ISSN:0003-2654    

DOI:10.1039/AN98914FX041

出版商:RSC    

年代:1989

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98914BX043

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989, VOL. 114 1357 Editorial November, 19641 GREENFIELD, JONES AND BERRY 713 High-pressure Plasmas as Spectroscopic Emission Sources BY S. GREENFIELD, I . LL. JONES AND C. T . BERRY (Alhnght C3 llilson (‘Zvg) Lttl., Oldbirq, Bimzznghanl) An account is given of the authors’ preliminary work in the use of high- pressure plasmas as both spectrographic and flame-photometric emission sources. Two types of plasma sources are discussed, 412z., the d.c.-arc type and the high-frequency induction type; the latter, having no electrodes, produces little background emission. Tables of the order of detection for some metallic elements are given, but these are not absolute detgction limits. The standard deviation obtained with calcium emission at 3933A is given for both sources.Professor S. Greenfield Twenty-five years ago in the November issue of The Analyst this paper appeared comparing the analytical performance of two types of high-pressure plasma, a d.c.-arc plasma jet based on the design of Margoshes and Scribner and a high-frequency induction plasma run in a torch similar to that previously described by Reed. It is interesting to speculate on how many readers of The Analyst at that time recognised that they were witnessing the birth of a technique that was to have a major impact on analytical chemistry in the coming decades. The answer is probably none because in the early stages of any project the researcher’s lot is essentially a lonely one. The publication of promising results is no guarantee of viability and the scientist must have, or acquire, the skills of a communicator and an evangelist.Happily in Stan Greenfield and others who followed, the inductively coupled plasma (ICP) was well served by its advocates. Even so, it was not until 1975 that the first commercial instruments became available. Looking back at this first publication, one cannot fail to be impressed at the authors’ perception in recognising the correct operating mode and the intrinsic merits of the ICP as a spectrochemical source. The potential user in 1964 was told that: (i) “The torch produces a flame-like plasma 1% inches long, -1/1 inch in diameter and annular in form, i.e., it has a hole or low-temperature region in its centre. The sample is injected through this low-temperature region, and if the radiation produced by the sample is in the visible range of the spectrum, a coloured “tail flame” is produced downstream of the main plasma’’; (ii) the analyte emission should be viewed in the tail flame, a region of low background dominated by the emission bands of the OH molecule; (iii) the source provides high sensitivity, good stability and is not prone to the classical chemical interferences of flame spectrometry; and ( i v ) the plasma is capable of handling samples in a variety of forms including aerosols, dry powders and slurries thereof.Under- standing, instrumentation and methodology have advanced, but the basic tenets established in the original paper remain unchallenged. Analytical chemistry is an applied science and many significant advances have been made by analysts attempting to solve practical problems.It was appropriate then that this work did not originate in the rarefied atmosphere of aca- demia, but in the laboratories of Albright and Wilson where problem solving and commercial reality were the natural order. From the outset, Greenfield’s group applied the ICP to practical analysis and how effective that discipline was in establishing the fundamentals of the technique. The Analyst, at that time still closely associated with the publication of analytical methods, was the natural vehicle for bringing this sensitive and robust technique to the attention of the analytical community. The subsequent development of the ICP is well documented and today a formidable array of derivative techniques are available, some involving different approaches to sample introduction such as chromatography, flow injection, hydride generation, electrothermal vaporisa- tion and laser or spark ablation and some alternative detection systems such as atomic fluorescence spectrometry and mass spectrometry.Hence as the science develops, so must the journals that record and propagate the new thinking. The Analyst has broadened its scope and now invites contributions on all aspects of analytical science, the Journal of Analytical Atomic Spectrometry was launched to provide new impetus in the rapidly expanding field of atomic spectrometry and both are supported by Analytical Proceedings and Analytical Abstracts. A family of journals serving the needs of the international analytical market. Twenty-five years of the ICP, it’s been an exciting time. Thank you Stan Greenfield, and The Analyst, and long may these pages contain the seed corn of new scientific adventures. Barry L. Sharp The Macaulay Land Use Research Institute Craigiebuckler A berdeen, UK

