 1.0 ml min-1 Sampleloop . . . . . . 100pl Eluent . . . . . . . . 1 m~ citrate - 2.5% methanol (pH 6.5) Dynamically coated columns were prepared using columns (50 x 4.6mm i.d) packed with ODS resin (TSK gel ODS-80TM, S pm particle size, Tosoh).The coating procedure was similar to that used by Duval and Fritz.6 The column was coated with about SO ml of 0.01 M cetylpyridinium chloride in 10% methanol at a flow-rate of 0.5mlmin-1 and then conditioned with the eluent before testing. The capacity of the ion-exchange resin was about 0.3 mequiv. g-1. The coated column allowed the analysis of about 70 samples. Regenera- tion of the column was then carried out by washing with methanol and coating with cetylpyridinium chloride. Procedure Thaw a frozen serum sample and take 0.5 ml of the sample in an ultrafiltration vessel. Dilute to 1 ml with de-ionised, distilled water.Filter the sample solution through an ultra- filter-paper (relative molecular mass 30 000 cut-off mem- brane) and transfer 0.5 ml of the filtrate into a 5-ml calibrated flask. Dilute to volume with de-ionised, distilled water and inject 100 pl of the sample solution on to the column. Identify the nitrite, sulphate, bromide and nitrate peaks on the chromatogram by comparison with the retention times of their standards and measure their peak heights. Calculate the concentrations of nitrite, sulphate, bromide and nitrate by using the corresponding calibration graphs which are con- structed daily. Results and Discussion Effect of Eluent pH on the Retention Times of Nitrite, Sulphate, Bromide, Nitrate and Chloride The effect of eluent pH on the retention times of nitrite, sulphate, bromide, nitrate and chloride was examined in the1202 15 C .- E '+z, 10 .E" t 0 C a, a, CE .- 4- + 5 - ANALYST, OCTOBER 1989, VOL. 114 - - 15 C .- E -.. .- E" - l o 0 a, [r .- +J 4- 5 A A E A I I I 5.0 6.0 7.0 PH Fig. 1. Effect of eluent pH on retention time. A, Nitrate; B, bromide; C, sulphate; D, nitrite; and E. chloride E . I I - I 0.50 1 .oo 1.50 Citrate concentration/mM Fig. 2. B, bromide; C, sulphate; D, nitrite; and E , chloride Effect of citrate concentration on retention time. A, Nitrate; - * * . E I I I 0 5 10 Methanol concentration, YO Fig. 3. Nitrate; B, bromide; C, sulphate; D, nitrite; and E, chloride Effect of methanol concentration on retention time. A, pH range 4.5-7.0. The results obtained are shown in Fig 1.The retention times of these anions decreased with increasing eluent pH; a marked decrease in the retention time was observed for sulphate. The extent of the decrease in the retention time of each anion was smaller in the range pH 6.0-7.0 compared with that in the range pH 4.5-6.0, as shown in Fig. 1. Hence an eluent pH of 6.5 was used, which gave a good separation of each anion. Effect of Citrate Concentration on the Retention Times of Nitrite, Sulphate, Bromide, Nitrate and Chloride The effect of the citrate concentration on the retention times of nitrite, sulphate, bromide, nitrate and chloride was 1 .o a, C (II 2 2 0.5 Q n 0 200 220 240 Wavelengthhm Fig. 4. Absorption spectra. I, Nitrite (10 pg ml-I); 11, nitrate (10 pg ml-1); and 111, bromide (10 pg ml-I) ~ Table 2.Relative value of the retention time for various anions ( t A ) relative to that of bromide (tBr) and relative sensitivity Anion Gluconate . . . . Aspartate . . . . Iodate . . . . . . Hydrogen carbonate Fluoride . . . . Lactate . . . . Phosphate . . . . Ascorbate . . . . Acetate . . . . Formate . . . . Pyrophosphate . . Uriate . . . . . . Chloride . . . . Pyruvate . . . . Bromate . . . . Nitrite . . . . . . Succinate . . . . Sulphate . . . . Butyrate . . . . Malonate . . . . Tartrate . . . . Oxalate . . . . Bromide . . . . Benzoate . . . . Nicotinate . . . . Nitrate . . . . . . Glucuronate . . . . a-Ketoglutarate . . Tannicacid . . . . Chlorate . . . . Iodide . . . . . . Phthalate . . . . Perchlorate . . . . Thiocyanate . . . . Thiosulphate ., * Negative peaks. . . . . . . . . . . . . , . . . . . . . . . . . a . . . . . . . . . . . . . . . . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . t AItB r 0.15" 0.2 0.2 0.2" 0.2" 0.2" 0.2" 0.25" 0.25" 0.3" 0.3" 0.35 0.4" 0.5 0.55 0.6 0.65" 0.75* 0.8 0.8* 0.9 0.9 1.0 1.2 1.5 1.55 1.9 2.2 2.2 ~ ~ 2 . 5 >>2.5 >>2.5 >>2.5 >>2.5 >>2.5 Relative sensitivity (anion : Br) 0.04 0.02 1.57 0.02 0.56 0.07 0.27 0.01 0.19 0.19 0.22 7.83 0.39 0.51 0.41 5.00 0.03 0.18 0.02 0.02 0.04 0.16 1 .oo 0.004 0.90 2.42 0.0006 0.21 0.30 - - - - - - examined in the concentration range 0.50-1.50 mM at pH 6.5. Fig. 2 shows the results obtained. The retention times of all five anions decreased gradually with increasing citrate concen- tration. In order to achieve a good separation of the five anions, 1.0 mM citrate was chosen as the optimum concentra- tion.Effect of Methanol Concentration on the Retention Times of Nitrite, Sulphate, Bromide, Nitrate and Chloride The effect of the eluent methanol concentration on the retention times of nitrite, sulphate, bromide, nitrate andANALYST, OCTOBER 1989, VOL. 114 1203 chloride was examined in the concentration range 0-10%. The results obtained are shown in Fig. 3. The retention times of nitrate and bromide decreased markedly whereas those of nitrite and chloride decreased gradually with increasing methanol concentration; little change was observed for the retention time of sulphate. Table 3. Recovery of nitrite, sulphate, bromide and nitrate in scrum Anion Addedlpg ml- 1 Nitrite .. . . . . 0.10 0.07 0.05 2.5 2.0 1 .o Bromide . . . . 0.60 0.40 0.25 0.20 Nitrate . . . . . . 0.20 0.15 0.04 0.03 Sulphatc . . . . 5.0 n 7 3 2 5 2 3 2 3 4 2 2 2 3 2 4 Recovery. % 107.7 108.0 110.5 106.6 103.0 100.3 94.0 110.6 107.6 106.5 107.0 95 .O 92.9 99.5 97.9 O Fig. 5. IO.001 A 5 10 15 Retent ion ti me/m i n Chromatogram of a serum sample For the separation of nitrite and sulphate, an eluent methanol concentration of less than 1% gave poorer results than 1-5% methanol. For the separation of bromide and sulphate or chloride and nitrite, a concentration of more than 5% methanol also gave poorer results than 1-5% methanol. Hence an eluent methanol concentration of 2.5% was used. Detection Wavelength The absorption spectra of nitrite, bromide and nitrate over the range 195-250nm are shown in Fig.4. The maximum absorbances of nitrite, bromide and nitrate were found to be at 215, 205 and 210 nm, respectively. The base line was less stable at 205nm. Hence 210nm was used as a suitable detection wavelength. The molar absorptivities of nitrite, bromide and nitrate at 210nm were 1500, 400 and 1600 1 mol-1 cm-1, respectively. There was little absorbance, however, in the range 200-250 nm for sulphate. Therefore, sulphate was determined by an indirect detection technique7 using the difference between the absorbance of the eluent and sulphate at 210 nm. Retention Times of Various Anions The retention times and the sensitivities of various inorganic and organic anions were examined using the proposed method.The retention times and the sensitivities relative to bromide are shown in Table 2. The relative retention times of nitrite, sulphate and nitrate were 0.60 k 0.03,0.75 k 0.06 and 1.55 k 0.05 [n = 10, mean k standard deviation (SD)], respectively. These anions showed no interaction with each other. Few peaks were observed near the nitrite, sulphate, bromide and nitrate peaks. The relative retention time of chloride was 0.40 and those of phosphate, lactate and hydrogen carbonate were each 0.20. Calibration Graphs, Detection Limits and Precision The calibration graphs were rectilinear in the range 0.03- 10.00, 0.5-10.0,0.05-10.00 and 0.05-50.00 pg ml-1 of nitrite, sulphate, bromide and nitrate, respectively. The detection limits of nitrite, sulphate, bromide and nitrate were 0.005, 0.15, 0.02 and 0.008 pg ml-1, respectively.The detection limits were defined as the concentration that produced a signal equal to three times the background noise level. The relative SDs ( n = 5 ) were 1.3% for a 0.1 pg ml-1 concentration of nitrite, 2.0% for a 4 pg ml-1 concentration of sulphate and Table 4. Analysis of serum Conccntratiodpg ml-1 (mean f SD) Anion SCX Nitrite . . . . . . . . M l- M + F Sulphate . . . . . . . . M F M + F Bromide . . . . . . . . M M + F Nitratc . . . . . . . . M F M + F * Lit. = literature value. i Whole blood. Proposed method (n = 34) Lit.*1 Lit.'- 0.19k0.14 - - 0.22 * 0.10 - - 0.20f0.13 - - 40.5 & 12.1 31.2 k 4.6 - 39.3 -t 6.4 31.2 f 5.6 - 39.9 f 9.8 - - 8.89 ? 2.41 5.35 f I .04 - 8.46 k 2.26 5.19 f 1.36 - 8.69f2.35 - 0-99.9 3.88 f 2.01 2.23 & 1.0 - 3.68 k 2.54 2.23f 1.2 - 3.79 i 2.28 - - Lit.4 Lit.5 Lit.8 Lit.', Lit.10 - - - - - - - 0.01-0.26t - - - - 2 7 .6 i 1.8 - - 29.4 i 1.2 27.6 f 1.2 28.6 & 0.9 - - - 30.4 kS.3 - - 28.7 f 2 . 4 - 20.6k33.4 - - - - - - - - - - - 3.5-10 - - - 4.65-35.80 - - - - -1204 60-79 M (0 = 4) F ( n = 3 ) ANALYST, OCTOBER 1989, VOL. 114 (a) * * 40-59 & M ( n = 6 ) 4 F ( n = 6 ) 20-39 F ( n = 7 ) I w% - M(n=8)- 1 I I 0.2 0.4 40 80 5.0 10.0 5.0 10.0 Co nee n t ra t i o n/iLg m I -- Fig. 6. (d) nitrate in serum for various age groups Concentration of ( a ) nitrite; ( h ) sulphate; ( c ) bromide; and IO.001 A Br- h 0 Retent ion t ime/m i n Fig. 7. Chromatogram of a saliva sample Table 5. Analysis of saliva Concentratiodpg ml-1 No.Sex 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 F 12 M 13 F Lit.Jrs . . . . M F M + F Lit.lI . . . . - Lit.12 . . . . - Lit.13 , . . . - Nitrite Sulphate Bromide 3.67 8.5 29.52 0.63 6.3 1.85 1.78 11.3 29.28 6.35 8.0 4.53 5.54 8.4 4.46 1.15 8.7 3.91 3.08 ND* 77.12 6.82 1.2 5.72 5.20 5.0 3.96 1.81 7.8 3.72 0.41 6.0 6.32 0.60 6.9 3.07 1.86 ND 8.08 - 6.8 4 0 . 6 - - 7.1 k 0 . 7 - - 6 . 9 t 0 . 4 - 6.7-7.0 2.76-11.67 - - 4.73-11.63 - - 1.38- 8.74 - - * ND = Not detected. + Lit. = literature value. Nitrate 20.76 4.56 5.68 34.13 8.49 18.19 20.92 31.19 11.09 5.95 26.39 9.25 28.36 2.0% for a 0.4 pg ml-1 concentration each of bromide and nitrate. Recoveries of Nitrite, Sulphate, Bromide and Nitrate in Serum The recoveries of nitrite, sulphate, bromide and nitrate were determined by spiking known concentrations into a serum sample that had been subjected to ultrafiltration.As shown in Table 3, the recoveries of nitrite, sulphate, bromide and Retention ti me/m i n Fig. 8. Chromatogram of a urine sample Table 6. Analysis o f urine Concentratiodpg ml- No. 1 2 3 4 5 6 7 8 9 10 11 SCX M M M M M M M M M M M Nitrite 0.63 ND* 0.93 ND ND ND 1.15 ND ND ND ND Sulphate 918.2 593.0 I 0 I 0. 8 778.9 1701.6 2(N). 8 617.3 618.3 807.3 958.6 1719.6 Bromide 8.91 34.26 75.36 5.00 11.77 21.40 140.46 54.95 24.19 3.02 ND Nitrate 126.89 54.13 66.36 118.59 196.43 10.61 204.82 120.89 78.12 88.47 106.43 _. 0.80-4.63 - Lit:fI() . . . . - - Lit.14 . . . . - - 4611 - - Lit.15 . . . . - 0.0254.136 - - 38.5-183.7 * ND = Not dctcctcd. 1 Lit. = literature value nitrate were 107-1 10, 94-106, 106-110 and 92-100% , respec- tively, and the rclative SDs of the four anions were 0.01- 0.08%.Chromatogram of Serum Fig. 5 shows a typical chromatogram o f a serum sample. The sulphate (retention time 7 min), bromide (9 min) and nitrate (14 min) peaks were free from interference by the other anions prcscnt in the serum. The nitrite (5.5 min) peak was little affected. The peak that appeared betwecn the chloride (3.5 min) and nitrite peaks was confirmed to be due to pyruvic acid, whereas the peak observed between the nitrite and negative sulphate peak could not be identified. The positive peak (1 min) may be caused by cations and the negative peak (2 min) may be a result of the overlapping hydrogen carbo- nate, phosphate and lactate peaks from the serum.Determination of Nitrite, Sulphate, Bromide and Nitrate in Serum Nitrite, sulphate, bromide and nitrate were determined simultaneously in 34 serum samples using the proposed method. The results obtained and those of other work- ers1J.4,s7*-10 are shown in Table 4. The average contents of nitrite, sulphate, bromide and nitrate in serum were found to be 0.20, 39.9, 8.69 and 3.79pgml-1, respectively. The average values of sulphate, bromide and nitrate were similar to those found in other studies.