|
11. |
Papers in future issues |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 170-170
Preview
|
PDF (66KB)
|
|
摘要:
170 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Future Issues will Include- Application of Inductively Coupled Plasma Source Mass Spectrometry (ICP- MS) to the Determination of Trace Metals in Organics-Robert C. Hutton Determination of Ionic Alkyllead in Water by Electrothermal Atomisation Atomic Absorption Spectrometry- Dipankar Chakraborti, Rudy J. A. Van Cleuvenbergen and Fred C. Adams An Overview of Recent Developments in the Determination of Aluminium in Serum by Furnace Atomic Absorption Spec trome t r y-W a1 ter Slavin Development of a Slurry Technique for the Determination of Cadmium in Dried Foods by Electrothermal Atomisation Atomic Absorption Spectrometry- Kehinde 0. Olayinka, Stephen J. Haswell and Roman Grzeskowiak A Variable Dispersion Flow Injection Manifold for Calibration and Sample Dilution in Flame Atomic Absorption Spectrometry-Julian F.Tyson, James R. Mariara and John M. Appleton Aerosol Deposition - Carbon Furnace Atomisation for Simultaneous Multi- element Atomic Absorption Spec- trometry-James M. Harnly Electrothermal Atomisation Atomic Absorption Conditions for Determining Ag, As, Au, Bi, Cd, Ga, In, Pb, Sb, Se, Sn, Te, Pt, Pd and Mo following Back- extraction of Organometallic Halide Extracts-J. Robert Clark Poly(dithi0carbamate) Chelating Resin Decomposition Procedures-R. S. Shreedhara Murthy, Zs. Horvath and Ramon M. Barnes A Comparison of Inductively Coupled Plasma Torch - Sample Introduction Con- figurations Using Simplex Optimisation- Robert Carpenter and Les Ebdon Photon Induced Fluorescence Cross- sections for K Shell X-ray Lines- Chander Bhan, Balwan Singh and N. Nath Fluoride Influence on the Solute Trans- port Efficiency of Boron in Methanolic Solutions-A. Canals, V. Hernandis and J. V. Sala Determination of Selenium in Steel by Flame Atomic Absorption Spectro- metry-Ivan JanouSek Atomic Spectrometry Update The Update in the August issue is- Chemicals, Iron, Steel and Non-ferrous Metals-David Littlejohn, Howard J. Ellis and Hugh Hughes
ISSN:0267-9477
DOI:10.1039/JA9860100170
出版商:RSC
年代:1986
数据来源: RSC
|
12. |
Thermal vaporisation for inductively coupled plasma optical emission spectrometry. A review |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 171-184
Henryk Matusiewicz,
Preview
|
PDF (2022KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 171 Thermal Vaporisation for Inductively Coupled Plasma Optical Emission Spectrometry A Review* Henryk Matusiewicz Technical University of Poznan, Department of Analytical Chemistry, 60-965 Poznan, Poland Summary of Contents 1 Introduction 2 Nomenclature 3 The rm a I Va po ri sa t i on Introduction Systems 3.1. Electrothermal Vaporisation (Em) 3.1.1. Em: metal heating devices 3.1.2. En/: graphite heating devices 3.2. Direct Sample Insertion Techniques (DSIT) 3.3. Solid Sampling Systems 3.3.1. Arc nebulisation/vaporisation 3.3.2. Spark nebulisation/vaporisation 3.3.3. Laser ablationhaporisation 4 Electrodeposition 5 Speciation 6 Summary of Instrumentation 7 Detection Limits 8 Applications 9 Conclusions 10 Suggestions for Future Studies 11 References Keywords: Thermal vaporisation; sample introduction; inductively coupled plasma optical emission spectrometrv; sensitivity; applications 1.Introduction In 1964 a stable energy source, the inductively coupled plasma (ICP), was suggested by Greenfield et al.1 and Wendt and Fassel,2 independently, for the generation of free atoms and the excitation of these released atoms. Over the last two decades, the ICP has been well characterised as a sensitive source for optical emission spectrometry (OES). After con- sidering all of the characteristics of ICP-OES that make the technique as popular as it is, one problem area that remains unresolved is the sample introduction process, particularly with regard to micro-sampling, which is crucial whenever sample size is limited.There is general agreement that the various aerosol generation techniques constitute the weakest link in ICP-OES. The modes of sample introduction into an ICP have been conveniently summarised by Fassel3 and Browner and Boorn.4 Traditionally, sample introduction has been effected by using pneumatic nebulisation. The popularity of nebulisation owes much to its simplicity, rapid sample changeover, relatively good stability and low cost. Negative aspects of the technique include low sample introduction efficiency, nebu- liser blockage problems and the requirement for sample volumes of greater than 1 ml. Overall, the most important disadvantage of pneumatic nebulisation as used for ICP-OES is poor sample transport efficiency, typically about 1 YO compared with 5-10% efficiency with flame atomic absorption spectrometry (FAAS).Whilst this causes no problems in routine work, in certain applications such as biochemical, clinical, forensic, environmental, toxicological and solid state, * This paper is a habilitating dissertation to be sumbitted to the Department of Chemistry, University of Warsaw, in partial fulfilment of the requirements for a degree of Doctor of Science (Assistant Professor) in Analytical Chemistry. micro- and/or ultra-trace analyses are however required. Microlitre volumes of solution can be nebulised reproducibily , relatively simply and conveniently (discrete sample nebulis- ation), as reviewed by Cresser,s the ICP source having been originally developed for the analysis of liquid samples only.In these situations conventional pneumatic nebulisation is inap- propriate. It is for this reason that an increasing number of researchers have developed and used thermal devices to vaporise micro-volumes of solutions, and to some extent solid and powder samples, into the plasma. Kantor6 and Ng and Carus07 have recently reviewed the important aspects and techniques of electrothermal vaporisation for sample intro- duction in atomic emission spectrometry in general. This review will discuss the current state and historical development of thermal vaporisation (TV) for sample intro- duction in ICP-OES and summarise past and possible future applications of the technique and the analytical aspects of this relatively new method. “Relatively new” here implies that the method has aroused general interest only in the last 5-8 years.2. Nomenclature Throughout this review the term thermal vaporisation will be used whenever thermal, electrothermal, electrical, direct or even inductive vaporisation are the single high temperature sources used to produce the appropriate free species for spectrometric detection. This author can see no ambiguity in the use of the expression, although electrothermal vaporis- ation or direct vaporisation may be considered by some to be more specific terms. It should be pointed out that terms such as electrothermal atomisation, thermal vaporisation, evapo- ration technique, plasma volatilisation and electrothermal volatilisation have been used by different authors and the expressions direct sample insertion device and direct sample introduction - vaporisation have also been suggested.Thus,172 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 when preparing this review paper on this particular subject it was necessary to choose arbitrarily from the terms used by other authors. This author prefers to see the term thermal vaporisation confined to where two separate systems are used, one to vaporise the sample and the other as the observation cell. The choice of the term vaporiser rather than atomiser throughout this paper is deliberate, as the main requirement for the thermal device in conjunction with the ICP is to produce a vapour or dry aerosol suitable for transport at atmospheric pressure by the carrier gas to the ICP source.For these reasons, the expression electrothermal vaporiser sugges- ted by several authors, will be used here. This is in marked contrast to the requirement in graphite furnace atomic absorption spectrometry (GFAAS) where the graphite fur- nace must produce a transient cloud of atoms as a precursor to absorption and is consequently described by the generally accepted term electrothermal atomiser. 3. Thermal Vaporisation Introduction Systems Several approaches to TV sample introduction for the ICP have been reported. These include electrothermal vaporis- ation, direct sample insertion and solid sampling systems. In the literature survey presented here, the main characteristics of each sample introduction method will be emphasised. A brief discussion of sub-groups of these methods, identified by the vaporisation unit, will be presented.3.1. Electrothermal Vaporisation (ETV) One of the most common techniques for converting liquid micro-samples into dry-vapour aerosols and introducing them into the ICP is electrothermal vaporisation (ETV). This approach involves the direct production of a dry-vapour aerosol from a micro-volume of solution (e.g., 5-20 pl) deposited on a metal filament or on a graphite electrode, a technique that has become well developed for GFAAS. Modified commercial electrothermal atomisers manufactured for AAS have been used with the ICP. The heating pro- gramme may include solvent evaporation, ashing and vapori- sation steps, but in contrast to GFAAS, ashing may be unnecessary and the final temperature may be lower if the sample is completely removed in molecular form.The resulting aerosol - vapour is transported into the ICP in a carrier gas stream, generally, by a minimum length of connecting plastic or glass tubing. 