ISSN:0003-2654    

DOI:10.1039/AN9891401357

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST,NOVEMBER 1989,VOL. 114 1359 Current Status and Prospects for the Use of Optical Fibres in Chemical Analysis A Review John 0. W. Norris Materials Development Division, B552 Harwell Laboratory, AEA Technology, Oxfordshire OX1 1 ORA, UK Summary of Contents I nt rod uction Perceived advantages and disadvantages Advantages Disadvantages Optical effects Chemical equilibria Instrumentation Method of classifying fibre optic sensors Description of some sensors Underlying principles of fibre optic sensors Extrinsic species-specific sensors Remote spectroscopy Sensors utilising immobilised reagents Evanescent wave devices Intrinsic species-specific devices N on-species-specif ic techniques lndi rect techniques Conclusions References Keywords: Review; chemical analysis; optical fibre Introduction Optical methods are some of the oldest and best established techniques for sensing chemical analytes.The development of inexpensive, high-quality optical fibres for the communica- tions industry has provided the essential component for a new technology, that of optical fibre sensors. In the last decade considerable research effort has been expended into develop- ing sensors based on optical fibres for both physical and chemical analytes, with many interesting schemes having been proposed. Potential application areas for chemical analysis include process plant, environmental and pollution monitoring, mili- tary applications, laboratory-based analysis, clinical diagnosis and other medical applications. Each particular application area has its own requirements for sensitivity, precision, selectivity, sensor lifetime and unit cost.The physical state of the analytes studied has encompassed gases, dissolved gases, liquids, ions in solution and solids. Physical sensors, most notably those for measuring pres- sure, temperature, fluid level and mass flow, are widely applied both to monitoring and controlling industrial processes. Currently, the chemical sensor most commonly used in industrial processes is the glass pH electrode. This system is relatively expensive, susceptible to electrical noise and not readily usable in the food industry because of the danger of breakage. There is now substantial interest in the development of miniaturised techniques that will overcome these problems and allow the measurement of a wide range of chemical parameters.Fibre optic techniques are likely to make a significant contribution towards satisfying this need. There have recently been a number of general reviews on optical fibre chemical sensors, ’--I reviews emphasising chem- ical sensors based on immobilised indicators”6 and optical fibre biomedical sensors.7.X This review is intended to give a current state of the art overview of optical fibre chemical sensors. It is structured so as to give some indication of the perceived advantages, disadvantages and applications for optical fibre chemical sensors. The transduction principles, how the optical measurement is related to the analyte concentration, and the type of experimental equipment required, are discussed. A convenient way of classifying these sensors is given and illustrated with a review of sensors developed to date.The principles underpinning the various sensor concepts are described briefly, and data on the performance of sensors is given where possible. It should be appreciated that the examples given are illustrative and not exhaustive, and that the principles described can be, and in the future undoubtedly will be, extended to additional analytes. In addition to optical fibre sensors some mention is included on integrated optic devices, because this area, although currently in its infancy, is a natural extension from optical fibres and will become increasingly significant in the future. Perceived Advantages and Disadvantages Advantages An essential component of any form of chemical analysis using optical fibres is the optical fibre itself.This consists of a light - guiding core, usually a silica-based glass although it can be made from an organic polymer [e.g., poly(methy1 methacry- late)] or a mixture of metal halides or chalcogenides, surrounded by an optically rarer cladding. The cladding can be of a similar composition to the core, but having a lower refractivc index, to give an all silica optical fibre, or of a1360 ANALYST. NOVEMBER 1989, VOL. 114 different material, e.g., to give a plastic clad silica fibre. Usually these two concentric cylinders are then sheathed in a polymer to provide further mechanical protection. Light incident on the end of the fibre within an acceptance cone, known as the fibre's numerical aperture, propagates down the fibre by repeated total internal reflection, with very little attenuation.(A more detailed description of the fundamental physics of optical fibres and some of the associated technical vocabulary, can be found in reference 5.) Several potential advantages arise from using an optical fibre as the basis of a chemical analysis technique. Small size andflexibility. The light guiding core of an optical fibre typically ranges from 3 to 1000 pm in diameter. The majority of chemical analysis applications use fibres of core dianicters SO-200 pm. Typically, a 200-pm plastic clad silica fibre can be bent round a 1 cm radius mandrel, illustrating that optical fibres can be used to sense in very small or otherwise inaccessible areas. Chemically and thermally stable.The basic material from which optical fibres are made is amorphous silica. This is chemically inert, i.e., it can be used in strongly acidic or moderately alkaline environments, although it is attacked by HF and strong alkalis (pH > 13). Pure silica softens a little above 1000 "C, with other glasses softening above 500 "C, and therefore the glasses are also thermally stable. However, the upper working temperature of the majority of optical fibres is limited by their surrounding polymers, and is often around 125 "C. The current development of polyimide, amorphous carbon and metallic coatings means that fibres capable of withstanding higher temperatures (2400 "C) are becoming increasingly available. Low loss. Typically, optical fibres have an absorbance of < 1 .O (a loss of only a few decibels) per kilometre.For sensors this means that the sensing point can be some distance from the servicing electlonics. In addition, one set of electronics, which might include expensive components such as lasers, can be multiplexed to many sensors. Remote in situ measurements. As a corollary to the preceding advantages, remote in situ chemical analyses can be undertaken using fibre optic sensors. Low mass and cost. The density of silica is less than a quarter of that of copper, and therefore a network of chemical sensors linked by optical fibres instead of copper wires offers significant mass savings. This is particularly important for the aerospace and offshore industries. Typically, fibres cost a few tens of pence per metre.The cost of an optical fibre sensor is therefore, rarely dominated by the cost of the fibre. Electrical isolation. Optical fibre sensors are, by their nature, electrically isolated from the interrogating electronics. This is especially important for in vivo medical sensors and application areas where flammable or explosive reagents are present. Freedom from electromagnetic interference. Similarly, glass fibres, being insulators, are immune to electromagnetic interference (EMI), and therefore fibre sensors can be used in electrically noisy environments. In contrast, the signals from potentiometric chemical sensors [e.g., the glass pH electrode and chemical ficld-effect transistors (chemFETs)] are affected by EMI. N o reference sensor. Optical fibre chemical sensors do not require a reference sensor, in contrast to potentiometric sensors where the potential difference between the sensing and a reference electrode is measured.However, referencing is usually required as optical sensors do not generally give an absolute reading. This is performed most frequently by comparing the optical intensity at two different wavelengths, only one of which is affected by the analyte whilst both are dependent on the optical characteristics of the remainder of the optical system. Potential of distributed sensing. In principle optical fibre chemical sensors can be made where the whole length of the fibre is sensitised to a particular analyte. The presence of the analyte at any point along the fibre can modify its optical properties, enabling a large area to be monitored simul- taneously.Alternatively, the sensitised fibre can be inter- rogated by an optical equivalent of radar (optical time domain reflectometry) to give a measure of the analyte concentration as a function of position over the extended area. Disadvantages Amhient light. In an optical fibre sensor ambient light can interfere with the signal of interest. This can be alleviated by pulsing the interrogating light and using phase sensitive detection to remove the background. Response time. In common with all multi-phase chemical sensing techniques the response time can be long because it might be determined by mass transport to and in the reagent phase. It can be reduced by designing probes that are small or incorporate only thin films of the reagent phase.Long term stability. Changes in the optical characteristics of the source, dctector, transmitting optical fibre or in the sensing transducer, e.g., due to fouling, can limit the long term stability of a system. In addition, for sensors utilising immobilised reagents, deterioration due to physical desorp- tion or chemical degradation of the reagent might limit the useful lifetime of the sensor. Limited dynamic range. Many optical sensors utilising immobilised reagents, e.g., acid - base indicators, obey the law of chemical equilibria. A plot of colour intensity against pH follows a sigmoid shape, with the majority of the change occurring over a rangc of 2-3 pH units, as discussed in the following section. In contrast, a potentiometric sensor follows the Nernst cquation, and gives an output voltage that is linear with pH over the range 0-14 pH units.Components. Chemical analysis using optical fibres offers the potential of miniaturised equipment based on recently developed electro-optic components. However, the com- ponents available are determined principally by the needs of the communications industry, and are not necessarily opti- niised for sensing applications. The relatively limited market size for components for chemical analysis means that this requirement alone is insufficient to initiate the development of new deviccs, e.g., lasers operating at a particular wavelength. Underlying Principles of Fibre Optic Sensors Optical effects Optical fibre sensing involves the probing of matter by photons.The photon can either be scattered or, if its energy is equal to that between the initially occupied and an excited state of the matter, it can be absorbed. Various parameters are associated with a photon flux, and changes in any one of these can give analytical information about the material being probed. These parameters, and some of the spectroscopic techniques used to measure their variation are given in Table 1. Absorption spectroscopy A beam of light, of initial intensity I(,, on passing a distance 1 (cm) through an absorbing (non-scattering) medium of concentration c (mol dm-3) will be reduced to an intensity I given by the Beer - Lambert law: log&)/I) = rlc . . . . . . (1) where E is the molar absorptivity, and is a constant for a particular species at a particular wavelength.The ratio loglo(k,/I) is, by definition, the absorptivity of the medium, and also, by definition, ten times the medium loss in decibels. This law is not followed by all species with increasing concentration because of subsequent, concentration depen- dent, reactions, e.g., polymerisation or complexation with the solvent. However it is generally applicable and is one of theANALYST, NOVEMBER 1989, VOL. 114 1361 Table 1. Classification of optical spectroscopies Parameter in the photon flux that varies Photon absorbcd Photon scattered Intensity . . , . . . . , Absorption and reflectance Turbidity and ncphelomctry Wavelength . . . . . . . . 1,umincscencespcctroscopy Raman spectroscopy Time charactcristics . . . . Lumincscencc lifetime Photon correlation Phasc or polarisation .. . . Polai-ised absorbance and circular Ellipsomctry spectroscopy yxctroscop y dichroism fundamental bases of many sensors. Consequently, measuring the optical density of a medium allows the concentration of a species to be calculated given E , the geometry of the system and no interfering absorbing species. More usually in fibre optic analyses F I is found by calibrating the system with standard solutions. Reflectance spectroscopy Reflection occurs when light meets a refractive index disconti- nuity. Specular reflection occurs from optically flat interfaces and is most important for intensely absorptive, crystalline analytes, whereas diffuse reflection occurs from non-optically flat boundaries. The reflectivity of an absorptive sample, in air, is given by equation (2) (2) where k is the material’s index of extinction and n is its refractive index.The latter is wavelength dependent, particu- larly in the vicinity of an absorption band. This results in the reflected light intensity increasing with increasing absorption coefficient. Virtually no fibre optic sensors use this basic reflectance technique as the transduction mechanism. However, many optical fibre chemical sensors incorporate relatively weakly absorbing species in conjunction with an optically denser, -‘white” material, the reflectance characteristics of which are virtually constant over the wavelength range interrogated. In this configuration the light passes through regions containing the absorbing species before and after being reflected by the second phase.Consequently the intensity of the reflected light decreases with increasing absorption coefficient. However, the relationship between the intensity of the reflected light and the concentration of the absorbing species is not linear. For diffuse reflectance the most widely used model is that of Kubelka - Munk.9 This assumes a semi- infinite scattering medium and relates the reflectance, R , to the concentration of the absorbing species on the scattering laver, C , by: ( 3 ) where S is a scattering coefficient; F(R) is commonly referred to as the Kubelka - Munk, or remission, function. This lack of linearity between reflectance, or absorptivity, and the analyte concentration means that sensors designed in this manner require extensive calibration and characterisation before they can be used analytically.Luminescence spectroscopy Following the absorption of the interrogating photon the excited species can lose its energy by one or more of several processes: (a) via a cascade of non-radiative deactivation steps, which could be intra- or inter-species, such that all the energy is converted into thermal energy; (b) by emitting a second photon; and (c) by inter-species transfer with an acceptor species that subsequently emits a second photon. Processes (b) and (c) usually occur in conjunction with process (a), both competitively and as part of the deactivation route of a single moicty. Consequently the photon< emitted are usually less energetic, shifted to the red end of the spectrum, relative to the initial excitation photons.This wavelength difference gives luminescence spectroscopy a sensitivity advantage relative to absorption spectroscopy because the former is measuring the appearance of photons of a particular wavelength relative to a low background, whereas the latter is measuring the reduction in intensity relative to the high background of the interrogating beam. As noise levels are depcndcnt on the square root of the background light level, it follows that lumincsccnce spectroscopy has a funda- mental signal to noise advantage. No distinction will be drawn between phosphoresccncc and fluorescence as the following generalities apply to both luminescence mechanisms, although lifetimes might range from around 1 ns to 1 ms.Luminescence spectroscopy can be used to obtain informa- tion on the concentration of an analyte in one of sevcral ways. Intensity. The intensity of the analyte luminescence is a fraction of thc amount of light absorbed, and thc latter is given by the Beer - Lambert law, equation ( l ) , for the majority of species. Rearrangement of equation (1) givcs I = I. e-cl( logelo, where I,) and I are the incident and transmitted light intensities, respectively. If the solution is only a weak absorber, i.e., EIC < 0.05 A, then I = I. (I - ELC log,lO) to within 0.7%. Whence I,) - I = I() czc log,l0 , 1 . . . . ( 3 ) where I. - I is the amount of light absorbed. Therefore for weakly luminescing species the luminescent intensity is proportional to the luminophore concentration, the constant o f proportionality being determined by factors including the quantum efficiency of the luminescence process, and the collection efficiency of the detector optics.For more strongly absorbing species the following further factors become increasingly important: (i) self absorption of luminescence light by the analyte, (ii) sclf quenching of the luminescing species and (iii) the variable volume illuminated by the exciting light and the variation in the collection efficiency with the distance from the end of the fibre. For example, if the majority of the incident light were adsorbed within 1 pm of thc end of the fibre this would result in a far higher detected luminescence signal than that obtained if the same amount of light were absorbed in a cone extending 1 cm from the end of the fibre.Quenching o f intrnrity. Some of the complicating factors dimmed above are reduced significantly if a constant concentration of luminescent species (L) is used with an analyte (A) that quenches this luminescence. The mechanism can be depicted by excitation luminescence - L + h’ L + hv - L” kl quenching by kc[ analyte, A . tt, L + A1362 ANALYST, NOVEMBEK 1989, VOL. 113 Under illumination with intensity Zo the appropriate differen- tial equation is d[L*]/dt = k,Z(,[I,] - k,[L*] - k,[L"][A] . . ( 5 ) At equilibrium under steady illumination, the steady-state luminescence intensity, k,[L*], is given by The ratio of this intensity with a quenching analyte concentra- tion of [A],L,, relative to the intensity when the concentra- tion of A is zero, Lo, is the well known Stern - Volmer equation &JLA = 1 + ks"[A] .. . . . . (7) where ksv is an amalgamation of constants known as the Stern - Volmer constant; ksv = k , / k l . Lifetime churucteristics. An alternative method of assessing the degree of luminescence quenching is to use a pulsed excitation source and to look at the temporal decay of the intensity. Solving the differential equation ( 5 ) gives When the concentration of A is zero this gives the simple exponential decay of the luminescent species. For other concentrations the luminescence half-life of the luminescing species is reduced from (In 2 ) / k l to (In 2 ) l ( k l + &[A]). This technique has a large advantage over simple intensity measurements for assessing the degree of quenching.As it involves comparing the luminescence intensity at various times it is self referencing, and compensates for factors such as the fouling of windows or changes in the efficiency of the source, detector or electronics. However it does incur the disadvantage of an increased capital cost for the controlling electronics, but given the advances in electro-optic com- ponents and electronics (for pulsing the excitation source and for signal processing) luminescence lifetime measurements are likely to become more widely used. Energy transfer. Another form of luminescence spectro- scopy involves energy transfer from the excited state of one species to the excited state of a second species, followed by luminescence from the second species.The luminescence intensity from this second species, S, is proportional to k,,[L*][S] where k,, is the rate of energy transfer between the two species, and is very dependent on the separation between the donor L* and the acceptor S. Also, as the intensity depends on the concentration of L", the factors that determine the intensity of luminescent light detected from the species L* are directly involved in determining the intensity of lumi- nescence from the second species, S. The principal advantage conferred by this technique is that it differentiates between that portion of species S in the very close proximity to L* and unassociated S, and can therefore be used to characterise the amount of bound S in a competitive equilibrium, e.g., in an immunoassay .Ram an spectroscopy The Raman process involves the inelastic scattering of photons from species with the loss (or gain) in energy being provided by vibrational energy of the scattering moiety, to give the Stokes (or anti-Stokes) Raman lines. This process has a much lower cross-section than absorption spectroscopy, typically a factor of lo9 less, however, because the scattered photon has a different energy to the incident photons the Raman technique has the same advantage over absorption spectroscopy that luminescence spectroscopy has. Its principal advantage is that it allows analysis based on the vibrational fingerprint of an analyte to be carried out at visible wavelengths, a region where silica fibres transmit light efficiently. The low cross-section for the Raman process has several important consequences for analysis using this tech- nique.(i) The interrogating light is not attenuated significantly due to the Raman process and therefore the intensity of a Raman band is directly proportional to the concentration of the analyte, assuming that the analyte absorbs neither the exciting laser nor the Raman scattered wavelength. (ii) Fluorescence from the analyte may swamp the Raman signal. (iii) Fluorescence or Raman scattering from the components comprising the optical fibre might be more intense than the Raman signal from the analyte. This can be reduced signifi- cantly by using different optical fibres to dclivcr the exciting light and to collect the Raman scattered signal. Whilst the first factor is an advantage, the remaining two are drawbacks.However, these may be ameliorated by using the rapidly developing technique of Fourier transform Raman spectroscopy. This typically employs 1.064 pm laser light, which substantially reduces the number of samples that fluoresce, making this technique more widely applicable. Chemical Equilibria The components of many optical sensors obey the law of chemical equilibria. This is particularly important for sensors using immobilised reagents, e.g., acid - base indicators, which are typically a weak acid where the acid form and its conjugate base have different optical characteristics. In such a system an equilibrium is established between the reactant, e.g., the undissociated indicator acid HI, and the reaction products, e.g., the hydrogen ion, H , and the conjugate base, I-.Hence for an acid - base indicator HI c H+ + I - . The law of chemical equilibria states that where K is the equilibrium constant of the reaction and [HI] is the concentration of HI etc. (strictly it is the activity of each component but in many circumstances replacing activity with concentration introduces little error). The amount of indicator present is usually a constant, c , such that the concentration of, for example, the conjugate base can be expressed by From the definition of pH it follows that [H+] = lO-pH, giving Therefore if I - is optically distinct from HI, then by measuring the absorbance at a convenient wavelength the concentration of I- can be found, as discussed previously. At low pH values equation (1 1) approximates to [I -1 = 0, whilst at high pH values [I-] = c.Consequently a plot of [I-], and the resulting absorbance strength, against pH follows a signioid shape, with the majority of the change occurring over a range of 2-3 pH units. In contrast, a potentiometric sensor follows the Nernst equation (Again, strictly it is the activity of H+ and not its concentration but this causes little difference for 0 < pH < 14). As pH = - loglo[H+] then the Nernst equation becomes E = E" - k'pH . . . . . . . . ( 13) illustrating that the ouptut voltage of a potentiometric sensor is linear with pH, over the dynamic range 0 < pH < 14. Although the different physical principles underlying poten-ANALYST. NOVEMBER 1989, VOL. 114 1363 tiometric and optical sensors based on equilibria does limit the dynamic range of the latter it also affords a high sensitivity over this limited dynamic range. Therefore, by choosing an indicator with the appropriate dissociation constant a highly sensitive probe can be made for a specific pH range.Immunological reactions The immune system of animals contains cells that, after being in contact with a “foreign” molecule, known as an antigen (Ag), secrete proteins, known as antibodies (Ab), which are shaped specifically to bind only to that antigen. The antibody - antigen binding is similar to other chemical equilibiria, and can be represented by Ag + Ab ; Ag-Ab. The equilibrium constant for this type of reaction, usually called the affinity by immunologists, is large, as can be anticipated, as usually both the antigen and antibody are present in low concentrations, < 10-6 moll-1, and a small equilibrium constant would mean negligible bound complex was formed.Such immune reactions form the basis of many clinical tests for the presence of antigens, e.g., human immunodeficiency virus and human chorionic gonadotrophin (the hormone used as a test for pregnancy), or antibodies, e.g., to rubella. These tests form the basis of a large commercial market, and consequently much research effort has concentrated on devising optical fibre analytical techniques suitable for measuring the degree of immune reactions. Instrumentation The equipment required for fibre optic chemical analysis can be subdivided into sources, optical fibres, detectors, wavelength selectors and othcr optical components.Specific examples of the equipment used for various configurations will be described later together with the performance of the system. Sources The sources can range from white light sources to lasers operating at unusual wavelengths, c.g., the 3.392-um line of a HeNe laser. Their size can range from 1.5 mm radius X 5 mm long light emitting diodes (LEDs) or laser diodes to eximer pumped dye laser systems occupying several cubic metres. Similarly their cost can range from a few pence to more than f100000. In fibre optic chemical analysis increasing use is being made of the small range of wavelengths, typically 20 nm, emitted by LEDs. These provide a cheap, robust, efficient, easily modulated source of light from 480 nm (for blue LEDs) to 4 pm (for infrared emitting diodes).Optical fibres The commercially available optical fibres used in chemical analysis range from single-mode single-strand fibres, e.g., with a 3 pm diameter light guiding core and 125 pm total glass diameter used in interferometers, to bundles of fibres that can be several millimetres in diameter. The composition of the light guiding core is most usually silica based, although increasing use is being made of plastic fibres, e.g., with a poly(methy1 methacrylate) core. The transmission windows for silica and plastic fibres are typically 350-1800 and 400-850 nm, respectively. Non-oxide glass fibres are being developed that transmit further into the infrared, with fluoride fibres transmitting over the region 0.4-4.0 pm being commercially available.Detectors As for light sources a range of detectors have been used for fibre optic chemical analysis. These vary from the small PIN photodiodes to the higher gain avalanche photodiodes and photomultiplier tubes. Similarly their size and cost varies from a few cubic millimeters and under &1 for the cheapest photodiodes, to larger than 10 x 20 x 30 cm cooled, high - speed, photomultipliers costing several thousands of pounds. Wavelength selectors and other optical components The wavelength selector appropriate to a given configuration depends on the proposed use of the system. When tunability is required, as a research tool or for many different analytes, a monochromator is used. For a system designed for a specific analyte a more economical and compact option is frequently a narrow band interference filter.The increased used of sources that emit light only over a narrow range of wavelengths means that for some applications no wavelength selector is required. The other components that can be used in an optical system include lenses, mirrors, and polarisers to focus, steer and polarise the light beam. Whilst these components are, in general, relatively cheap, they do add complexity to an optical system, requiring careful alignment, and the current trend is towards systems with a minimal number of such components. Fibre optic systems reduce the need for mirrors significantly, the flexibility of the fibre providing the spatial guiding of the light. Solid-state sources and detectors can be “pigtailed” to an optical fibre providing rugged, aligned emitter and detector conditions and eliminating the requirement for lenses to couple the light into and out of the fibre. All fibre optical polarisers are now commercially available, and integrated optics offers the potential of reducing further the number of discrete optical components required.Method of Classifying Fibre Optic Sensors Fibre optic sensors for both physical and chemical parameters can be subdivided into the following two types: (i) extrinsic sensors, where the optical fibre merely acts as a light guiding link between the measurement point and the interrogating and display electronics and (ii) intrinsic sensors, where the fibre, probably in some modified form, is the. sensing transducer. One type of extrinsic sensor, which will not be covered in this review, is the hybrid sensor, where the transduction mechanism produces a non-optical output that is then converted into an optical signal for transmission along an optical fibre to a receiver/display unit.An example of chemical analysis using such a device would be a glass pH electrode associated with electronics to produce a digitally-, intensity-, or wavelength-coded optical signal. Such a sensor would have some of the advantages discussed earlier, and could be used advantageously to save mass, gain freedom from EM1 or for safety reasons. Optical fibre chemical sensors, and consequently chemical analysis techniques utilising optical fibres, can be classified as follows: (i) species-specific sensors, (ii) non-species-specific sensors and (iii) indirect techniques.Species-specific sensors consist of remote spectrometry, where the optical properties of the analyte are measured directly, and sensors using immobilised reagents. Non-species-specific sensors involve measuring some optical property directly. but where that property might be perturbed by any one of a number of analytes. One example is a fibre optic refractometer, for which a change in, e.g., transmitted intensity merely indicates a change in the refractive index of the surrounding medium, and not the specific species that caused it. Finally, the indirect techniques involve using an optical fibre sensor to measure some non-optical parameter, e.g., strain or temperature, and relating the measurement to the analyte of interest.The non-species-specific sensors and indirect techniques predomi- nantly involve the intrinsic type of sensor. Description of Some Sensors Considerable research effort has been spent on demonstrating various fibre optic sensors concepts. The examples that follow are illustrative, demonstrating the range of sensors and1364 ANALYST, NOVEMBER 1989, VOL. 114 0.5 r n - source Scanning Fabry - Perot Detector band pass filter rf I output Sweep I Co ntro I signal --- electronics Received signal ' 1 Fig. 1. a Fabry - Perot filter Schematic diagram of an optical fibre methane sensor, using sensing techniques that have been studied and the advantages to be gained from using fibre optic sensors. The sensors are grouped according to the classification just described.Extrinsic Species-specific Sensors Remote spectroscopy In this type of sensor the fibre acts as a light guide coupling a light source to the analyte volume, where it is modified by the optical characteristics of the analyte, and returning the light from the sample to the detector. It can be described simply as remote conventional spectroscopy, and is one of the more developed areas of fibre optic chemical analysis. A scheme using absorption spectroscopy to detect gases has been proposed, using two lines from a laser system, one of which is absorbed by the analyte of interest whilst the other is not. 10 This allows referencing against potential interferences such as smoke particles or steam. The advantage to be gained from using optical fibres is that the absorption cell can be placed at, for example, the top of a chimney, whilst the laser and the detector are kept in a suitable laboratory on the ground.Also the instrumentation can be multiplexed to several absorption cells, allowing the output of several chimneys to be monitored. The principal disadvantages are the requirement of a laser and that the species of interest must have an absorption band within the transmission window of the optical fibre. For silica this extends from around 400 to 1800 nm. Although this is suitable for some gases, e.g., methane with its absorption bands at 1.33 and 1.66 pm, there are many gases with no absorption band in the transmission window of silica. Some of these, e.g., CO (2.3 and2.4 pm), HF (2.75 pm), NH3 (3.0 pm) and hydrocarbons (3.2-3.4 pm) have absorption bands that are accessible using the heavy metal fluoride fibres currently under development, the transmission windows of which extend to around 4 pm.The concept described above has been demonstrated using the 1.66-pm absorption bands of methane, a pulsed infrared LED, and a narrow band ( 3 nm) interference filter, detecting that part of the signal in phase with the light pulsing frequency." The detection limit was quoted as 700 p.p.m. (1.3% lower explosive limit, LEL). A slightly less sensitive version, with a detection limit of around 25000 p.p.m. (5% LEL) has been demonstrated where the sensing point was 10 km from the source and the detector.12 One potential limitation with both of these schemes is the lack of stability as no reference wavelength was used.There is a commercially available instrument that detects methane using this general principle, but including a reference wavelength. This is sold by ASEA as an optical fibre gas alarm. One elegant technique for both improving sensitivity and providing a reference against changes in sensitivity has been described by Dakin et a1.13 and is shown schematically in Fig. Sheathing Sealing tube Guide tube I I I Indicator and support polymer / \ I Membrane Optical fibre Fig. 2. Cross-sectional schematic diagram of a fibre-optic pH probe 1. Again the analyte of interest was methane, but the narrow band interference filter was replaced by a scanning Fabry - Perot etalon, a tunable multi-layer interference filter. The characteristics of this filter are a series of narrow (<0.001 vm), evenly spaced lines whose separation is virtually constant as their absolute position is swept over a small range (0.03 pm).The 1.66-pm absorption band of methane consists of a central Q branch with well resolved rotational structure in the neighbouring P and R branches. The important features of this rotational structure is that the absorption lines are very narrow, and are very evenly spaced (there is little rotational anharmonicity). By tuning the Fabry - Perot etalon so that the separation between neighbouring output fingers is equal to the separation between adjacent rotational bands, a sensor specific to methane is obtained. As the Fabry - Perot etalon is scanned its output lines move relative to the methane absorption peaks such that it scans through these, simul- taneously detecting the absorption of many bands.For this configuration any background absorption, scattering or obscu- ration merely appears as a constant offset on the output (plus a slight deterioration in the signal to noise ratio). The noise limited resolution of the technique has been reported as 100 p.p.m.,l3 making this a sensitive and selective technique. Within the United Kingdom Atomic Energy Authority (UKAEA) absorption spectroscopy is being used to analyse the composition of the various streams during fuel reprocess- ing.14 Used fuel from a nuclear reactor contains very radioactive fission products together with unused fuel: ur- anium and plutonium. The actinide elements have their own distinctive absorption spectra, and their concentration can be monitored using absorption spectroscopy.Optical fibres are useful for this application because of the extremely high radiation levels present, allowing the spectrometer to reside in an area where the radiation levels are low. Similar configura- tions but using a dye laser, have been described by workers at Karlsruhe15 where on-line measurements gave plutonium concentrations within 0.14 g 1-1 for concentrations up to 50 g 1-1. In France the CEA has also undertaken similar research, developing several instruments that are now manu- factured under licence16 principally for the determination of actinide species. Simple extrinsic species-specific sensors can also be used for remote luminescence measurements. This has been demon- strated by Wolfbeis et al.17 for the determination of aluminium in the range 1-800 p.p.m. by monitoring the fluorescence intensity of the aluminium - morin [2-(2,4-dihydroxyphenyl)- 3,5,7-trihydroxy-4H-benzopyran-4-one] complex on titrating with 1,6-diaminohexane-N,N,N',N'-tetraacetic acid. 17 The use of a fibre optic configuration gives good precision even when the solutions are coloured or turbid. Remote absorption and luminescence spectroscopy has also been used to detect minor species in combustion products. Using a specially designed water-cooled probe a detection limit for OH radicals of 30 p.p.m. was obtained for a pre-mixed air - methane burner.18 Another illustration of the use of extrinsic luminescence sensing in inaccessible areas was its use to establish whether a radioactive waste repository, in the USA, was leakiag.19 It was known that the repository contained uranium, which fluoresces in its common chemicalANALYST, NOVEMBER 1989, VOL.114 1365 state, U022+. Therefore Klainer et al.19 at the Lawrence Livermore Laboratory introduced optical fibres down very small bore holes, close to the repository. They shone blue light from an argon-ion laser down the fibres and looked for the characteristic green fluorescence from any uranium. In the biomedical field the thickening of arteries has been assessed by introducing an optical fibre, via a catheter, and examining the luminescence spectrum, because normal arteries and atherosclerotic arteries give different lumi- nescence spectra? The ratio of the luminescence intensity at 600 nm relative to 550 nm is particularly useful, with diseased arteries showing much lower intensity at 600 nm.Remote Raman spectroscopy using optical fibres has also been demonstrated. A detailed study of the parameters governing the design of an optical fibre system for use in Raman spectroscopy has been described,21 illustrating that such a configuration can significantly reduce background scattering and luminescence levels relative to conventional Raman spectroscopy. This technique has been used to monitor an alcoholic fermentation, giving information on concentrations of ethanol, glucose and fructose.22 Although the level of accuracy obtained, e.g., 9 g 1-1 for glucose, indicates that further optimisation of this configuration is required, reference 21 indicates that such improvements are possible.A further advance in the field of Raman spectro- scopy has been the recent emergence of Fourier transform (FT) Raman spectroscopy, using 1.064-pm excitation from a Nd:YAG laser. Fibre optic bundles have been used to deliver the laser light, and to collect the scattered light.23 A good quality Raman spectrum of the bright yellow, naturally occurring antibiotic, Amphotericin A , illustrates a major advantage of this technique relative to conventional Raman spectroscopy using visible excitation, when the spectrum is dominated by an overwhelming fluorescence background. However, currently the state of instrumentation for FT Raman spectroscopy is immature, with instruments only just becoming commercially available, but in the near future it is anticipated that this will become a further powerful tool in the spectroscopists armoury .Optical fibre sensing heads are commercially available from Oriel Scientific for absorption, fluorescence, turbidity and refractive index. Whilst these are not miniature, the fibre is a 5 mm diameter bundle, they are ruggedly constructed, e.g. , for use in a process plant.24 Other companies sell fibre optic absorption spectrometers, e.g., Guided Wave International (El Dorado Hills, CA, USA), and the increasing availability of such systems will promote their use further. Sensors utilising immobilised reagents A recent, very thorough review has been published on this type of sensor,5 with other publications concentrating on particular analytes, e.g., dissolved ionic chemical species,25 or work in particular laboratories.26 One such device is described in the patent of Elf UK,27 invented by Kirkbright et al. ,28 and is shown schematically in Fig. 2. A plastic optical fibre bundle of nominal diameter 1 mm was used, at the end of which was a sensitive tip consisting of a styrene - divinyl benzene copolymer, on to which the indicator bromothymol blue was adsorbed. This was retained at the end of the probe by a PTFE membrane. The complete device had a diameter of about 2 mm. Changes in pH in the vicinity of the sensitive tip caused a variation in the absorption spectrum of bromothymol blue as it changed from its yellow acid form to a blue basic form. The measured response was the attenuation of the light reflected back up the optical fibres at 593 nm relative to that at 800 nm, a suitable reference wavelength.The bromothymol blue probe is useful over the pH range 7-9. When subjected to a step change of 1 pH unit this early probe took around 65 s to reach 63% of its final value, (this corresponds to the l/e I End seal ’ Optical f Cellulose dialysis tubing ibres 0.3 mm Cellulose dialysis tubing Polystyrene microspheres (1 pm diameter) Polyacrylamide ’ microspheres (5-10 pm diameter) + phenol red Fig. 3. use. (b) Encircled part of (a) shown in greater detail (a) Construction of a fibre-optic pH probe for physiological characteristic lifetime assuming an exponential response function) and showed a stable response after 5 min.This relatively long response time was believed to be due to the rate of diffusion of solvent and ions across the PTFE membrane and through the reagent polymer sensing matrix. The group at the University of Manchester Institute of Science and Tech- nology have been successful in improving the design of the probe, reducing its response time.29 Similar probes have been described using other acid - base indicator systems to cover the pH range 3-12.30 By using a thin, 3-pm, porous cellulose acetate film, formed by spin coating, as the support for an immobilised pH indicator other workers31 have fabricated pH sensors which take 0.32 k 0.03 s to reach 63% of its final value when subjected to a pH change of 8 units. A similar probe using a low-cost LED and a PIN diode has been described, illustrating how a cheap miniature instrument can be constructed.32 This basic concept could be adapted for other analytes.For example, the enzyme urease (which selectively oxidises urea forming ammonia as one of the reaction products) could be incorporated with the pH-sensi- tive dye. The ammonia produced, being a base, would cause a change in the local pH. The sensitivity of this device would be dependent on the rate of mass transfer of urea to the sensing volume, its rate of oxidation and the rate of mass transfer of the ammonia formed away from the sensing volume. Altern- atively, if the product of the enzyme reaction is coloured an immobilised enzyme can be used alone.33 Commercially exploitable application areas range from process plant, environmental sensing, as demonstrated by a system designed to measured the pH of rainwater,34 to biomedical uses.A miniaturised pH probe was described several years ago and is shown in Fig. 3. This probe was designed for in vivo use,35 and so is configured to be sensitive over the pH range of physiological interest, 7.0 < pH < 7.4. Its quoted sensitivity is 0.01 pH units. In many ways it is similar to the device described earlier. However, one of the major differences is the use of 1 pm diameter polystyrene microspheres, which increase the amount of light backscat- tered into the fibre. The device has been licensed to a commercial company for product exploitation, Cardiovas- cular Devices (Irvine, CA, USA). Few extrinsic species-specific sensors using absorptionhe- flectance of immobilised reagents have been reported for use with gases or immunoassays.1366 ANALYST, NOVEMBER 1989.VOL. 114 Many research groups have described extrinsic species- specific sensors that use luminescence to obtain information about the concentration of an analyte. The earlier sensors tended to use the intensity dependence of pH-sensitive luminophores to monitor pH. A typical example is reported by Saari and Seitz36 who used fluorescein amine, immobilised covalently either on to glass having a controlled pore size or on to cellulose, held at the end of a 4.5 mm diameter bifurcated fibre, to demonstrate the concept. The fluorescence intensity at 488 nm was found to be pH dependent over the pH range 3-9. The lack of any reference meant that the device could not be used quantitatively without an immediately preceding calibration step, and the response time of the device was found to be 15-30 s.Subsequent papers have described further developments of this basic concept.37-s9 For example, a recent paper describes the chemical linking of fluorescein isothiocyanate to silanised, controlled porosity, glass beads.39 These had an average pore diameter of 50 nm and an average bead diameter of 150 pm. A single bead was stuck to the end of an all-silica fibre (core diameter, 105 pm; cladding diameter, 125 pm) using an ultraviolet curable epoxy resin. This miniature pH probe gave a change in fluorescence intensity at 525 nm over the pH range 2.5-7.5, with a response time of around 20 s. No claim was made regarding its accuracy but relative intensities were quoted k0.025, corresponding to k0.25 pH at pH 4 and kO.1 pH at pH 7.The fluorescence intensity was also found to be temperature sensitive, changing by the equivalent of 0.8 pH units between 8 and 33 "C, and also dependent on the other ions present. The workers state that each individual probe must be calibrated. Notwithstanding the limitations discussed above the fact that the dye is attached covalently significantly improves the long-term stability relative to devices where indicators are simply adsorbed on to a substrate, except at high pH when hydrolysis of the covalent linkage occurs. The pH range that can be measured using this type of probe can be extended by using other pH-dependent luminescent indicator^.