*.'.~.5.*-l0 The data were classified into six groups relating to the age and sex (M, male;ANALYST, OCTOBER 1989, VOL. 114 1205 F, female) of the subjects, as shown in Fig. 6. N o significant differences were found in terms of age and sex for the concentrations of these anions.Determination of Nitrite, Sulphate, Bromide and Nitrate in Saliva and Urine Saliva and urine samples were centrifuged and 0.5-ml aliquots of their supernatant solutions were diluted to 10 and 50m1, respectively. The diluted solutions were injected on to the column and analysed under the conditions given in Table 1. Fig. 7 shows a chromatogram of saliva. The nitrite, sulphate, bromide and nitrate peaks suffer little interference from the other anions present in saliva. The nitrite peak in saliva was higher than that in serum, whereas the sulphatc peak in saliva was much lower compared with that in serum. The results obtained are shown in Table 5 . The concentrations of nitrite and sulphate in saliva samples were similar to those found by other workers.’,ll-’3 Fig.8 shows a chromatogram of urine. The sulphate, bromide and nitrate peaks were free from interference by the other anions present in urine. However, the peak for nitrite in urine was only detected in a few instances (Table 6). The sulphate and nitrate peaks in urine were generally much higher than those in serum. The results obtained are shown in Table 6 and arc compared with those found by other The proposed method may therefore be applied to the determination of nitrite, sulphate, bromide and nitrate in various biological samples. workers. 10,14,15 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References De Jong, P., and Burggraaf, M., Clin. Chim. Acta, 1983, 132, 63. Versicck, J., and Cornalis, R., Anal. Chim. Acta, 1980, 116, 217. Michigami, Y., Takahashi, T., He, F., Yamamoto, Y., and Ueda, K., Analyst, 1988, 113, 389. Miller, M. E., and Cappon, C. J., Clin. Chem., 1984,30, 781. Cole, D. E. C., and Landry, D. A., J. Chromatogr., 1985,337, 267. Duval, D. L., and Fritz, J. S., J. Chromatogr., 1984, 295, 89. Gjerd, D. T., and Fritz, J. S., “Ion Chromatography,” Second Edition, Huthig, New York, 1987, p. 28. Tanaka, A., Nose, N., Saito, S., Masaki, H., and Watanabe, A., Bunseki Kagaku, 1981, 30, 269. Gamoh, K., and Yagi, T., Anal. Sci., 1988,4, 433. Vallon, J. J., Pegon, Y., and Accominotti, M., Anal. Chim. Acta, 1980, 120, 65. Funazo, K., Tanaka, M., and Shono, T., Anal. Chem., 1980, Ohta, T., Arai, Y., and Takitani, S., Anal. Sci., 1987, 3, 549. Tanabe, S., Kitahara, M., Nawata, M., and Kawanabe, K., J. Chromatogr., 1988, 424, 29. Anderson, C., Clin. Chem., 1976, 22, 1424. Cox, R. D., and Frank, C. W., J. Anal. Toxicol., 1982,6,148. 52,1222. Paper 910121 01 Received March 21st, 1989 Accepted May 24th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401201

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. OCTOBER 1989. VOL. 114 1207 Gas Chromatographic Determination of Volatile Anaesthetic Agents in Blood Part 1. Preparation of Standard Gas Mixtures of Volatile Anaesthetic Agents John Flynn, J. Declan O'Keeffe and William S. Wren Department of Anaesthesia, Children's Research Centre, Our Lady's Hospital for Sick Children, Dublin 12, Ireland Imelda M. Shanahan* School of Chemical Sciences, Dublin City University, Glasnevin, Dublin 9, Ireland A method for preparing standard gas mixtures of the volatile anaesthetics halothane, enflurane and isoflurane is described. Static mixtures of gases of known concentration can be prepared manometrically by measuring the required pressure of anaesthetic gas into a bulb and diluting to atmospheric pressure with air. Standard gas mixtures in the concentration range 0 4 % V/Vcan be prepared with an accuracy of +0.01% V/V, and the relative standard error of measurements of a single standard concentration is less than 0.8%.Significant adsorptive losses in the gas sampling valve were observed for gas standards prepared in the absence of any diluent gas. These losses were not detected for measurements of standards made up to atmospheric pressure in air. A comparison with calibration procedures currently in practice is presented. Keywords: Calibration gas mixture; standard gas mixture; volatile anaesthetic; halothane, enflurane and is0 flurane standard gas mixtures; gas chromatography The extensive use of 2-bromo-2-chloro-1 ,I ,I-trifluoroethane (halothane), 2-chloro-l,l,2-trifluoroethyl difluoromethyl ether (enflurane) and l-chlor0-2,2,2-trifluoroethyl difluoro- methyl ether (isoflurane) as general anaesthetics has led to the development of numerous methods for their determination in body fluids and tissues.Although the selectivity and sensitivity of gas chromatography make it an ideal technique for this purpose, it is not widely used owing to the difficulties associated with preparing standard gas mixtures for calibra- tion. Primary standard gas mixtures can be prepared by a variety of static and dynamic methods, which have been reviewed by Nelson1 and by Barratt.' Static mixtures are easier to prepare and are particularly useful when relatively small volumes are required. Gravimetric methods potentially offer the greatest accuracy and have been recommended in standard procedures for preparing reference gas mixtures,3 but they are not suitable for routine laboratory use.Volumetric methods involving the addition of known volumes of gases to vessels of fixed dimensions have also been described.2.4 Gray4 described such a method for the preparation of standard mixtures of volatile anaesthetic agents based on a volumetric addition of volatile agents to a 500-cm3 gas sampling bulb fitted with a septum for addition and withdrawal of gases and liquids. The main disadvantages of this method are that it is time consuming and requires considerable care in the manipula- tions involved in the preparation of gas mixtures. The purpose of this work was to develop a rapid and easy to use method for the routine preparation of gas mixtures for calibration purposes.The calibration method will ultimately be used to allow accurate determinations of volatile anaes- thetic agents in venous and arterial blood of patients under- going surgical procedures. This paper describes a method based on the measurement of partial pressures of the components of the mixture which is convenient to use and is suitable for routine laboratory use. The major advantages of this method over alternatives are the speed and ease with which gas mixtures may be prepared. The method is also less * To whom correspondence should bc addressed. sensitive to the skill and care with which the manipulations are performed, thus leading to greater precision than that attainable with alternative methods. Experimental Materials The method has been used for the preparation of standard mixtures of halothane (Fluothane, TCT; Hoechst) , enflurane (Abbott Laboratories) and isoflurane (Forene, Abbott Lab- oratories) in air.For each of the volatile agents investigated, the standard mixtures were prepared in medical-quality air (Cryogas) . Apparatus The apparatus, shown in Fig. 1, consists of a conventional mercury-free high-vacuum line made of Pyrex glass and fitted with greaseless PTFE taps (J. Young, Acton). The vacuum is achieved by means of a single-stage rotary pump (Edwards, Speedivac 2) and is measured by a Pirani gauge (Edwards, PRElOK). The volumes of bulb A (2417.0 cm3), bulb B (1715.0 cm3) and bulb C (522.2 cm3) were measured prior to attachment to the main line. The volumes of line 1 (66.9 cm3), i K To pump Line 3 To GC Fig.1. Schematic representation of the vacuum system. A, B, C, gas storage and mixing bulbs; D, E, F, G, liquid reservoirs and gas inlets; H, pressure transducer; J , Pirani gauge; K, liquid nitrogen trap; 1, air admittance valve; 2, isolation valve; 0, Teflon taps1208 ANALYST, OCTOBER 1989, VOL. 113 line 2 (76.6 cm3) and line 3 (11.4 cm3) were measured by sharing known precsures of air into them from bulb A. Gases are introduced into the system from reservoirs at D, E, F and G. The acceptable working vacuum conditions are determined by the application for which the system is intended. For the present application, the working limit is defined by the senstivity of the gas chromatograph detector. At an ultimate vacuum of 1 0 - 2 mmHg, the number of molecules present in the gas phase at 20 "C is 3.3 X 1014 crrir3, which is well below the limit of detection of the flame-ionisation detector (FIII) under these operating conditions ( I .h x 101' molecule\ cm 3 ) ; hence this represents an acceptable working limit for the vacuum system.Absolute gas pressures are measured on a pressure trans- ducer (MKS Baratron, 122H). 'The accuracy with which the pressure measurement may be made depends on the pressure range in which measurements are required. The (k1000 mmHg model was used in this study to measure pressures in the range 0.0-1000.0 mmHg with an accuracy of i0.3% at full-scale and kO.1 mmHg at pressures less than 30 mmHg. Thus standard gas mixtures in the range 0-4740 V/V can be prepared with stated concentrations +0.01% V/V.When more accurate determinations in the Crl% V/V concentration range are requircd, the (b100.00 and the OL10.000 mmHg models may be used to achieve accuracies of k0.001 and rtO.0001'% V/V, respectively. The apparatus is connected to a Perkin-Elmer 8500 gas chromatograph via &in stainless-steel tubing. Gases are transferred from the vacuum system on to the chromato- graphic column by means of a gas sampling valve (Valco Instruments) fitted with a 0.t-cm7 sample loop. Three different gas sampling valves were used. Valve A is a six-port manual gas sanipling valve. The valve body is made of 300 Series stainless steel and the rotor is made of a graphite-filled fluorocarbon polymer incorporating 20% d i n of the filler.Valve B is a ten-port automatic gas sampling valve, the rotor of which is made of the same material as that in valve A. Valve C is also a ten-port automatic gas sampling valve, the rotor of which is made of a fluorocarbon-filled polyimide incorpor- ating 50% d m o f t h e filler. Valvec B and C were operated at 50 "C and valve A at room temperature. Operating Procedures The apparatus is pumped down with liquid nitrogen around the trap (K) to a pressure of ca. lo-' mmHg as measured by the Pirani gauge. The system is then isolated from the pump by closing tap 2. Liquid reservoirs containing the volatile agents of interest are attached by means of greaseless cones and sockets (J. Young) at I), E, F and G. The liquids in these reservoirs are thoroughly degassed prior to use by a series of freeLe - pump - thaw cycles. Bulbs A, B and C are used for the preparation of gas mixtures.Static mixtures of gases of known concentration may be prepared manometrically by measuring the required pressure of pure gas into bulb C and making up to atnwspheric pressure with air. Thus, to prepare a 1 % V/V mixture of the vapour, e.g., halothane, a pressure equivalent to I "/o of atmospheric pressure is measured into bulb C and made up to atmospheric