3.1.1. E TV: metal heating devices The first reported work combining electrothermal sample vaporisation with ICP atomisation and excitation was by Nixon et a1.,8 who used a tantalum filament. The sample was introduced into a small depression in a tantalum strip contained in a quartz dome with a volume of ca. 120 ml. The dome was provided with an inlet for argon (1.2 1 min-1) at the base and an outlet port at the top connected to the sample introduction orifice of the plasma torch. The length and material of the transfer tube were not detailed and the possibility of deposition of elements on the tube walls was not examined.Detection limits in the pg 1-1 to fractional pg 1-l range were obtained for 16 elements from 100-p1 samples. These workers considered the main advantage of this type of introduction system to be the fact that a single set of paramaters should suffice for the vaporisation of many types of sample, therefore the potential for multi-element trace analysis in real samples was considered promising. Smytheg briefly described a multi-coil tungsten filament passing through a small graphite furnace for the introduction of aqueous micro-samples (1-5 pl) into the ICP. Inter-element effects, matrix effects and real sample applications were not examined and discussed. KitazumelO reported filament (primarily platinum and tungsten) vaporisation of 10-p1 samples for ICP-OES.In this technique, the sample solution was vaporised on a filament heated by a momentary condenser discharge in a small quartz evaporation chamber (ca. 4.5 ml in volume). The vaporised specimen was introduced into the ICP torch through poly- propylene tubing (20 cm) and a three-way stopcock. Detection limits for B, Ge, P, Pb, Sn and Zn were measured and the effects of Na, K and Li on analyte emissions were briefly studied. Tikkanen and Niemczykll" have described the incorpora- tion of an ETV system into a commercial ICP direct reader based on the concept of rapid vaporisation of the sample solution (5 pl) from a tantalum boat. The distance the sample must travel from the tantalum boat to the plasma was kept rather short, ca.30 cm. The system could readily be switched between the ETV mode and the conventional pneumatic nebulisation mode of sample introduction, However, they were limited to monitoring a single analyte channel for each firing of the ETV. A further extension of Tikkanen and Niemczyk's worklla has very recently been reported.11hJc In the first paperllb they detailed the manner in which simultaneous signal versus time profiles could be obtained. When a multi-element solution is introduced to an ETV-TCP system, the retardation of plasma appearance time with decreasing volatility for the various components is demonstrated. However, they used a multi-element solution of the trace elements at a rather high concentration, 1 p.p.m. of each element, which is an artificial situation and does not reflect the real trace element content of different materials.In the latter paperllc the use of an ETV system to sequence in time the arrival of various components of a sample into an ICP direct-reader system is discussed. By using a multi-channel time gated detection sequence, they demonstrated the elimination of some well-documented spectral interferences noted for A1 on As determinations and the easily ionised element Na (as the chloride or the sulphate salt) on Fe, Mn and Pb determinations. In this reviewer's opinion, it is a good start and one of the approaches that can be utilised to eliminate some interferences, as long as the introduction of the analyte and the interferent can be separated in time.However, in these results, the ratios of analyte to interferent, e.g., Fe in the presence of Na, range fromcomparable to at most about 200-fold higher, andone must therefore question the value of this approach, especially as the ratios of Na to analyte in natural matrices (e.g., biological or clinical samples) are much higher, for example, 1 + 3500 for Fe and Na in serum, respectively. In addition, these workers did not take into account natural ratios and did not present any acutal matrix effects on real sample applications to demonstrate the practical utility of the time-gating concept. At this stage of development, only a limited improvement in detection limit has been reported for the vaporisation of practical samples from the metal surfaces of electrothermal devices in the ICP, compared with conventional nebulisation.3.1.2. ETV: graphite heating devices Although tantalum filaments were initially employed, the most widely used and preferred material for constructing vaporisation cells is carbon, for example graphite or pyrolytic- ally coated graphite. An early approach by Dahlquistl2J3 used graphite yarn as the substrate for the vaporisation of liquid micro-samples (5-50 pl) into an ICP. No investigations were reported concerning matrix and residue effects resulting from real samples or inter-element effects and other interferences. This technique has been applied with a commercial spectrometer (Plasma-Spec Spectrometer, Leeman Labs. Inc.). I 4 The operation of this equipment is based on the electrothermal drying of small liquid sample aliquots on a graphite filament prior to analyte vaporisation and excitation in an ICP.173 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Although early attempts to use ETV for ICP sample introduction showed promise, the efforts of Kirkbright and co-workers clearly established the features and limitations of this combination.A graphite rod ETV device, contained in a 1-1 cylindrical glass manifold, has been used15-18 for the introduction of microlitre (10 1.11) liquid samples into an ICP via a polyethene connecting tube (0.5 m). They examined the following parameters: sample transport, effect of viewing height and plasma operating power, detection limits and precision. They concluded that the transport systems had good capabilities for distances up to 20 m.In spite of the dilution of pg amounts of vaporised sample in the large-volume manifold, adequate signals were produced in the plasma. In later they continued the investigation of matrix, inter- element and sample transport effects. The studies mainly concerned the effects of matrix and concomitant elements on the determination of As and Cd and variations in associated pre-vaporisation loss and transport efficiency of the analyte. These workers concluded that the primary interference was actually caused by pre-vaporisation losses of analyte, depo- sition of analyte during transport to the plasma and the formation of refractory carbides with the graphite rod. They23 circumvented this latter problem by the addition of a halocarbon (0.1% trifluoromethane) to the argon carrier gas to form preferentially the volatile halides.This resulted in an improvement in detection limits to the sub-nanogram level for elements such as B, Cr, Mo, W and Zr. It was proposed that this approach would be suitable for multi-element investi- gation. It should be noted that the studies23 were carried out with aqueous solutions and there were no studies made using real samples. Dean et a1.23a have assessed the accuracy of graphite rod vaporisation (GRV), the same device that was used by Kirkbright and co-~orkers,15~21 for sample introduction into an ICP. Based on the results for the concentration of Ag, Cd, Cu, Mn and Pb in Bowen’s kale they judged that GRV for sample introduction into an axially oriented plasma *as an accurate technique.The method of standard additions was used to compensate for evident matrix effects. A similar design to that of Nixon et a1.8 and Gunn et a1.I5 was described by Ng and Caruso.24 This system involved vaporising the sample electrothermally in a carbon cup followed by atomisation and excitation of the vapour cloud in an ICP. Compromise conditions were used for the ICP but the furnace conditions were varied from element to element. The electrothermal vaporiser assembly (a glass dome with an inner volume of ca. 280 ml) was positioned directly underneath the ICP torch. The two were connected via 18 cm of PTFE tubing. The detection limits reported for 21 elements in 10 1-11 of aqueous sample are at the ng ml-1 and sub-ng ml-1 level. Pyrolytic graphite and impregnation of the graphite sample holder with tantalum salts was found to be advantageous for certain elements. In later papers they continued the investi- gation of matrix effects in synthetic ocean water25 and the improvements obtained by the preferential formation of the halides of refractory elements (Cr, U, V and Zr) in the electrothermal carbon cup,26 and they then utilised their system for the introduction of organic solvents into a low-power ICP.27 In the latter paper for example, the addition of iodine to spiked gasoline samples allowed tetramethyllead and tetraethyllead to be determined without difficulty.Combining an ICP or an ETV source with mass spec- trometry (MS) provides a convenient method for the introduc- tion of liquid or solid samples and is particularly attractive for isotope ratio determinations on small samples.A system similar to that described by Gunn et a1.15 has been used to introduce 5-1.11 samples28 into a mass spectrometer. The sample was desolvated in the usual way at low temperature and then vaporised into the injector gas flow. A short pulse of ions lasting a few seconds was obtained. As scan times of as little as 20 ms may be used, this provides an ample number of scans over the changing signal to enable isotope ratio measurements to be made. Excellent agreement, according to these workers, has been obtained between isotope ratios determined on microsamples and nebulised solutions. Efforts to improve ETV devices for introducing samples into an ICP in recent years have been directed towards increasing the efficiency of sample transport into the plasma and as a consequence improving detection limits.Many versions of ETV devices have been explored for use with the ICP. Several analytical researchers have modified Perkin- Elmer (PE) graphite furnaces for micro-volume sample introduction into an ICP. Crabi et aL29 modified a PE