^" Sensing schemes have been described that give the quantita- tive simultaneous measurement of pH and ionic strength using various indicators and methodologies.Opitz and Lubbers41 described the use of 8-hydroxypyrene-l,3,6-trisulphonic acid and analysis of the luminescence resulting from variable wavelength excitation, uhilst Wolfbeis and OffenbaeheF used 7-hy droxycoumarin-3-carboxylic acid immobilised on to two differently prepared surfaces where both sensors are interrogated. In addition to pH, several papers describe sensors for measuring concentrations of dissolved oxygen, usually based on luminescence quenching. One such sensor is described by the same workers who developed the pH probe for physiologi- cal use, Fig. 3, using a similar probe design with the dye perylene dibutyrate, which was the best of 70 dyes examined for this application.-'3 Experimentally, it was found that a plot of fluoresence intensity \'emus the partial pressure of dissolved oxygen was close to that predicted by theory [the Stern - Volmer plot, equation ( 7 ) ] .This probe was designed initially Polymer enclosure Thermocouple / pH sensor pC0, sensor Optical / fibres ..,- \ p02 sensor \ Tip coating Fig. 4. Schematic diagram of an intravascular blood gas probe for use in vivo, particularly to measure the partial pressure of dissolved oxygen in the blood of patients under anaesthesia. However, it has been charaeterised in both aqueous and gaseous environments, and was found to be sensitive over the range 0-20% oxygen. An accuracy of around 0.130/0 was claimed within this range for solutions and over most of the range in a dry gas stream.Significant, and progressively increasing, errors occurred after 80 min of use in whole blood during an in vivo experiment due to fouling, i.e., the formation of blood clots and other proteinaceous layers. The probe was removed, cleaned, re-inserted and found to be restored to its original sensitivity. This paper clearly demon- strates that such probes can be used usefully for physiological monitoring. A similar sensor, using the quenching of luminescence from pyrenebutyric acid and the Stern - Volmer relationship to relate the luminescence intensity to the partial pressure of oxygen, has been developed to monitor the concentration of oxygen in a gas stream between 300 and 500 OC.4-l The device includes a thermocouple to provide data for the temperature compensation.A recent paper describes how the reduction in fluorescence lifetime of tris(2,2'-dipyridyl)ruthenium(IT) dichloride hydrate, caused by fluorescence quenching, can be used to measure the partial pressure of oxygen.45 This was done using a relatively cheap blue LED, instead of a pulsed laser, modulated at 455 kHz and measuring the relative phase shift of the luminescence signal with respect to the driving current of the LED, to give a precision and reproducibility of 2 ns. This corresponded to the determination of the lifetime to +1%, and the measurement of the partial pressure of oxygen to k0.3770 in the range 0-20%0 oxygen. As discussed earlier this change in lifetime is independent of the concentration of the luminophore, and as anticipated the sensor described showed negligible drift due to the leaching and bleaching of the indicator.Another advantage to be gained from using time-resolved luminescence studies is the significant reduction of interfer- ents in favourable circumstances. In biological analyses there is frequently a mixture of many luminescing species of which the majority have short, < I ps, lifetimes, and give a large background obscuring the luminescence signal from the species of interest. Rare earth metal chelates, which have long luminescent lifetimes, typically 0.1-1 .0 ms, can be used as luminescent labels on the species of interest. By using a pulsed excitation source, and looking only at the luminescence that occurs some time, a few microseconds, after the excitation pulse, the background can be discriminated against.A recent paper using this technique for a model immunoassay demon- strated an increase in the limit of detection of nearly three orders of magnitude, relative to an assay using a label with a short lifetime.4h Another physiologically important gas is carbon dioxide. This has been measured by luminescent techniques, making use of its equilibrium with water, i.e., CO2 + HZO H+ + HC03-. The CO2 generates a change in pH, which can be measured using the techniques described earlier. The average venous and arterial partial pressures of carbon dioxide (pC02) entering and leaving the lungs are 6.1 and 5.3%, respectively. To cover the pCOz range 3.9-7.9% an indicator sensitive to the pH range 4.4-4.7 is required.Two configurations have been described by Hirschfeld et aZ.47 using either a porous glass bead or a liquid plug. The above concepts for the determination of pH, p 0 2 and pCO2 have all been combined into a single probe by Cardiovascular Devices,@ as shown in Fig. 4. The perform- ance figures for a typical prototype, as taken from reference 48 arc given in Table 2. Other analytes for which luminescence-based sensors have been described include ammonia (using the pH-dependent equilibrium of ammonia in water linked to a dye the luminescence intensity of which is pH dependent+)), lactateANALYST, NOVEMBER 1989. VOL. 114 1367 Table 2. Performance of a multianalyte intravascular probe (taken from reference 48) Parameter PH pcoz, % p a , Yo Range .. . . . . . . . . . . . . . . . . 6.8-7.8 1.3-13 2.5-29 Calibration accuracy . . . . . . . . . . . . 0.034 0.42 0.43 displayed value, one standard deviation) . . . . 0.027 0.25 0.37 Stability (maximum change in 12 h) . . . . . . . . 0.042 0.37 0.59 a step change of ca. 45% of the range)/s . . . . . . 54 54 42 Precision (short-term fluctuations about mean Response time (to reach 63% of final value for I I T 3 mm I I I 3 mm Fig. 5. Schematic diagram of a competitive fluorescence glucose sensor. C = Concanavalin A; D = FITC-labelled dextran; and = glucose. Free FITC-labelled dextran is denoted by an encircled D and pyruvate (utilising an iminobilised dehydrogenase enzyme and the spectrofluorimetric detection of consumed or gener- ated nicotinainide adenine dinucleotide, NADHSO), and ethanol (using the enzyme alcohol dehydrogenase and NAD+ to form acetaldehyde and NADH, again with spectroflu- orimetric detection of the NADH generated").Another gas that has been detected by fluorescence quenching is SO2. Using the dye benzofluoranthene the luminescent intensity was found to follow the Stern - Volmer equation over the range 0.014% S 0 2 . 5 2 Wolfbeis and Schaffar53 have described an ion-selective optode for the continuous determination of potassium. This used the fluorescence intemi ty of a lipid-soluble, modified rhodamine B dye, incorporated in a Langmuir - Blodgett deposited bilayer lipid membrane (BLM), to sense electrical potential. The fluorescent, hydrophilic end of the dye is in the central region of the BLM insulated from the external solution. Specificity was obtained by incorporating valinomy- cin, an ionophore specific to potassium, into the BLM.A plot of the reduction in fluorescence intensity relative to the fluorescence intensity in the absence of potassium ions gave an approximately linear dependence against loglo[ K+] over the range 10 mM-10 p ~ . The relative response to sodium ions, a potential interferent, was shown to be ca. 2 x 10-3, a 10 mM NaCl solution giving a similar response to a 20-VM KC1 solution, and it is described how this can be reduced further by using the output from a second reference sensor containing no valinomycin. This interesting technique was actually demon- strated using a BLM deposited on to a microscope slide instead of an optical fibre.However, there is no inherent reason why this should not be extended to optical fibres and to sensing other ionic species. Another luminescence sensing concept based on competi- tive binding within a semipermeable membrane has been described54*ss and patented by Schultz and co-workers.56 The sensor measures glucose concentration and is shown schemat- ically in Fig. 5. The sensor involves immobilising concanavalin A, a reagent that binds to sugars such as glucose and dextran, on to the inside walls of a short length of dialysis tubing. Dextran labelled with the fluorescent dye fluorescein isothioc- yanate (FTTC), is bound to the immobilised concanavalin A. This relatively large molecule will not diffuse through the walls of the dialysis membrane, whereas glucose can.When the sensor is placed into a solution containing glucose some of these molecules pass through the membrane and displace some of the FITC-labelled dextran from the concanavalin A into solution. The sensor is illuminated with blue light, to coincide with the absorption spectrum of the FITC, and the intensity of the resulting green fluorescence is monitored. Because the walls of the dialysis membrane are substantially outside the volume illuminated by the fibre, in the absence of any glucose little fluorescence intensity is observed. The sensor was found to be linear in the range 2.8-22 mM glucose, and had a typical response time of 5-7 min. In this type of device the response time is determined not only by the diffusion and permeability characteristics of the membrane, but also by the forward and reverse rate constants of the binding reactions.Changes of the chemical equilibria, the membrane thickness and composition could be used to modify the time characteristics of the device. Fluorescence energy transfer has been used as a trans- duction mechanism in a modified version of the above sensor.~7~~8 Labelled concanavalin was used in conjunction with labelled dextran, one as a donor and the other as an acceptor. In the absence of glucose efficient energy transfer occurs between the bound dextran and concanavalin A. In the presence of glucose some of the dextran is displaced from concanavalin A by glucose, and consequently there is a reduction in the luminescence intensity from the acceptor species. Fibre optic probes have also been demonstrated for chemildminescence .sq Chemiluminescence reactions generate photons as a result of a chemical reaction. Consequently no interrogating photon source is required, but merely a detec- tor.The probe demonstrated59 contained the immobilised enzyme peroxidase, in a polyacrylamide gel on the end of the fibre. The enzyme catalyses the oxidation of luminol by hydrogen peroxide, giving chemiluminesccnce with a maxi- mum intensity at 430 nm. The sensor was configured so as to provide an analysis for hydrogen peroxide. The experimental data and theoretical treatment show that the emitted light intensity is proportional to the concentration of hydrogen peroxide when luminol is present in excess, and certain other conditions are met.The advantage of this type of sensor is its optical simplicity. The principal disadvantages are that there are only a very limited number of chemiluminescent reactions, and that in the configuration described an excess of reagent, luminol, has to be added. Intrinsic Species-specific Devices For intrinsic sensors the interrogating light remains guided. Therefore interaction with the analyte, or an immobilised analyte sensitive reagent, can only occur within the waveguide or in its vicinity by an evanescent wave interaction. The idea of propagating light through the reagent phase was first proposed by Hardy et al. ,60 although in practice it suffers from several drawbacks. It requires an optical guiding region of the reagent phase, which invariably implies fabricating special cylindrical (fibre) or planar waveguides, and that the analyte be able to penetrate into this reagent phase.For configurations where the waveguide is a solid, the ease of mass transport into the sensitive region will usually be dependent on the size and number of pores or channels in the waveguiding region. Circumstances that increase the rate of mass transport1368 ANALYST, NOVEMBER 1989, VOL. 114 will degrade the optical quality of the waveguide, principally through scattering losses from the pores. Despite these disadvantages this type of sensor has been demonstrated using a photopolymerised channel of di- methacrylate oligomer containing uranyl perchlorate as the photoinitiator (refractive index 1.6) on a glass substrate (refractive index 1.5).61 The intensity of HeNe laser light through this guiding structure was found to alter in the presence of toluene, nitrobenzene or ammonia, but to be insensitive to water, acetone, methanol and propanol.Elunescent wave devices A much more experimentally convenient configuration for an intrinsic species-specific chemical sensor is to have the analyte, or a reagent sensitive to the analyte, in close proximity to the waveguide, and to measure its interaction with the evanescent field to determine the presence of the analyte. When light undergoes total internal reflection a portion of the energy extends a little way beyond the interface, into the optically rarer medium. This is known as the evanescent wave, and it can interact with the optically rarer medium, being transmitted, absorbed (possibly causing luminescence), re- flected or scattered by it. If the optically rarer medium is absorptive then the amount of power being transmitted by the optically denser guiding core will be attenuated.Similarly, a fluorescence species just beyond the interface can be excited by the evanescent wave interaction, and the luminescent photons can couple back into the waveguide by the reverse effect. The electric field intensity of the evanescent wave falls off exponentially with the distance from the interface, with a characteristic l/e decay distance, d,, given by . . . . . . h 2nVnd2sin2@ - n,2 d, = where h is the wavelength of the light propagating in the waveguide, nd and n, are the refractive indices of the optically denser core and optically rarer surrounding medium, respec- tively, and 8 is the angle of the guided mode relative to the waveguide/surroundings interface.For typical values of h = 500 nm, nd = 1.46, n, = 1.33, and 0 = 75", corresponding to a silica - aqueous media interface, d, = 170 nm. The consequences of this for sensing applications are that the light need not be coupled out of the guiding medium in order to sense the analyte, and also the spatial volume probed is restricted to that within 1 pm of the waveguide. These properties have been exploited in fibre-optic chemical sens- ing, principally in the biomedical area to measure antibodies or antigens. Antibodies are nature's method of selectively identifying, and binding strongly to, foreign materials, known as antigens, that are frequently proteins such as bacteria or virus.In immunological assays the specificity required to identify a given antibody or antigen is obtained by using the conjugate of the antigedantibody pair. This reagent is immobilised on a waveguide surface, and the evanescent wave interaction is used to probe specifically this immobilised layer. However, the presence of any fouling on the interface will prevent the analyte from being at the interface and will cause a significant reduction in Sensitivity. The underlying theory of evanescent waves, and their properties, was first published over two decades ago.62 A recent quantitative treatment has calculated the theoretical sensitivity of a variety of guiding structures to the presence of absorbing chemical species.63 Both planar and cylindrical waveguides were considered, as also was the effect of buffer layers (fouling) at the surface of the waveguide.The reduction in sensitivity caused by a layer of thickness s, is given by exp -(2s/d,), where d, is the penetration depth, as defined above. Therefore, for the typical values used above where d, = 170 nm, a fouling layer only 10 or 50 nm thick reduces the sensitivity from 100 to 89 or 56%, respectively. Recent reviews have discussed optical immunoassays at continuous surfaces64 and biosensors based on evanescently excited fluoroimmunoassays.65 Some of the earliest patents and papers were published in 1975 describing the use of evanescent absorption spectro- scopy. Sodium picrate, co-deposited on a 1-mm silica rod with poly(viny1 alcohol), was demonstrated to give a selective test for cyanide ions, detecting <0.1 pg of cyanide.6" Absorption spectroscopy, and the same experimental configuration, was used to detect ammonia, by its reaction with iron(II1) sulphate (which is off-white) to produce violet ammonium iron(II1) sulphate.66 This Monsanto patent also describes the use of scattering by antigen-coated polystyrene spheres upon immune binding to the appropriate antibody immobilised on the glass rod.One difficulty with the above reactions is that the equilib- rium constant for each is very large, i.e., the reactions are essentially irreversible. A reversible optical waveguide sensor for ammonia has bcen described67 that uses the pH-sensitive dye oxazine perchlorate.Concentrations of ammonia as low as 60 p.p.m. were detectable, and the device had a time constant of < 1 min. One aspect of this work was the use of a coated capillary tube instead of a solid glass rod, to increase the number of reflections with the outer surface, and hence sensitivity . A humidity sensor has been described, again using a coated capillary tube geometry, where the variation in absorbance characteristics of the reversible hydration of cobalt(I1) chloride was used as the transduction mechanism.68 This device was most sensitive over the range 60-95% relative humidity. Evanescent wave absorption sensors have been demon- strated for gaseous and liquid analytes, without the addition of any further reagent.For gases a relatively strong absorption band must be selected to provide adequate sensitivity. Methane gas has been detected by evanescent absorption spectroscopy using its strong absorption of the 3.392-ym line of a HeNe laser, which corresponds to a fundamental C-H rotational vibrational absorption band. Ten millimetres of a single fibre, made by drawing a step index silica fibre down to 1.8 pm diameter, was able to detect below 5% methane in air.69 This does not represent a commercially viable configur- ation because the very small diameter fibre is extremely fragile, and the intrinsic loss of silica at 3.392 pm, due to Si-0 overtone absorptions, limits its sensitivity. However, the development of fluoride fibres, showing much less loss at this wavelength, alleviates this problem.Similar studies, using the bare silica core of a 200-ym plastic clad silica fibre and rhodamine dye as the absorber, have demonstrated that liquid analytes can be sensed.7" However, the fibre in such a configuration becomes fragile when its protective polymer coating is removed. It has been demonstrated that non-polar solvents containing dissolved analytes can penetrate into the silicone rubber coating of a plastic clad silica fibre,71 thereby allowing spectroscopic analysis without severe degradation of the mechanical properties of the fibre. The sensitivity measured was lower than for the bare fibre,70 due partially to the larger diameter of the coated fibre, and partially to the presence of the coating. An interesting observation was that coiling the sensing region increased the sensitivity by up to a factor of two, due to a redistribution of the light into higher order modes (for which the penetration depth of the evanes- cent field is greater).The effect of coiling the fibre also provides a longer interaction length within a compact probe, increasing the sensitivity further without increasing the size of the sensing volume significantly. In addition to abswrption spectroscopy scattering of the evanescent wave has also been used as a basis for chemical sensing. A silica fibre surrounded by a microporous silica cladding, consisting of corpuscles up to 0.2 ym, shows a transmitted power level that is dependent on the surrounding relative humidity.72 This sensor was demonstrated to respondANALYST, NOVEMBER 7989.VOL. 114 3 1 % Flow cell '> I 1369 PMT Light source Sample out t Optical fibre I Filter I SamDle in Fig. 6. sensing Diagram of a fibre-optic assembly for cvanesccnt wave Sensitive coating Waveguide of W03 + Pd I I Substrate - LiNb03 Fig. 7. Integratcd optic cvancscent wave hydrogen sensor to 2045% relative humidity. Several of these were joined via an unmodified optical fibre to give a total length of 130 m. This was interrogated using optical time domain reflectometry to give a quasi-distributed sensor. Various researchers have patented different aspects of evanescent wave sensing. Some generic patents are held by Buckles73J4 and the Battelle Memorial In~titute.75~76 Whilst these cover many different aspects of evanescent wave sensors the Battelle patents in particular quote results based on absorption and luminescence spectroscopic determinations of immunological reactions.Fig. 6 illustrates an experimental arrangement, based on an optical fibre, that was used to verify the feasibility of this concept.77 (The optical fibre can be replaced by a waveguide supported on an optically rarer substrate, in contact with the analyte.) The antibody to the antigen of interest was immobi- lised on the outer surface of a bare 600 pm diameter optical fibre core. A known amount of fluorescently labelled antigen was added to the analyte and flowed through the cell. There is competition for the binding sites on the immobilised antibody between the added fluorescent antigen and that already present in the analyte solution.The higher the concentration in the original analyte the fewer the number of fluorescently labelled moieties that bind. A1 ternatively , a sandwich immunofluorimetric assay can be used. This was demonstrated using the antibody to immuno- globulin G (IgG), a blood seven protein, immobilised on the fibre, adding a standard solution of IgG, and then incubating for 10 min. This solution was then flushed out, and a second fluorescein isothiocyanate labelled antibody to IgG was added; this fluorescent antibody bound selectively to those sites already containing an IgG molecule. Therefore, this assay gave an increase in fluorescence signal with an increase in analyte concentration. The limits of detectability quoted were 3.0 and 1.5 mg of IgG per litre for the planar and fibre optic waveguides, respectively (corresponding to 2.0 X l0kx or 1.0 x 10-8 ~).77.78 It must be emphasised that the configurations just outlined are experimental, aimed at demonstrating the concept.Their significance is both technical, in the use of the spatial selectivity of the evanescent wave interaction to discriminate between bound and unbound analyte, and commercial, as Analyte in \ Flow cell Analyte out H Pyrex substrate / Laser beam n Fibre to readout detector Fig. 8. Schcrnatic diagram of an integrated optical biosensor there is a large potcntial market for simple, disposable clinical biosensors. Two more commercially attractive configurations have been patented, both using a capillary gap to meter the quantity of analytc sampled and therefore making the devices simpler to use.One is based on an optical fibre within a surrounding capillary tube79 and the other is based on a planar waveguide configuration. 80 Another interesting evanescent wave sensor is an integrated optical device designed to sense hydrogen. Integrated optics are the optical equivalcnt of the ubiquitous silicon chip, in which a single substrate is masked and modified to give the optical characteristics required. Nippon Sheet Glass have made and demonstrated a device based on an LiNb03 substrate with a thin sensitive coating of W03 + Pd on one arm of the substrate817gz as shown schematically in Fig. 7. If hydrogen is present it dissociates o n the surface of the palladium and reduces the W03, to form a hydrogen tungsten bronze, H,W03, which is deep blue in contrast to the pale yellow of W03.It is this colour change that is monitored to indicate the presence of hydrogen. The attenuation at 1.3 pm was found to obey the expression attenuation [HZ]0588 and levels of hydrogen as low as 20 p.p.m. were detected, although the response time of the device was around several minutes. Non-species-specific Techniques Changes in refractive index can be used to measure variations from an optimal composition, e . g . , of a feedstock in a process plant. Although such measurements indicate a change in the chemical composition generally they do not reveal which species is involved; this has to be found by other techniques, or by a knowledge of the process involved. A fibre optic refractometer has been described in a patent of the Battelle Memorial lnstitute.83 This uses a series of curves with different radii o f curvature, and i n different directions, in preferably a step index optical fibre.The presence of the cladding does not fundamentally modify the phenomenon of light loss, and consequently polymer-coated fibres can be used without having to remove the protective polymer coating, obviating the problem of fragility that occurs when bare fibres are used. In an example using a step index, all plastic fibre, the coefficient of contrast varied by a factor of 120 when the refractive index changed from 1.40 to 1.45, indicating that this is a sensitive configuration. A guided wave device designed to measure small changes in refractive index, is described by its inventors as an integrated optical biosensor (IOBS) ,8435 and is shown schematically in Fig. 8.This uses a grating coupler to couple light from a laser into a planar waveguide, with the efficiency of the coupling being monitored by a fibre optic probe. This efficiency is very dependent on the refractive index of the analyte solution, and the device can resolve changes in refractive index as small as 5 x 10-5. This is a miniature device, with the cell volume being1370 Silver film (38 nm) Liquid or vapour phase analyte ANAI>YST, NOVEMBER 1989, VOL. 114 Al2O3 film (0.2 pm) MgF2 film (0.2 pm) Ag ion exchanged t 1 microscope slide Fig. 9. device Cross-section of a waveguide surface plasmon resonance 0.2 ml. It can be configured with two devices in parallel, being fed by the same analyte, one having an enzyme column preceding it.In such a configuration the difference in the signal from the two devices will be a measure of the change in refractive index induced by the enzyme column, and the selectivity of the enzyme will make such an embodiment species-specific. An optical fibre refractive index sensor has been described where two fibres, with their protective coatings intact, were twisted together. The fibres consist of a 200 pm diameter fused silica core, 100-pm silicone rubber cladding and a 70-pm nylon jacket .86 The amount of light coupled from one into the other was found to be a function of the refractive index of the surrounding medium. Whilst there are some uncertainties about the exact mechanism involved, and the sensitivity is significantly less than the IOBS outlined above, the simplicity of this technique makes it attractive.Another evanescent wave coupled fibre optic chemical sensor has been demonstrated as a pH sensor.87 This uses a coaxial directional coupler that senses a change in the refractive index of an outer polymer coating. In the configura- tion described the polymer contained the pH-sensitive dye Phenol Red, the refractive index of which at the wavelength of the interrogating HeNe laser, 633 nm, changes with pH due to anomalous dispersion. Although the accuracy of this device was less than that quoted for some end-on intrinsic, species- specific pH sensors, its speed of response was rapid, around 5 s, to changes in pH of a gas stream. Surface plasmon resonance can also provide the basic transduction mechanism for a non-species-specific chemical sensor.A surface plasmon is a particular form of electromag- netic wave that propagates along the surface of a metal. It can be excited optically by light undergoing total internal reflec- tion at the surface of a glass substrate on to which the metal film has been deposited. With a proper choice of metal, usually silver or gold, and its thickness, usually a few nm, excitation occurs at a particular angle of incidence, leading to a sharp dip in the intensity of the reflected beam at that particular angle. This angle is related to the resonant frequency of the surface plasmon, and is very sensitive to variations in the refractive index of the medium immediately adjacent to the metal surface.The sensitivity of the resonant frequency as a function of distance from the metal surface, falls exponentialy, having the same form and a similar decay constant to the evanescent field interaction discussed pre- viously. This principle of analysis has been shown to be suitable for sensing anaesthetic gas concentrations88 and for detecting immune bir1ding.8~ In the latter instance the spatial selectivity conferred by the technique makcs it particularly suitable for this type of analysis. References 88 and 89 used bulk optics, measuring the angular dependence of the resonance on a metal coated prism. An ingenious extension has been reported by Kreuwel et af.90 using the planar waveguide configuration shown schematically in Fig. 9. The thin waveguiding layer only supports a few modes. By injecting white light selectively into one mode the different propagation characteristics of the Fig.10. Schematic diagram of a Mach - Zehnder interferometer different wavelengths make only a narrow range of wavelengths capable of exciting the surface plasmon. As the frequency of the surface plasmon alters, demonstrated by placing reagents having different refractive indices on the surface of the device, the wavelength absorbed altered. Hence this invention converts the angular dependence into a wavelength-selective filter. It has been demonstrated on a planar waveguide, and could be adapted to an optical fibre. Indirect Techniques These measure, optically, a physical parameter that varies as a consequencc of the presence of a chemical.Interferometers can be used to measure changes in optical path length of one arm relative to a reference arm. These changes can result from a change in the physical length of the arm, or in the refractive index of the core. A typical experimental arrangement for a Mach - Zehnder interferometer is illustrated in Fig. 10. Such configurations have been used as the basis for chemical sensors. When a portion of one arm is coated with a thin film of palladium in the presence of hydrogen it undergoes a small change in length, due to the formation of non-stoicheiometric palladium hydride.91 Consequently such a device is a specific hydrogen sensor. It has been suggestcd that coating an optical fibre with a catalytic coating, to catalyse the exothermic oxidation of a combustible gas, in conjunction with a technique to measure the temperature of the fibre, would give a fibre-optic combustible gas sensor.92 This concept has been demonstrated using an interferometer to sense the increase in temperature.93 In this instance the change in temperature causes both a change in the refractive index of the fibre and its length. For silica the former effect is dominant.When a 100 mm long sensing element was used, made by evaporating a 3-pm coating of platinum on this region, a temperature rise of 1 "C was observed for 2% C4Hlo and 7% CH4,93 the different sensitivities reflecting the different rates of catalytically enhanced oxidation. The quoted resolution is 10-4 "C, indicat- ing that this technique could detect these gases at concentra- tions << 1%.Another indirect non-species-specific chemical sensor has been developed by British Gas in collaboration with Pilking- ton Security Equipment (Denbigh, Clwyd, UK).y4 By the appropriate choice of materials it is possible to arrange for the core and cladding refractive indices of a special fibre to cross over at around -55 "C. Hcnce if any part of the fibre is below this temperature the refractive index of the cladding exceeds that of the core, and previously guided core modes move out into the cladding where they are either absorbed or scattered by the outer coating of the fibre. This fibre is used as a distributed cryogenic leak detector system, where several kilometres can be distributed beneath a cryogen, e.g., a liquefied natural gas (LNG) store.In the event of a leak the LNG, b.p. -165 "C, is detected by the fibre and triggers an alarm. This is a particularly elegant way of sensing a large areaANALYST, NOVEMBER 1989, VOL. 114 1371 with one, intrinsically safe sensor, and demonstrates the potential benefit to be gained from distributed sensors. Conclusions The examples given illustrate that the subject of optical tibre chemical sensors is still in its infancy, and is currently receiving a significant amount of attention. A large number of different sensing concepts have been demonstrated in the laboratory, although as yet few are in the market place. The majority of effort in the future is likely to be in transforming the demonstrated concepts into commercially viable instruments. For sensors using immobilised reagents the success of this will depend critically on being able to dcvelop reagents and techniques that give, reproducibly, the sensors fabricated the required sensitivity and lifetime.Therefore, in the next few years the sensors that are most likely to become available are either those not using immobi- lised reagents, i. e., an increasing use of remote spectroscopy, or those for which optical fibre sensors offer very significant technical advantages over existing sensors, especially where safety considerations are important. These include in vivo medical sensor$, where safety is vital and a long sensor lifetime frequently is not important, immunological assays, where the spatial discrimination of evanescent wave spectroscopy pro- vides a large technical adavantage, and sensors for use within fire hazard environments.Also the commercial advantages to be gained from being able to monitor analytes continuously, and in situ, as distinct from single laboratory-based measure- ments, will provide the impetus for further developnient of the concepts already demonstrated. It must also be remembered that the success of optical fibre sensors in the market place will depend on how they perform relative to other, principally electrochemical, sensors. There- fore significant developments providing improved electrode systems will have a direct detrimental effect on the attractive- ness of fibre sensors. Overall, fibre optic chemical sensors are a reality, offering the potential of a valuable complement to existing chemical analytical techniques and sensors.This work is part of the long-term research carried out within the Underlying Programme of the UKAEA. 1 . 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References Angel, S. M., Spectrosc.. 1988, 2, 38. Harmer, A. L.. in Chester, A N., and Martellucci, S . , Editor?, ”NATO Advanced Study Institute Series E: Applied Science,” Volume 132, 1986. p. 201. Guthrie, A. J . , Narayanaswamy, R . , and Rus\ell, D . A., Trans. Inst. Meus. Control (London), 1987, 9, 71. Narayanaswamy, R., and Sevilla 111, F., J. Phys. E , 1988, 21, 10. Seitz. W. R.. CRC Crit. Rev. Anal. Chem.. 1988, 19, 135. Wolfbeis, 0. S . , 2. Anal. Chem.. 1986, 325, 387. Pctcrson, J . I . , and Vurck, G. G., Science, 1984.224, 123. Wolfbeis. 0. S . , Purr Appl. Chem., 1987, 59, 663. Guthric. A. J . , Narayanaswamy, R., and Russell, D. A., Amlysr, 1988, 113, 457. Inaba, H.. Kobayasi, T., Hirama, M., and Hamza, M., Electron. Lett.. 1979. 15. 749. Chan. K., Ito, H., and Inabc, H.. J . Lightwuve Tethnol., 1984, Chan, K., Ito, H . , Inabc, H., and Furuya, T., Appl. Phys. B, 1985, 38. 11. Dakin. J . P . , Croydon, W. F., and Hedges. N. K., Proc. SPIE-lnt. Soc. Opt. Eng., 1988, 949, 200. Brown, M. L., Mills, C. L.. and Kyffen, -1. W., UKAEA Groll, P . , Romer, J., Roder, L., Persohn, M., and Schlosser, B., Anal. Chirn. Acta, 1986, 190, 265. Boisde, G.. Blanc, F., and Perez, J.-J., Talanta, 1988. 35, 75. Wolfbeis, 0. S . , Schaffar, B. P. H., and Chalmers. R. A , , Tulanta, 1986, 33. 867.LT-2. 234. RCpo1-t ND-R- 1266(D), 1986. 18. 19. 20. 21. 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. Kimball-Linnc, M. A., Kychakoft, G., and Hanson, R. K., Combust. Sci. Technol., 1986, 50, 307. Klainer, S. M., Hirschfeld, T., Bowman, H., Milanovich, F.. Perry, D., and Johnson, D . , “A Monitor for Detecting Nuclear Waste Leakage in a Subsurface Repository,” Lawrence Ber- keley Laboratory Report LBL-11981, California, 1981. Kittrcl. C., Willett, R. L., dc 10s Santos-Pancheo, C., Ratcliff, N. B., Kramer, J. R . , Malk, f3. G., and Fcld, M. S . , Appl Opt.. 1985, 24, 2280. Hendra, P. J., Ellk. G . , and Cutler, D. J.. J . Ramun Specirosc., 1988, 19, 413.Gomy, C . , Jouan, M., and Uuy Dao, N . , C . K . Seances Acud Sci., Ser. 2, 1988,306, 417. Lewis. E. N., Kalasinsky, V. F., and Levin, I. W.. Anal. Chem.. 1988, 60, 2658. Knight, J., Control Instrum., 1986, March, 41. Narayanaswamy, R., and Russell, D. A., Sem. Actuators, 1988, 13, 293. Milanovich, F. P., Hirschfeld, T. B., Wang, F. T., Klaincr, S. M.. and Walt, D., Proc. SPIE-lnt. SOC. Opt. Eng., 1984, 494, 18. Elf UK, Eur. Pat.. 126 600, 1984. Kirkbright. G. F., Narayanaswamy, R . , and Welti, N. A . . Analyst, 1984, 109, 1025. Narayanaswamy, R., per’jonal communication. Kirkbright. G. F., Narayanaswamy, R., and Welti, N. A.. Anulyst, 1984, 109, 15. Jones, T. P., and Porter, M. D.. Anal. Chem., 1988, 60, 404. Bcnaim, N.. Grattan, K. T. V., and Palmer. A. W., Analyst, 1986, 111.1095. Arnold, M. A.. Anal. Chem., 1985, 57. 565. Woods, B. A., R&Zii-ka, J.. Chrktian, G . D., Rose, N. J . , and Charlson, R . J.. Analyst. 1988. 113, 301. Peterson, J. I . , Goldstcin, S. R., Fitzgerald, R. V., and Buckhold, D. K., Anal. Chem., 1980, 52, 864. Saan, L. A., and Seitz, W. R . , Anal. Chern., 1982, 54, 821. Offenbacher, H., Wolfbcis, 0. S . , and Furlingcr, E . , Senc.. Actuators, 1986, 9, 73. Wolfbeis, 0. S . , Schaffar, B. P. H . , and Kaschnitz, E., A n a l y t , 1986, 111, 1331. Fuh, M.-R. S., Burgess. L. W , Hirschfeld, T. B., Christian, G. D., and Wang, F. T., Analyst, 1987. 112, 1159. Wolfbeis. 0. S . , and Marhold, H., Frewnius Z. Anal. Chem., 1987, 327, 347. Opitz. N . , and Lubbers, D. W., Sen\. Actuarous, 1983, 4, 473.Woltbcis, 0. S . , and Offenbacher. H., Sens. Actuators, 1986.9, 85. Peterson. J . I . , Fitzgerald, R. V., and Buckhold, D. K., Anal. Chem., 1984, 56, 62. Opitz, N., Graf, H.-J.. and Lubers, D. W., sen^. Actuators. 1988. 13, 159. Lippitsch, M. E . , Pusterhofer, J.. Leiner. M. J . P., and Wolfbeis, 0. S . , Anal. Chim. Acfa, 1988, 205, 1. Petrea. R. D . , Sepaniak, M. J.. and Vo-Dinh, T., Talantu, 1988, 35, 139. Hirschfeld, T. B . . Miller, F. S.. Thomas, S . N. B., Miller. H. H.. Milanovich, F. P., and Gavcr, R. W., .I. Lightwuve Technol., 1987, LT-5, 1027. Gehrich. J. L., Lubbers, D. W., Opiti. N., Hansmann, D. R., Miller, W. W., Tusa, J . K., and Yafuso, M., IEEE Trans. Biomed. Eng., 1986, 33, 117. Wolfbeis, 0. S . , and Posch, H. E., Anal. Chim. Acta, 1986. 185, 321. Wangsa, J.. and Arnold, M. A., Anal. Chem., 1988,60, 1080. Walters, B. S . , Nielsen, T. J., and Arnold, M. A , , Talanta, 1988, 35, 151. Wolfbeis, 0. S . , and Sharma, A., Anal. Chim. Acta, 1988,208, 53. Wolfbeis, 0. S . , and Schaffer, B. P. H . , Anal. Chim. Acfa, 1987. 198, 1. Lui, B. L., and Schultz. J. S., 16EE Trany. Riomed. Eng., 1986, 33, 133. Srinivasan, K. R., Manwuri, S . . and Schultz, J. S . , Biotechnol. Bioeng., 1986, 28, 233. Schultz, J . S., U.S. Pat., 4344438. 1982. Warner, M. D., A n d . Chem., 1986.58, 766A. Meadows, D . , and SchultL. J . S . . Tulantu, 1988, 35, 145. Freeman, T. M., and Seitz, W. R., Anul. Chem.. 1978, 50, 1242.1372 ANALYST, NOVEMBER 1989, VOL. 114 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. Hardy, E. E . , David, D. J . , Kapany, N. S . , and Unterleitner, F. C., Nature (London), 1975, 257, 666. Giuliani, J . F., J. Opt. Sens., 1987, 2, 11. Kapany, N. S . , “Fiber Optics,” Academic Press, New York, 1967. Stewart, G., Norris, J. 0. W., Clark, D. F., Tribble, M. J . , Andonovic, I., and Culshaw, B., Proc. SPIE-Int. Soc. Opt. Eng., 1988, 990, 188. Place, J . F., Sutherland. R. M., and Dahne, C., Biosensors, 1985, 1, 321. Andrade, J. A . , Vanwagenen, R. A., Gregonis, D. E., Newby, K., and Lin, J.-N.. ZEEE Electron Devices, 1985, 32, 1175. Monsanto, Rrir. Put., 1530997, 1978. Giuliani, J . F., Wohltjen, H., and Jarvis, N. L., Opt. Lett., 1983, 8, 54. Ballantine, D. S., and Wohltjen, H., Anal. Chem., 1986, 58, 2883. Tai, H . , Tanaka, H . , and Yoshino, T., Opt. Lett., 1987, 12, 437. Paul, P. H., and Kychakoff, G . , Appl. Phys. Lett., 1987,51,12. DeGrandprc, IM. D.. and Burgess, L. W., Anal. Chem., 1988, 60, 2582. Ogawa, K., Tsuchiya, S . , Kawakami, H., and Tsutsui, T., Electron. Lett., 1988, 24, 43. Bucklcs, R. G., U.S. Put., 4321 057, 1982. Buckles, R. G . , WO Put., 81/00912, 1981. Battelle Memorial Institute, Eur. Pat., 075 353, 1983. Battelle Memorial Institute, WO Put., 86/03004, 1986. Sutherland, R. M., Dahne, C.. and Place, J. F., Anal. Lett.. 1984, 17(B), 43. Suthcrland, R. M., Dahne, C., Place, J . F., and Ringrose, A. S . , Clin. Chem., 1984. 30, 1533. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. Block, M. J . , Eur. Put., 103426, 1984; Block, M. J . , and Hirschfeld, T. B., Brit. Pat.. 2 180338A, 1986. Unilever plc, WO Put., 86/00135. 1986. Nishizawa, K., Sudo, E., Yoshida, M., and Yamasaki, T., “Proceedings of the 4th International Conference on Optical Fibre Sensors,” Tokyo, October 1986, p. 131. Hochiki Kabushiki Kaisha, Brit. Put., 2 144 849, 1985. Battelle Memorial Institutc, U. S. Put., 4240747, 1981. Tiefenthaler, K., WO Put., 86/07249, 1986. Scifert, M., Tiefcnthaler, K . , Heuberger. K . , Lukosz, W., and Mosbach, K., Anal. Lett., 1986, 19, 205. Smela. E . , and Santiago-Aviles, J. J . , Sens. Actuators, 1988, 13, 117. Attridge, J . W., Leaver, K. D., and Cozens, J . R., J. Phys. E . , 1987, 20, 548. Nylander, C., Liedberg, B., and Lind, T., Sens. Actuutors, 1982, 3, 79. Liedberg, B., Nylandcr, C., and Lundstrom, I . , Sens. Actuu- tors, 1983, 4, 299. Kreuwal, H. J. M., Popma, Th. J . A., Lambeck, P. V., and Gerritsma, G . J . , “Proceedings of the 2nd International Meeting on Chemical Sensors,” Bordeaux, 1986, p. 7. Butler, M. A., Appl. Phys. Lett., 1984, 45, 1007. UKAEA, Brit. Put. Appl., 2192710, 1988. Farahi, F., Akhavan, P., Jones, J . D. C., and Jackson, D. A . , J . Phys. E, 1987, 20, 435. Pinchbeck, D., Trans. Inst. Meus. Control (London) 1986, 19, 46. Paper 8103772H Received September 26th, 1988 Accepted April 26th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401359