pressure with air. The gases are allowed to mix and the gas mixture is injected on to the column via an automatic gas sampling valve. The procedure is repeated for different pressure5 (and hence concentrations) o f vapour until a calibration graph has been prepared for the relevant concen- tration ranges.The shape of bulb C was carefully chosen to allow rapid and efficient mixing of gas mixtures prepared according to the procedure described above. It was found that a spherical round-bottomed bulb promoted rapid mixing by diffusion of gas molecules and bv the turbulence produced by the admittance of air into the mixing bulb. The gas mixtures were allowed to equilibrate for periods up to and including 1 h in order to determine the optimum time required for mixing. For gas mixtures at total pressures in the range 760-850 mmHg, no change in concentration (i.e., no change in area of peaks from the gas chromatographic analyses) was observed for mixing times in excess of 45 min. I t was also found that the response of the FID to the volatile agents of interest is independent of total pressure in the range 100-760 mmHg.Gas mixtures prepared at total pressures of 100 mmHg were shown to be completely mixed after only 10 min. Calculation of Concentration The concentrations of gases are almost always expressed in units of per cent. by volume, which may be calculated by where v,,, vh, . . . , v,, and y , , P h , . . ., y,, are the volumes and partial pressure of components a, h , . . ., ri at constant temperature. If the partial pressures of all components are accurately known, then the concentrations are readily calcu- lated. Analytical Procedures Samples were analysed on a Perkin-Elmer 8500 gas chromato- graph equipped with a flame-ionisation detector and fitted with a 2 m x 4 mm i.d. glass column packed with 80-100-mesh Chromosorb W HP coated with 15% Apiezon L.The carrier gas was nitrogen at 35 cm? min-1 and the air and hydrogen flow-rates were 430 and 35 cm3 min-1 respectively; all gases were supplied by Cryogas. The injector was maintained at room temperature when a manual gas sampling valve was used and at 50 "C when the automatic gas sampling valve was used. The column temperature was 130 "C and the detector was operated at 250 "C. The peak areas were determined using the Perkin-Elmer integration system and were used as a measure of the response of the detector to individual compounds. Results The samples were prepared as described previously and were introduced on to the chromatographic column via one of three gas sampling valves used for this study. Peak areas were used as a measure of the response of the chromatographic detector.It was found that in the absence of air, all three of the volatile agents were retained to various extents in the valves following injection. The extent of the retention of the volatile agents in the valve bodies was measured by a blank injection following the initial injection of the standard, and was shown to increase as the contact time between the anaesthetic agents and the valve bodies increased. For valves A and H , the results for standard gas mixtures containing at least SO mmHg air showed that if adsorption did occur, the amount of volatile agent adsorbed was below the limit of detection of the FID (0.05 mmHg using a 0.1-cm3 sample loop). Valve C was found to be unsuitable for use with the agents of interest as significant adsorptive losses were observed even in the presence of a diluent gas.Despite the fact that significant adsorptive losses occurred in the low concentration ranges, it was possible to remove all traces of volatile agents from the valve by pumping for 4 min following injections; therefore, there was no possibility that successive injections would give erroneous results as a consequence of adsorption from previous injections. The results of the analyses of halothane standards are shown in Fig. 2. Data points for measurements of gas standards made up to atmospheric pressure in air may be fitted to a goodANALYST, OCTOBER 1989. VOL. I14 1209 2000 1500 c 2 s N r 0 m 2 1000 Y IT) a 500 I 1 I I 0 0.25 0.50 0.75 1.00 Halothane concentration, % V N Fig. 2.FID response to halothane in the 0-1% V/V concentration range using the automatic gas sampling valve (valve B) for introduc- tion of the sample 011 to the column. a, Total pressure approximately equal to atmospheric pressure; 0. no diluent gas present. Each data point represents the mean o f six measurements with a relative standard deviation of 0.50"/6 for each data point straight line with a correlation coefficient of 0.9999, a slope of 3.03 IL 0.02 and an intercept of 0.07 -t 0.112. In the absence of any dilucnt gas, significant adsorptive losses for satnples in the Ckl% V/V concentration range were observed and the results are clearly non-linear. The results for the analysis of standard mixtures of the other two anaesthetic agents of interat were similar to those obtained for halothane.In the presence of diluent gas, the calibration graph5 for isoflurane and enllurane were linear with correlation coefficients of0.9999 in both instances. I n the absence of any diluent gas, significant deviations from linearity were observed at low concentrations. The precision of the method was studied by analysing ten individually prepared gas mixtures of the same concentration. The coefficients of variation of the results for halothane, isoflurane and enflurane mixtures were 0.52, 0.50 and 0.78%, respective I y . Discussion One of the major disadvantages associated with the prepara- tion of standard gas mixtures by static methods is the possibility of significant changes in concentration with time as a result of adsorption on, or dissolution in, the materials from which the storage and handling system are constructed. The glass manifold fitted with Teflon taps used here is free from these types of effects over the time scale of the experiments conducted for this study-no measurable changes in concen- tration occurred for standards stored in the manifold over- night.However, significant adsorptive losses did occur when the volatile anaesthetic agents were stored in the gas sampling valve for time periods >30 s in the absence of any diluent gas. These losses were attributed to the adsorption on and/or dissolution in the materials from which the valve rotor\ were constructed. The results indicated that valves A and B showed lower adsorptive losses than valve C.The improved performance of valves A and R compared with valve C is probably due to the difference between the rotor materials, as this is the only difference between the valves. Although valves A and B are operated at different temperatures (cu. 20 and 50 "C, respectively), no difference i n behaviour is observed with respect to the retention of volatile agents in these valves. The absence of adsorptive losses in the presence of diluent gas may be due to displacement of organic molecules by competitive adsorption of air molecules at the same site. The reverse process of displacement of air by an organic vapour as the concentration of the latter increases has been discussed.' We have not found any reference in the literature to adsorptive losses of volatile anaesthetic agents in gas sampling valves.Jonsson et ul.0 discussed the adsorption of aliphatic and aromatic hydrocarbons on the inner wall of the sample loop, which causes systematic errors in the injected amounts. They also found that adsorption in the valve itself was negligible. In this study, changing the volume o f the sample loop did not alter the extent of adsorptive losses, thus ruling out the possibility that adsorption was occurring in the sample loop as opposed to the valve body and rotor. Adsorptive losses of volatile anaesthetic agents in gas sampling valves are most significant in the ()-I% V/V concentration range, which, in clinical terms, is a most important range. Evidently, the suitability o f materials of the type discussed earlier for use with volatile anaesthetic agents must bc established if accurate quantitative measurements are t o bc madc at low concentrations.We have initiated a series of studies in our laboratories to quantify the extent of adsorption on each of the materials of interest and to evaluate a number of commercially available gas sampling valves for use with volatile anaesthetic agents. The results in this paper show that the materials used i n some gas sampling valves (valves of type C described above) make these valves unsuitable for use with the halogenated anaesthctic agents used in this study. The available methods for the preparation of standard gas mixtures are based on gravimetricl-3 or volumetric3 dispensing of volatile agents into a mixing vessel, and the standards are transferred into the gas chromatograph using a gas-tight syringe.Each method involves a detailed sequence of steps which is relatively easy to perform, but is time consuming, and each operation potentially contributes to losses in accuracy. As daily calibration is essential if accurate quantitative dctcrminations arc to be madc, the calibration method must be rapid and convenient to use without sacrificing either accuracy or precision to achieve this aim. In particular, the uncertainties associated with syringe calibration and random variations in injected volumes may significantly detract froin the over-all accuracy of the method. Gray4 and Zbinden Pi ~ 1 . ~ suggested that data collected on three separate days may be combined to construct a calibra- tion graph, but the validity of this approach is questionable.Although they have reported very low relative standard deviations of results (2.0%4 and 2.16%; for halothane, compared with O.S2% reported in this work), it is possible that such precision is fortuitous considering the assumptions that have been made. We have found that it is essential to construct calibration graphs each day, hence the necessity for a rapid and casy to use calibration method. The method described here is inherently more accurate than the available calibration methods. There is only one operation involved in the preparation of each standard, so there are fewer possible sources of error, and the single pressure measurement required is made with an accuracy of kO.1 mmHg in the 0-32 mmHg range; any variation in concentra- tion of this order of magnitude is almost outside the limit of detection of the FID (0.05 mmHg under the present experimental conditions). The accuracy of the pressure measurements allows us to quote gas concentrations in the M Y o V/V range within ?O.Ol% V/V.For a volumetric concentration of I % , one can expect to introduce the volatile agent at a pressure of 8.3 mmHg. Hence the relative error on this measurement is 1.2?0, and the volumetric concentration of such a mixture will be 1 .OO -t 0.01% V/V. This compares with an accuracy of kO.1%0 V/V quoted by Zbinden PI al.71210 In addition to the accuracy of the method, the use of a gas-tight vacuum apparatus and gas sampling valves elimi- nates the need for gas-tight syringes, and these features contribute greatly to the over-all precision of the method, as is evident from the comparison of the precision of this method with those of the alternative methods of Gray4 and Zbinden et al.7 The procedure described here is also more convenient to use than existing methods as only one operation, a pressure measurement, is required for the preparation of each standard mixture.Although the mixing times are relatively long, 45 min for each mixture when the total pressure is atmospheric, the apparatus includes numbers of mixing bulbs to permit the simultaneous preparation and mixing of all of the gas standards required. Further, our results have shown that adsorptive losses are below the limit of detection when as little as 50 mmHg air is included in the standard mixture, and mixing times are hence reduced to 10 min. ANALYST, OCTOBER 1989, VOL. 114 J. D. O’Keeffe was supported by a grant from the Children’s Research Centre at Our Lady’s Hospital, Dublin. 1 . 2. 3. 4. 5 . 6. 7. References Nelson, G. O., “Controlled Test Atmospheres. Principles and Techniques,” Ann Arbor Science, Ann Arbor, MI, 1971. Barratt, R. S . . Analyst, 1981, 106, 817. “Methods for the Preparation of Gaseous Mixtures,” RS 4559 : 1983, British Standards Institute, London, 1983. Gray, W. M., Br. J. Anaesth., 1986, 58, 345. de Boer, J. H.. “The Dynamical Character of Adsorption,” Oxford University Press, London, 1953, p. 86. Jonsson. J. A., Vejrosta, J . , and Novak. J . , J . Chromatogr., 1982, 236, 307. Zbinden, A. M . , Frei, F. J., Funk, B., Thomson. D. A., and Westenskow, D.. Br. J. Anaesth., 1985, 57, 796. Paper 81041 74A Received October 20th, 1988 Accepted March 28th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401207