HGA-500 graphite furnace and used it with a L’vov platform to introduce samples into an ICP. They stated that background correction is necessary. In a later paper in 1985, Casetta et al.30 continued this line of investi- gation using similar instrumentation (HGA-500) for the determination of sulphur in solid rubber. They used the graphite tubes without holes and therefore overcame the need29 to enclose the furnace head for collecting sample vapours (the dead volume of the vaporisation cell is also reduced).Recently, Christian and CO-workers31-33 modified a PE HGA-2000 graphite furnace and transported the sample aerosol through 20 cm of tubing to a spray chamber, which was connected to an ICP torch. They used this system for single31 and simultaneous multi-element32J3 analysis of aqueous solutions, and obtained rather poor detection limits probably caused by the loss of vapours in the spray chamber. It should be recognised that internal standards were ~ s e d 3 ~ to correct for errors due to changes in the flow-rate of the argon carrier gas, the observation zone, the observation period, the sample volume and graphite tube deterioration. Aziz et al.34 used a PE HGA-74 graphite furnace with a special aerosol transport system that transports the sample aerosol through a 30-cm glass tube to the base of an ICP torch.They have examined the matrix effects from biological samples on analyte emission and generally found them to be significant, which required the use of a standard-additions technique. Evaluation of a novel configuration and a new furnace design for ETV-ICP was reported by Matusiewicz et a1.35 Modification of a PE HGA-500 furnace, which allowed vertical mounting of the graphite tube, insertion of a graphite cuvette and a direct, shortest practical connection to the base of an ICP torch was described. The operational characteristics including the effect of transport tube length to the ICP torch, vaporisation temperature, carrier argon flow-rate, obser- vation height above the coil and plasma power have been investigated.The effects of major matrix constituents (Ca, Fe, K, Mg, Na and P) on the determination of trace elements (Be, Cd, Co, Cu, Mn, Pb and Zn) by ETV-ICP were also investigated. It was found that significant enhancement and/or suppression of the analyte emission occurs in the presence of these major (matrix) elements. Unfortunately, no results for actual analyses were presented to demonstrate the practical uses of the system. Also, even though matrix effects have been characterised, there was no suggestion about how to overcome them by, for example, altering furnace or plasma conditions. Other workers have modified Varian carbon rod atomisers, which were combined with glass or quartz chambers, for ICP sample introduction.Some of them have been discussed previously in this revie~.15-2~,~%2~ In a novel approach, commercial ETV electrodes were replaced by an optimised electrode and a novel double-walled quartz chamber (ca. 30-ml volume) was added to prevent analyte aerosol losses owing to leakage and to reduce pressure surges produced during heating of the argon carrier gas.36 The chamber exhaust was attached to the ICP torch aerosol tube by means of a 1 m length of Tygon tubing. A number of significant modifications have been made37 since the original design was174 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 described.36 These include: (a) reducing the chamber volume from ca. 30 to 7 ml, damping the negative background transient signal considerably, (6) reducing the argon flow- rates, (c) connecting the chamber to the ICP torch through a glass tube (55 cm long) and (d) modifying the graphite electrode dimensions so that they are similar to the design of conventional graphite electrodes employed for d.c.arc spectrography . More recently, a Varian carbon rod atomiser CRA-63 has been modified38 to give a system that consists of a normal Varian type carbon rod - cup (the upper half of the cup was cut off and discarded) enclosed in a 50-ml quartz cell between water-cooled electrode holding blocks, in a manner similar to that described by Barnes and Fodor.36 A 30-cm intermediate tube was used to connect the cell to the ICP torch. This relatively simple and rather large cell (chamber) is not likely to prevent the negative background signal (peak) caused by the pressure pulse created upon rapid argon heating.Other workers preferred to modify various models of Instrumentation Laboratory (IL) controlled-temperature fur- nace atomisers (CTF). The ETV graphite furnace Model IL655 CTF was modified for ICP application by Matusiewicz and Barnes.39 Some of the existing designs have the drawback of being of large scale8,15>24J6,38 leading to the problem of “dead space” and to memory effects owing to the plating-out of the analyte on the cold surfaces. The CTF chamber was adapted so that an argon flow swept the furnace and carried the aerosol to the ICP. An important feature of this design is the very small internal volume (0.8 ml) of the graphite cuvette used as the vaporisation chamber.A small vaporisation chamber volume minimises the volume of hot argon gas produced during vaporisation as well as aerosol dilution, but the temperature must therefore be higher and the pressure pulse sharper. This reduces variations in ICP emission background from pressure pulses and broadening of the analyte temporal emission peak signal. A variety of graphite cuvette geometries with and without microboats and a new carbon tube - platform arrangement were examined.40 The same workers extended their investigations to the evaluation of discrete nebulisation using aerosol deposition into the furance as a procedure for sample introduction in ETV for ICP-AES using commercial instruments.41 Small volumes of solution (ca.50 pl) were introduced manually from a PTFE microsampling device or f ~ n n e l , ~ l a or automatically by a flame sampler system into a pneumatic nebuliser and deposited under controlled conditions on the surface of a graphite platform. The entire system could be easily automated; however, automation was not adopted for routine use. No sample preparation was required. Matrix effects, although not completely eliminated by the above method, become more consistent from sample to sample. The multiple peaks often observed in ETV-ICPlOJ6-29J4 can be eliminated by modifi- cation of the graphite tube. For example, use of a contoured tube provided for more even heating of the tube ends resulting in a single pulse of sample vapour.42 Blakemore et aZ.43 recently used a modified IL555 atomiser for sample introduction with the ICP in multi-channel mode.The carbon rod was used in place of the graphite tube and samples were weighed or injected in IL pyrolytic microboats and placed on the flat portion of the carbon rod. The furnace was mounted in the ICP below the quartz torch so that only 12 cm of PTFE tubing were required to connect them. Heating of argon in the rather large cell, however, resulted in variations in background emission throughout the heating cycle with a consequent loss of sensitivity. Also, these workers stated that no ashing step was needed for the direct simultaneous determination of major and trace elements in biological materials. However, this statement was not confirmed experimentally and matrix effects caused by major elements were not studied. No detection limit data were listed and data show relatively large uncertainties for certain analyte - matrix combinations.It is clear that an electrothermal vaporiser can be easily incorporated as an accessory into an existing ICP system as has been suggested,3+4* and indeed commercial accessories are now available. An Allied Analytical Systems EVA (elec- trothermal vaporisation accessory) system, a graphite furnace - aerosol deposition system combined with the ICP 9000 simultaneous plasma emission spectrometer, was presented at the 1985 Pittsburgh Conference.44,45 The design has adopted suggestions made in the literature,3+42 although the manu- facturer used graphite tubes instead of platforms or contoured tubes for sample vaporisation and long tubing (ca.80 cm) connecting the IL graphite furnace with the ICP torch, which can cause transport losses. The furnace could have been mounted in the ICP box below the quartz torch so that only a few centimetres of tubing would be required to connect them, as has been shown previously.43 Very recently, a versatile accessory for ICPs, an electrothermal vaporiser specially designed for use with ICPs and ICP-MS has been announced by PSA Analytical Ltd., Orpington, Kent, UK. 3.2. Direct Sample Insertion Techniques (DSIT) The use of a direct sample insertion graphite and/or metal device for introducing small amounts of liquids or solids into the ICP is an alternative to the ETV technique. The sample is placed in a graphite rod, the top of which is hollowed out to form a cup, or on a metal wire-loop. This rod or loop is then inserted manually, mechanically or pneumatically, instead of the gas injection tube of the plasma torch, directly into the lower portion of the plasma.Direct inductive and thermal heating of the cup or wire-loop occurs and direct vaporisation and excitation of the material ( e . g . , 5 pl or 0.5-20 mg) into the plasma can take place with high efficiency. Transport losses do not occur as in the ETV technique where the generated aerosol must be transported into the plasma by a stream of carrier gas. By suitable control of the height of the electrode solvent drying and sample (liquid or solid) ashing steps may be incorporated prior to the final vaporisation step. An early approach by Kleinmann and Svoboda46 used a 30-p1 sample that was evaporated by ohmic heating from a graphite support and excited by a low-voltage, high-frequency inductively coupled discharge in argon.The