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989, VOL. 114 1373 High-performance Liquid Chromatographic Method for the Determination of 9-(4-Hydroxy-3-hydroxymethylbut-l-yl)guanine (BRL-39123) in Human Plasma and Urine Susan E. Fowles and David M. Pierce* Drug Metabolism and Pharmacokinetics Department, Beecham Pharmaceuticals, Medicinal Research Centre, Harlow, Essex CM19 5AD, UK A rapid, sensitive and reliable reversed-phase high-performance liquid chromatographic (HPLC) method with UV detection has been developed for the assay of a novel anti-herpes agent, 9-(4-hydroxy-3- hydroxymethylbut-I-y1)guanine (BRL-39123), in human plasma and urine. The drug and the internal standard, the structural analogue BRL-42377, were extracted from the biological matrix by adsorption on a cation-exchange column and were subsequently eluted under alkaline conditions prior t o HPLC.The method is reproducible, with coefficients of variation of ca. 5%, and linear from 0.1 t o at least 30 yg ml-1 in plasma and from 50 t o at least 2000 pg ml-1 in urine. The method has been used extensively t o measure BRL-39123 in plasma and urine samples generated during clinical studies and is adequate for defining pharmacokinetics at projected therapeutic doses. Keywords: 9-(#-Hydroxy-3-h ydroxymeth ylbut- I-y1)guanine (BRL-39 123); anti-herpes agent; high- performance liquid chromatography; solid-phase extraction The novel antiviral drug 9-(4-hydroxy-3-hydroxymethylbut-l- y1)guanine (BRL-39123; I) has been shown to be a potent inhibitor of the herpes simplex virus (HSV types 1 and 2) in vitro. It has also shown activity against HSV in mice following vari6us routes of administration.1~ In order to carry out pharmacokinetic studies to evaluate BRL-39123 in man, a specific and adequately sensitive method for the determination of BRL-39123 in human plasma and urine was required. This paper describes a high-perfor- mance liquid chromatographic (HPLC) method that has been developed for this purpose. The method uses a structural analogue of BRL-39123 as an internal standard (BRL-42377; 11) and employs UV detection. 0 CH2CH2CH I CH20H I BRL39123 0 CH~CH I CHzCH20H ll BRL42377 Experimental Apparatus The HPLC system consisted of a reciprocating pump (Model 510), a sample injector (WISP 710R), a fixed wavelength * To whom correspondence should be addressed.detector (Model 441) [all obtained from Waters Associates (Harrow, Middlesex, UK)] and a dual pen recorder (Model PM 8222, Philips, Cambridge, UK). Quantification was carried out by measuring peak-height ratios using a computer data-capture system (Multichrome, VG Laboratory Systems, Altrincham, Cheshire, UK). Reagents Methanol (HPLC grade) was obtained from Rathburn Chem- icals (Walkerburn, Peeblesshire, UK), sodium dihydrogen orthophosphate ( AnalaR grade) was supplied by Fisons (Loughborough, Leicestershire, UK) and potassium di- hydrogen orthophosphate (AnalaR grade) and trichloroacetic acid (AnalaR grade) were obtained from BDH (Poole, Dorset, UK). Both RRL-39123 and the internal standard, BRL-42377, were obtained from the Chemotherapeutic Research Centre, Beecham Pharmaceuticals (Betchworth, Rrockham Park, Surrey, UK).Cation-exchange columns (“Bond-Elut” SCX, 1 ml) were obtained from Jones Chromat- ography (Hengoed, Mid-Glamorgan, UK). Sample Storage Plasma and urine samples containing BRL-39123 were stored at -70 “C when not in use. In addition, quality control samples were prepared, stored and analysed with the unknown samples to monitor the Performance of individual analyses. Solid-phase Extraction Procedure Aliquots of plasma (0.5 ml) were placed in 7-ml glass vials and spiked with the internal standard (50 p1 of an approximately 100 pg ml-1 aqueous solution) before the addition of 0.5 ml of 16% mlV trichloroacetic acid to precipitate proteins and inactivate any viruses present. Calibration standards [pre- pared by spiking control human plasma (0.5 ml) with aliquots of a standard solution of BRL-391231 were prepared by the same method.Following centrifugation (3000 rev min-I for 5 min in an MSE Centaur centrifuge, with a 12-cm rotor arm), the supernatants obtained were transferred into 3-ml vials containing 0.5 ml of 0.001 M sodium dihydrogen phosphate buffer (pH 7). Meanwhile, cation-exchange columns (SCX,1374 4- 5 , Q c 0 % 0 - ANALYSl 1 ’ 23 NOVEMBER 1 1 1 1 1 1989, VOL. 114 1 ml) were each primed with methanol and de-ionised water (1 ml each) before the application of 1 ml of 0.001 M sodium dihydrogen phosphate buffer (pH 7). The buffered samples were then loaded on to the columns and the columns washed with further 1-ml portions of 0.001 M sodium dihydrogen phosphate buffer (pH 7) before, in each instance, the analytes were eluted with 0.25 M potassium dihydrogen phosphate buffer (pH 11) - methanol (4 + 1; 0.5 ml).The eluate was transferred into tapered vials and subjected to chromato- Aliquots of urine (20 pl) were also treated with 20 p1 of 16% mlV trichloroacetic acid to inactivate any pathogenic organ- isms. Calibration standards [prepared by spiking control human urine (20 pl) with aliquots of a standard solution of BRL-391231 were prepared by the same method. After the volume had been made up to 0.5 ml with de-ionised water, the internal standard (50 pl of an approximately 400 vg ml-1 aqueous solution) and 0.5 ml of 0.001 M sodium dihydrogen orthophosphate buffer (pH 7) were added. The analytes were then extracted from the buffered urine sample by the same method used for the plasma samples, except that the final elution from the cation-exchange column was achieved with 1 ml of potassium dihydrogen phosphate buffer (pH 11) - methanol (4 + 1).graphy. Chromatography Aliquots of the plasma extracts (20-50 pl) were injected on to a Spherisorb ODS 2 ( 5 pm) reversed-phase analytical column (25 cm X 4.6 mm i.d.; Jones Chromatography) and eluted with 0.01 M sodium dihydrogen orthophosphate buffer (pH 7) - methanol (9 + 1) at a flow-rate of 1 ml min-1. Similarly, portions (15-50 PI) of the urine extracts were injected on to an Apex ODS 2 (3 pm) analytical column (15 cm x 4.6 mm i.d.; Jones Chromatography) and eluted with 0.01 M sodium dihydrogen orthosphosphate buffer (pH 7) - methanol (47 + 2) at a flow-rate of 1 ml min-1.Both mobile phases were filtered and de-gassed under vacuum before use. In order to extend the useful life of the analytical columns, a “Newguard” RP-18 (7 pm) guard column (15 X 3.2 mm i.d.; Anachem, Luton, Bedfordshire, UK) was inserted on-line between the injector and the analytical column. Both BRL-39123 and the internal standard were detected by UV absorption at 2.54 nm. Human plasma and urine samples extracted using this method were also submitted €or analysis by HPLC - mass spectrometry to confirm the peak homogeneity of BRL-39123 and BRL-42377. Results and Discussion The strong cation-exchange material employed in the solid- phase extraction was chosen to achieve maximum selectivity for the adsorption of BRL-39123 and the internal standard from plasma and urine.The additional selectivity required for the separation of the analytes from endogenous compounds in the urine extracts was provided by employing a different column (Apex ODS 2) from that used for the chromatography of the plasma extracts (Spherisorb ODS 2). However, for the same reason the Apex ODS 2 column was unsuitable for the chromatography of plasma extracts. Using the method described, the retention times of BRL- 39123 and the internal standard in human plasma were usually 9 and 10.5 min, and in urine 6.2 and 7.8 min, respectively. Typical chromatograms obtained for plasma and urine are shown in Fig. 1. At the defined lowest detection setting (0.01 a.u.f.s.), no interference on BRL-39123 or the internal standard was observed in either the blank plasma or urine extracts.Calibration graphs (usually consisting of a minimum of six standards) for BRL-39123 in human plasma and urine were linear from 0.1 to at least 30 pg ml-1, and from SO to at T a, C 0 n a, [r 0 2 4 6 8 1 0 ( b ) 1 4 6 8 1 0 5 2 4 6 8 0 2 4 6 8- Time/mi n Fig. 1. Chromatograms of extracts of ( a ) blank human plasma, ( h ) human plasma containing BRL-39123 (2 pg ml-l) and internal standard (1 pg ml-I), ( c ) blank human urine and (d) human urine containing BKL-39123 (306.6 pg ml-’) and intcrnal standard (150 pg ml-l). Chromatographic conditions as in text. 1, Solvent front; 2, BRL-39123; and 3, internal standard Table 1. Accuracy of the quantification of three nominal concentra- tions of BKL-39123 in human plasma and urine Nominal Mcan measured concentration/ concentration Biological tluid pg ml-1 (k SD)*/pg ml-’ Accuracy,% Plasma .. . . 3.06 3.07 (20.03) +0.33 7.13 6.79 (k0.21) -4.8 12.2 12.15 (k0.17) -0.41 Urinc . . . . 52.8 49.2 (k0.9) -6.8 1056 1022 ( t 7 0 ) -3.2 528 533 (k3.6) -0.9s * n = 5 . least 2000 pg ml-1, respcctively, with relative standard deviations about the regression line generally < 10% . The reproducibility of the method was investigated by analysing, on five consecutive days, five replicate samples of human plasma and urine, spiked at each of three different concentrations of BRL-39123. In plasma, the mean within-day coefficients of variation for peak-height ratios at concentra- tions of 3.1, 7.1 and 12.2 pg ml-1 were 3.0, 2.6 and 3.6%, respecitvely.Similarly, at concentrations of 52.8,528 and 1056 pg ml-1 in urine, the mean within-day coefficients of variation were 4.0, 3.3 and 6.1%, respectively. Between-day coeffi- cients of variation for peak-height ratios in human plasma and urine extracts were <7%. The accuracy of the quantification was examined by spiking five aliquots of plasma and urine at three different nominal concentrations of BRL-39123 and evaluating the difference between the nominal and mean measured concentrations obtained using this analytical method. The results of these analyses are shown in Table 1 and demonstrate accuracy to within 7%. The peak ascribed to BRL-39123 in samples of human plasma and urine obtained after administration of BRL-39123 to humans was confirmed as containing the ( M + H)+ ion corresponding to BRL-39123.Similarly, the internal standard peak was shown to be homogeneous by the same method. The control plasma and urine showed no evidence of a peak or relevant ion at the retention time of the BRL-39123 peak, or that of the. internal standard, BRL-42377. The method described is rapid, allowing a sample through- put of up to 70 samples per day. It has been shown to be validANALYST. NOVEMBER 1989, VOL. 114 1375 " 1 2 3 4 5 6 7 8 Time after dosinglh Fig. 2. Plasma concentration versus time profile for BRL-39123 in a healthy volunteer given a single 60-min constant rate intravenous infusion of 10 mg kg-1 of BRL39123A (sodium salt) for the parent drug in plasma and has proved sufficiently sensitive for monitoring plasma levels after doses in the predicted therapeutic range.On this basis, the lower limit of sensitivity in plasma (0.1 pg ml-1) would permit the plasma concentration - time profile to be monitored for approxi- mately 8 h (i.e., at least three half-lives). A typical plasma concentration - time graph, obtained using this method, for a healthy subject administered a single 60-min intravenous infusion of 10 mg kg-1 of BRL-39123A (sodium salt) is shown in Fig. 2. As can be seen, the maximum concentration observed 1 h after dosing was 16.6 pg ml-1 and concentrations fell to approximately 1 pg ml-1 8 h after dosing. Urinary concentration data for the same healthy subject showed that ca. 72% of the administered BRL-39123 was excreted unchanged up to 72 h after dosing. No metabolites of BRL-39123 are known at present and hence 30% of the administered dose remains to be accounted for. No drug- related products of BRL-39123, other than the parent compound, are extracted or detected by this method. This robust HPLC method is precise with coefficients of variation of ca. 5% and is accurate to within 7% of the nominal concentration. In routine use it has adequate sensitivity for the assay of samples from clinical studies. The authors thank Dr. D. Staniforth and Miss R. Corbett for providing the human plasma and urine samples, Miss J. McMeekin for excellent technical assistance, Dr. R. A. Vere Hodge, Mr. C. Winton and Mr. B. Zussman for helpful discussions and Mr. G. D. Allen and Mr. M. Nash for HPLC - mass spectrometry support. References 1. Boyd, M. R.. Bacon, T. H., Sutton. D., and Cole, M., Antimicrob. Agents Chemother., 1987, 31, 1238. 2. Boyd, M. R., Bacon, T. H., and Sutton, D., Antimicrob. Agents Chemother., 1988, 32, 358. Paper 910021 7K Received January 1 Oth, 1989 Accepted June 9th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401373