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. OCTOBER 198Y. VOL. 114 121 1 Gas Chromatographic Determination of Volatile Anaesthetic Agents in Blood Part 2.* Clinical Studies John Flynn, Salman Masud, J. Declan O'Keeffe and William S. Wren Department of Anaesthesia, Children's Research Centre, Our Lady's Hospital for Sick Children, Dublin 12, lrela n d Imelda M. Shanahant School of Chemical Sciences, Dublin City University, Glasnevin, Dublin 9, Ireland A method is described for the direct determination of the volatile anaesthetics halothane and isoflurane in blood by gas chromatography with flame-ionisation detection. The method is accurate and precise and allows rapid measurements of blood levels of anaesthetic agents. Headspace concentrations of anaesthetic agents in the concentration range 0-3% V/V are determined with an accuracy of 20.01% V/V.The relative standard deviation of these results is less than 4.0%. A relatively small volume of blood is required for each determination, a factor of great significance in the treatment of children. The need for separate blood calibration graphs for each patient is discussed, further emphasising the need for a rapid calibration procedure. The results from the clinical application of this method show conclusively its suitability for the management of anaesthetised subjects. Keywords: Volatile anaesthetic; halothane and isoflurane; blood; gas chromatography The insufflation technique of anaesthesia involves blowing an anaesthetic vapour with a carrier gas into the pharynx or trachea, and the patient is ventilated through a wide-bore tube which is passed into the trachea.End tidal vapour concentra- tions in expired air from the anaesthetised subject are measured at the outlet of this tube and the values provide an indication of the depth of anaesthesia o f the subject. Under certain circumstances, however, end tidal vapour concentra- tions do not give accurate information on the level of anacsthesia. When surgery is required in or near the trachea, the anaesthetist must share a confined space with the surgeon, maintain an adequate airway and keep the patient asleep. In such cases, a small-bore venturi jet or catheter is used for insufflation, and mixing with air in the pharynx is unavoidable. Consequently, end tidal vapour concentrations no longer represent accurately the level of anaesthesia of the subject, and an alternative tnethod of assessing this level must be used. The most reliable alternative is to monitor the level of anaesthetic vapour in the blood, as it has been established that this value is always 2-5% less than the end tidal concentra- tion.' Techniques for the determination of the concentration of volatile anaesthetics in blood have received considerable attention. The most widely used method €or this purpose is gas chromatography.The methods that have been proposed include procedures such as direct injection ,3-7 extraction with organic solvents8-17 and headspace techniques. 13-17 These procedures, with the exception of the headspace technique, have proved to be time consuming with extensive sample pretreatment being required before analysis, and they are not readily applicable to the routine establishment of the depth of anaesthesia in the management of anaesthetised subjects.The headspace technique, in which the concentration of the volatile anaesthetic in the blood is derived from the measured concentration in the headspace over the blood, is a relatively simple method and has the added advantage over alternative methods of measuring the partial pressure of anaesthetic, * For Part 1. see referencc I . f To whom correspondence should be addressed. which is physiologically more relevant than the total content. To calculate the partial pressure of anaesthetic in blood from the total content, the blood - gas partition coefficient must be known. This partition coefficient is not constant, varying from patient to patient and even within one patient.18 Its value may depend on haemoglobin concentration,l6.Iv albumin to globu- lin ratio,l" osmolarity,20 temperature21322 and particularly serum triglyceride concentration.23 The choice of a headspace technique €or determinations of volatile agents in blood circnmvents the need for a knowledgc of the partition coefficient, as this is readily determined at the same time as the quantitative measurements are made.17 This paper describes a method for the determination of the concentrations of volatile anaesthetics in blood by gas chromatography with flame-ionisation detection. The method is based on a headspace analysis, and allows more accurate, precise and rapid determinations of anaesthetic concentra- tions in blood than existing methods.Experimental The method was applied to determinations of 2-bromo-2- chloro-1 ,l,l-trifluoroethane (Halothane, Hoechst) and l-chloro-2,2,2-trifluoroethyl difluoromethyl ether (isoflurane, Abbott Laboratories) in blood. For each of the volatile agents studied, standard mixtures for calibration were prepared in dinitrogen oxide - oxygen - carbon dioxide (60 + 35 + 5 ) (Air Products). Apparatus The vacuum apparatus that was used for handling the volatile agents and preparing the standard mixtures was described in detail in Part 1.1 The apparatus is connected to a Perkin-Elmer 8500 gas chromatograph via $-in stainless-steel tubing. Gases are introduced on to the chromatographic column by means of a ten-port automatic gas sampling valve (Valco Instruments) fitted with a O.l-cm3 sample loop.Blood samples placed in the sample chamber of a tonometer (Instrumentation Laboratories, Model 237) were exposed to known concentrations of volatile agent introduced into the1212 ANALYST, OCTOBER 1089, VOL. 114 tonometer from a vaporiser (Ohmeda, Fortec Tec 3 for isoflurane; Drager for halothane). Gas-tight glass syringes fitted with PTFE plungers (J. Young, Acton) and PTFE valves (Mininert valves from Supelco) were used to handle blood samples. Analytical Procedures Gas samples were analysed on a Perkin-Elmer 8500 gas chromatograph equipped with a flame-ionisation detector, and fitted with a 2 m x 4 mm i.d. glass column packed with 15% Apiezon L on 80-100-mesh Chromosorb W. The carrier gas was nitrogen at 35 cm3 min-1 and the air and hydrogen flow-rates were 430 and 35 cm3 min-1, respectively.The injector temperature was maintained at 50 "C, the column temperature was 130 "C and the detector was operated at 250 "C. Peak areas were determined using a Perkin-Elmer data station and were used as a measure of the response of the detector to individual compounds. Calibration Procedure Standard mixtures of volatile anaesthetics were prepared in dinitrogen oxide - oxygen - carbon dioxide (60 + 35 + 5 ) and were analysed as described in Part 1 . 1 A calibration graph relating concentration to peak area was thus established. A clinically relevant range of concentrations of the volatile agents of interest in the calibration gas mixture was obtained from a vaporiser.The gas mixture was warmed to 37 "C and humidified in a tonometer. The exact concentration of volatile agent in the gas phase was determined by sampling and analysing the gas in the tonometer. A 3.5-cm3 volume of pre-operative heparinised arterial blood was placed in the sample chamber of the tonometer and equilibrated at 37 "C for 20 min with the gas mixture of known concentration at a flow-rate of 500 cm3 min-1. The blood was then withdrawn into a gas-tight syringe and 0.6-cm3 aliquots were dispensed accurately into 5.0-cm3 glass vials fitted with PTFE-coated Neoprene septa. The contents of the vials were equilibrated with moderate shaking in a water-bath main- tained at 37 "C for 5 min, after which a 100-pl headspace sample was withdrawn in a gas-tight syringe and analysed on the gas chromatograph.The procedure was repeated for each of the vials, and an average of the five measurements was used to derive the concentration of volatile agent in blood from the calibration graph. This procedure was repeated with different concentrations of volatile agent exposed to the blood sample, and a calibration graph relating the concentration of volatile agent to which the blood has been exposed to the peak areas from the headspace analyses was prepared. The blood samples from anaesthetised subjects were obtained in the operating theatre using 5.0-cm3 gas-tight syringes and were stored at 4 "C for no longer than 2 h prior to analysis. Aliquots of 0.6 cm3 were dispensed into vials as before and equilibrated, with shaking, at 37 "C for 5 min, after which 100-p1 samples of the headspace were withdrawn and analysed as before.The results from the five vials were averaged to give a determination of the concentration of volatile agent in the test blood. End tidal concentrations of volatile agent in inspired and expired gas of the patient were routinely determined in the operating theatre using a Normac infrared analyser at the patient's mouthpiece. These values were determined whenever a blood sample was taken. The results obtained using this technique were compared with those obtained using gas chromatographic analysis of expired air in a separate series of experiments, and an experimentally determined correlation factor was used to relate the end tidal concentrations measured in the theatre by infrared analysis to the gas chromatographic results.Results The response of the flame ionisation detector to each of the volatile agents of interest was calibrated using standard gas mixtures as described in Part 1.1 The calibration graphs were linear over the concentration range 0.00-2.00% V/V, which spans the range of headspace concentrations under clinically relevant conditions. The accuracy and precision of the calibration method were discussed in Part 1,' and did not change under thc conditions of this study. The optimum time required for equilibration of blood with gas in the tonometer was shown to be 20 min. For shorter exposure times, it was found that equilibration was not complete; the measured concentration of anaesthetic in the blood increased until equilibrium was finally reached for exposure times of 20 min.Equilibration of the blood and gas in the vials was shown to be complete after 5 min in a water-bath maintained at 37 "C during equilibration. The concentration of anaesthetic agent in the gas phase to which the pre-operative blood samples were