sample was placed on top of graphite supports with tantalum holders inside a discharge chamber. Detection limits were measured for 15 elements, but the method was not a practical prop- osition for routine analyses of real samples. More recently, Salin and Horlick4’ have developed an elegant direct sample insertion device (DSID) , which can handle solids (1-20 mg), powders and liquid samples (5 pl). The device consists of a conventional d.c. arc electrode drilled axially to form a cup and the insertion of the electrode into the load coils is used to ignite the plasma for each sample.A disadvantage in their experiments was that the plasma had to be ignited and extinguished on each insertion. Sommer and Ohls48.49 achieved continuous operation of the ICP with a similar system dubbed the “sample elevator technique.” They used a hydraulic ram device with a graphite crucible on a silica rod to introduce small samples (10 mg or 10 pl) directly into a Greenfield type torch supporting an Ar - N2 ICP at a power of 3 kW. Compromise conditions for routine simultaneous multi-element analysis by ICP were reported. Kirkbright and Waltonso and Kirkbright and Li-Xingsl were also able to introduce sample solutions into a low-power (1.5 kW), continuously running Ar ICP via the injector tube of a demountable torch. They deposited 5 yl of solution on to a flat-ended graphite rod and used a heat gun to dry the sample prior to insertion into the plasma.Significant improvements in the detection limits over the pneumatic nebulisation technique were obtained. Li-Xing et al.52 developed a microprocessor-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 175 controlled system for the direct introduction of a graphite rod into an ICP operated in a demountable torch. Provisions for drying and ashing steps were made. Automation of the graphite rod insertion device, in the manner described, improves the precision attainable by this technique. The system was shown to be suitable for trace element determi- nation in small liquid samples or directly with small solid samples. The d.c. arc standard carrier distillation procedures3 has been adapted for selective volatilisation of analyte elements from a refractory matrix by injecting analyte vapours into the central zone of the plasma using Scribner - Mullin electrodes covered with a graphite lid having a central opening.54 The results indicate that the carrier - distillation - ICP combination shows promise as a sensitive and precise method for analysis of trace metallic elements in a refractory matrix with a complex spectrum.The direct insertion of a graphite sample cup55356 similar to the previously described ~ystem5~ and the use of a graphite lid on the top of the cup57 has been investigated for the analysis of microlitre volumes of liquid samples. The solution was dried and ashed inside the injector tube below the ICP and then the cup was introduced into a continuously operating Ar ICP for sample vaporisation.The univariate search method was employed for optimisation of cup size, rod height and plasma operating parameters and the technique was optimised for simultaneous multi-element analysis. Most previous applications of DSID used a graphite electrode, although Salin and Sings* have reported using a wire-loop system suitable for microsample liquid (10 p1) analysis. The sample insertion system was pneumatically powered by a separate argon gas cylinder. The graphite electrode used in previous experiments47.48 has been replaced by a tungsten wire with two vertical loops on the end. The atom population is produced more rapidly by the wire-loop system because of its low thermal mass and the direct exposure of the sample to the argon plasma.The method of Salin and Sing, although not suggested directly, could also be xdapted for the electrodeposition technique. The trace metals could be deposited on a tungsten electrode prior to excitation. 3.3. Solid Sampling System (SSS) Considerable impetus exists for the development of an ICP sample introduction technique that can be used directly with small amounts of solid samples without prior dissolution. Several aproaches have been described in the literature. Solid sampling will now be discussed; reference 59 only discusses solutions, not solids. 3.3.1. Arc nebulisationlvaporisation A high-voltage interrupted arc has been used for producing vapours to be transported by the argon carrier gas flow into the ICP torch of a high power (3.3 kW) N2 - Ar ICP sy~tem.49~60 The vapours were produced by using a simple flow chamber containing an electrode gap.For metallic samples or briquetted mixtures of Cu powder and oxides, a point-to- plane technique with a graphite counter electrode was used. Oxidic materials were placed in a graphite cup electrode. Many elements could be determined directly without special sample preparation. The “microarc,” a high-voltage, low-current, pulsating d.c. arc was developed59 to vaporise efficiently discrete micro- volumes of sample solution (0.1-10 p1 range) from a tungsten wire-loop into the plasma. Desolvation of sample solution deposited on the cathode was accomplished by ohmic heating of the cathode wire prior to arcing.Matrix and ionisation interference effects were absent or easily overcome. In a later paper, Farnsworth and Hieftje61 described a radiofrequency arc technique for the analysis of solids and microsamples by ICP-OES. The new method parallels the direct insertion methods mentioned above; however, instead of the sample being transported to the plasma, the plasma is brought to the sample. The r.f. arc technique requires the positioning of an electrode below the plasma inside the sample tube of a modified torch. The quartz torch differs from conventional ICP torches in that its central sample tube is flared at the base to form a small bell jar. A brief discharge of a Tesla coil initiates an arc between the plasma and the electrically grounded electrode.The method is capable of sampling a wide range of elements in solid form and can also be applied to solution microsampling. Internal standardisation and signal integration were not employed to counter the effects of preferential vaporisation of elements and instabilities of the arc. 3 -3.2. Spark nebulisationlvaporisation The use of conventional high-voltage spark elutriation oper- ating at 50 Hz, combined with ICP-AES was first described by Human et a1.62 and applied to the analysis of compact metallic samples. Argon gas, fed through the spark chamber, trans- ports the solid (metal) particles into the plasma. In a later paper, continued the investigation of spark elutriation of powder into an ICP. The method used a high-voltage controlled-waveform spark discharge between two graphite electrodes situated above the powder sample, which was placed in a glass or plastic sample container (vial). In this method thermal dispersion (which is based on the formation of a highly dispersed aerosol6) may be involved to some extent.Recently, Aziz et al.64 used a medium-voltage spark for the direct nebulisation of compact metallic samples and non- conducting powders in which the elutriated material was excited in a high-power (3 kW) Ar - N2 ICP. The length of the transport tube may be increased from 0.5 to 6 m without deterioration in the detection limits by more than a factor of two. From a characterisation of the aerosol, it was found that limitations on the power of detection were due to the low number of small particles arriving in the ICP.The influence of the various working parameters on the detection limit were investigated, i.e.. , voltage, capacity, repetition rate, argon carrier gas flow-rate, ICP power, length of transport tube and observation height. It should be mentioned here that Jarrell-Ash introduced the first ICP spectrometer and sampling system for direct analysis of solid samples.65 This solid sampling system uses a high- voltage electronically-controlled waveform spark that pro- vides extremely rapid and precise sampling of solids. A series of high-voltage sparks erode the surface of conducting and non-conducting solid samples, producing a representative solid aerosol. A stream of argon carries the solid aerosol into the ICP torch for excitation and analysis. The above mentioned commercially available solid sam- pling device has been utilised in a multi-element emission spectrometer with an ICP source for the direct analysis of complex nickel alloys.66 The incorporation of the solid sampling device required only minor modifications to the instrument, and alternating between solid and solution sampling was rapidly and easily accomplished.In this method, as opposed to other analogous combined methods, the sample aerosol is transported from the spark sampling chamber, not to the ICP directly, but first to an intermediate chamber where the aerosol is additionally nebulised together with water and introduced into the injection tube of the torch by another independent argon flow. Thus, uniform nebulisation and the introduction of the homogeneous, finely dispersed, alloy aerosol fraction are accomplished simultaneously.Recently, Applied Research Laboratories described a conductive-solids nebuliser (CSN) to be used with their ICP instruments in the analysis of metals and other conductive samples.67.68 The CSN system, which is a very similar design to that of Jarrell-Ash ,65 uses a spark discharge source that sputters material from the surface of the solid metal sample.176 JOURNAL OF The aerosol formed is swept by the argon stream into the plasma, where the aerosol is vaporised and excited. 