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989, VOL. 114 1377 Analysis of Basic Drugs by High-performance Liquid Chromatography Using On-line Solid-phase Extraction and an Unmodified Silica Column Mary T. Kelly and Malcolm R. Smyth* School of Chemical Sciences, Dublin City University, Glasnevin, Dublin 9, Ireland Darioush Dadgar IWF Research Laboratories Inc., 6075 Kestrel Road, Mississauga, Ontario L5T I Y8, Canada A method is described for the high-performance liquid chromatographic analysis of a number of basic drugs on unmodified silica gel, which employs an aqueous methanol-rich eluent of pH 8. The analytical system was coupled to a column-switching assembly in order to facilitate on-line solid-phase extraction of the drugs from plasma. The method is reproducible, with a coefficient of variation of 3% or less for most compounds, even without internal standardisation. Drug recovery from plasma is efficient, being in excess of 90% in most instances. The drugs were detected by UV absorption at 254 nm.Keywords: Basic drug; high-performance liquid chromatography; unmodified silica; plasma analysis; on-line solid-p h ase extraction The use of bare (unmodified) silica columns for the separation of basic analytes was first reported by Jane' in 1975, and has since gained steadily in popularity owing to its ability to produce greater column efficiencies and improved peak shapes over the more rugged bonded-phase materials. The problem of poor chromatographic characteristics for basic compounds on hydrocarbon-bonded stationary phases (par- ticularly CIS) has been reported extensively.This type of chromatographic behaviour has been shown to be due to the basic amino function interacting with residual silanol moieties on the modified silica surface.2.3 As a result, much of the work in this area has focused on the elimination of these unreacted silanols through the use of secondary bonding reactions, i.e., end-capping. Other methods employed to circumvent this problem include ion-pair chromatography, ion suppression4 or the use of organic ammonium compounds at high pH.596 By employing an unmodified silica column, however, the amine - silanol interaction can be exploited, and efficient separations for basic compounds have been obtained using the type of eluent more commonly associated with reversed-phase chromatography, i.e., aqueous methanolic solvents containing a buffering salt. The mechanisms involved in these polar separations are complex and multi-functional, but are known to consist, at least in part, of ion-exchange interactions.7-9 The silanol groups are weakly acidic and are only ionised at neutral or akaline pH*; therefore, at high pH they are available to partake in ionic reactions with oppositely charged protonated bases. Recently, Cox and Stout9 addressed a number of questions with respect to the retention mechanisms on modified and unmodified silica. It was known that interactions other than simple ion exchange were taking place, and they showed that for compounds which are ionised under the chromatographic conditions employed, ion exchange was the main mechanism of retention, and for non-ionised species hydrophobic interactions dominated.Separate hydrophobic interactions were observed on alkyl-bonded and unbonded silica, and evidence suggests that the latter involve siloxane bridges rather than silanol groups. The influence of pH, ionic strength, nature of the cation and origin of the silica has been widely st~died.4~7-12 The aim of this work was to develop a chromatographic system using an unmodified silica column, on which a number of basic drugs could be separated and determined. It was also decided * To whom correspondence should be addressed. to couple the chromatographic separation to a column-switch- ing assembly in order to facilitate on-line solid-phase extrac- tion of the drugs from plasma. This expedites the extraction process considerably and would have particular application in a laboratory where blood analyses were performed routinely.Schmid and Wolf7 have previously carried out on-line solid-phase extraction of some tricyclic drugs and tamoxifen in plasma. They used a C2-concentration column in place of the injection port and effected manual clean-up of the sample by injecting plasma followed by water and then a 10% methanolic solution. In the present study, which employs a dual pump column-switching procedure, the concentration column is mounted independently of the injector and the second pump delivers the washing eluent (water). Switching between the two columns is achieved using a six-port switching valve and hence the concentration column can be equilibrated with water while separation of the previous sample is completed on the analytical column.In contrast to the single-pump system employed by Schmid and Wolf,7 this arrangement permits injection of the next sample immediately after (or, in some instances, before) elution of the last peak, with a consequent saving in labour and analysis time. Experimental Reagents and Solvents The drugs used were received as a gift from the Institute of Clinical Pharmacology, Dublin, Ireland. Ammonium nitrate (analytical-reagent grade) was obtained from BDH (Poole, Dorset, UK) and analytical-reagent grade ammonia solution (25%) from Riedel de Haen (Seelze, Hannover, FRG). High-performance liquid chromatography (HPLC) grade methanol was supplied by Labscan Analytical Sciences (Dublin , Ireland).De-ionised water was obtained by passing freshly distilled water through a Millipore Milli-Q water purification system. Dried human plasma from the Blood Transfusion Board, Dublin, was dissolved in de-ionised water and used within 7 d of reconstitution. Drug Standards Stock solutions equivalent to 1 mg ml-1 of the drugs were prepared in methanol. Working standards of1378 ANALYST, NOVEMBER 1989, VOL. 114 50-10 000 ng ml-1 (depending on the detector response of the drug) were prepared in mobile phase for direct injection or in water for column-switching. Plasma Standards Aliquots of blank plasma were spiked with stock solutions to produce the required concentrations. These plasma solutions were then diluted 1 + 1 with water and 500 PI introduced into the chromatographic system through a loop.Instrumentation and Operating Conditions The drugs were separated on a LiChrosorb silica 60 ( 5 pm) column (250 X 4.6mm i.d.) supplied by HPLC Technology (Macclesfield, UK). It was protected by a Waters (Waters Associates, Milford, MA, USA) Guard Pak module, fitted with a silica insert. A pre-column was placed between the pump and the injector; it was packed with LiChrosorb silica (10pm) (Merck, Darmstadt, FRG) in order to saturate the mobile phase with silica and so prolong the life of the analytical column. Stock solutions of ammonia and ammonium nitrate (both 1 M) were mixed to produce the required pH. The resulting solutions were diluted with de-ionised water to produce the desired ionic strength. Mobile phases were made by mixing the aqueous component with methanol to produce solutions containing 80% organic phase.The mobile phase was passed through a 0.45-pm filter and delivered by a Waters Model 501 HPLC pump. The drugs were detected by ultraviolet (UV) absorption at 254 nm using an Applied Chromatography Systems (Macclesfield, UK) Model 75011 1 fixed-wavelength detector and chromatograms were recorded with a Linseis (Selb, FRG) recorder at a chart speed of 200 mm h-1. For the purposes of column-switching a second Waters Model 501 pump and the concentration column were connec- ted to the analytical assembly via a Rheodyne (Cotati, CA, US.4) Model 7000 six-port switching valve. The 10 X 1.5 mm i .d. concentration columns were dry-packed with either Corasil (Waters Associates) RP-18 (37-50 pm), Sepralyte (Analytichem International, Harbour City, CA, USA) RP-8 (40 pm) or Supelco (Bellefonte, PA, USA) cyano pellicular packing.Pump A eluent was de-ionised water filtered through a 0.45-pm membrane and de-gassed under vacuum. Column-switching Procedure The instrument arrangement (incorporating a six-port switch- ing valve) is shown in Fig. 1. The spiked plasma sample is introduced via the injector port and swept on to the concentration column by the water from pump A. The drugs are retained by the concentration column while the plasma is eluted to waste. At the same time the mobile phase is passed by pump B through the analytical column, and on switching the valve the drugs are swept in a back-flush mode from the concentration column on to the analytical column where they are separated.Results and Discussion Development of Chromatography In order to obtain reproducible chromatography, it is neces- sary that the pH and ionic conditions at the silica surface be defined, stable and in equilibrium with those of the eluent.7 For this purpose the aqueous component should contain buffering salts that do not precipitate in methanol-rich solvents and which preferably do not tend to dissolve the silica packing. Phosphate salts would be unsuitable as they have low solubility in organic solvents and ammonium salts are prefer- able to sodium salts as their solutions show a reduced tendency to dissolve silica. Examination of the relevant literature Concentration column 7 Injector m 4- Pump A H20 - Drain m To detector Fig.1. ing valve Column-switching assembly incorporating a six-port switch- Table 1. Effect of pH on retention. Mobile phase: 0.025 M ammonium nitratehmmonia solution - methanol (20 + 80) Retention time/min Drug PK, pH5 pH7 Benzylsulphamidc . . NA* Chlorpheniramine . . 8.9 Clcmastine . . . . NA Codcinc . , . . . . 8.2 Dextromethorphan . . 8.3 Diltiazem . . . . NA Fluphenazine . . . . 8.1 Lignocaine . . . . 7.9 Mcpivacaine . . . . 7.7 Perphenazine . . . . 7.8 Phentoamine . . . . 9.1 Protriptyline . . . . 10.0 Trimethoprim . . 7.2 Tripelennamine . . 9.0 Vcrapamil . . . . NA 4.2 8.4 4.2 7.5 5.7 5.7 5.0 5.1 5.4 5.1 5.7 3.6 4.2 7.2 4.8 4.5 8.7 5. I 9.6 7.8 5.7 5.1 4.8 5.1 5.4 6.0 4.2 4.2 7.2 4.8 * NA = Not available in standard referencc texts.PH8 6.9 19.7 8.4 11.7 9.8 6.3 5.4 4.8 5.1 5.7 11.2 9.8 4.4 11.5 6.3 revealed that even within the constraints of these criteria, a number of options were available, ranging from ammonium nitrate as originally used by Jane,' to ammonium formate11 or ammonium perchlorate.8 The chosen starting point in this study was 0.005 M ammonium nitrate/ammonia solution, pH 10 - methanol (20 + SO). Under these conditions, the tricyclic drug protriptyline eluted with a retention time greater than 60 min. It was found that by increasing the ionic strength to 0.025 M NH4+, the retention time was reduced to 29.1 min, an observation that is consistent with the ion-exchange theory of retention. Lowering the pH to 9.5 caused the drug to elute earlier (i.e., at 21.0 min), which may be explained in terms of reduced silanol ionisation on the column surface; the degree of dissociation of this strongly basic amine (pK, = 10.0) will be enhanced in media of low basicity.The other less basic compounds studied were found to elute too early to permit their eventual quantification in plasma. Hence, retention times were compared at the lower eluent pH values of 5,6 and 8. As different relative amounts of ammonium nitrate and ammonia solution would be required to generate these different pH values, the nitrate concentration varied from one solution to the next. The cation (NH4+) concentration was, however, kept constant, and as the proposed retention mechanism is one of cation exchange, it was assumed that a variation in the nitrate concentration would not seriously affect changes resulting from an alteration in pH.The results of this experiment, and the chromatographic conditions, are presented in Table 1; they show that most of the drugs, except for mepivacaine and lignocaine, exhibit the greatest affinity for the column at pH 8, which was eventually selected as theANALYST, NOVEMBER 1989, VOL. 3 14 1379 Table 2. Retention times, coefficients of variation and recoveries of drugs extracted from plasma. Mobile phase: 0.05 M ammonium nitratdammonia solution, pH 8.0 - methanol (20 + 80) Con- Mean centration/ retention Recovery, Drug ng ml-1 time/min CV. % 74" Benzylsulphamide Benzylsulphamidc Clemastine . . Clemastine . . Protriptyline . . Protriptyline . . Phentoamine . . Phentoamine . . Tripelennamine Tripelennamine .. 2500 . . 5000 . . 2500 . . 5000 . . 125 . . 500 . . 625 . . 2 so0 . , 125 . . 500 7.0 6.9 8.8 8.7 6.5 6.5 9.7 9.6 11.2 11.1 2.1 2.1 3.4 4.7 1.8 4.2 2.3 2.1 4.0 1.3 85.5 97.9 40.7 51.7 72.7 76.3 72.9 85.0 89.3 99.1 Chlorpheniramine . . 625 17.4 1.8 92.0 Chlorpheniramine . . 2500 17.3 2.3 96.5 Diltiaxm . . . . 50 7.2 3.0 92.0 Diltiazem . . . , 250 7.1 1.4 76.8 Codeine . . . . . . 2000 15.5 4.9 89.7 Codeine . . . . . . 10000 15.5 1.8 109.8 C I I I 9 15 21 3 9 15 21 Time/m i n Fig. 2. Sample chromatograms of extracted plasma standards. Mobile phase: 0.05 M ammonium nitratdammonia solution, pH 8 - mcthanol (20 + SO). PR, Protriptyline; PT, phentoamine; T, tripelcnnamine; and C, chlorpheniramine. ( u ) PR, 500; PT, 2500; T, 500: and C , 2500ngml-1. ( h ) PR, 125; PT, 625; T, 125; and C, 625 ng ml 1 operating pH.These findings can be explained by the increase in silanol ionisation with increasing pH, as most of these compounds would be at least partially dissociated under these conditions. It does appear, however, that for those com- pounds with lower pK, values, the increase in retention at higher pH is less pronounced than for the more basic amines, an observation which may be accounted for by the fact that the increase in silanol dissociation is partially offset by decreased ionisation of the less basic solutes as the pH begins to exceed their pK, values. Column-switching Procedure Having developed a suitable mobile phase for the drugs under study, the next stage was to couple the system to a column-switching assembly with a view to performing on-line solid - liquid extraction of the drugs from plasma.It was necessary to determine which type of packing in the concentration column would prove to be the most suitable in terms of drug recovery and sample clean-up. Hence, aqueous solutions of the drugs were introduced via the concentration column and back-flushed on to the analytical column. The washing solution delivered by pump A was filtered and de-gassed, de-ionised water, and eluent B was ammonium nitrate/ammonia solution, pH 8.0 - methanol (20 + 80). Three different types of packing material were investigated: Sepra- lyte C8, Corasil CIS and Supelco CN. Overall, the recovery from the C8 column was slightly better than from the CI8 column, and with a few exceptions (notably fluphenazine and perphenazine), the recovery was poor or non-existent from the CN packing material.This is due to the fact that water has a stronger eluting ability on the cyano packing material than on the C8 or CI8 materials. The difference in the recovery between the C8 and Clx packing materials is due to the greater hydrophobicity of the latter and its inability to retain the more polar analytes under the experimental conditions employed. Hence, whereas it appears that the C8 packing offers the best match for most drugs, another factor that must be considered is how well the packings behave with respect to plasma clean-up. It was found that the CI8 packing retained fewer plasma components and produced cleaner blank plasma chromatograms than the C8 packing, and for this reason, the octadecyl material was chosen for performing on-line solid- phase extractions from plasma.Analysis in Plasma As water had been established to be a suitable solvent for the wash phase, and based on previous work in this area,13,14 a wash time of 1 min (at 1 .S mi min-1) was deemed a suitable regimen for the removal of plasma while retaining the compounds of interest. It has previously been shown13 that increasing the wash time will not usually remove endogenous interferents which are retained on the concentration column under the above conditions, but that it will cause band spreading in all the retained species, including the drug peaks. The ionic strength of the aqueous component of the mobile phase was increased to 0.05 M ammonium nitratdammonia solution and the flow-rate was increased from 1.0 to 1.3 ml min- I in order to reduce the run times.These measures did not cause merging of the last plasma interferent and the first compound of interest. It was then decidcd to gain some measure of the reproduci- bility of the method with respect to each drug and to determine the percentage recoveries from the plasma matrix. Each compound was injected five times at two concentration levels in spiked plasma (ten injections) and in duplicate as authentic (aqueous) standards. The mean, standard deviation (SD) and coefficient of variation (CV) were calculated for the five spiked plasma standards at both concentrations. Recovery of the drugs from plasma was calculated by comparing the peak heights of the quintuplicate extracted standards with the peak heights of the duplicate authentic standards.Results of the plasma analyses are presented in Table 2 and sample chromatograms are shown in Fig. 2. For each compound the mean retention time and the CV (n = 5 at both concentrations of each drug) are given. Also shown is the recovery based on peak-height measurements as described above. As can be seen from the results, the CV did not exceed 5% and was frequently less than 3%. The low levels of variability, even in the absence of an internal standard, are possible because of the inherent reproducibility of the column-switch- ing technique, particularly when compared with liquid - liquid extractions which can involve many different steps and may permit the introduction of artefacts.The recovery from plasma using the column-switching technique is generally high. As the results show, the recovery is usually in excess of 70% and is frequently greater than 90%. Some drugs do,1380 ANALYST, NOVEMBER 1989, VOL. 114 however, exhibit particularly small percentage recoveries, especially at the lower concentrations. The most likely explanation for these observations is that some drugs are protein bound and would require more vigorous extraction methods such as liquid - liquid extraction or protein precipita- tion in order to liberate them from their binding sites. It is possible, of course, to subject a sample to column- switching subsequent to protein precipitation, but acid or base precipitants would first have to be neutralised and organic precipitants evaporated, as they would effect rapid elution of the drug from the concentration column.For one of the drugs (clemastine), the recovery was substantially lower than that of its sister drug, benzylsulpham- ide, to which it is closely related structurally. It was found that the recovery of this drug depended strongly on the batch of pooled blank plasma into which it was spiked. This may, in fact, be the situation for many drugs although it was beyond the scope of this study to test the recovery of all the drugs from a number of plasma sources. It does, however, illlustrate the point that in biopharmaceutical analysis, the condition of the plasma and factors affecting its composition may play a significant role in determining the amount of drug recovered, and seemingly low subject levels of a drug may be a result of the poor yield from that particular subjects’ plasma rather than inefficient bioavailability of the preparation.Conclusion It has been shown in this and other studies that a silica column with a methanol-rich buffered aqueous eluent can be applied to the separation of basic compounds of medicinal and forensic interest. A method has been developed that can accommodate the determination of a number of thera- peutically important compounds: it involves extraction of the drugs using an on-line column-switching assembly to permit rapid separation of the drugs from plasma. For the drugs studied, the method was shown to be reproducible, with a CV of 3% or less for most compounds, without the need for internal standardisation.The method offers drug recoveries that are comparable, and in many instance superior, to conventional liquid - liquid extraction methods. Apart from being less tedious and time consuming, column-switching has the advantage of being less hazardous and expensive than liquid extraction as it does not require any organic solvents. The only consumable (apart from the eluents) is the concen- tration-column packing, of which very little is required. It has been shown previouslyl3.14 that a single concentration column can accept up to 7.5 ml of plasma without significant deteriora- tion in column performance, back-pressure or peak width. In theory it should be possible to measure a much larger number of basic compounds using this technique, and with the ability to perform rapid plasma analyses, such a system has the potential to form an important component of any laboratory engaged in the routine analysis of samples originating from forensic or toxicological studies. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Janc, I., J . Chrornatogr., 1975, 111, 275. Papp, E., and Vigh, Gy., J . Chromatogr., 1983, 259, 49. Papp, E., and Vigh, Gy., J . Chromatogr., 1983, 282, 59. Bidlingmeyer, B. A., Del Rios, J. K., and Korpi, J., Anal. Chem., 1982,54, 442. Wehril. A., Hildenbrand, J. C., Keller, H. P., and Stampfli, R., J . Chromatogr., 1978, 149, 199. Kraak, J. C., and Bijster, P., J . Chromatogr., 1977, 143, 499. Schmid, R. W., and Wolf, C., Chrornatographia, 1987,24,713. Flanagan, R . J., and Jane, I., J . Chromatogr., 1985, 323, 173. Cox, G. B., and Stout, R. W., J . Chrornatogr. 1987.384,315. Smith, R. M., Hurdley. T. G.. Gill, R., and Osselton, M. D., J . Chromatogr., 1987, 398, 73. Sugden, K., Cox, G. B., and Loscombe, C. R., J . Chroma- togr., 1978, 149, 377. Law, B., Gill, R., and Moffat, A. C., J . Chromatogr., 1984, 301, 165. Dadgar, D., and Power, A., J. Chrornatogr., 1987, 416, 99. Dadgar, D., and Power, A., J . Chromatogr., 1987,421,216. Paper 8104870C Received December 12th, 1988 Accepted May 26th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401377