exposed was determined by measuring the gas concentration at the inlet to the tonometer sample chamber. The apparatus was designed and modified so that sampling via the gas sampling valve was possible, thus eliminating the need for sampling via gas-tight syringes. The results obtained were plotted against the measured headspace concentrations in the equilibrated pre- operative blood, thus establishing a calibration graph from which headspace concentrations in the test blood could be determined.The blood calibration graphs were linear over the clinically relevant concentration ranges (0--2% V/V) with slopes of 1.93 5 0.1 1 and 3.65 Tfr 0.20, and correlation coefficients of 0.9999 for isoflurane and halothane, respec- tively. Each data point represented the mean of eight separate determinations with a relative standard deviation of 2.0%. Representative results from the analyses of headspace concentrations of volatile anaesthetics in arterial blood from ten anaesthetised subjects are shown in Table 1 together with the end tidal concentrations determined in the operating theatre by infrared analysis. Discussion The headspace technique described in this paper allowed us to measure the concentrations of volatile anaesthetics in blood with an accuracy and precision greater than those attainable with existing methods.Headspace concentrations of anaes- thetic agents were measured in this study with an accuracy of 50.01% compared with +0.1% reported by Zbinden et al.17 The relative standard deviation of these results is always less than 4.0%, compared with 4.77% reported by Zbinden et al. 17 In addition, the method is easy to perform and overcomes many of the disadvantages associated with earlier methods. The method requires that blood calibration graphs relating concentration of volatile anaesthetic to which the blood has been exposed and the measured headspace concentrations following equilibration of the blood be prepared. This is achieved by exposing blood to known concentrations of volatile agent in the sample chamber of a tonometer, while ensuring adequate humidification of the gases t o avoid dilution or desiccation of the blood sample.This is guaranteed if a tonometer is used for the exposures. It is also essential that adequate time is allowed for equilibration o f the blood and gas mixture. Under the conditions used in this study, it was shown that for 3.5 cm3 of blood, equilibration was complete after 20 min with a gas flow of 500 cm3 min-1. The data used to generate blood calibration graphs for clinical studies were fitted to excellent straight lines, as stated under Results, and the precision of the replicate measure- ments was excellent considering the large number of manipu- lations required for each measurement. It is particularly important to emphasise the need for separate calibration graphs for each patient studied.The solubility of volatileANALYST, OCTOBER 1989, VOL. 114 1213 Table 1. Concentrations of anaesthetic agents in arterial blood and expired air of patients under general anaesthetic Relative Concentration End tidal No. of Average peak Standard standard of anaesthetic concentration blood area. headspace, deviation, deviation, in blood, of anaesthetic, Time*/ samples GC counts 0 % Yo v/v Yo v/v min 5 5 5 5 5 5 5 5 5 4 2.12 2.44 3.19 2.86 2.44 2.87 4.0s 2.81 1.57 1.61 0.08 0.08 0.10 0.11 0.08 0.08 0.10 0.02 0.03 0.05 3.6 3.4 3.1 4.0 3.4 2.7 2.5 0.7 2.0 2.9 0.5 1 0.58 0.76 0.60 0.62 1.31 1.87 0.58 0.73 0.71 * Represents the length of time elapsed after commencing delivery of anaesthetic to patient. i.This sample was taken after the patient had received one unit of blood. 0.54 0.62 0.80 0.68 0.72 1.45 1.94 0.66 0.79 1.78 162 177 149 100 177 105 160 175 110 190t Anaesthetic agent Halothane Halothane Halothane Halothane Halothane Isoflurane Isoflurane Isoflurane Isoflurane Isoflurane anaesthetics in blood varies widely from patient to patient, and even within patients on occasion. The decisive factor governing solubility is the serum triglyceride concentration,23 hence the importance of using blood from the subject to be anaesthetised for calibration purposes. This significant result has not been established previously. It is also essential that the pre-operative blood sample be obtained as close to the start of the operation as possible in order to minimise differences in triglyceride concentration caused by further metabolism.Considering the need for daily calibration of the gas chromatograph and separate blood calibration graphs for each patient to be studied, the importance of a rapid method for these measurements is evident. Further, for ethical reasons it is desirable that as little blood as possible be required for the analyses. The method described here requires as little as 10 cm3 of pre-operative blood for calibration, and 3.5-cm3 aliquots of test blood for each subsequent determination. This factor assumes overriding significance when patients are very small, as in the children’s hospital where this study was carried out. The measured concentrations of anaesthetic agent in blood and expired air from patients anaesthetised with isoflurane or halothane are given in Table 1.With the exception of the last entry, the results show clearly that the measured blood levels are in the expected concentration range relative to the end tidal concentrations. The last measurement was made after the patient had received one unit of blood, representing ca. 10% dilution of blood in this 45-kg subject. Under such circumstances, it is not surprising that the measured blood level is significantly lower than the end tidal concentration. Obviously, great caution must be exercised in interpreting the results of measured blood concentrations when the subject has received a blood transfusion. The data in Table 1 are representative of the results obtained in this laboratory using the methods described here for measurement of headspace concentrations of volatile anaesthetics.To date, analytical reJults for 33 additional case studies have been obtained, and the accuracy and precision of the method does not differ from the limits given above. The varying trends in concentrations of volatile anaesthetic agents in blood and expired air which are evident from the data in Table 1 have important clinical significance. Such trends are expected as the concentration of volatile anaes- thetic delivered to the patient varies frequently during the operation, and from patient to patient, according to the nature of the surgical procedures in progress at a given time. The results of this study show conclusively that the method described is suitable for the routine establishment of the depth of anaesthesia of anaesthetised subjects.The method is rapid and easy to use, requires small volumes of blood for each measurement and is accurate and precise. J. D. O’Keeffe and S. Masud were supported by a grant from the Children’s Research Centre at Our Lady’s Hospital, Dublin. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 1s. 16. 17. 18. 19. 20. 21. 22. 23. References Flynn, J.. O’Keeffe, J . D., Shanahan, I., and Wren, W. S . , Analyst, 1989, 114, 1207. Vickcrs, M. D., Wood-Smith, F. J., and Stewart, H. C., “Drugs in Anaesthetic Practice,” Butterworths, London, 1978, Yokota, T., Hitomi, Y., Ohta, K., and Kosaka, F., Anesthr- siology, 1967, 28. 1064. Cousins. M. J., and Mazze, R. I . , Anesthesiology, 1972, 36, 293.Douglas, R., Hill, D. W., and Wood, D . G. L . , Br. J. Anaesth., 1970, 42, 119. Lowe. H. J . , Anesthesiology, 1964, 25, 808. Hill, D. W., and Newell, H. A., Nature (London), 1965, 206, 708. Butler, R. A., and Freeman, J . , Br. J. Anaesth., 1962,34,440. Jacobs, E. S . , Anesth. Analg. (Cleveland), 1964, 43, 177. Allott, P. R., Steward, A . , and Mapleson, W. W., Br. J . Anaesth., 1971, 43, 913. Atallah, M. M., and Gcddes, I. C . , Br. J . Anaesth., 1972, 44, 1035. Davies, D. D., Br. J . Anaesth., 1978, 50, 147. Yamamura, H . , Wakasugi, B . , Sato, S . , and Takebe, Y.. Anesthesiology, 1966, 27, 311. Butler, R. A.. Kelly, A. B., and Zapp, J . , Anesthesiology, 1967, 28, 760. Fink, B. R., and Morikawa, K., Anesthesiology, 1970,32,451. Cowles, A. L., Borgstedt, H. H., and Gillies, A. J., Anesthe- siology, 1971, 35, 203. Zbinden, A. M., Frei, F. J . , Funk, B., Thomson, D. A., and Westenskow, D., Rr. J. Anuesth., 1985, 57, 796. Munson, E. S . , Eger. E. I., 11. Tham, M. K., and Embro, W. J., Anesth. Analg. (Cleveland), 1978, 57, 224. Laasberg, L. H., and Hedley-Whyte, J., Anesthesiology, 1970, 32, 351. Lerman, J . , Willis, M. M., Gregory, G. A., and Eger, E . I., 11, Anesthesiology, 1983, 59, 554. Stoelting, R. K., and Longshore, R. E., Anesthesiology, 1972, 36, 503. Knill, R. L., Lok, P. Y. K., Strupat, J. P., and Lam, A. M., Can. Anaesth. Soc. J., 1983, 30, 155. Saraiva, R. A., Willis, B. A., Steward, A . , Lunn, J . N.. and Mapleson, W. W., Br. J. Anaesth., 1977, 49, 115. p. 125. NOTE-Reference 1 is to Part 1 of this series. Paper 81041 7 8 0 Received October 20th, 1988 Accepted March 28th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401211

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, OCTOBER 1989, VOL. 114 1215 Determination of Underivatised Efaroxan and ldazoxan in Blood Plasma by Capillary Gas Chromatography With Mass-selective Detection John D. Nichols, Neil A. Hyde and Keith Sugden Pharmaceutical Division, Reckitt & Colman plc, Dansom Lane, Kingston-upon-Hull, Hull HU8 7DS, UK Sensitive and specific methods for the determination of efaroxan and idazoxan in blood plasma have been developed based on solvent extraction, chromatographic separation and quantification by selected-ion monitoring using a quadrupole mass-selective detector. The use of a short non-polar bonded-phase capillary gas chromatography (GC) column facilitated rapid separation of the compounds of interest from internal standards, metabolites and endogenous material. Of equal significance was the ability to chromatograph these basic compounds without prior derivatisation.The application of bonded-phase capillary GC coupled to selected-ion monitoring resulted in robust analytical procedures with sub-ng ml-1 sensitivity and high selectivity. Keywords: Efaroxan; idazoxan; plasma assay; capillary gas chromatography; mass-selective detection Efaroxan. 