3.3.3. Laser ablationhaporisation Laser ablation has the advantage of being able to vaporise both conductive and non-conductive solid materials. In addition, because of the excellent focusing characteristics of a laser beam it may be used in microprobe analysis of the sample surface. Discrete sample injection by laser ablation has been conveniently summarised by Thompson and Walsh69; they include descriptions of the Lasertrace, LMAlO Laser-ICP microprobe, and the analysis of soil, rock and discrete mineral particles. Laser ablation as a means of volatilising material for subsequent injection into the ICP was first described by Abercrombie et aZ.70 in a system specially constructed for the analysis of airborne particulates collected on adhesive tape.They utilised the prototype of the first commercially available ICP-AES system, the ARL ICPQA137 instrument. Thompson and co-workers69~71-73 have developed a laser - ICP system for the analysis of solid samples with appreciable success.They combined a commercially available LMAlO Laser Microspectral Analyzer (Carl Zeiss Jena, GDR) and an ICP spectrometer, the ARL 34OOOClICP (Applied Research Laboratories, Sunland, CA, USA) for multi-element analysis. The interface between the laser ablation glass cell and the ICP torch was made with flexible PVC tubing. Steel discs and chips were vaporised by a ruby laser. Recently, they demonstrated the potential for quantitative heavy mineral grain analysis with the LMA10-ICP system? ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Carr and Horlick74 briefly reviewed laser vaporisation and described the design and capabilities of a laser vaporisation system developed in their laboratory. To ensure that the connection of the sample chamber to the ICP torch was kept as short as possible, the system was constructed so that the glass sample chamber fits inside a shielded plasma chamber directly below the torch. A ruby laser was employed to vaporise aluminium and copper alloys.Kawaguchi et aZ.75 constructed a laser - ICP microprobe system by employing a low-energy Nd : YAG laser (maximum 0.1 J per pulse) and a commercial laser optical system. The length and diameter of the vapour-carrying tube between the sample chamber and the ICP torch effected not only emission signal responses but also reproducibility of intensities. Linear working curves were obtained for Cr, Mn and Ni in low-alloy steels, but the analysis of copper alloys was unsuccessful because the amounts of samples evaporated changed con- siderably with their composition.For the purpose of constructing an optimum laser ablation cell - ICP torch system, Ishizuka and Uwamino76 investigated the effects of varying the length of PVC tubing interface between the laser ablation cell and the ICP torch, the effect of varying amounts of laser-ablated sample (by varying the laser focusing) and analytical practicality. A ruby laser was used and it was found that the interface tubing should be kept as short as possible. In ail instances69.71-76 the emission signals of various elements were measured with multi-channel analysers. Very recently, Gray77 described some preliminary studies using a ruby laser to ablate the sample into an ICP source for mass spectrometry. Standard rock samples were studied and Table 1.Operating parameters for TV-ICP studies A. ICP instrument and operating parameters- ICP system Power/ kw Ebert mount, 0.5-m Jarrell-Ash 82000 . . . . . . . . . . . . . . . . 1 Hilger monospek 1000 . . . . . . . . . . . . . . . . . . . . 1 Jarrell-Ash 975 . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Jarrell-Ash 965 Plasma Atom Comp . . . . . . . . . . . . . . . . 1 Hilger Monospek 1000 . . . . . . . . . . . . . . . . . . . . 1 Spex Industries 0.85-m 1402 . . . . . . . . . . . . . . . . . . 1 Spex Industries 1.26-m 1269 . . . . . . . . . . . . . . . . . . 1.2 Jarrell-Ash 955 Plasma Atom Comp . . . . . . . . . . . . . . . . 1 0.9-m Czerny - Turner monochromator . . . . . . . . . . . . . . 3 1-m Czerny - Turner monochromator . . . . . . . . . . .. . . . . 0.35-1 1-m Czerny - Turner monochromator . . . . . . . . . . . . . . . . 0.55 McPherson 0.35-m Czerny - Turner system and photodiode array spectrometer 1-m Czerny - Turner monochromator . . . . . . . . . . . . . . . . 0.55 1.5 Observation height/mm 18 1 Jarrell-Ash 1160 Plasma Atom Comp . . . . . . . . . . . . . . Spectrograph . . . . . . . . . . . . . . . . . . . . . . Photodiode array, Heath EV-700 and Bausch & Lomb 2-m spectrograph ARL 15000 and Labtest V25 . . . . . . . . . . . . . . . . Monospek 1000 . . . . . . . . . . . . . . . . . . . . . . Jarrell-Ash 1100 Atom Comp . . . . . . . . . . . . . . . . Jarrell-Ash Mk I1 . . . . . . . . . . . . . . . . . . . . Jarrell-Ash Mk I1 . . . . . . . . . . . . . . . . . . . . Jarrell- Ash 1 -m spectrometer .. . . . . . . . . . . . . . . Kontron Plasmaspec 2000 . . . . . . . . . . . . . . . . . . GCA McPherson monochromator EU-700 . . . . . . . . . . . . 0.35-m McPherson monochromator EU-700 . . . . . . . . . . . . Jarrell-Ash monochromator 82000 . . . . . . . . . . . . . . Baird Atomic Spectrovac 1000 and 1-m Czerny - Turner monochromator Jarrell-Ash 975 Atom Comp . . . . . . . . . . . . . . . . ARL 3560 . . . . . . . . . . . . . . . . . . . . . . . . ARL ICPQA 137 . . . . . . . . . . . . . . . . . . . . ARL 34000 . . . . . . . . . . . . . . . . . . . . . . Plasma-Therm Inc. . . . . . . . . . . . . . . . . . . . . Shimadzu ICPS-2H . . . . . . . . . . . . . . . . . . . . Nippon Jarrell-Ash 1000s . . . . . . . . . . . . . . . . . . Plasma-Therm Inc., ICP 2500 . .. . . . . . . . . . . . . . Jarrell-Ash 1160 Plasma Atom Comp . . . . . . . . . . . . . . Jarrell-Ash 1100 Atom Comp . . . . . . . . . . . . . . . . . . 0.85 . . 1 . . 2 . . 1.5-3.0 . . 1 . . 1.08 . . 1.4 . . 0.9 . . 0.95 . . 1.75 . . 3.3 . . 1.25 - . . 1.5 . . 1 . . 3 . . 1.4 . . 1.2 . . 1.6 . . 1.25 . * 1.5 . . 1.2 . . 1.2 . . 1.2; 1.5 - . . - 16 20 20 18 14 10-20 3.5-18.5 - 16 16 15 6-10 - - - 10 19 6 15 16 16 16 - -2.5-7.5 15-22 20 4-8 18 15 16 14 - - 15 10; 12 Electrothermal atomiser device, electrode power supply or laser General Electric transformer Shandon Southern A3370 Condenser Varian-Techtron CRA-90 Shandon Southern A3370 Varian-Techtron CRA-63 Perkin-Elmer HGA-500 Perkin-Elmer HGA-2000 Perkin-Elmer HGA-74 Varian-Techtron CRA-90 Varian-Techtron CRA-90 Varian-Techtron CRA-63 Instrumentation Laboratory IL655 Instrumentation Laboratory IL555 Perkin-Elmer HGA-500 - - - - - - - - - Interrupted arc Microarc Radiofrequency arc Spark Spark Spark Spark C0,-TEA laser LMAlO ruby laser Ruby laser NEC Nd : YAG laser Ruby laser AJ.K.2000 ruby laserJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 177 Table l.-continued B. Thermal vaporisation sample introduction system- Type of electrode Tantalum filament Platinum filament Tantalum . . . . Graphiterod . . Carboncup . , L’vovplatform . . Graphite tube . . Graphitetube . . Graphiterod . . Carbonrod . . Tungsten . . . . Graphite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphite platform or microboat Carbon rod and microboat .. Graphite rod or platform . . Graphitesupport . . . . Graphiteelectrodes . . . . Graphiteelectrode . . . . Graphitecup . . . . . . Graphiterod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . Scribner - Mullin electrode with a lid Graphite cup . . . . . . . . Graphite cup . . . . . . . . Tungsten loop . . . . . . . . Graphite cup or material electrode Tungsten loop . . . . . . . . Graphite cup . . . . . . . . Materialelectrode . . . . . . Materialelectrode . . . . . . Materialelectrode . . . . . . Materialelectrode . . . . . . Vaporisation cell Sample size Carrier argon . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Tempera- tureiOC 1800 2300 1400 2400 2200-2900 2400 2400 1600-2500 2500 2500 2700 2800 - - - 1800 3000 - - - - - - - - - - - - - - - - - - - - Time/ S - 3 3 3.5 1.5 1.5 4 5-10 6 1.5 1 5 7 4 - - 12.5 5-30 - 30 45 30 15 0.1 5-20 5 9.2-55.2 2-5 min 20 40 - - - - - - - P1 1-200 1-10 10 5 10 10 20 10 50 5 5 15-20 5 5-10 3-5 30 5 10 5 - - 5-10 10 10 0.1-10 - - - - - - - 5 - - - - flow-rate/ 1 min-1 1.2 0.5 1 .o 0.8 0.8 0.4 1.2 1-2.5 4.5 0.8-1.6 0.6 0.8-1 .0 0.3-1.2 (optimum 0.5) 0.5 1.3-2.0 (optimum 1.65) 0.21 m3 hk] - - 0.2 0.6 1.0 1.4 0.1 0.8 0.7-1.0 1 .0 0.9 1 .0 3.5 0.5 0.8 0.9 0.5 0.8-1.2 0.8 1.1 0.5-1 .0 Sample transport tube length/ cm - - - 5 50 18 90 20 30 100 55 30 45 12 2 Reference 8 9 10 11 15 24 29 31 34 36 37 38 40 43 35 46 47 48 50 53 54 55 56 ’ 58 60 59 61 62 90 64 50 66 67 70 92 71 74 140 75 40 76 150 77 - - - - - - - - - - - - - - - - were pelleted with a binder into the form of a disc.Laser pulse energies of 0.3-1 J were used in the fixed Q mode and the ablated material transferred from the ablation cell into the plasma torch by means of the plasma injector gas flow. The mass spectrometer was used in the fixed ion mode using mean ion current detection to evaluate the reproducibility of pulses. It can be concluded that an ICP can be used in place of the spark source with laser ablation, resulting in an overall increase in performance. Judging from the precision data obtained in published works, the existing laser - ICP systems are still unsuitable for quantitative analysis of solid samples, although good precisions were demonstrated71 if the sample material was homogeneous.Internal standardisation or cor- rection by laser power measurement should be useful in improving precision; this might make the system capable of quantitative analysis. 