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989. VOL. 114 1381 Solid-phase Extraction Method for the Separation of Amphotericin B From the Lipids Dimyristoyl Phosphatidylcholine and Dimyristoyl Phosphatidylglycerol Kenneth R. Stewart, Joseph Gentile and Thomas B. Platt" Analytical Research and Development Department, Bioanalytical Section, E. R. Squibb & Sons, New Brunswick, NJ 08903, USA Kathy A. Stewart and Christine E. Swenson The Liposome Company, 7 Research Way, Princeton, NJ 08540, USA A solid-phase extraction method is described for separating amphotericin B from the lipids used in a liposomal amphotericin B formulation. The drug was bound on to a silica sorbent extraction column and the lipids dimyristoyl phosphatidylcholine and dimyristoyl phosphatidylglycerol were removed by using a solvent consisting of chloroform, methanol and ammonia solution.The lipid-free amphotericin B was eluted from the column by using a slightly acidic methanol solution. Drug recoveries averaged 95% as measured by microbiological assay and high-performance liquid chromatography. The precision of the method was demonstrated by a coefficient of variation of 1.38-1.65%. Keywords: Lipid removal; solid-phase extraction; liposomal amphoptericin B; microbiological assay; lipid-free amphotericin B Amphotericin B is the drug of choice for the treatment of many systemic fungal infections and is becoming increasingly important for use in immunocompromised individuals, such as cancer or acquired immunodeficiency syndrome patients. ,2 Much effort is currently being focused on mitigating the toxicity associated with amphotericin B.One approach is to incorporate the drug in a phospholipid (liposome) matrix. This lipid formulation has been shown to be effective in combating fungal infections in animal models and appears to decrease the toxic side effects of the drug.3.4 In humans, one formulation of liposomal amphotericin B has been used for compassionate therapy in 12 patients whose systemic fungal infections were not responding to conventional antifungal medications, including the deoxycholate preparation of amphotericin B (Fungizone). Three patients had complete remission, five had partial remission and four showed no improvement .* The phospholipid complex currently undcr development employs the lipids dimyristoyl phosphatidylcholine (DMPC) and dimyristoyl phosphatidylglycerol (DMPG).The formula- tion contains approximately 5.0 mg ml-1 of amphotericin B, Fig. 1, and 6.5 mg ml-1 of lipid (7 mol of DMPC + 3 mol of DMPG) in a 0.9% saline vehicle. When the amphotericin B potency was measured by a microbiological assay, unusually low values were obtained; we speculated that the lipids probably interfered with the assay and that for a successful microbiological assay of this formulation, complete lipid removal from the drug was required. For lipid removal, a conventional lipid extraction procedure (using hexane) was tried. This extraction method failed to remove the DMPC and DMPG lipids; hence, a different method was required. Thin-layer chromatography (TLC) experiments were per- formed to assess the effectiveness of different methods for lipid removal.These experiments showed significant differ- ences in migration behaviour between the two lipids and the drug when silica-gel plates were developed with a solvent consisting of chloroform - methanol - acetone - distilled water - ammonia solution (65 + 30 + 5 + 2 + 2). DMPC had an RF of 0.34 and DMPG an RF of 0.67, while amphotericin B failed to move from the point of application. These findings suggested * To whom correspondence should be addressed. that there were significant differences in polarity between the lipids and the drug and that these differences might allow the separation of the components in a silica-based extraction system. This paper describes the separation of the lipids from amphotericin €3 on an unbonded silica column.Experimental Solid-phase Extraction Materials Silica sorbent extraction columns (1 cm3) and accessories for solid-phase extraction were purchased from Analytichem International (Harbor City, CA, USA). Accessories from the same source used for extraction included a vacuum processing station, column adaptors, reservoirs and a calibrated flask rack. Reagents All reagents were purchased from Fisher Scientific (Spring- field, NJ, USA). The methanol was of high-performance liquid chromatography (HPLC) grade; the chloroform was certified ACS grade and the acetic acid and ammonia solution were of analytical-reagent grade. The solutions listed below were prepared fresh daily. Alkaline methanol solution. A 2.0-ml aliquot of ammonia solution was added to 98 ml of methanol. Acidic methanol solution.A 1.0-ml aliquot of 0.1 M acetic acid was added to 99 ml of methanol. Column pre-treatment solution. Prepared by mixing 20 ml of alkaline methanol with 80 ml of chloroform. Lipid removal solution. Prepared by first adding 18 ml of methanol to 80 ml of chloroform and mixing the resulting solution thoroughly. Then, to this mixture, 2.0 ml of ammonia solution were added. Thc final solution was mixed well. Sample Dilution One millilitre of the thoroughly mixed amphotericin B - lipid preparation containing approximately 5.0 mg ml-1 of amphotericin B was placed in a 100-ml calibrated flask. Nineteen millilitres of alkaline methanol were added and the resulting mixture was agitated by gentle swirling. Chloroform was then added to yield 100 ml of diluted sample solution.1382 ANALYST, NOVEMBER 1989, VOL.114 OH OH Fig. 1. Structure of amphotericin B Standard Preparation To prepare the lipid-free amphotericin B standard, a dried (60 "C for 3 h under vacuum) sample of amphotericin B (without deoxycholate) was sonicated for 1 min in 0.9% saline to produce a homogeneous suspension containing 5.0 mg ml- 1 of amphotericin B. A 1-ml sample of this suspension was then dissolved and diluted in alkaline methanol - chloroform in the same manner as for the amphotericin B - lipid complex sample. Extraction Procedure Approximately 2 ml of the column pre-treatment solution were added to the silica sorbent extraction column and about 1.5 ml were drawn through the column by applying a mild vacuum, leaving a pool of the solution above the sorbent surface.The column must not be allowed to dry. The column was then fitted with an adaptor and a 25-ml reservoir. Sample loading With the pre-treatment solution in place, exactly 5.0 ml of the diluted sample containing a total of approximately 250 pg of amphotericin B were placed in the reservoir. A mild vacuum (the flow-rate was approximately 1.0 ml min-1) was then applied to pass the solutions through the column and facilitate binding of the amphotericin B to the silica sorbent. During this time a narrow yellow band was visible in the upper portion of the sorbent column. Lipid removal The lipids were removed from the amphotericin B by a thorough rinsing of the column with the lipid removal solution.Approximately 25 ml of the lipid removal solution were placed in the column reservoir and a moderate vacuum was applied to pass all of the solution through the column at a flow-rate of approximately 2 ml min-1. Recovery of amphotericin R Amphotericin B was eluted from the column with 4.5 ml of acidic methanol solution at a flow-rate of approximately 2mlmin-1 and was collected in a 5.0-ml calibrated flask. A moderate vacuum (the flow-rate was approximately 2.0 ml min- 1) was used for drug recovery. Sufficient methanol (unmodified) was then added to bring the contents of the flask to 5.0 ml with a theoretical amphotericin B concentration of 50 pg ml-1. Microbiological and Chemical Assays The United States Pharmacopeia (USP) small-plate agar diffusion microbiological assay for amphotericin B5 was employed.For this assay, 2.0 ml of lipid-free amphotericin B, obtained by the procedure described above, were diluted with the assay buffer specified [supplemented with 5.0% dimethyl sulphoxide (DMSO)] to 50 ml such that the concentration of amphotericin €3 was theoretically 2 pgml-1. As the final antibiotic test solution contained 5.0% DMSO and 4.0% methanol, amphotericin B standards for microbiological assay were diluted with buffer of the same composition. This modified buffer had no adverse effect on the microbiological assay. For the HPLC assay of amphotericin B, the drug was diluted with methanol and determined by the method of Joseph.6 Lipids were measured by a phosphate assay.7 Results and Discussion To determine the fate of the amphotericin B and lipids during the extraction procedure, column effluents were tested for the presence or absence of these components. The results of this evaluation are shown in Table 1.Analysis of the column effluents for phosphate during processing of an amphotericin B - lipid sample indicated that approximately 23% of the phospholipid passed through the column during the sample loading procedure. The remaining 77% was washed from the column during the lipid removal step. No phospholipids were detected in the amphotericin B recovery effluent, indicating that complete phospholipid separation from the drug was achieved. When an amphotericin B standard was processed in the same way, no phospholipid (phosphate) was detected in the three column effluents. Analysis of the sample load effluents from either the lipid-free house standard or the lipid complex sample for amphotericin B showed no evidence for the presence of the drug, indicating complete retention on the sorbent.This complete retention was largely due to the high pH of the alkaline methanol used for column pre-treatment. Previous experiments had shown that when the column was pre-treated with unmodified methanol, approximately 2-3% of the drug always passed through the sorbent during the sample loading procedure. Analysis of the effluents obtained from the lipid removal step showed that 5% of the amphotericin B was washed from the sorbent when either the amphotericin B - lipid complex or the house standard (no lipid present) was processed. We believe that this small amount of drug passed through the column during this step because it was only loosely bound to the sorbent, whereas 95% of the drug was firmly attached.We do not believe that the column had been saturated (all binding sites occupied) as it was possible to bind at least 500 pg (twice the amount generally present in our samples) of drug in other experiments (data not shown). To determine the method variability resulting from column to column differences a single drug - lipid sample was prepared and processed nine times using nine silica sorbent columns. The same evaluation was performed using an amphotericin BANALYST. NOVEMBER 1989, VOL. 114 1383 Table 1. Amount o f lipids (DMPC and DMPG) and amphotericin B in column effluents after each analytical step for lipid removal.An amphotericin B - lipid sample and a lipid-free amphotericin B standard (formulated to resemble the lipid preparation) were evaluated Test for lipid Amphotericin B - lipid sample Test for amphotcricin R Amphotericin B - lipid sample Amphotericin B standard Amphotericin B Analytical step pg ml-1 Yo standard CLgml ' Yo pg ml-I Yo Sample load effluent . . 93.38 23 0.00 0.00 0 0.00 0 Lipid removal effluent . . 313.60 77 0.00 13.50 5 13.00 5 Amphotcricin B recovery ctflucnt . . . . . . 0 . ( 10 0 0.00 237.00 95 236.00 95 Totals . . 406.98 100 0.00 250.50 100 249.00 100 Table 2. Recovery of amphotcricin B from the amphotcricin l3 - lipid formulation and from an amphotericin B house standard preparation. The potency of the amphotericin B - lipid formulation was 5.10 mg ml-1; the amphotericin B potency of the \tandad was 4.40 mg ml-1.All potencies were dctermined by HPLC Amphotcricin B recovcred/mg ml-I Amphotericin B - lipid Trial No. formulation standard Amphotericin B 4.97 4.79 4.79 4.96 4.84 4.81 4.85 4.75 4.78 4.14 4.17 4.10 4.12 4.23 4.23 4.16 4.25 4.25 Strrtisticul t)\,riluntion Average . . . . . . . . . . 4.84 mg ml 4.18 mg ml-1 Recovery . . . . . . . . 95% 95 % Standard deviation . . . . . . 0.08 mgml-1 0.05 mg ml ~' Coefficient of variation . . . . 1.65% 1.38% 95% rclativc confidence limit . . 1.27 1.06 Table 3. Amphotcricin B potencies of the amphotericin B - lipid \ample rcndcrecl lipid-free. Amphotericin B standards were processed in the same way as the sample. The drug - lipid amphotcricin B potency was 5.00 mg ml-I Amphotericin B potency/mg ml-l Microbiological assay 5.02 4.98 5.29 4.83 4.96 4.98 4.94 4.98 Stutisiic-ul c \durrtion Average .. . . . . . . . . Standarddeviation . . . . . . Coefficientotvariation . . . . 95% relative confidence limit . . HPLC 5.02 5.02 4.90 4.99 5.02 4.90 3.90 4.99 5.00mgml-1 4.97mgml-I 0.13 mg ml-' 0.06 mg ml-I 2.60% 1.21% 2.17 1.01 standard prepared to resemble the drug - lipid formulation. The results of this experiment are shown in Table 2. Both the standard and the drug - lipid formulation yielded statistically identical 95% recoveries of amphotericin B and similar coefficients of variation (1.38% for the amphotericin B standard preparation and I .6S% for the amphotericin B - lipid formulation). Recovery Improvement Efforts to improve the recovery of amphotcricin from the lipid complex yielded mixed results. Large (3 cm-i) silica sorbent columns were tried, which allowed 99% of the drug to be recovered, as measured by HPLC, but the microbiological assay results were low.Subsequent TLC experiments indi- cated that lipid removal was incomplete. With these large columns, the lipids could not be completely removed when five times the usual amount of lipid removal solution was used for a single sample. Other approaches for improving the recovery of lipid-free amphotericin B included: (i) the use of two l-cm3 columns (one placed on top of the other); (ii) sonicating the diluted sample preparation to mediate drug - lipid separation; (iii) using Tween 80 to weaken the strong drug - lipid attraction; and (iv) recycling the column effluents.These approaches did not lead to an improvement in the recovery beyond 95%. To correct for this 5% loss incurred in the lipid removal step, amphotericin B standards were processed in the same manner as for the drug - lipid preparation. Amphotericin B Degradation An experiment was performed to investigate the possibility of amphotericin B degradation in the highly alkaline lipid removal solution. An amphotericin B standard was prepared to resemble the drug - lipid formulation and was diluted with the lipid removal solution and maintained at room tempera- ture for 120min. No degradation could be detected after a 90-min exposure period and only a 3% loss was attributed to degradation when the drug was exposed to the alkaline solution for 120min.With the method described here, the amphotericin B was in contact with the alkaline solvent for between 10 and 15 min. Analysis of an Amphotericin B - Lipid Sample Rendered Lipid-free For microbiological and HPLC assays of the recovered drug, amphotericin B standards were processed using the same extraction method employed for the drug - lipid samples. This was carried out to ensure that the sample percentage recoveries would be experienced in the standard and test preparations. Data from a microbiological and HPLC evalua- tion of an amphotericin B - lipid preparation rendered lipid-free are given in Table 3. Both methods yielded the same potency of the drug - lipid formulation, approximately 5.0 mg ml-1, indicating that the procedures used did not affect the chromatographic retention time or the activity of the drug.The two methods yielded acceptable statistical evaluations, with the HPLC method having less variability, as expected. Conclusions The proposed method is an effective and reliable technique for separating the lipids DMPC and DMPG from the drug amphotericin B. Using this method, lipid-free amphotericin B (from an amphotericin B - lipid formulation) can be prepared and assayed by the USP small-plate microbiological assay or by HPLC with almost identical results.1384 ANALYST, NOVEMBER 1989, VOL. 114 The authors thank Jonathan Karten (E. R. Squibb & Sons) for his assistance with the HPLC equipment, Frank Marziani (The Liposome Company) for his assistance with the TLC experiments and Michelle Tomsho (The Liposome Company) for performing the lipid assays. References 1. 2. Ratafia. M.. and Scott, F. I., Am. Clin. Prod. Rev., 1987,5,29. Lopei-Berestein, G.. Fainstein. V.. Hopfer. R., Mehta, K., Sullivan, M. P., Keating, M., Rosenblum, M. G.. Mehta, R., Luna, M., Hersh, E. M., Reuben, J . , Juliano, R. L., and Bodcy, G. P., J. Infect. Dis., 1985, 151, 704. Juliano. R., Lopez-Berestein, G., Mehta, R., Hopfer, R., Mehta, K., and Kasi, L., Biol. Cell, 1983, 47. 39. 3 . 4. Tremblay, C., Barza, M., Fiore, C., and Szoka, F., Anri- microb. Agents Chemother., 1985, 26, 170. 5. “The United States Pharmacopeia.” Twenty-first Revision, Mack, Easton. PA, 1985, p. 1160. 6. Joseph, J . M.. in Ahuja, S . , Editor, “American Chemical Society Symposium Series. No. 297. Chromatography and Separation Chemistry: Advances and Developments,” Ameri- can Chemical Society, Washington, DC, 1986. p. 83. Rouser, G., Fleisher, S . , and Yamamoto, A., Lipids, 1970, 5, 494. 7. Paper 8/0471346 Received December 5th, 1988 Accepted June 9th, I989