2-[(2-ethyl-2,3-dihydrobenzofuran-2-yl)]-2-imid- azoline hydrochloride, and idazoxan, 2-[ (1,4-benzodioxan-2- yl)]-2-imidazoline hydrochloride, are potent and highly selec- tive a2-adrenoreceptor antagonists currently under investiga- tion for a range of therapeutic indications.1.2 Initial animal studies and early Phase 1 studies with idazoxan in man utilised a high-performance liquid chromatographic (HPLC) assay to determine plasma concentrations of the drug.3 The minimum levels that could be quantified with this technique were in the range 5-10 ng ml-1 in plasma.Subsequent single-dose studies to evaluate pharmacokinetic parameters required a specific assay procedure with an order of magnitude improvement in assay sensitivity to characterise fully the absorption and terminal elimination phases of the drug profile. Efaroxan was recognised early in its development as being more potent than idazoxan and this necessitated that a highly sensitive assay be developed before Phase 1 human studies could commence. This prompted the investigation of potentially more sensitive assay procedures which in turn led to the development of the methods described here.Experimental Apparatus Chromatographic separation was performed using a Hewlett- Packard 5890A gas chromatograph fitted with a 15m X 0.25 mm i.d., 0.25-pm film thickness DB1 fused-silica capil- lary column (J&W Scientific, Folsom, CA, USA). Sample introduction was via a Hewlett-Packard 7673A autosampler with fast injection into an inert injection port liner containing a glass-wool plug. A Hewlett-Packard 5970 mass-selective detector (MSD) was used for peak detection. Data handling, including integration and calculation of peak-area ratios, was performed on a Hewlett-Packard 59960 MS Chemstation. Reagents Acetate buffer (0.2 M, p H 5.0). Prepared by mixing 0.2 M sodium acetate solution and 0.2 M acetic acid. Carbonate - hydrogen carbonate buffer (0.4 M, p H 10.5).Prepared by mixing 0.4 M sodium carbonate and 0.4 M sodium hydrogen carbonate solutions. Diethyl ether. Glass-distilled grade (Rathburn Chemicals, Walkerburn, UK). Chloroform. Laboratory-reagent grade , special for chroma- tography (BDH, Poole, Dorset, UK). Standard Solutions for Efaroxan Assay Efaroxan solution. Prepared by dissolving the drug in distilled water and diluting the solution to give 0.1 and 0.01 pg ml-1 solutions. Internal standard solution, RX83 1001 { 2-[ (2,3-dihydro-2- propylbenzofuran-2-yl)]-2-irnidazoline hydrochloride} . Pre- pared by dissolving the salt in distilled water and diluting the solution to give a 0.1 pg ml-1 solution. Standard Solutions for Idazoxan Assay ldazoxan solution. Prepared by dissolving the drug in distilled water and diluting the solution to give 1.0, 0.1 and 0.01 pg ml-1 solutions.lnternal standard solution. Prepared by dissolving efaroxan in distilled water and diluting the solution to give a 50 ng ml-1 solution. Plasma Standards for Calibration Control plasma (1 ml) and an appropriate volume of the internal standard solution (SO pl) were placed in a round- bottomed 15-ml test-tube. For the efaroxan assay 25, 50, 100 or 200 pl of 0.01 pg ml-1 or 50, 100 or 250 pl of 0.1 pg ml-1 efaroxan solutions were added (0.25,0.5,1,2,5,10 or 25 ng of efaroxan, respectively). For the idazoxan assay 50, 100 or 250 pl of 0.01 pg ml-1 or 50, 100 or 250 p1 of 0.1 pg ml-1 or 50 pl of 1.0 pg ml-1 idazoxan solutions were added (0.5,1,2.5, 5 , 10, 25 or 50 ng of idazoxan, respectively).Test Samples Sample plasma (1 ml) and internal standard solution (50 pl containing 5 ng of RX831001 or 2.5 ng of efaroxan, respec- tively) were placed in a round-bottomed 15-ml test-tube. Extraction Procedure Acetate buffer (1 ml) was added to each sample or standard test-tube. The tube was stoppered and vortex-mixed for 15 s. Diethyl ether ( 5 ml) was added followed by further mixing for 1 min and centrifugation for 5 min to separate the phases. The tubes were frozen in a cardice - acetone bath and the diethyl ether was decanted off and discarded. The tube contents were thawed and mixed with l m l of carbonate - hydrogen carbonate buffer (15 s) prior to further addition of diethyl1216 cn w .- 5 40000 2 P ._ I) & 30000 6 C m -0 3 c 20000 z ANALYST.OCTOBER 1989, VOL. 114 - - - - 80 -100 :::: 50 100 rnlz 150 187 h -1 200 Fig. 1. line) and RX831001 peak (inverted below zero line) Comparison of mass spectra from efaroxan peak (above zero 12 000 ._ E 10000 c 3 > 5 8000 c ._ e 6000 0 C m 4000 3 a 2000 I I I I I I 5.5 6.0 6.5 7.0 7.5 8.0 Time/min Fig. 2. Chromatogram of the ion at m/z 187 for a 2-p.1 injection of the test sample. Efaroxan elutes at 6.1 min and the internal standard at 6.5 min. The sample was calculated to contain 2.4 ng ml-1 of efaroxan 80 70 a; c m -0 C 3 I) Q 50 100 150 200 mlz Fig. 3. Mass spectrum of idazoxan peak 12 000 v) c .- 5 10000 2 F 2 8000 a; 2 6000 .I- m m -0 S 2 4000 Q 2ooo t 0- 6.0 6.5 7.0 Tim eim i n 50 000 i" L 10 000 0- 6.0 6.5 7.0 Time/m i n Fig. 4. ( a ) Chromatogram of the ion at m/z 174; and ( h ) chromato- gram of the ion at m/z 187, for a 2 4 injection of a test sample found to contain I .1 ng ml-1 o f idazoxan. ldazoxan elutes at 6.5 min and the internal 3tandard at 6.2 min ether ( 5 ml). The tubes were vortexed (1 min), centrifuged and frozen as before, and the diethyl ether was decanted off into a 12-ml conical test-tube. The aqueous layer was re-extracted with a further 5-ml aliquot of diethyl ether and the extracts were combined and allowed to reach ambient temperature. Any aqueous phase carried over was removed by freezing the tip of the tube and decanting the diethyl ether off into a further clean conical tube. The diethyl ether was then evaporated under a stream of nitrogen and the residue reconstituted in chloroform (50 PI), transferred into a low- volume insert in an autosampler bottle, capped and taken for analysis.Chromatography Injections were made while the injection port was switched to the splitless mode, the split valve being opened after 3 min. The gas chromatograph was temperature programmed as follows: inlet pressure, 50 kPa helium; injection port temper- ature, 250 "C (constant) (efaroxan method), 200 "C (constant) (idazoxan method); column oven temperature, 45 "C for 3 min, 45-180 "C at 70 "C min- , 180 "C for 5 min, 180-280 "C at 70 "C min-1, 280 "C to end of run (16 min in total); and transfer line temperature, 275 "C (constant). Data Acquisition The gas chromatography (GC) column was interfaced directly to the MSD which was run in the selected-ion monitoring (SIM) mode.Acquisition parameters for the efaroxan assay were: ion, mlz 187; dwell time, 200 ms; and acquisition cycles per second, 3.7. Acquisition parameters for the idazoxan assay were: ion, mlz 187 and 174; dwell time, 25 and 250 ms, respectively; and acquisition cycles per second, 2.8. The MSD was switched on only during the expected elution of the peaks of interest, i.e. , 5-8 min. For the efaroxan procedure the raw data were processed by the Chemstation and the resulting chromatogram was inte- grated. The efaroxan peak area was divided by the internal standard peak area and the peak-area ratio reported together with the chromatogram and sample details. For the idazoxan procedure the raw data were processed to give two extracted ion chromatograms due to the ions at mlz 187 and 174, respectively.The chromatograms were inte- grated and the peak-height ratio was calculated by dividing the idazoxan peak height in the m/z 174 chromatogram by the internal standard peak height in the mlz 187 chromatogram.ANALYST. OCTOBER 1989, VOL. 114 1217 Table 1. Calibration data for the efaroxan procedurc Efaroxan concentration/ ng ml- 1 0.25 0.50 1 .o 2.0 5.0 10.0 25 .o Peak-area ratio (cfaroxan to internal standard) I 2 3 4 5 0.0745 0.0669 0.0741 0.05 14 0.O6O7 0.0928 0.0959 O.IO96 0.1007 0.1048 0.1856 0 . I745 0.1814 0.1748 0.1716 0.3143 0.3001 0.3400 0.3 142 0.3967 0.9165 0.8408 0.8173 0.9125 0.8876 1.955 1.936 1.794 I .875 2.163 4.745 4.685 4.505 4.263 4.299 Table 2. Calibration data for the idazoxan proccdurc Idazoxan concentration/ Peak-height ratio (idazoxan to internal standard) ng ml-1 0.25 I .o 2.5 5.0 10.0 25 .o 50.0 1 0.0692 0.1266 0.2583 0.5 102 0.9750 2.161 3.095 2 3 0.091 1 0.0765 0.1078 0. I36 1 0.2706 0.2708 0.43 19 0.5289 0.850 1 1.01 1 1.909 2.221 3.980 4.397 The peak-height ratio was reported together with the two chromatograms of interest and the sample details. Results The feasibility of the above methods was initially investigated using 1-ul injections of a 0.1 mg tnl-1 chloroform solution of efaroxan or idazoxan base. The GC conditions were optimised to those described above with additional temperature ramping being included to avoid possible late-running peak contamina- tion from plasma extracts. The MSD was operated in the scan mode, giving full mass spectral information for each eluting peak.The spectra of the peaks of interest were studied in order to determine which was the best ion or ions to use in the analysis. Operation in the SIM mode was necessary to achieve the desired assay sensitivity. In each instance several com- pounds were considered as possible internal standards. Analogues of efaroxan were available and their suitability as an internal standard was systematically tested. RX831001 was found to give good peak shape and was well separated from efaroxan. Comparison of the mass spectra of these two compounds (Fig. 1) showed a common base peak and as a result offered the opportunity to maximise the assay sensitivity by acquiring data for a single ion (mlz 187) only. A typical STM chromatogram is shown in Fig.2. The mass spectrum of idazoxan (Fig. 3) showed a base peak at mlz 174; however, despitc running many analogues of idazoxan, no compound could be found which had a common base peak. Efaroxan was eventually chosen as the internal standard owing to its availability and good chromatographic separation from idazoxan. Fig. 4(u) and (b) shows typical SIM chromatograms for idazoxan and the internal standard, respectively. The efficiency of the extraction procedure was determined for both efaroxan and idazoxan by spiking plasma with a radiolabelled drug of known purity and specific activity and determining the recovery by liquid scintillation counting of the extract. Replicate extracts ( n = 6) gave recoveries of 84.0% [coefficient of variation (CV), 3.4%] and 85.0% (CV, 2.6%) for efaroxan and idazoxan, respectively, at concentrations appropriate to the assay (10 ng ml-1).'The extraction proce- dure was the rate-limiting step in sample analysis; however, up to 30 samples per day could easily be analysed by one technician. 4 0.0715 0.1297 0.2500 0.574 1 0.981 1 1.834 5.092 5 0.0908 0.1347 0.2563 0.5895 1.084 2.489 4.900 Mean k SD 0.0655 k 0.0097 0.1008 k 0.0067 0.1776 * 0.0057 0.3331 rf: 0.0384 0.8749 k 0.0441 1.945 2 0.137 4.499 k 0.2 I8 Mean k SD 0.07% k 0.0105 0. I270 * 0.01 14 0.2612 t 0.0092 0.5269 i 0.0622 0.9802 k 0.0847 2.123 k 0.262 4.293 k 0.799 cv, % 14.8 6.6 3.2 11.5 5.0 7.0 4.8 cv, Yo 13.2 9.0 3.5 11.8 8.6 12.3 18.6 l o I 0 .- +- 2 1 2 ? m Y m a, a 0.1 1 I I 0.1 1 10 Concentratiodng ml-1 Fig. 5. Calibration graph for the determination of efaroxan in blood plasma.Calculated linc of best fit (solid line) and 95% confidence limits (broken lines) 0 .- +- p 1 f +- L G7 a) Y m a .