4. Electrodeposition It has become quite common to apply electrochemical deposition as a pre-concentration and separation method for instrumental determinations of metals in complex matrices such as biological fluids, waters, food and mineral digests, and in general, in any samples in which there is a high content of interfering matrix elements. This method should be well suited for the determination of trace and ultratrace elements that are soluble in mercury. Also, the method should provide a good simultaneous separation - pre-concentration step prior to TV-ICP analysis.Electrodeposition on a thin film (layer) of mercury de- posited on a graphite rod electrode78 extended ICP capabili- ties. Copper was selected as a test element and electro- deposition was used, from an aqueous solution, with an apparatus functionally identical to the demountable torch system described previously.47 The goal was to pre- concentrate the sample electrochemically by cathodic depo- sition on to a graphite electrode prior to ICP analysis. Salin and Habib7* have recently reported79 a further extension to their work. The aim was an evaluation of the technique for the simultaneous determination of heavy metals (Co, Cd, Cu, Ni, Pb and Zn) in aqueous solution as well as an evaluation of the performance with a difficult sample matrix (artificial sea water).In the reviewer’s opinion, detection limits obtained for the elements studied were not as good as could have been expected. The application of controlled-potential electrolysis for the determination of trace metals in biological standard reference materials with both graphite electrodes previously coated with mercury80 and a hanging mercury drop electrodes1 as a separation and pre-concentration technique for ICP using ETV has been described.178 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 ~~ Table 2. Comparison of reported detection limits for TV-ICP, pneumatic nebulisation ICP and GFAAS TV-ICP Element Ag . . . . A1 . . . . As . . . . Au . . . . B . . . . Ba . . . . Be . . . . Bi . . . . Ca . .. . Cd . . . . c o . . . . Cr . . . . c u . . . . Fe . . . . Ga . . . . Ge . . . . Hg . . . . In . . . . K . . . . Li . . . . Mg . . . . Mn . . . . Mo . . . . Na . . . . Ni . . . . P . . . . Pb . . . . Sb . . . . Se . . . . Si . . . . Sn . . . . Sr . . . . Ti . _ . . T1 . . . . v . . . . Zn . . . . Zr . . . . ng ml-1 or ng g-’ , , 0.1-30 . . 1.5-13000 . . 20-5000 . . 1-5 . . 0.1-23200 . . 0.0003-20000 . . 0.02-4300 . . 2-10 . . 0.002-70000 . . 0.2-1640 . . 3-650 . . 0.3-790 . . 0.07-530 . . 2-9 560 . . 1 . . 1-6 . . 0.8-6 . . 2 . . 110-83400 . . 0.01-3 . . 0.01-4910 . . 0.002-1080 . . 10-100 . . 80-38000 . . 0.9-1050 . . 10-14800 . . 2- 1 400 . . 1-250 . . 6-600 . . 10-500 . . 2-1 250 . . 1-5 430 . . 1.15-52000 . . 5-86 . . 2-60 . . 0.05-1770 . . 1-4 Pg 1-300 7.5-26 000 200-10000 10-20 10-46 400 0.03-40 000 200-100 0.02-139000 1-3 280 2-8 600 30-1 300 1.5-1 580 0.35-1 060 20-19 100 10 10-60 4-60 20 550-167 000 0.05-20 0.1-9 820 0.01-2 160 100-1 000 400-76 000 4.5-2 100 100-29 600 20-2 800 100-2 500 600-3 000 100-2 500 20-2 500 5-10900 5.7-104 000 50-4 300 10-300 0.25-3 540 10-20 Reference* 15,33 37,57 24,57 15,25 8,57 8,57 8,57 8,24 18,57 40,57 24,57 40,57 40,57 18,57 18 24,lO 85,15 15 40,57 40,25 24,57 40,57 48,23 86,57 41,57 20,57 20,57 8,33 8,36 48,37 10,57 86,57 36,57 24,34 86,26 40,57 23,26 GFAASR3 1 ~ ~ 8 3 1 ng ml-I 6 3 60 30 3 0.15 0.15 0.015 1.5 3 1.5 1.5 6 9 1% 60 60 60 91 0.6 0.15 0.6 3 0.3 6 60P 6 151 60 9 30 0.06 1.5 151 6 1.5 3 ng ml-1-t 0.12 1.2 1.2 0.3 60 6 0.12 1.2 0.9 0.03 1.2 0.6 0.9 6 0.3$ 60 6 0.3 0.9 0.02 0.2 6 1.8 3 0.3$ 0.6 1.2 0.8 3 1.8 0.6 1200 3 12 0.02 - - Pgs 6 60 60 18 3 000 240 6 60 45 1.8 60 30 45 300 - - 3 000 240 15 45 1 10 300 90 150 15 30 60 90 150 90 30 60 000 150 600 1 - * The pairs of references refer to the lowest and highest reported detection limits, respectively t On the basis of a consumed sample volume of 50 pl.$ Authors calculation: 10/3.3 (30) based on given reported limit of determination ( 1 0 ~ ) . § Reference 84. 5. Speciation One of the more intriguing areas of atomic spectrometry at present is the possibility of deriving information regarding the various forms in which elements are chemically present in a sample. “Speciation” refers to the ability to discriminate between various forms in a mixture containing many chemical forms of the same element, or indeed several elements.Because chemical compounds have different vaporisation temperatures, by gradually raising the heating temperature of the TV system, the speciation of the analyte compounds is possible. In other words, with temperature programming of the furnace heating rate, it should be possible to distinguish between different chemical forms of the trace components of a sample, on the basis of their characteristic vaporisation temperatures. Controlled TV coupled with the ICP has the potential to extend the capabilities of ICP analysis. One area where this combination shows promise is the speciation of metals. However, in spite of very successful laboratory studies, there is very little widespread use of TV-ICP applied to organic and inorganic materials.To date, there has been only one published application of a direct sample insertion device (DSID) with an ICP to metal speciation.82 Prack and Bastiaans82 described an evolved gas analysis - emission spectrometer system capable of speciating inorganic compounds (vanadium compounds and Cd, Hg and Pb salts) in solid samples. Samples are gradually heated to 2300 “C in a graphite sample probe that is moved in a controlled manner into an ICP discharge, as described previously.47 As the sample is heated, its components vaporise, at characteristic temperatures, into the supporting Ar of the ICP, which acts as an atomic emission source from which the atomic emission from the evolved vapours can be detected. Quantitation was found to be possible.It could be suggested that if thermal and/or electrothermal furnace - rod - cup interfaces could be designed and developed that were truly continuous and on-line, with real-time analyte determinations, then such a pre-concentration and speciation method might readily provide outstanding detection limits and make practical applications possible. 6. Summary of Instrumentation Table 1 presents instrumental information and the method- ology commonly used in TV-ICP for atomic emission spec- trometry that has been discussed and summarises several ICP-OES systems.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 179 Table 3. Applications of TV-ICP systems to the analysis of waters Amount of Sample type Elements determined TV device mode Sample pre-treatment sample/$ Reference Acidicwatersolution .. Pb ETV-ICP Direct determination with 5 42 resin pre-concentration Wastewater . . . . . . Ca, Cd, Cu, Fe, K, Mg, ETV-ICP, carbon Direct determination 10 43 Syntheticseawater . . . . Cu DSID-ICP Direct determination with 25 ml 79 Na, P, Pb, Zn rod - microboat CUP Synthetic ocean water . . As, Au, Cd, Li, Sn, Zn ETV-ICP, carbon Direct determination 5 25 electrochemical pre- concentration and separation (NH4)*S modifier Drinkingwater . . . . Hg ETV-ICP Direct determination with 5 85 Table 4. Applications of TV-ICP systems to the analysis of mineral substances and technical products Sample type Elements determined Uranium . . . . . . . . Ag, Al, Be, Cd, U30,powder . . . . . . Cd, Cu, K, Na, Ni, Cr, Cu, Fe, Zn Pb U,O,powder .. . . . . B, Be, Cd, Co, Cr, Cu, Fe, K, Li, Mn, Na, Ni, Pb As, Cr, Cu, Fe, Mn, Ni, Pb, V, Zn Al, As, Ba, Be, Ca, Cd, Ce, Co, Cr, Cu, Eu, Fe, Ga, Mn, Mo, Na, Ni, Pb, Rb, Sb, Si, Sr, Th, Ti, TI, U, V, Zn NBS SRM 1632 Coal NBS SRM 1633a Coal . . . . Hf, Hg, K, Mg, Cadmium mercury . . Ag, Al, Co, Cr, Cu, telluride Fe, In, Mn, Ni, Pb, Zn High-puritygraphite . . Cu Purezirconium . . . . Cd MineraloilCONOSTAN . . Cd Airborne particulates . . Ca, Fe, V US Geological Survey rocks . , . , Ba, Cu, Ni, Sr, Zn Heavy mineral grains . . Ag, Al, As, Ba, Be, Bi, Ca, Cd, Cr, Co, Cu, Fe, Mg, Mn, Mo, Ni, S, Sb, Se, Si, Sn, Ti, V, W, Zn, Zr Solidrubber . . . . . . S Motoroil . . . . . . Zn Gasoline . . . . . . . . Pb TV device mode ETV-ICP, graphite rod DSIT-ICP, graphite carrier distillation electrode DSIT-ICP, graphite carrier distillation electrode - graphite lid graphite DSIT-ICP, undercut R.f.arc-ICP, graphite cup Sample pre-treatment - Direct determination of U30, containing 5% AgCl as carrier Direct determination of U308 containing 5% AgCl as carrier Direct qualitative analysis Direct qualitative analysis ETV-ICP, graphite rod Dissolution in hot aqua regia ETV-ICP, graphite rod - cup Graphite containing 5% Mg(N03)2 as ashing aid was dissolved in HCI - HNO, (3 + l), resin pre-concentration and addition of H3B03 Vacuum impacted on to the adhesive surface of a flexible Mylar type tape ETV-ICP, graphite rod - cup Dissolution in HF - HN03 DSIT-ICP, graphite crucible Direct determination SSS-ICP, C0,-TEA laser SSS-ICP, C0,-TEA laser Vacuum impacted on to the adhesive surface of a flexible Mylar type tape Semiquan ti tative direct determination from sample grains set on to a polyethylene base with polystyrene cement SSS-ICP, ruby laser ETV-ICP, graphite tube ETV-ICP, pyrolytically Direct determination ETV-ICP, pyrolytically Chemical stabilisation Direct determination of solid samples coated carbon cup coated carbon cup with iodine Reference 21 53 54 47 61 87 88 89 90 70 70 73 30 27 27180 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL.1 ~~ Table 5. Applications of TV-ICP systems to the analysis of metals, alloys and other metallurgical samples Sample Steel . . . . . . . . Stahlspanen . . . . . . Steelreference . . . . Iron . . . . . . . . Steelworkslag . . . . Alalloy, brass .. . . Al, A1 - Mn, A1 - Mg - Si, A1 - Si - Cu, A1 - Si - Cu Nialloys . . . . . . A1203, CaC03 . Nickel-base superalloys (NBSstandards) . . Steel standards (NBS 1260 series and BCS 400 series) Low-alloysteel . . . . Aluminiumbase . . . . Standard low-alloy steel (BCS402) . . . . . . Alluminium (high and low alloy Alcoa standards) Naval brass (NBS standards) Steel . . . . . . . . Brass . . . . . . . . Aluminiumalloy . . . . Titanium-base alloy standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . * . . . Low-alloy steels, copper alloys Standard rock samples (BGS graniteGN1) . . . . . . Elements determined Ni c u Al, Co, Cu, Mn, Ni, Si Mn Al, Ca, Fe, Mg, Ti Cu, Fe, Mg, Zn Cu, Fe, Mg, Mn, Si