ISSN:0003-2654    

DOI:10.1039/AN9891401381

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989, VOL. 114 1385 Adsorption of Metal Ions From Ethanol on an Iminosalicyl-modified Silica Gel Lauro T. Kubota and Jose C. Moreira lnstituto de Quimica, Unesp - Araraquara, CP 174, 14800 Araraquara SP, Brazil Yoshitaka Gushikem" lnstituto de Qu-imica, Unicamp, CP 6154, 13081 Campinas SP, Brazil lminosalicyl groups attached to a silica gel surface have been used for the adsorption of metal ions from ethanol. This organofunctionalised silica adsorbs iron, nickel, cobalt, copper, zinc and cadmium from ethanol by both batch and column techniques. The average distribution coefficient determined for each metal ion was as follows (ml g-I): Feiil,4.5 x 102; Coil, 1.4 x 102; Ni", 1.1 x 102; Cull, 3.6 x 102; Znll, 3.0 x 102; and Cd", 2.2 x 102. Keywords: Iminosalicyl-modified silica; metal ions in ethanol; silica gel; adsorption of metal ions from ethanol; ethanol analysis The direct determination of the trace metals found in ethanol is difficult and in many instances it has been necessary to use a pre-concentration procedure employing a very time-consum- ing method, such as the evaporation of the solvent to dryness, prior to analysis.' The immobilisation of chelating groups on siliceous surfaces has been employed successfully to produce a variety of organofunctional-modified silica gels.These have been used for the batch and column adsorption of metal ions from non-aqueous solvents .2,3 Of particular interest are the organo- functional groups that have been used to determine the trace metals present in ethanol.'' From the literature it is apparent that Schiff bases were largely used in the past to prepare various stable transition and non-transition metal complexes.5 Procedures for the immobi- lisation of the iminosalicyl group on the silica surface have been developed and, in addition, the structure of the surface complexes with CU" has been described in some detail.6 This paper describes the application of a silica-based iminosalicyl complexing agent to the batch and column adsorption of various transition and non-transition metals from ethanol solution. Experimental Silica gel of internal surface 500 m2 g- 1 and an average pore diameter of 0.6 nm was used.Immobilisation of the organofunctional group was perfor- med in two steps. (1) Activated silica gel (60 g) was suspended in 200 ml of 10% V/V 3-aminopropyltriethoxysilane in dry toluene.The mixture was refluxed for 10 h in a nitrogen atmosphere with constant stirring. The resulting product, aminopropyl silica gel (APSG), was filtered, washed consecutively with toluene, ethanol and acetone and then heated at 348 K for 8 h in a vacuum line. (2) The APSG (10 g) was reacted with 10 ml of salicyl- aldehyde in 30 ml of anhydrous diethyl ether with constant stirring. The mixture was filtered and washed with ethanol and the solid, iminosalicyl silica gel (ISSG), was heated at 330 K in a vacuum line. * To whom correspondence should be addressed. Step (2) can be written as Y-(CH2)3NH2 + HOC - Y-(CH2)3N=CH APSG ISSG where Y represents the solid support. Optimisation of Adsorption Time The time required for the solid - liquid system to attain the equilibrium conditions was determined by placing 50 ml of 1 x 10-3 M ethanol solutions of the metal ions in the various flasks and shaking with 0.06 g of ISSG.At different time intervals, the supernatant from each flask was separated off and the metal determined. 'The amount of the metal ion adsorbed by the solid phase was calculated using the following equation: N f = ( N , - N,)/rn, where N , is the initial amount of the metal ion (mmol), N , is the amount of the metal ion (mmol) in the supernatant after equilibrium has been achieved and rn is the mass of the ISSG (8). Adsorption Isotherms The adsorption isotherms were determined in anhydrous ethanol using the batch technique at 298.3 2 0.2 K. Solutions of metal ions in the concentration range 0.2 x 10-4-2.5 x l o - 3 ~ were shaken for 30 min with 0.06 g of ISSG.The supernatant from each flask was separated off, the metal determined and the amount of metal adsorbed calculated using the equation given above. Maximum Adsorption Capacity, N p a x . , of ISSG To obtain N p x - data for each metal, 50 ml of the 0.1 M ethanol solution of each metal ion were shaken with 0.5 g of ISSG for 1 h at 298 K. The mixture in the flask was filtered and washed with pure ethanol, the filtered solution was transferred into a calibrated flask, the volume was made up to 200 ml and then the metal was determined.1386 t i ar t m 4 d .w .- F !- 2000 ANALYSl. NOVEMBE,R 1989, VOL. I14 \ \ \ , \ B 1 r b J I I I 1800 1600 1400 Pre-concentration and Recovery of the Metal Ions This study was carried out using a 13 x 0.6 cm i.d.glass column packed with 1 g of ISSG. Initially the column was washed with pure ethanol and then 100 ml of the individual ethanol solutions of 1 X 10 -5 M FeC13 or MC12 (M = Co, Ni, Cu, Zn or Cd) were percolated through the column with a flow-rate of 2.0 ml min-1. The column was then washed with 50 ml of pure ethanol. The metal ions were eluted from the column with 10 ml of a solution of citric acid in ethanol (0.1-2.5 M) and the metal ion was determined directly in this solution by complexometric titration with 0.0 1 M ethylene- diaminetetraacetic acid (EDTA) according to the procedure described in reference 7. The experiment for each metal was carried out in triplicate in order to determine the precision of the method.Infrared Spectra and Surface Area The infrared spectra of the samples were obtained in the region 2000-1300 cm-1 using the pressed-disc technique, according to a previously described method.8 The surface area was determined using the small-angle X-ray scattering (SAXS) technique. The SAXS measurements were obtained using Cu Ka. radiation and Kratky collimation, and a position-sensitive detector coupled to a multi-channel an a1 y se r . ') - 1 0 Salts Used for Preparation of Solutions The following pure salts were used without further purifica- tion: FeCI3, CoCI2.2H2O, NiCI2.6H20, CuCI2.2H20, CuBr2, Cu( 02CMe)2. H 2 0 , C U ( N O ~ ) ~ .3H20 , Cu( C104)2. 6H20, ZnC12.2H20 and CdCI2.2H20. Chloride salts were used to prepare the solutions unless stated otherwise. Results and Discussion Characteristics of the Material The preparations were made in duplicate and the amount of organofunctional groups attached to the silica surface was determined by a method based on the nitrogen content (see Table 1).The surface areas of both preparations of ISSG are very similar, i.e., 296 k 24 and 294 k 29 m2 g-1 for the products obtained from preparations 1 and 2, respectively. Assuming that the organofunctional groups cover the surface uniformly, the average intermolecular distance is 0.67 nm for both samples. The infrared spectrum shown in Fig. 1 confirms the prcscnce of iminosalicyl groups bound to the silica surface. The absorption band observed at 1650 cm-1 is due to the C-N stretching frequency of the imino group" whereas the Table 1.Chemical analysis of the functionalised silica gel Preparation Nitrogen No*/ st/ Sample No. content, % mmol g- 1 m?- g 1 APSG . . 1 1.61 k 0.08 I . 15 2 0.06 - 1SSG . . I 1.52 -t 0.05 1 .OX t 0.04 296 f 24 APSG . . 2 1.37 f 0.06 0.98 f 0.05 - ISSG . , 2 1.26 2 0.07 0.90 f 0.05 294 IL 29 * No = amount of functional group (mmol)/mass of 1SSG (8). 1- Surface area. remaining bands at 1585, 1535, 1500, 1460, 1415 and 1345 cm-1 are due to the benzene ring vibration and CH2 deformation modes. The stability of ISSG in purified ethanol was tested by shaking the solid with the solvent for 30 h, prior to use in a batch or column experiment. This procedure was carried out because the iminosalicyl group is sensitive to water (present in the solvent) with which it can react and regenerate the APSG and liberate free salicylaldehyde to the solution.12 As salicyl- aldehyde shows a strong absorption band at 380 nm, various spectra of the supernatant were run in this region during the stability experiments. The results indicated that the amount of salicylaldehyde was negligible after 30 h. An important aspect of the material type is the velocity with which the solid phase adsorbs metal ions from the solution and attains the equilibrium conditions. Fig. 2 shows plots of N f versus time for Cu" and Zn". For these and all remaining metals, the time necessary for the systems to reach equilibrium was about 10 min. Adsorption Isotherms The adsorption isotherms are shown in Fig. 3. By defining the distribution coefficient, D , as D = Nf/c and expressing the values of Nt in mmol 6-1 and c in mmol ml-1, the following average values of D were calculated for each metal (ml g- I ) , in the concentration range shown in Fig.3: Fell', 4.5 x 10'; Co", 1.4 X 102; Ni", 1.1 X 102; Cu", 3.6 X 102; Zn", 3.0 X 10': and Cd", 2.2 x 102. Assuming that only the organofunctional groups are responsible for metal adsorption, the maximum value of Nf is determined by (i) the number of iminosalicyl groups attached 0.4 c o) 0.3 - I 1 I I I 0 2 4 6 8 10 12 Time/min 0.1 Fig. 2. Plots of Nf vecsus time at 298 K. A, Cu"; and B, ZnIlANALYST, NOVEMBER 1989, VOL. 114 0 6 500 1387 - 0.5 I 1 I A I 0.4 r m 0.3 - E" 5 0.2 0.1 0 0.5 1.0 1.5 2.0 2.5 dmmoll-1 Fig. 3. 298 K. A, FclI1: B, C'o"; C , N F ; D, Cu"; E, Zn"; and F, Cd" Adsorption isotherms of metal ions from ethanol solution at to the surface, (ii) the nature of the surface complexes and (iii) the stabilities of the complexes.From the literature it is apparent that Schiff bases can form stable complexes with transition and Group IIB metals. Normally the complexes formed are of the type metal : ligand = 1 : 2,' assuming that the ligand is bidentate. As No = 1.08 mmol g-1 [No = amount of functional group (mmol)/mas5 of ISSG (g)J, the maximum value of Nt should be approximately 0.5 mmol g-1 if a 1 : 2 complex is formed. Except for FeI11, the isotherms in Fig. 3 apparently do not show N t tending to a constant value as c increases. The maximum adsorption capacitj, i.e., N f l l l a x = [Nf],,,, is attained for each metal when the solution concentration is as high as 0.1 M.At this concentration the following values of Nflnax were obtained for each metal (mmol g-1): Fell1, 1.16 k 0.01; Co", 0.68 k 0.01; Ni", 0.27 k 0.02; Cull, 1.12 &- 0.02; Zn", 1.07 k 0.02; and Cd", 1.10 k 0.02. As Ntm"x/No =1 for all metals, with the exception o f Co11 and Nil[, a 1 : 1 complex can be formed at the surface. However, it is not possible to infer from isotherm data alone that the 1 : 2 complex is present at the surface even with Thc average intermolecular distance of the attached mole- cules, 0.67 nm, is favourable with respect to formation of the 1 : 2 complex; however, on the basis of steric requirements, complex formation appears unlikely. On the other hand, this stoicheiometry has been suggested for the Cu" - iminosalicyl complex bound to the silica surface, with the geometry around the metal ion having a distorted tetrahedral symmetry." NJNo < 1.Temperature Influence It has been observed that the adsorption of metal ions from ethanol is affected by temperature, with the equilibrium constant being decreased as the temperature was lowered.8 The temperature range for the Cull adsorption study was 273-308 K. The results are shown in Fig. 4. Although D is not a constant, as it changes with concentra- tion, it is clear from Fig. 4 that the temperature plays an important role in the adsorption process. The average values o f D at each temperature in the concentration range studied are as follows: 273 K, 1.4 x 102; 283 K, 2.4 x 102; 298 K, 3.6 x 10': and 308 K, 5.2 X 102 ml g-I.Anion Influence The influence that an anion (A) could have in the adsorption process is related to its tendency to associate with a cation (M) in the solvent medium. In aqueous solution, the stability constants of the CuI1 complexes with the anions studied, log K , where K = [MA]/[Mj[A], at zero ionic strength I , are as follows: CH3COO-, 2.22; NO3-, 0.5; C1-, 0.4; and Br-, -0.03.13313 In solvents of lower dielectric constant the ions are Iooo /t"rl I I I I 0 0.5 1 .o 1.5 2.0 drnrnoll-1 Influence of temperature on the adsorption of Cu" from Fig. 4. ethanol solution. A, 308; B, 298; C, 283; and D, 273 K Iooo 3 D (;2 n l I I I I 0.5 1 .o 1.5 2.0 drnrnol I-' Fig. 5. Influence of various anions on the adsorption of Cu'I from ethanol solution at 298 K.A, Cl-; B, CH3COO-; C , NO3-; D, Br -; and E, CIOj- associated much more and, therefore, the adsorption process could be influenced more by changes in the co-ordination ability of different anions. However, Fig. 5 shows that this influence is small and there is no clear correlation with the stability constants in aqueous solution given above. The average values of D in the experimental concentration range used are as follows: CI-, 3.6 X 102; Br-, 3.4 X 102; CH,COO-, 2.6 x 102; C103-, 2.2 X 102; and NO3-, 2.1 X 10' ml g-1. The smaller influence of the anion on the adsorption process in comparison with the temperature effect is presumably due to the high stability of the surface complex. The cation - anion interaction in the solution phase that could decrease the transfer of metal ion from the solution to the solid phase is a secondary effect.Pre-concentration and Recovery of the Metal Ions Under the experimental conditions, the mass, rn, of the organofunctionalised silica and the amount of metal added, N , , were controlled in order to maintain the ratio (rn x No)IN1 = 103 and to ensure that 100% of the metal was adsorbed when the solution containing it was passed through the column. Desorption of the metals adsorbed on the column could not be achieved using an aqueous solution as eluent, as the iminosalicyl group was hydrolysed to a certain extent by water. The use of citric acid - ethanol as eluent was very convenient because the attached organofunctional groups were not lixiviated by this solution after various metal adsorption - desorption cycles.The results of the pre-concentration and recovery of the metal ions using the column method are shown in Table 2. It can be seen that in every instance the recovery is almost 100%. The volume of eluent was 10 ml for all metals, except for FelI1 for which 20 ml of 2.5 M solution were used.1388 ANALYST, NOVEMBER 1989, VOL. 114 Table 2. Prc-concentration and recovery of metal ions using the column method at ca. 298 K Concentration of metal, Ion p.p.m. FelI1 . . 0.5 Co” . . 0.6 NiII . . 0.6 Cu” . . 0.7 Zn” . . 0.7 CdlI . . 1.0 Volume of of eluent/ ml 20 10 10 10 1 0 10 Concentration of eluent/ moll- 2.5 2.0 1.2 1.2 1.2 2.0 Recovery, Yo 97.7 k 2.5 98.1 * 1.6 99.3 ? 0.9 99.8 iz 0.8 99.7 -t 0.7 99.5 * 1.1 Conclusions Iminosalicyl groups attached to a silica surface can readily be used to adsorb metal ions from ethanol solution.Its relatively high chemical stability in ethanol, and in particular the velocity with which the metal ions are adsorbed, make the iminosalicyl-modified silica gel potentially useful for analytical purposes. L. T. K. is indebted to CAPES and CNPq for a fellowship and to FINEP for financial support. References I . Bruning, I. M. R. A . , and Malm, E. B., HOE. Tec. Petrobras. 1982, 25, 217. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11 12. 13. 14. Airoldi, C., Gushikem, Y., and Espinola, J . G. P., Colloid Surf., 1986, 17. 317. Iamamoto, M. S . , and Gushikem, Y., J. Colloid Interface Sci., 1989, 129, 162. Moreira, J. C., and Gushikem, Y., Anal. Chim. A m , 1985, 176, 263. Holm. R. H., Everett, G. W., and Chakravorty, A., Prog. Inorg. Chem., 1966, 7, 83. Karpenko, G. A., Filippov. A . P.. and Yatsimirskii, K. B., Theor. Exp. Chem. (Engl. Transl.), 1980, 15, 440. Kubota, L. T., Ionashiro, M., and Moreira, J . C., Ecl. Quinz., 1988, 13, 19. Gushikem, Y . , and Moreira, J . C.. J. Colloid Interface Sci.. 1985, 107, 70. Luzatti, V., Acta Crystallogr., 1960, 13, 939. Glatter, O., and Kratky, O., “Small Angle X-ray Scattering,” Academic Press, London, 1982. Mabad, B. M., Cassoux, P., Tuchagues, J . P . , and Hendrick- son, D. N., Znorg. Chem., 1986, 25, 1420. Yatsimirskii, K. B., Chuiko, A. A., Filippov, A. P., Karpenko, G. A . , Thertykh, V. A., Yanishpol’skii. V. V., Grinenko, S. B., and Belousov, V. M., Dokl. Phys. Chem. (Engl. Trans/.), 1978, 237, 1182. Smith, K. M., and Martell, A. E., “Critical Stability Con- stants,” Volume 3. Plenum Press, New York, 1976. Smith, R. M., and Martell, A. E., “Critical Stability Con- stant5,” Volume 4, Plenum Press, New York, 1977. Paper 9/01 183H Received March 20th, 1989 Accepted June 9th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401385

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, NOVEMBER 1989, VOL. 114 1389 Use of Light Scattering to Detect Nebulisation Transport Interferences in Analytical Atomic Spectrometry Paul Gordon Department of Chemistry, Meston Building, University of Aberdeen, Aberdeen A69 2UE, UK Malcolm Cresser" Department of Plant and Soil Science, Meston Building, University of Aberdeen, Aberdeen A69 2UE, UK The spray chamber of a flame atomic absorption spectrometer has been fitted with two perpendicular side- arms with end windows. The aerosol was irradiated with white light through one window and the scattered light from the sample aerosol monitored through the other. Scattered light provided a useful method for monitoring changes in aerosol transport and provided a basis for automatic error warning. However, the range of use was confined to conditions under which the aerosol size distribution was reasonably constant.Keywords: Analytical atomic spectrometry; transport interference; error detection; aerosol light scatter; transport In spite of the steady improvements in the design and case of operation of analytical atomic spectrometers over recent years, it is still possible to obtain results with significant systematic error as a consequence o f unsuspected inter- ference. Earlier work in this laboratory demonstrated that a pressure transducer could be used to monitor aspiration rate by exploiting the observation that the pressure drop along a capillary tube is, within limits, proportional to the rate of solution flow in thc tube.',' If the signal from the transducer was connected to a microcomputer, immediate warning of partial blockage of a nebuliser or of aspiration rate related interferences was possible.Although useful, such a system is unfortunately insensitive to changes in transport efficiency not related to aspiration rate. It was therefore decided to investigate the potential, as an error detectiodwarning system. of monitoring light scattered by the sample solution aerosol when the latter is irradiated by a simple white-light sourcc, an idea suggested recently by Sharp.3 The magnitude of the scatter signal should be a function of the number of droplets per unit volume and the aerosol size distribution. Experimental Apparatus The concept was evaluated using a Baird A3400 atomic absorption spectrometer. The spray chamber of the spec- - JI 'Side-arm Fig.1. Apparatus for measuring light scattered by aerosol * To whom correspondence should be addressed. trometer was modified as shown in Fig. 1. The 40 mm long side-arms were made from 5 mm i.d. tubing and were centred 85 mm from the nebuliser end of the spray chamber. The external ends of the tubes were fitted with glass windows cut from microscope slides. The light source used was a KeyMed Industrial KLS-311 source. This contains a lS-V, 150-W quartz - halogen type A1/232 lamp, and the white light ouput is via a fibre-optic probe. The detector was an inexpensive photodiode (Radiospares) with a laboratory-built housing. 0 ' I I L I I I 0.3 0.5 0.7 0.9 Capillary diametedmm Fig. 2. absorbance or (c) scatter and aspiration capillary i.d.Relationship between ( a ) aspiration rate, ( b ) iron atomic1390 ANALYST, NOVEMBER 1989. VOL. 114 The detector output was fed to a chart recorder via a laboratory-built adjustable gain amplifier. The atomic ab- sorbance signal was also fed to a chart recorder. Aspiration rates were measured by timing the uptake of 5 ml of distilled water with a stop-clock. Discrete sample nebulisation was from the conical bottom of Technicon AutoAnalyzer sample cups. Methodology The effects of changing the aerosol amount and size distribu- tion were studied by changing the diameter of a fixed length (440 mm) of aspiration tubing over the range 0.4-0.8 mm, by rotation of the impactor bead or by aspiration of a concen- trated sodium chloride solution to cause deliberate nebuliser clogging.For the discrete nebulisation mode, integrated absorbance x time and scatter x time signals were quantified by cutting out and weighing the appropriate recorder traces; the results were normalised to show the closeness of the relationship between the two integrated signals. Results and Discussion Fig. 2 shows the influence of changing the aspiration capillary diameter on aspiration rate, absorbance and scatter signal from a 10 ug m - * iron(1lI) standard solution. Although the three graphs are superficially similar i n form, there are distinctive differences. When the aspiration rate decreases from its maxinium to its lowest rate on using narrow-bore tubing [Fig. 2 ( a ) ] , neither absorbance nor scatter decrease pro-ratu with aspiration rate. This discrepancy occurs because aerosol size distribution becomes more favourable at low aspiration rates, resulting in improved transport efficiency.4 The irregularity of the scatter signal graph [Fig.2(c)] was reproducible, and suggests, as might be expected, that aerosol size. in addition to aerosol amount, influences the size of the Nebulisation rate/$ s-1 0 20 40 L. 75 P +! m' 50 c) ._ - m C 07 .- 25 a, m + 4- v) Fig. 3. t W C m +? v) n m -0 al c P WJ C c - 0 1 2 3 4 Aspiration rate/ml min-1 Scattcr versus ncbulisation rate plot I At 200 400 600 800 1000 Vol u me/pl Fig. 4. and discrete sample size Relationship between 0, integrated scatter o r e , absorbancc scattered light signal. The unusual shape of the graph is almost certainly an artefact of the scattered light measurement optical configuration used, with an aerosol illumination path length of ca.40 mm but detection over only ca. 5 mm. There is also a further long path containing aerosol between the illuminated aerosol being measured and the detector window. In spite of the differences between the three graphs shown in Fig. 2., there is a linear relationship between scatter signal and aspiration rate over a limited range, i.e., where the capillary diameter exceeds 0.5 mm. This relationship is shown in Fig. 3, which suggests that, provided aerosol size distribu- tion does not change substantially, scattered light measure- ment can provide a quantitative indication of aerosol concen- tration. To test the above hypothesis, it was decided to study discrete sample nebulisation over the range 100-1000 pl, as aerosol size distribution is only significantly changed during ncbulisation of sample volumes smaller than 100 yl.5,h For volumes of sample solution >lo0 PI, therefore, integrated scatter should give a reliable quantitative measure of the total aerosol transported, which should be proportional to the integrated absorbance.Fig. 4 confirms that this is indeed the case. The integrated scatter x time scale was chosen to facilitate comparison. To provide further evidence that, with the system used, size distribution of the aerosol influences the scatter signal, the effect of impactor displacement o n absorbance and scatter signal was also studied. The absorbance and scatter signals were measured as a function of the angle of rotation of the impactor support arm from its normal vertical position.The results presentcd in Fig. 5 show that, although the graphs are similar in general shape, as in Fig. 2 there are significant differences. To ascertain how useful the system was for detecting nebuliser blockage, a solution of iron(ll1) in concentrated / 0.4 II 0 10 20 30 40 50 60 Rotation from VerticaVdegrees Fig. 5 . scatter signal Effect of impact bead displacement on 0, absorbance and 0, Time - Time -+ Fig. 6. and ( b ) scatter signal Effect of clogging nebuliser on ( u ) iron atomic absorbanceANALYST. NOVEMBER 19x9, VOI,. 114 1391 I I I I I I I 0 2 4 6 8 1 0 1 2 1 4 Time/min Fig. 7. into a drq spray chamber Change in scatter signal with time when water is ncbuliscd sodium chloride was nebuliscd and the absorbance and scatter were plotted as the nebuliser progressively clogged.The rclationship betwccn the two signals shown in Fig. 6 is obvious. I t is interesting to note that the sodium chloride itself produced a small but significant enhancement of the scatter signal, probably through a droplet size effect or via inhibition of small (<2 llm diameter) droplet evaporation in the presence of the high salt contcnt.7 More dilute but highly coloured solutions (e.g., potassium permanganate solution) gave the same scatter signal as water. During this work it was observed that, except during discrete sample nebulisation, the scatter signal tended to drift with time, and a 10-min stabilisation period was required before good reproducibility was attained.Fig. 7 shows how the scatter signal varies with time after the start of nebulisation into a dry spray chamber. The steady decrease is probably attributable to Condensation, and equilibrium is only reached after drops of water have formed and fallen from the windows, which then remain uniformly wet. The break in the curve is probably associated with the point in time at which individual deposited droplets coalesce to form a more transparent water film. Conclusions Measurement of the amount of light scattered by the sample aeroml provides a reliable indication of the amount of aerosol being transported, provided that the aerosol size distribution is fairly constant. In future designs, more thought should be given to shortening the non-contributing absorption path lengths for the incident and scattered light beams and to condensation problems. It should then prove possible to detect even fairly small changes in the transport rate. I . 2. 3. 4. 5 . 6. 7. References O’Grady, C. E., Marr, 1. L., and Crcsscr, M. S . , Analyst, 1985, 110, 431. Crcsscr, M. S . , O’Grady, C. E . , and Marr. 1. L., Prog. Anal. At. Sprctrmc., 1985, 8, 19. Sharp, B., J . Anal. At. Spectrom., 1988, 3, 939. Lope7 Garcia, I., O’Grady, C., and Cresser, M., J. Anal. At. Spetwom.. 1987, 2, 221. Malloy, .I. M., Kcliher, P. N . , and Cresyer, M. S . . Spectrochim. Actu, Purr B , 1980, 35, 833. Crcsscr, M. S . , Prog. And. At. Spectrosc., 1981, 4, 219. Crcsscr, M. S . , and Browner, R. F . , Spectrochim. Actu, Part B , 1980, 35, 73. Paper 9/01 6921 Received April 24th, I989 Accepted June 9th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401389