- 0.1 1 I I 0.1 1 10 Concentratiodng ml-1 Fig. 6. Calibration graph for the determination of idazoxan in blood plasma. Calculated linc of best fit (solid linc) and 95% confidence limits (broken lines)1218 ANALYST, OCTOBER 1989, VOL. 114 For both assay procedures multiple calibration points (five) were obtained for each of seven standard concentrations (Tables 1 and 2). The relationship between concentration and peak-area ratio was defined using least-squares regression of the log - log transformed data. Rartlett’s test4 was used to check the variance stability over the calibrated range and the goodness-of-fit was tested using the F-value for each poly- nomial following an analysis of variance.The line of best fit for efaroxan was described by four regression coefficients. This is shown, together with the calculated 95% confidence limits, in Fig. 5. The multiple correlation coefficient of tl e line was 0.998. Assay precision (CV, Yo) was calculated as 11% at 20 ng ml-1 and 13% at 0.5 ng ml-1. The idazoxan assay gave a linear calibration graph; this is shown together with the 95% confidence limits in Fig. 6. The multiple correlation coefficient of the line was 0.996 and the precision (CV, YO) was calculated to be 15.5% over the calibrated range. Test samples were run concomitantly with quality control (QC) standards to ensure the accuracy of the methods.For the test data to be valid, the QC results had to fall within the defined 95% tolerances. Concentrations in the test samples were computed directly from the calibration graphs. Discussion The availability of bonded-phase fused-silica capillary column technology has resulted in a renaissance in GC in many areas where HPLC had become the separation method of choice. It is now possible to chromatograph many compounds by GC without employing time-consuming and laborious derivatisa- tion procedures. Efaroxan and idazoxan are two such com- pounds, their basic nature and chemical structure offering few opportunities for derivatisation. Gas chromatography of the native compounds on packed columns proved very difficult, with degradation problems and poor peak shapes.However, using a short non-polar bonded-phase capillary system, good, consistent peak shapes were obtainable with the additional benefit of improved chromatographic resolution. The most specific detection technique for GC is mass spectrometry. Specificity and sensitivity were major consider- ations in the development of a bioanalytical procedure, and when used in the STM mode mass-selective detection offered the potential for both high sensitivity and specificity. By combining the unique properties of bonded-phase capillary GC with the convenience and power of mass-selective detection in the SIM mode, robust, sensitive and specific analytical procedures for the determination of efaroxan and idazoxan in blood plasma samples were successfully devel- oped. Automation of the chromatographic procedures to facilitate the analysis of large numbers of samples over an extended period was also a key objective achieved in method development. To date, over 3000 samples have been analysed using these procedures, with a single column handling approximately 700 samples over a 4-month period. References 1. 2. Chapleo, C. B., Doxey, J. C., Myers, P. L., and Roach, A. G., Br. J . Pharmacol., 1981, 74. 842. Chapleo, C. B.. Myers. P. L., Butler, R. C. M., Davis, J. A., Doxey, J . C., Higgins, S. D., Myers, M., Roach, A. G., Smith, C. F. C., Stillings, M. R., and Welbourn, A. P.,J. Med. Chem., 1984, 27, 570. Muir, N . C., Lloyd-Jones, J . G., Nichols, J. D., and Clifford. J . M., Eur. J . Clin. Pharmacol., 1986, 29, 743. Snedecor, G. W., and Cochrane. W. G., “Statistical Methods,” Sixth Edition, Iowa State University Press, Amcs, 1967. 3 . 4. Paper 9100048H Received January 4th, 1989 Accepted May 23rd, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401215

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. OCTOBER 1989, VOL. 114 1219 Determination of Propranolol and 4-Hydroxypropranolol in Human Plasma by Hig h-performance Liquid Chromatography Chauhwei J. Fu Kansas City Analytical Services, 12700 Johnson Drive, Shawnee, KS 662 16, USA William D. Mason Division of Pharmaceutical Science, University of Missouri-Kansas City, Kansas City, MO 64 108, USA A high-performance liquid chromatographic method was developed to determine the levels of propranolol and its major metabolite, 4-hydroxypropranolol, in human plasma. The limits of determination are 10 ng ml-1 of propranolol and 5 ng ml-1 of 4-hydroxypropranolol using a 0.5-ml plasma sample. The stability of plasma samples stored at -30°C for up to 2 months was also tested. No stabilising antioxidants were added to the samples.Keywords: Propranolol; 4-h ydroxypropranolol; high-performance liquid chromatography; human plasma Propranolol, an important drug widely used in the treatment of cardiac arrhythmia, hypertension, angina and thyrotoxico- sis, has been shown to be almost completely metaboliscd in humans. 1 Naphthoxylactic acid and 4-hydroxypropranolol were reported to be the major metabolites in humans, with 4-hydroxypropranolol subsequently being identified as an equipotent p-adrenergic blocker to the parent Hence pharmacodynamic assessments of propranolol should include the determination of both propranolol and 4-hydroxy- propranolol concentrations in plasma. A method for the simultaneous determination of propran- 0101 and 4-hydroxypropranolol in plasma by mass fragmento- graphy was established by Walle et ul.0 in 1975.Our laboratories have been involved in the study of propranolol in both humans and animals in recent years, and developed the first method to detect propranolol and 4-hydroxypropranolol s i m u 1 t a ne ous 1 y i n h u m a n p 1 asm a by high - per form ance 1 i q u i d chromatography (HPLC) in 1977.' Since then, further HPLC methods have been developed by ourselves8 and others.')-27 These methods have been employed in clinical studies of propranolol and its metabolites. A concern associated with the determination of propranolol and 3-hydroxypropranolol in biological tluids has been the reported rapid degradation of the latter. The addition of several reagents, such as sodium ascorbate, sodium metabisul- phite and sodium dithionite, as antioxidants has been recom- mended to slow the breakdown of 4-hydroxypropranolol during the analytical procedures and/or the storage of plasma samples.11-22.26 However, the addition of antioxidants is not without problems. Alteration of the sample pH and the possible effects of redox reactions on other analyte5 in multiple drug studies require evaluation and validation. Hence these reagents should be added only if it is essential. Recently, in preparing to investigate the drug interaction between propranolol and propafenone, another antiarrhyth- mic drug, we developed a method for determining propranolol and 3-hydroxypropranolol simultaneously in human plasma without an added antioxidant. Experimental Reagents Propranolol hydrochloride (99% mlm) was obtained from Aldrich (Milwaukee, WI, USA), 4-hydroxypropranolol hydrochloride (99.1% rrzlm) from Ayerst Laboratories (Rouses Point, NY, USA) and indalpine from Maybridge Chemicals (UK).Acetonitrile, methanol, butanol and butyl chloride were purchased from Burdick and Jackson Labora- tories (Muskegon, MI, USA), phosphoric acid (85% VIV) and sodium hydroxide from American Scientific Products (North Kansas City, MO, USA) and boric acid from J . T. Baker (Phillipsburg, N J , USA). All other chemicals were of analy- tical-reagent grade. The structures o f propranolol, 4-hydroxy- propranolol and indalpine are presented in Fig. 1. Instrumentation The HPLC system consisted of an Altex Model LLOA pump, a Micromeritics 725 autoinjector, a Chromanetics Spherisorb C8 (5-pm) column (150 x 4.6 mm i.d.), a Shimadzu RF-535 fluorescence detector and an Altex CRlA integrator. The excitation wavelength was set at 240 nm and the emission wavelength at 410 nm.Prepared Solutions A stock standard solution of propranolol (SO pg ml--l) was prepared with 5%" V/V methanol and a stock standard solution of 4-hydroxypropranolol (25 pg ml-1) with 10% V/V methanol, to which 0.05 ml of concentrated phosphoric acid was added per 100 ml. The stock standard solution of the internal standard (IS), indalpine (100 pg mi-'), was prepared with 10% V/Vmethanol with a working solution (500 ng ml-l) being prepared by dilution with HPLC-grade water. All stock solutions were stored at 4°C for up to 1 month without any deterioration. OH OH I OCHZCHCHZN HCH (CH3)Z I OCHzCHCHzNHCH(CH3)z 1 I m Fig.1 pine Propranolol I OH 4-hydroxypropranolol Q-$ rJ H lndalpine Structures o f propranolol, 4-hydroxypropranolol and indal-1220 ANALYST, OCTOBER 1989, VOL. 114 Calibration Standards Calibration standards of 10,20, 50, 100, 167 and 250 ng ml-1 of propranolol and 5, 10, 25, 50, 83 and 125 ng ml-1 of 4-hydroxypropranolol were obtained by diluting the stock standard solutions with pre-screened blank plasma. Handling of Samples The calibration standards and quality assurance pools were prepared at the beginning of the study and stored with the unknown study samples at -30 "C until analysis. In addition, fresh standards were also prepared after 1 , 2 and 4 weeks and 2 months, relative to the quality assurance pools, to test the stability of frozen samples. Analytical Procedures Unknown plasma samples (0.5 ml) together with calibration standards and quality assurance samples were pipetted into conical test-tubes with 0.10 ml of 1.92 M sodium hydroxide solution, 0.15 ml of internal standard solution and 0.4 ml of 0.8 M boric acid.The samples were then extracted with 8 ml of 3% V/V butanol in butyl chloride. After centrifugation, the organic layer was transferred into a second corresponding set of test-tubes and back-extracted with 0.5 ml of 50 mM phosphate buffer (pH 2.9) containing 10% V/V methanol. After aspiration of the organic layer, the acid solution was stored at 4°C until analysis and 0.20 ml was injected into the HPLC system for analysis. Chromatographic Conditions Analyses were performed by reversed-phase HPLC on a C8 column (150 x 4.6 mm i.d.).No pre-column was used. The mobile phase was acetonitrile - 50 mM phosphate buffer (pH 2.9) (21 + 79 V/V) at a flow-rate of 2.0 ml min-I. Calibration and Calculation After a batch of calibration standards, quality assurance samples and study samples had been chromatographed and peak heights for all analytes and internal standard had been measured, a calibration graph was generated by linear regression ,28 and the concentrations of unknown samples and quality assurance samples were calculated. Quality assurance samples were prepared by the quality assurance officer in our laboratories, and analysed "blind" by the analyst. Results Specificity Plasma from 12 subjects taking propafenone and also from subjects not taking drugs showed no interfering peaks.The chromatographic separation of propranolol, 4-hydroxypro- pranolol and indalpine was accomplished with an efficiency of about 5900, 4400 and 4400 theoretical plates, respectively (Table 1). Fig. 2(a) shows a typical chromatogram where propranolol and indalpine have resolutions of 14.61 and 9.95, respectively, from 4-hydroxypropranolol. Fig. 2(h) shows a typical chromatogram for a plasma sample with no propran- 0101, 4-hydroxypropranolol or indalpine (i. e., blank plasma). The retention times were about 5, 12 and 18 min for 4-hydroxypropranolol, indalpine and propranolol, respec- tively. No significant alterations were observed with a column being used continuously for at least 1 month. Sensitivity The limit of determination was set at 10.0 ng ml-* for propranolol and 5.0 ng ml-' for 4-hydroxypropranolol in plasma for a 0.50-ml sample.Fig. 2(c) presents a typical chromatogram of a sample containing 10.0 ng ml-l of propranolol and 5 .0 ng ml-1 of 4-hydroxypropranolol. The peaks for both analytes are at least three times the back- ground. Table 2 gives the statistics for six samples containing both analytes at this lower limit, which were assayed as unknowns on one occasion. The mean relative recoveries were 103.80 and 101.8O% for propranolol and 4-hydroxypropranolol, respectively, the coefficients of variation (CV) being 14.74 and 4.72%, respectively. Linearity Linearity was observed in the range 10.0-250.0 ng ml-1 of propranolol and 5.0-125 ng ml-1 of 4-hydroxypropranolol in plasma, as shown in Table 3.( a ) 1 2 i 0 4 8 12 16 20 Time/m in Fig. 2. Typical chromatograms: ( a ) instrument validation solution; ( b ) blank plasma; and (c) calibration standards with 10 ng ml-1 propranolol and 5 ng ml- 4-hydroxypropranolol. 1. 4-Hydroxy- propranolol; 2, indalpinc (IS); and 3, propranolol Table 1. Specificity for assay of propranolol and 4-hydroxypropranolol Mean values rt SD ( n = 4) Relative Capacity Theoretical Analyte retention" Rcsolu tion" factor" plates" 4-Hydroxypropranolol . . 1 .0 k 0.00 - 3.99 * 0.06 4426 k 100 Indalpine . . . . . . 2.45 rt 0.02 9.95 k 0.37 9.77 rt 0.08 4395 k 68 Propranolol . . . . . . 