Fe, Mg, Mn, Pb, Si, Sr, V, Zn Al, B, Co, Cr, Cu, Fe, Hf, Mn, Mo, Nb, Si, Te, Ti, V, W Cr, Cu, Mn, Mo, Ni, C, Cr, Mn, Mo, Ni, P, Cr, Cu, Fe, Mg, Mn, P, S, Si, V S, Si Si, Ti, Zn Fe, Cr, Cu, Ni, Si, Ni/Fe Cr, Cu, Fe, Mg, Mn, Pb, Si, Sn, Ti, V, Zn Al, Cu, Fe, Mn, Ni, Pb, Sn, Zn Al, Co, Cr, Cu, Mn, Mg, Ni, V Fe, Ni, Sn Cr, Cu, Fe, Mg, Sn, Zn Cr, Mo Cr, Mn, Ni Bi, Hg, Pb, Th, T1, U Amount of sample/ TV device mode Sample pre-treatment mg Reference DSIT-ICP, graphite crucible Direct determination 5 48 DSIT-ICP, graphite crucible Direct determination 5 48 DSIT-ICP, graphite crucible Direct determination 0.03 48 59 SSS-ICP, interrupted arc Direct determination from - metallic surface, sample - Cu powder mixture (1 + 5 ) , briquetted (1 + 5), briquetted metallic surface SSS-ICP, interrupted arc Sample - Cu powder mixture - 59 SSS-ICP, spark Direct determination from - 62 64 SSS-ICP, spark Direct determination from - SSS-ICP, spark Direct determination from - metallic surface, a point- to-plane technique metallic surface, a point- to-plane technique 64 66 SSS-ICP, spark Direct determination from - metallic surface SSS-ICP, spark Faced metal sample on spark - 67 Petrey table SSS-ICP, spark Faced metal sample on spark - 67 Petrey table SSS-ICP, spark Faced metal sample on spark - 67 Petrey table SSS-ICP, ruby laser Direct determination from - 71,72 flat surface prepared on a finishing machine 74 SSS-ICP, ruby laser Direct determination from - discs 74 76 SSS-ICP, ruby laser Direct determination from - SSS-ICP, ruby laser Direct determination from - discs flat and smooth surfaces prepared by polishing with a 400-mesh silicon carbide paper flat and smooth surfaces prepared by polishing with a 400-mesh silicon carbide paper flat and smooth surfaces prepared by polishing with a 400-mesh silicon carbide paper flat and smooth surfaces prepared by polishing with a 400-mesh silicon carbide paper 76 SSS-ICP, ruby laser Direct determination from - 76 SSS-ICP, ruby laser Direct determination from - 76 SSS-ICP, ruby laser Direct determination from - 75 SSS-ICP, low-energy Direct determination from - SSS-ICP-MS, ruby laser Direct determination from - Nd : YAG laser chips 77 discs 7.Detection Limits In the literature, it is usual practice to quote the detection limit as the result for a particular technique or method, and to draw comparisons between the detection limits observed using similar techniques.The ranges of detection limits for the TV-ICP technique are summarised in Table 2, instead of just the lowest value reported, and compared with those obtained using ICP pneumatic nebulisation and GFAAS.83.84 This approach was adopted because the range of reported values is a reflection of differences in instrumental applications and a variability in the TV-ICP capability. The values of the detection limits were obtained by calculations based on the given limit of determination (loo), but using 30 and are presented in terms of both mass and concentration to simplify comparison with other work (or between different works). Whilst any direct comparison of detection limits is misleading owing to the use of different systems, operating conditions andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL.1 181 Table 6. Applications of TV-ICP systems to the determination of elements in biological materials I Amount of sample Sample pre-treatment Dilution with Triton X-100 (1 + l), pyrolysis with Me4NOH solution (1 + 4) Dilution with Herrmann None. direct determination 4 10 50 5 5 5 5 10 ml 50 5 ml 10 15 5 10 - 5 5 ml - 50 10 - 50 10 5 10 Reference 91 34 37 43 37 36 80 41 81 92 93 37 94 90 86,95 81 47 34 90 56 57 34 56 43 23a Biological material Elements determined Human whole blood . . Mn, Ni TV device, mode ETV-ICP, graphite rod Test serum . . . . . . Mn, Zn ETV-ICP, graphite tube ETV-ICP, graphite ETV-ICP, carbon rod electrode and microboat Human serum . . . . . . Al, Si Human blood plasma .. Ca, Cd, Cu, Fe, K, Mg, Na, P, Pb, Zn Pooled urine . . . . . . Al, Si None, direct determination ETV-ICP, graphite electrode None, direct determination NIOSH-NBS “Elevated” freeze-dried urine As, Cr, Cu, Ni, Se ETV-ICP, graphite electrode Reconstituted with water, direct determination except Se (resin pre- concentration) . . NIOSH-NBS “Normal” and “Elevated” freeze- driedurine . . . . Cd, Cu, Cr, Mn, Ni, Pb ETV-ICP, graphite rod Direct determination with electrochemical pre- concentration and separation . . NIOSH-NBS “Elevated” freeze-dried urine Cr, Cu, Ni Cr, Cu, Ni ETV-ICP, graphite tube - platform ETV-ICP, graphite cuvet te Reconstituted with water, direct determination Direct determination with electrochemical pre- concentration and separation Dissolution with Me4NOH Wet digestion (H2S04 - HN03), dithizone extraction None, direct determination Humanmilk .. . . Ni Ni ETV-ICP, graphite rod ETV-ICP, graphite rod Haemodialysis solution A1 ETV-ICP, graphite ETV-ICP, graphite cup DSIT-ICP, graphite ETV-ICP, graphite electrode crucible tube - platform and graphite electrode Incubation solution . . Grass . , . . . . None, direct determination None. direct determination . Au . Cd Treeringwood . . . Al, As, Ba, Ca, V, Cu, Fe, Ge, K, Mg, Mn, Na, Si, Sr, Zn Cd, Co, Cu, Mn, Ni, Pb, Sb, Zn 50% H202 - PTFE bomb digestion NBS-SRM 1576 Rice flour ETV-ICP, graphite cuve tte Direct determination with electrochemical pre- concentration and separation NBS-SRM 1571 Orchard leaves . . . . . . Be, Ca, Cr, Cu, K, Fe, Mg, Mn, Na, P, Zn Mn, Pb, Zn DSIT-ICP, undercut graphite cup electrode None, direct determination ETV-ICP, graphite tube Wet digestion (HN03 - None.direct determination H202 - HC104) DSIT-ICP, graphite ETV-ICP, graphite cup crucible Cd Al, Ca, Cd, Co, K, Cr, Cu, Fe, Mg, P, Mn, Na, Ni, Pb, Zn K, Mg, Fe, Mn, Na, P, Pb, Zn, As, Cu, Cr, Ni, Cd Wet digestion (HN03 - HC104) DSID-ICP, graphite cup - lid Mixture of sample solution and cellulose powder, dried, residue was weighed NBS-SRM 1577 Bovine liver . . . . . . . . Cd, Mn, Zn Wet digestion (HN03 - Wet digestion (HN03 - HC104 - H202) HC104 - HF) ETV-ICP, graphite tube ETV-ICP, graphite cup NBS-SRM 1570 Spinach . . Al, Ca, Cd, Co, K, Cr, Cu, Fe, Mg, P, Mn, Na, Ni, Pb, Zn NBS-SRMBovineljver . . Ca, Cd, Cu, Fe, K, Mg, Na, P, Pb, Zn Bowen’s kale .. . . . . Ag, Cd, Cu, Mn, Pb None, direct determination ETV-ICP, carbon rod - microboat ETV-ICP, graphite rod Wet digestion [HN03 - HC104 ( 5 + l)]182 Table 6.-continued Amount of sample JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Biological material Elements determined TV, device mode Sample pre-treatment P1 mg Reference IAEA H-5 Animal muscle Mn, Ni ETV-ICP, graphite rod Dissolution with Me4NOH 10 - 91 IAEA V-9 Cotton 86 - cellulose material . . . . Al, As, Ba, Ca, K, ETV-ICP, graphite 50% H202 - PTFE bomb 5 Cu, Fe, Mg, Mn, tube - platform digestion V, Na, Pb, Si, Sr, Zn 96 - Humanbone . . . . . . Cu ETV-ICP, graphite Wet digestion (HN03 - 5 Formulation . . . . . . Cu, Ca, Fe, Mg, Zn ETV-ICP, graphite rod Slurry 10 - 97 Whole capsule .. . , . . Ca, Cu, Fe, Mg, P, ETV-ICP, graphite rod Slurry 10 - 97 H202) or resin pre- concentration electrode Zn modes, it is clear that detection limits obtainable with TV-ICP are one to two orders of magnitude superior to pneumatic nebulisation, and tend to approach those obtained with electrothermal atomisation AAS. The current detection limits that can be obtained in realistic matrices under compromise operating conditions for a broad range of elements are thus of the order of 1 ng mi-’. In this author’s opinion, there is no indication that the sensitivity and hence detection limits can be further improved without additional pre-concentration steps. A linear dynamic range of 3-4 orders of magnitude was initially obtained by Nixon et a1.8 and this has now been extended to 5-6 orders of magnitude.59 At the current state of development then, the linear dynamic range that has been demonstrated is six orders of magnitude in single-element solutions.It is not yet clear how this varies with TV mode, plasma power or carrier gas flow-rates. 8. Applications There are many applications of the TV-ICP method for qualitative and quantitative measurements (elemental deter- minations) in a wide range of samples. For clarity, illustrative applications have been tabulated in the Tables 3-6, with descriptions of the elements determined and sample matrix. It should be noted that Matusiewicz and Fricke98 reviewed the application of TV-ICP/MIP to the analysis of biological materials and dealt primarily with the usefulness of TV-ICP/ MIP in determining trace elements, and Zil’bershtein99 has recently surveyed the application of the ICP discharge to the AES analysis of solid samples.9. Conclusions In recent years increasing attention has been placed on the development of atomic discharge systems that segregate the sampling from the excitation step. It is clear from this review that the TV technique provides an efficient method (the assumption of 20-60% vaporisation and transport efficiency) for the introduction of small samples into an ICP for excitation for optical emission spectrometry and progress in evaluating the technique for routine elemental analysis should be rapid. Although, in general, interfacing a TV to an ICP is mechanic- ally straightforward, tandem operation of the two components (or sometimes three) has required careful manipulation of the instrumental variables. However, some practice is needed to enable a new operator to become skilled. The advantage of the method is its freedom from time-consuming steps and potential contaminations associated with sample preparation ( i .e . , digestion). The TV-ICP combination should be well suited for analyses where only small volumes of sample or amounts of solids are available, such as biological, environ- mental and forensic applications, but where high