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. NOVEMBER 1989. VOL. 114 1393 Quantification of Arsenic Species in a River Water Reference Material for Trace Metals by Graphite Furnace Atomic Absorption Spectrometric Tech n iq ues Ralph E. Sturgeon, K. W. Michael Siu, Scott N. Willie and Shier S. Berman Division of Chemistry, National Research Council of Canada, Ottawa, Ontario KIA OR9, Canada Direct injection graphite furnace atomic absorption spectrometry (GFAAS) and hydride generation GFAAS techniques have been used to quantify As species in the National Research Council of Canada river water reference material SLRS-1 following their concentration and separation on a strong cation-exchange resin. Arsenic(lll), As” and dimethylarsinic acid account for 71% of the total As. Negligible amounts (<5%) of arsenocholine, tetramethylarsenic, monornethylarsonic acid and organically bound As were present.A significant fraction of the As (ca. 22%) exists as a relatively inert, unidentified species, which has similar but not identical properties to arsenobetaine. The species integrity of the sample has remained unaltered for the past 13 months of storage. Keywords: Graphite furnace atomic absorption spectrometry; h ydride generation; arsenic; speciation; river water Few data exist on the species distribution of As in natural waters.’-7 The major thrust of such studies has been related to the characterisation of As in fauna and flora, primarily because o f toxicological interest in food chain species and the difficulty of isolating As fractions when the total As content in natural waters might lie in the sub-ng inl-1 range.The behaviour of As in terrestrial and marine waters is largely controlled by biochemical processes involving planktonic alga&-s with methylated forms correlating with photosyn- thetic activity. The production of organoarsenic compounds from inorganic As is now generally regarded as a de-toxifica- tion mechanism, allowing algae to regulate the cellular levels of arsenate ~ which would otherwise interfere with phosphory- lation processes. Analogous de-toxification mechanisms are believed to be active in higher mammals.q noted that abiotic oxidation processes involving sediments might also occur in aquatic environments, potentially alleviating the toxicity of AsIII. A plethora of lipid- and water-soluble arsenicals have been isolated from the fauna and flora of natural water habitats (primarily saline), including identified species such as monomethylarsonie acid (MMA), dimethylarsinic acid (DMA), trimethylarsine, trimethylarsine oxide, arseno- betaine, arsenocholine, trimethylarsonium lactate, trimethyl- 0-phosphatidylarsonium lactate and several arseno-sugars.which emphacises that the most widely observed biochemical fate of ‘4s in the environment is methylation.”ll-Ia Organo- arsenic compounds such as these are released into the aqueous environment, thereby becoming available to higher trophic levels of the food chain. A knowledge of the cpeciation and transformations of As in natural waters is important because the bioavailability and the phy~iological/toxicological effects of As depend on its chem- ical form.li Human toxicity to As species decreases in the following order: arcenite > arscnate > rnonomethylarsonic acid > dimethylarsinic acid > arsenobetaine.16 Numerous analytical techniques have been applied to differentiation studies, including various forms of chromato- graphy, viz., ion chromatography,l7 high-performance liquid chromatography (HPLC) ,1X-2* gas chromatography or selec- tive volatilisation of As derivatives,2-6.**-*~ anion exchange25 and selective liquid - liquid extraction.26 These techniques are often coupled with element-specific atomic spectrometric detection methods [e.g., graphite furnace atomic absorption spectrometry (GFAAS) and inductively coupled plasma atomic emission spectrometry].Oscarson rt al. Ideally, the experimental procedure should be sensitive, simple and species selective. This paper describes a method that utilises cation exchange and the in situ concentration of volatile arsines in a pre-heated graphite furnace and its application to the determination of soluble inorganic and organic forms of As in a National Research Council of Canada (NRCC) river water reference material for trace metals, SLRS-I. To date, speciation studies undertaken on terrestrial and marine waters have revealed the presence of only As111, AsV, DMA and MMA. Mono-, di- and trimethyl-arsine and trimethylarsine oxide, which were specifically sought by A n d r e a ~ , ~ could not be detected. No reports have appeared that suggest the presence of “inert” (non-hydride active) organoarsenic species in natural waters such as arsenobetaine although potential arsenobetaine precursors have been iso- lated from algae.2”*7.2* Problems encountered with the certification of SLRS-1 for total As content revealed that a significant fraction (>20%) was bound in a highly inert form that required vigorous acid oxidation or UV photolysis to rcnder it reactive towards NaBH4.Experimental Apparatus All atomic absorption measurements were made using a Perkin-Elmer Model 5000 spectrometer fitted with a Model HGA-500 graphite furnace and a Zeeman-effect background correction system. An As electrodeless discharge lamp (Perkin-Elmer) operated at 6 W was used as the source. Absorption was measured at the 193.7-nm line with a spectral band pass of 0.7 nm.The introduction of aqueous wmples (20 PI) into the furnacc was achieved with a Perkin-Elmer AS-40 auto- sampler. Pyrolytic graphite coated graphite tubes and L’vov platforms were used t o atomise the samples. Arsenic was also determined by hydride generation GFAAS with in situ concentration in the graphite furnace. The hydride cell has been described in detail elsewhere.*q The borosilicate glass columns used to support the resins for analyte pre-concentration have also been described pre- vi o us1 y. 30 Reagents All reagents were purified prior to use. Concentrated hydrochloric acid ( I 0 M), nitric acid (16 M ) , perchloric acid1394 ANALYST, NOVEMBER 1989, VOL. 114 (12 M ) , acetic acid (12.5 M), ammonia solution (12 M) and methanol were prepared by sub-boiling distillation in a quartz still.Aqueous stock solutions of As111, AsV, MMA and DMA were prepared by dissolving arsenic trioxide (Fisher) in 1 M HCI, sodium arsenate (Baker) in 1 M HN03, and disodium methylarsonate hexahydrate (Pfaltz and Bauer) and dimethyl- arsinic acid (Pfaltz and Bauer) in de-ionised, distilled water (DDW), respectively. The last three were standardised against As111 using GFAAS and inductively coupled plasma mass spectrometric techniques.?' Aqueous stock solutions of arsenobetaine (AB), arsenocholine (AC) and tetramethyl- arsenic (TMA) were prepared from arsenobetaine hydrate, arsenocholine iodide and tetramethylarsonium iodide, respec- tively. These were also standardised against As111 and their purity was verified by HPLC - inductively coupled plasma mass spectrometry.31 Dowex 50W-X8 strong cation exchanger (5G100 mesh, Baker) and CI8 bonded silica gel (200-400 mesh, Waters Associates) were used for column work.Procedures All sample and analytical manipulations were conducted in a class 100 clean room fitted with laminar-flow fume hoods and benches. Hydride generation GFAAS Aliquots of SLRS-1 (2.0 ml) were placed in the hydride generation cell with 8 ml of 0.5 M HCI and the hydride-active arsenic species formed on the addition of 1% mlV NaBH4 (Alfa Ventron) were transferred to the preheated graphite tube.29 The response was quantified against the signal from a similarly generated As111 standard and provided a measure of the As"*, AsV, DMA and MMA in the sample. UV photolysis and oxidative digestion Aliquots (300 ml) of SLRS-1 were acidified further by the addition of 450 pf of HN03 and 450 p1 of 30% V/VH202 and placed in a quartz spiral UV photolysis system where the sample was irradiated for 7 h by a 500-W high-pressure Hg - Xe arc lamp.The DDW blanks were treated similarly. Aliquots of the photolysed water and the blank were analysed using the above hydride generation GFAAS technique29 to yield a measure of the total As in the sample. Aliquots (15 ml) of SLRS-1 were placed in pre-cleaned PTFE beakers and 1 ml of HCIO4, 1 ml of HZS04 and 1 ml of HNO3 were added. The samples were taken to fumes of SO3 over a I-h period and analytical blanks were run concurrently. The samples and blanks were analysed by hydride generation GFAAS to give a measure of the total As.Cation-exchange pre-concentration Aliquots (500-1000 ml) of SLRS-1 (pH 1.6) were passed through pre-cleaned tandem columns of CI8 (1-cm bed) and Dowex 50W-X8 (5-cm bed) using gravity feed (2-3 ml min-I). The Dowex resin was previously washed with 2 M HN03 and rinsed with DDW until the effluent was neutral. Organic species sorbed on the C18 column were subsequently eluted using 20 ml of methanol. The methanol fraction was slowly evaporated to dryness following the addition of 200 pl of HN03 and the residue was re-dissolved in 2 ml of 0.1 M HN03 and subjected to direct analysis with calibration against standards of As"'. Palladium (20 pg) was used as a modifier and a thermal programme recommended by the manufacturer was selected. Atomisation was from a L'vov platform.The Dowex 50W-X8 column was washed once with 25 ml (five column volumes) of DDW and eluted sequentially with 25 ml of 4 M NH40H, 25 ml of DDW and 25 ml of 4 M HCl. The NHJOH and HCI fractions were gently evaporated to near dryness and reconstituted using 5.0 ml of DDW for the NH40H fraction and 2.0 ml of DDW for the HCI fraction. The NH40H fraction contained DMA and AB species; this was subjected to analysis by both hydride generation GFAAS to yield a measure of the DMA content and by direct injection to yield a measure of the total As in the fraction. The HCl fraction contained AC and TMA and was analysed by direct injection with quantification against standards of AslIl. After passage through the Dowex column, the sample solution contained species not ion exchanged at pH 1.6, including Aslll, AsV and MMA.This effluent was subjected to analysis by hydride generation GFAAS. With a 0.5 M HCI medium in the generator, signals from AsiL1, AsV and MMA could be obtained quantitatively. Separate aliquots of this solution were also used to generate the hydrides of As"1 at pH 4.61.',6 by adding 300 pl of 1 M NH40COMe buffer and 50 pl of HCl to 10 ml of DDW and 1.0 ml of sample in the generator. Under these conditions, no response was obtained from AsV and a response of less than 6% from MMA. Results and Discussion Andreaei has reported that the methylarsenicals are kinetic- ally very stable and natural water samples can be rendcrcd suitable for long-term storage if they are either acidified or sterile-filtered through 0.2-pm filters.The preparation of the NRCC river water reference material SLKS-1 included both acidification to pH 1.6 with HN03 and \terile filtration (CHC13 - HN03 pre-rinse of the homogenisation tank and filters) through 0.2-pm membrane filters. The synthetic spikes of As"', AsV, MMA, DMA, AB, TMA and AC added to DDW (pH 1.6) and aliquots of SLKS-1 were subjected to separation by passage through the Dowex 5OW-X8 column and the recovery was found to be near quantitative. The effluent contained 100% of the As"', AsV and MMA. The AB and DMA were recovered in the 4 M NH40H fraction with 100 k 2 and 94 k 3% efficiency, respectively, whereas 96 k 3% of the AC was recovered in the 4 M HCI fraction. The relative peak-height response of the various As species to quantification with the hydride generation GFAAS system was studied and the results were found to be similar to those reported earlier,'g i.e., As111 = As" = MMA = 1.2DMA.No signals could be obtained from AC, TMA or AB. At pH 4.6 only As"' remained hydride-active and the response from MMA dropped to 6% and that of AsV to zero. Alkaline hydrolysis of AB has been reported to result in the production of trimethylarsine oxide,1"3' a compound reactive towards NaBH4. In this study, attempts to hydrolyse AB to produce hydride-active species were relatively unsuccessful. The digestion of 800 ng of AB in 25 ml of 4 M NaOH at 90 "C for 72 h resulted in only 35% conversion to a hydride-active form. The remaining AB was recovered (64% of the original) after acidification, passage through Dowex 50W-X8 and elution with 4 M NHJOH.Greater than 86% conversion of AH to a hydride-active form could be obtained by heating 800 ng of AB to yield fumes of SO3 in a mixture containing 5 ml of H2S04, 5 ml of HN03 and 5 ml of HCI04. Quantitative conversion was also obtained by subjecting the sample to UV photolysis for 7 h in the presence of HNO? and H20z. Quantification of Species Fig. 1 illustrates the various steps used to isolate and quantify the As species in SLRS-1. The identification of species was, with the exception of AH, based solely on elution characteristics from Dowex 50W-X8 through comparison with known standards. Oxidative photolysis of SLRS- 1 presumably converts all species to As". Analysis of the photolysed sample gave a yield of 0.54 k 0.05 ng ml-1, in good agreement with the certified value of 0.55 k 0.08 ng ml-1. Similarly, an oxidative acidANALYST, NOVEMBER 1989, VOL.114 1395 H202 - HNO,; UV oxidative 4 M HCI r- Direct GFAAS Organically bound As Hydride (l)-* generation with HCI Aslll AsV MMA Dowex generation 50W-X8 \ 1 Direct GFAAS AC TMA AB DMA DMA DMA As111 generation AsV MMA DMA AB AC TMA A AB AB AC TMA 1 generation A aeneration Fig. 1. AB. arsenobetainc; AC, arsenocholine; TMA, tetramethylarsenic; and A , difference Fractionation/quantification scheme for As species in NRCC SLRS- 1. DMA, Dimethylarsinic acid; MMA, monomethylarsonic acid; digestion followed by hydride generation GFAAS resulted in a value of 0.56 k 0.04 ng ml-I. Analysis of SLRS- 1 by hydride generation GFAAS prior to any form of sample pretreatment gave a value of 0.39 k 0.03 ng ml-1.This reflects the concentration of all NaBH4-active species in the sample, i.e., As"', AsV, MMA and DMA, leading to the conclusion that at least one non-hydride active form of As, with properties similar to arsenobetaine, is present at a concentration of 0.15 k 0.06 ng ml-1. The ratio of hydride-active to hydride-inactive forms of As has remained unchanged over the past 24 months. Reversed-phase CI8 will sequester neutral organic material from the SLRS-1. Direct GFAAS analysis of the acid-digested methanolic eluent from this column revealed that no signifi- cant amounts of As (41.01 ng ml- l) were associated with this fraction of material. Direct GFAAS analysis of the fraction of the As eluted from the Dowex column with 4 M NH40H provided a measure of the DMA and any other arsenobetaine-type species present.This fraction contained 0.17 f. 0.02 ng ml-1 of As. By subjecting this fraction to analysis by hydride generation GFAAS, an apparent As content of 0.05 k 0.01 ng ml-1 was found. The hydride-active form of As in this fraction is DMA. The difference in the above two values, 0.12 t- 0.02 ng ml-1, reflects the concentration of non-hydride active species and compares reasonably well with the earlier estimate of 0.15 k 0.06 ng ml-1 as its concentration. Direct GFAAS analysis of the fraction of the eluent stripped from the Dowex column with 4 M HCI revealed that no significant amounts of AC or TMA (<0.01 ng ml-1) were present in SLRS-1.After passage through the Dowex column the sample solution should contain As'", AsV and MMA species. Quantification of the As fraction by hydride generation GFAAS in 0.5 M HCI gave a concentration of 0.34 -t 0.02 ng ml-I. This value compares well with the concentration of As measured by hydride generation GFAAS in 0.5 M HCI using untreated samples of SLRS-1 that have been corrected for DMA content (0.05 k 0.01 ng nil-I), i.e., 0.34 Ifi 0.03 ng ml-1. N o direct or indirect measurement of MMA concentration was possible using the protocol outlined in Fig. 1. However, this determination was attempted earlier in a separate study in which MMA was coprecipitated with hydrated iron(II1) oxide, derivatised with 2,3-dimercaptopro- panol and quantified by gas chromatography.33 The concen- tration was found to be below 0.02 ng ml-l.The MMA is generally present in natural waters at concentrations compar- able to or lower than those of DMA, possibly as a conse- quence of it being an intermediate in the As methylation sequence. 1 Quantification by hydride generation GFAAS of the As in the sample after treatment of the sample on the Dowex cation exchanger at pH 4.6 yielded a measure of As111 only, i.e., 0.16 t- 0.01 ng nil-1. The difference in values obtained with the above two hydride procedures is the concentration of As", i . e . , 0.18 -t 0.02 ng ml-I in SLKS-1. When this measurement was repeated after 13 months, within experimental precision, identical results for As111 and AsV were obtained. Table 1 summarises the results of the analyses of the various fractions of As separated according to the scheme shown in Fig.1.1396 ANALYST, NOVEMBER 1989. VOL. 114 Table 1. Forms of As in SLRS-1 Species Concentratiodng ml- 1 Asti1 . . . . . . . . 0.16 * 0.01 AsV . . . . . . . . 0.18 * 0.02 MMA . . . . . . . . <0.02 DMA . . . . . . . . 0.05 i 0.01 AC, TMA . . . . . . <0.01 Inactive” . . . . . . 0.12 * 0.02 Totalt: 0.52 k 0.03 Organically bound . . <0.01 * Fraction of As that is unreactive towards NaBH4 in 0.5 M HCI. T Certified value, 0.55 f 0.08 ng ml-l of total As. Nature of the Non-hydride Active Species A significant fraction (22%) of the total As in SLRS-1 is present in a chemical form that is non-reactive towards NaBH4 in 0.5 or 1.0 M HCI, elutes from Dowex 50W-X8 in the 4 M NH40H fraction and cannot be extracted into organic solvents or sorbed on to hydrophobic surfaces.All these characteristics point to the occurrence of one or several species that are water soluble, inert and probably charged, i.e., of the arsenobetaine tY Pe. Further characterisation of thidthese species by established techniques for arsenobetaine related substances, i. e., solid- probe electron impact mass spectrometry, thin-layer chroma- tography and HPLC - inductively coupled plasma mass spectrometry,” was inconclusive; a major obstacle is the low concentration of As and the intensive sample work-up required. At present, the identity of thesehhis species is unknown. The conversion of arseno-sugars released from phytoplank- ton to dimethylarsinoylethanol under anaerobic conditions has been reported.34 The latter species is apparently similar to arsenobetaine in terms of solubility, size, polarity and basicity and might be a candidate for the unidentified “inert” As fraction in SLRS-1. However, its stability with respect to oxidation under extended storage conditions is not known.Conclusion Speciation of As in NRCC river water reference material SLRS-1 has revealed the presence of a significant fraction of analyte (22%) in an inert, unidentified form. Acidification to pH 1.6 and sterile filtration of the sample through membranes of 0.2 pm porosity have resulted in stabilisation of the species distribution in which the relative amounts of As”’ to As” and hydride-active to hydride-inactive forms have remained un- altered for at least 13 months.The authors thank W. R. Cullen, Department of Chemistry, University of British Columbia, Vancouvcr, Canada, for supplying the arsenobe taine, arsenocholine and tetrame thyl- arsonium standards. NRCC No. 30648. References 1. 2. Braman, R. S., and Foreback, C. C., Science, 1973,182. 1247. Braman, R. S.. Johnson, D. L., Foreback, C. C., Ammons, J . M., and Bricker, J. L., Anal. Chem., 1977, 49, 621. 3. 4. 5. 6. 7 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Andreae, M. O., Anal. Chern., 1977, 49, 820, Andreae, M. O., Deep-sea Krs., 1978, 25, 391. Andreae, M. O., Limnol. Oceanogr., 1979, 24, 440. Howard, A. G., and Arbab-Zavar, M. H., Analyst, 1981, 106, 213. van Cleuvenbergcn, R.J. A., van Mol, W. E . , and Adams, F. C., J. Anal. At. Spectrom.. 1988, 3, 169. Lunde, G., Environ. Health Perspect., 1977. 19. 47. Vahter. M., in Fowler, B. A., Editor, “Biological and Enkironmcntal Effects o f Arsenic,” Elsevier, Amsterdam, 1983, p. 171. Oscarson, D. W., Huang, P. M., and Liaw, W. K., J. Environ. Quai., 1980, 9, 700. Zingaro, R. A., and Rottino, N. R., in Lederer, W. H., and Fensterheim, R. 9.. Editorx. “Arsenic: Industrial, Biomedical, Environmental Perspectives,” Van Nostrand. New York, 1983, p. 327. Bcnson, A. A.. Cooney, R. V., and Herrera-Lasso, J. M., J . Plant Nutr., 1981, 3, 285. Edmonds, J. S . , and Francesconi, K. A., Nuture (London), 1977, 265, 436. Andreae. M. 0.. in Lederer, W. H., and Fensterheim, K. J . , Editors, “Arscnic: Industrial, Biomedical, Environmental Per- spectives,” Van Nostrand, New York.1983. p. 378. Penrose, W. R., Crit. Rev. Environ. Control, 1974, 4, 465. Squibb, K. S . , and Fowler, B. A., in Fowler, B. A., Editor. “Biological and Environmental Effects of Arsenic,” Elsevier. Amsterdam, 1983, p. 233. Foa, V., Colombi, A., Maroni, M., Buratti, M., and Cal- zaferri, G., Sci. Total Environ.. 1984. 34, 241. Fish, R. H., Brinckman, F. E., and Jcwett. K. L., Environ. Sci. Trchnol., 1982, 16, 174. Tyc, C. T.. Haswell, S. J . , O’Ncill, P., and Bancroft, K. C. C.. Anal. Chim. Acta, 1985. 169, 195. Chana, B. S . , and Smith, N . J . . Anal. Chim. Actu, 1987, 197, 177. Spall, W. D., Lynn, J. G., Anderson, J . L., Valdez, J . G., and Gurley, L. R., Anal. Chem.. 1986. 58. 1340. Ebdon, L., Hill, S . , Walton, A. P., and Ward. R. W., Analyst. 1988, 113, 1159. Andreae, M. 0.. and Klumpp, D., Environ. Sci. Trchnol.. 1979, 13, 738. Haraguchi, H., and Takatsu. A., Spectrochirn. Actu, Part B , 1987. 42, 235. Aggett, J . , and Kadwani, R., Analyst, 1983, 108, 1495. Amankwah, S. A., and Fasching, J . L.. Talanta, 1985,32,111. Maher, W. A., in Craig, P. J . , and Glockling, F., Editors, “The Biological Alkylation o f Heavy Elements,” Royal Society of Chemistry, London, 1988, p. 120. Benson, A. A., in Craig, P. .I., and Glockling, F., Editors, “The Biological Alkylation of Heavy Elements,” Royal Society o f Chemistry, London, 1988. pp. 127 and 132. Sturgeon, R. E,., Willie, S . N., and Berman, S . S . , J . Anul. At. Spectrorn., 1986, 1, 115. Sturgeon, R. E., Bcrman, S . S . , Willie, S . N . . and Desaulniers, J . A. H., Anal. Chem., 1981, 53, 2337. Beauchemin, D., Rednas. M. E., Berman, S. S., McLxen, J. W., Siu, K. W. M., and Sturgeon, R. E., Anal. Chrm., 1988, 60, 2209. Crccelius, E. A., Environ. Health Perspect., 1977, 19, 147. Siu, K. W. M., Roberts, S. Y., and Berman, S. S . , Chrornato- graphia, 1984. 19, 398. Edmonds, J . S., and Francesconi, K. A., Exprrientiu, 1987,43, 553. Paper 9/01 633C Received April 18th, 1989 Accepted June 14th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891401393

出版商:RSC    

年代:1989

数据来源: RSC