3.44 k 0.02 14.61 k 0.53 13.71 k 0.08 5930 k 303 * Calculation is based on the equations presented by Yost et uLZyANALYST, OCTOBER 1989, VOL.114 1221 Table 2. Assay results for plasma spiked with propranolol and 4-hydroxypropranolol 4-H ydroxypropranolol Quality Propranolol Quality assurancc assurance concentration/ SD ( n = 6)/ concentration/ ng m1-l ng ml-1 CV, YO Recovery, YO ng ml- 10 10.38 t 1.53 14.74 103.80 5 3 0 30.79 f 2.07 6.72 102.63 15 80 79.64 t 3.44 4.32 99.55 40 200 195.64 Ifr 5.68 2.90 97.82 100 pool: Mean k pool : Mean t SD (n = 6)/ ng ml-I CV, Yo Recovery, Yo 5.09 k 0.24 4.72 101.80 15.02 zk 1.18 7.86 100.13 40.25 + 1.25 3.11 100.63 99.07 L 5.40 5.45 99.07 ~~ ~ Table 3. Summary of regression statistics for calibration graphs IS average peak height/ Batch vv Standard error estimate/ ng ml-I Correlation coefficient 0.997 0.999 1 .000 1.000 0.999 0.999 Slope/ ng ml-1 213.72 199.32 194.59 196.80 198.38 200.45 Intercept/ ng ml-1 1.302 2.216 1.918 0.769 0.731 - 1.248 Compound Propranolol .. . . . . . . . . 2945 2465 2396 2314 2387 1965 7.765 4.539 3.319 2.754 3.982 4.185 Mean SD cv, Yo 1 2 3 4 5 6 2412 315 13 2945 2465 2396 2314 2387 1965 0.999 0.001 0.100 1.000 0.999 0.999 0.998 0.999 0.999 200.54 6.77 3.38 119.13 127.18 128.68 134.89 128.61 104.94 0.948 1.231 4.391 1.784 1.350 1.478 2.165 2.867 1.449 2.805 1.441 2.260 2.391 3.446 1.662 2.623 4-Hydroxypropranolol . . . . . . Mean SD cv. Yo 2412 3 15 13 0.999 0.001 0.100 123.91 10.57 8.53 2.019 0.696 - 2.304 0.717 - Table 4. Stability test for propranolol at -30 "C Control sample 25 ng ml- 200 ng ml-I Mean t S D ( n = 6)/ng ml-1 CV, Yo Recovery, % 25.63 k 1.42 5.54 102.52 24.41 f 1.00 4.10 97.64 24.72 -t 0.44 1.78 98.88 25.10 t 0.76 3.03 100.40 24.04 ? 1.73 7.20 96.16 Mean +SD ( n = 6)/ng ml-1 CV, YO Recovery, YO 195.45 f 1.47 0.75 97.73 191.98 k 1.36 0.71 95.99 201.27 t 6.67 3.31 100.64 201.80 t 2.40 1.19 100.90 196.79 k 4.45 2.26 98.40 Time Dayl .. . . Week1 . . . . Week2 , . . . Week4 . . . . Month2 . . . . Table 5. Stability test for 4-hydroxypropranolol at -30 "C Control sample 12.5 ng ml-1 100 ng ml-1 Mean t SD (n = h)/ng ml-l CV, 5% Recovery, YO 102.83 t 1.30 1.26 102.83 99.84 zk 2.51 2.51 99.84 105.83 t 4.47 4.22 105.83 101.66 t 3.62 3.56 101.66 107.55 k 4.06 3.77 107.55 Mean t SD Time ( n = 6)/ngml-l CV, % Recovery, Yo Dayl . . . . 12.57t0.61 4.85 100.56 Week 1 . . . . 12.00zk0.28 2.33 96.00 Week2 .. . . 12.66-tO.17 1.34 101.28 Week4 . . . . 12.92t0.49 3.79 103.36 Month2 . . . . 11.63 t 1.00 8.60 93.04 Accuracy and Precision Analysis of quality assurance pools to which propranolol and 4-hydroxypropranolol were added "blind" to the analyst was employed as a form of internal validation. Table 2 presents the results of thesc assays, for which the relative recoveries of propranolol ranged from 97.82 to 103 .80% for four different concentrations, the CV being 2.9&14.74%. For 4-hydroxy- propranolol, the relative recoveries ranged from 99.07 to 101.80%, the CV being 3.11-7.86%. Stability Freshly prepared solutions for both instrument validation and plasma calibrations showed no evidence of deterioration for either analyte or the internal standard indalpine.Tables 4 and1222 ANALYS'I', OCTOBER 1989, VOL. 114 5 show the results of stability tests for two stability control samples which were detected by different sets of calibration standards prepared fresh on different dates. No significat degradation was observed for either analyte for up to 2 months of storage at -30 "C. Discussion The method described provides a rapid and sensitive means of determining the plasma levels of propranolol and its major metabolite 4-hydroxypropranolol in human plasma for phar- macokinetic andor pharmacodynamic studies. The mean plasma levels in the steady state of both analytes from 12 subjects following the last dose of an 80-mg dosage three times per day are presented in Fig. 3. The level of 4-hydroxypropranolol in plasma is low, the mean maximum concentration being close to 20 ng ml-1.Therefore, an assay with a sensitivity for 4-hydroxypropran- 0101 of 5 ng ml-1 or less is necessary in order to establish the pharmacokinetic andor pharmacodynamic characteristics in clinical or other studies. To overcome this sensitivity problem, in several previous assays9-'8 a large volume (1-3 ml) of plasma was used to obtain the necessary limit of dctcrmina- tion. This technique is not suitable for multiple drug studies as different analytes may be detected by separate methods with limited blood samples being collected. Compared with other methods, if the volume of plasma used is factored, the limit of determination in this method is almost identical or even lower. Both propranolol and 4-hydroxypropranolol exhibit natural fluorescence, allowing their detection with a fluorescence detector.The excitation wavelengths (205 or 295 nm) are the same for both compounds, but the fluorescence emission spectra of propranolol and 4-hydroxypropranolol differ in that the wavelength maxima for emission in aqueous solution are 360 and 435 nm, respectively.14 Therefore, in our first method7 and other assays,9,10,13,14,22 dual detectors connected in series with an optimum setting for each analyte were required. The present method uses a compromised setting of excitation and emission wavelengths (240 and 410 nm) to balance the loss of sensitivity for both analytes; however, it will still allow the measurement of 10 ng ml-1 of propranolol and 5 ng ml-1 of 4-hydroxypropranolol simultaneously in human plasma for a 0.5-ml sample.Another concern associated with the determination of propranolol and 4-hydroxypropranolol in biological fluids has been their instability, especially the reported rapid degrada- tion of 4-hydroxypropranolol. As mentioned earlier, the addition of antioxidants as stabilisers is not without problems. Alteration of the sample pH and the possible effects of redox reactions on other analytes in multiple drug studies require evaluation and validation. Hence these agents should be 100, 1 0 4 8 12 16 20 24 Time/h Fig. 3. Mean plasma levels in the steady state of A, propranolol and B, 4-hydroxypropranolol from 12 subjects following the last dose of an 80-mg dosage three times per day added only if it is essential. In this method, 10% V/Vmethanol in the phosphate buffer for back-extraction has been shown to stabilise 4-hydroxypropranolol during the entire chromato- graphic procedure (approximately 12 h per batch).In addi- tion, no significant degradation was observed for propranolol or 4-hydroxyprppranolol which was added to plasma and stored for up to 2 months at -30°C without any antioxidant being added. This 2-month stability is consistent with the results of previous assayslh.24 in which antioxidants were used, and differs from the results presented by Smith etal.25 in which 4-hydroxypropranolol was stated to be stable for only 1 d at 0°C with 15% or more loss on the second day if antioxidants were not added. The difference between our results and those found by Smith et ~ 1 .2 5 is probably due to the different storage temperatures (0 versus -30°C). Stability for only 1 d is not sufficient in most studies owing to the large numbers of samples involved. A very small peak appears in the chromatogram with a retention time of 10.3 min. This peak is considered to be an impurity in the internal standard as it occurs in the chromato- grams from all samples, including prc-dose samples, except the blank plasma extracted without internal \tandard. As shown in Table 1 , the separation of the analytes of interest is good. However, to avoid interferences from propafenone and its metabolites in this drug interaction study, no attempt was made to alter the mobile phase to reduce the run time and optimisc the limits of determination. In addition to the system described above, in which a Shimadzu RF-535 fluorescence detector was used, another system was also evaluated with a Kratos 970 fluorescence detector with a wavelength of 240 nm and a cut-off filter at 389 nm. This system exhibited almost identical chromato- graphic characteri5tics with respect to specificity, sensitivity, linearity, accuracy and precision when cornpared with the dual monochromator system.Conclusion The method described permits the determination of as little as 10 ng ml-1 of propranolol and 5 ng ml- 1 of 4-hydroxypropran- 0101 simultaneously in human plasma with a 0.5-ml sample. Only one detector is used, the excitation and emission wavelengths of which are set, achieving sensitivity for both analytes. No antioxidant is added as a stabiliscr, which avoids possible redox reactions of other analytes in multiple drug studies.This method was originally developed for a propran- 0101 and propafenone interaction study but, due to its consistency and low deviation during the assay, it can also be used for other propranolol studies. 1. 2. 3 . 4. 5. 6. 7. 8. 9. 10. 11. 12. References Paterson, J . W., Conolly, M. E., Dollcry, C. T., Hayes, A., and Cooper, R . G., Pharrnacol. Clin.. 1970, 2, 127. Bond. P. A . , Nature (London), 1967, 213, 721. Fitzgerald. .I. D., and O'Donnell, S . R., Br. J . Pharrnucol., 1971. 43, 222. Cleaveland, C. R., and Shand, D. G., Clin. Pharrnucol. Ther., 1971. 13. 181. Tindell, G. L., Walle, T., and Gaffney, 1'. E.. i,ife Sci., 1972, 11, 1029. Walle, T., Morrison, J . , Walle, K ., and Conradi, E., J . Chromarogr., 1975, 114, 351. Mason, W. D., Amick, E. N., and Weddle, 0. H.. Anal. Lett., 1977, 10, 515. Li, W., and Lanman, K. C . , Anal. Lett., 1987, 20, 603. Taburet, A.. Taylor, A. A , , Mitchell, J . R., Rollins, D. E., and Pool, J. L., Life Sci.. 1979, 24, 209. Rao, P. S . , Quesada. L. C., and Mueller, H. S . , Clin. Chirn. Acta, 1978, 88, 355. Nation, R. L., Pcng, G. W.. and Chiou, W. L., J . Chrornutogr., 1978, 145, 429. Schncck, D. W.. Pritchard, J. F., and Hayes, A. H . , Jr.. Res. Cornrnun. Chem. Pathol. Pharrnacol., 1979, 24, 3 .ANALYST, OCTOBER 1989, VOL. 114 13. 14. 1s. 16. 17. 18. 19. 20. 21. 22. Gyselinck, P., Remon, J . P., van Severen, R., and Braeckman, P., Br. J . Clin. Pharmacol., 1980, 10, 406. Drummer, 0. H., McNeil, J., Pritchard, E., and Louis, W. J . , J . Plzarm. Sci., 1981, 70, 1030. Rosseel. M. T., and Bogaert, M. G., J . Pharm. Sci., 1981, 70, 688. Lo, M., Silber, B., and Riegelman, S . , J. Chromatogr. Sci., 1982, 20, 126. Albani, F., Riva, R., and Baruzzi. A., J. Chromatogr., 1982, 228. 362. Harrison. P. M., Tonkin, A . M., Cahill, C. M., and McLean, A . J . , J. Chromatogr., 1985, 343, 349. Pritchard, J. F . . Schneck, D. W., and Hayes, A. H., Jr., J . Chromatogr., 1979, 162, 47. Lo, M., and Ricgelman, S . , J . Chromatogr.. 1980, 183, 213. Bahr. C. V., Hcrmansson, J., and Lind, M., J . Pharmucol. Exp. Ther., 1982, 222. 458. Tamai, G., Morita, I . ? Masujima, T., Yoshida, H., and Imai, H., J. Plzarm. Sci., 1984, 73. 1825. 1223 Wilson, M. J., and Walle, T., J . Chromatogr., 1984,310,424. Kwong, E. C.. and Shen, D. D., J . Chromatogr., 1987, 414, 365. Smith, K. A., Wood, S . , and Crous, M., Analyst, 1987, 112, 407. Koshakji, R. P.. and Wood, A. J . J . , J . Chromatogr., 1987, 422, 294. Sood, S. P., Green, V. I., and Mason, R. P., Ther. Drug Monit., 1988, 10, 224. Dixon, W. J . , and Massey, F. J., Jr., Editors, “Introduction to Statistical Analysis,” McGraw-Hill, New York, 1983, p. 211. Yost, R. W.. Ettre, L. S . , and Conlon, R. D., “Practical Liquid Chromatography, An Introduction.” Perkin-Elmer, Nonvalk, CT, 1980, p. 31. 23. 24. 25. 26. 27. 28. 29. Paper 8104806A Received December 6th, I988 Accepted May 8th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401219

出版商:RSC    

年代:1989

数据来源: RSC