sensitivity is desirable or necessary. The detection limits are better than those obtained with conventional pneumatic nebulisation. The reported overall precision is satisfactory, 1-15%, and is similar to GFAAS, but in most instances is worse than that of solution ICP-OES analysis.Because TV-ICP is used in an emission mode, simultaneous multi-element detection is possible even for discrete sample analysis, as has been demonstrated mainly with the laser vaporisation - ICP tech- nique. The results obtained with electrochemical pre- concentration methods indicate that the mercury thin layer and hanging mercury drop electrodeposition - ICP-OES technique with TV can considerably extend the analytical capabilities of the ICP method by combining sufficient pre-concentration/separation with simultaneous multi- element analysis. Relatively few observations have been reported concerning matrix interference effects (when the analyte vaporisation coincides with that of the matrix), requiring, in some instances, that the method of standard additions or matrix matching with background correction be utilised.The fact that the operating conditions of the furnace and/or ICP were often optimised individually for each element (especially for solid samples), and that standard-additions techniques were used suggests that the TV-TCP separation of the vaporisation and analysis functions did not eliminate all of the troublesome analyte specificities or matrix effects observed in GFAAS. In practice, in the reviewer’s opinion, it is unlikely that absolute or complete separation of the two processes (vaporisation and excitation) is possible. Encouraging preliminary results were obtained for the speciation of solid samples; however, there is little widespread use of TV-ICP with application to practical samples for speciation.Although this technique is appealing due to its simplicity, it seems that mineralogical history and element volatility will have an adverse effect on accuracy. These interferences can be compensated for by the use of internal references. 10. Suggestions for Future Studies It seems probable that most future developments will come in the area of specialised application studies on a broad range of sample types and instrumental developments. For example, an ETV system should be designed specifically for use with an ICP (optimised for its specific function) and not be just a modification of existing commercially available furnaces designed for use with AAS, as has been shown by PSA Analytical Ltd.(see section 3.1.2.). Future improvements lie in further optimisation of the interface between the ICP and electrothermal vaporiser in terms of analyte vapour transport time and efficiency (the length of the tubing or connection and streaming parameters would be expected to affect the efficiency of transportation). Ideally the operating conditions of the electrothermal vaporiser should have minimal effect on the excitation conditions of the ICP and the ramp time of the temperature should be as short as possible (to ensure that theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 183 rate of sample introduction is independent of the carrier gas flow-rate). Also, the tube - platform configuration40 should act like a L’vov platform in a constant-temperature vaporiser (graphite, pyrolytically coated graphite or even glassy carbon) and is worthy of further consideration to eliminate or reduce matrix effects in the analysis of practical samples.Owing to matrix interference effects, additional research is required to differentiate between sample vaporisation effects, sample transport effects and atomisation and excitation processes in the ICP discharge. Probably, a more important question is whether the chemical form of the species evolved has an influence on the transport efficiency. Because the non-selective volatilisation of the matrix in TV methods will often discount an apparent improvement in detection limits in practical analysis it is another question how meaningful the detection limits that are appearing in the literature are.There is also much scope for improvement in pre-concentration techniques because commonly used procedures for “optimis- ation of experimental variables” are not likely to produce the required improvement in detection limits for some trace elements. Future studies should involve controlled potential electrodeposition on a hanging mercury drop electrode, thin-film layer of mercury or graphite or metal electrodes. This should provide for contamination-free pre-concentration and separation from the matrix and compare favourably with other commonly used solvent extraction or ion-exchange procedures. The use of internal standards could be employed to compensate for variations in sample introduction and in instrumental parameters. This might improve analytical preci- sion and allow for simultaneous multi-element analysis at less than optimum operation conditions for each element.Theoretically, automation of the sample pipetting (liquids) or deposition (solids) should also improve the overall precision of the technique, but in the reviewer’s opinion, practically it is not an easy task (the loading, pipetting and/or weighing and transfer operations of samples can be found to be difficult and delicate). On the other hand, the search for compromise TV-ICP conditions for simultaneous multi-element analysis needs to be studied in more detail. In addition, work should be carried out on ways of improving the measurement electronics to ensure that they are fast enough not to distort the transient signal produced with TV and should also be capable of measuring the background with the same speed, and simul- taneously.Also, because the photodiode array possesses a number of distinctive features that make it a particularly attractive detector, as a simultaneous multi-element measure- ment capability, it should be investigated for potential use with TV-ICP-OES in practical applications. Present laser - ICP systems are still unsuitable for quantita- tive analysis of solid samples, but it is possible to obtain spatially resolved information on sample surfaces. There is potential for the incorporation of lasers into commercial ICP analytical instrumentation. Also, combining laser vaporis- ation and the ICP source with mass spectrometers should have potential for the introduction of liquids and/or solid samples with increased sensitivity and selectivity and enlarging the analytical scope by providing isotopic capability.It should be stressed that ICP-MS provides a unique capability for the direct determination of isotope ratios for elements in solu- tion.1m Thermal vaporisation sample introduction techniques developed for ICP-OES are readily adapted for ICP-MS and some initial applications studies have been carried out, but much remains to be done to characterise and optimise the technique. Finally, emphasis should be placed on increased exploi- tation of the unique properties of TV-ICP for speciation studies of elements in biological materials, which will help scientists in the biomedicai field to obtain a better understand- ing of the role of metals, especially toxic metals, in life processes.In conclusion, it should be noted that each of the sub-groups of TV sample introduction for the ICP-OES described in this review has its own attributes and specialised applications that cannot be combined in a single, universal device (usually these two units are connected in some way). The critical factors, in the reviewer’s opinion, in the construction and use of those TV-ICP systems (combinations) for attaining the best analy- tical performance are: (a) the interface between the thermal vaporiser and the ICP; (b) vaporisation temperature steps and/or ramp-rate; (c) capabilities for simultaneous multi- element vaporisation and introduction of the vapours into the ICP with the possibility of detection systems that can measure the very short residence time of atoms in the ICP source; and (d) minimum matrix and transport effects on the excitation conditions of the ICP, which should be considered as the optimum design parameters.The author is grateful to Karen A. Wolnik for assisting with the preparation of this manuscript. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 l a . 11. References Greenfield, S., Jones, I. L., and Berry, C. T., Analyst, 1964, 89, 713. Wendt, R. H., and Fassel, V. A., Anal. Chem., 1965,37,920. Fassel, V. A., Anal. Chern., 1979, 51, 1290A. Browner, R. F., and Boorn, A. W., Anal. Chem., 1984, 56, 786A and 875A. Creqser, M. S., Prog. Anal. At. Spectrosc., 1981, 4, 219. Kgntor, T., Spectrochim. Acta, Part B, 1983, 38, 1483. Ng, K. C., and Caruso, J. A., Appl.Spectrosc., 1985,39,719. Nixon, D. E., Fassel, V. A., and Kniseley, R. N., Anal. Chem., 1974, 46, 210. Smythe, L. E., ICP Inf. Newslett., 1980, 6, 224. Kitazume, E., Anal. Chem., 1983, 55, 802. Tikkanen, M. W., and Niemczyk, T. M., Anal. Chem., 1984, 56 1997 - - 7 - _ _ ‘ . l l b . Tikkanen, M. W., and Niemczyk, T. M., Anal. Chem., 1985, llc. Tikkanen, M. W., and Niemczyk, T. M., Anal. Chem., 1986, 12. 13. 14. 15. 16. 17. 18. 19. 20. 57, 2896. 58, 366. Dahlquist, R. L., US Pat., 3832060, 1974. Fassel, V. A., Am. SOC. Test. Muter., Spec. Tech. Publ. No. 618, 1977, p. 22. Instrument Column, Spectrochim. Acta, Part B , 1982,37,359. Gum, A. M., Millard, D. L., and Kirkbright, G. F., Analyst, 1978, 103, 1066. Kirkbright, G. F., Millard, D. L., and Snook, R. G., Anal.Proc., 1980, 17, 16. Kirkbright, G. F., Pure Appl. Chern., 1982, 54, 769. Long, S. E., Snook, R. D., and Browner, R. F., Spectrochim. Acta, Part B, 1985, 40, 553. Millard, D. L., Shan, H. C., and Kirkbright, G. F., Analyst, 1980, 105, 502. Kirkbright, G. F., in Barnes, R. M . , Editor, “Developments in Atomic Plasma Spectrochemical Analysis,” Heyden, London, 1981, p. 223. Kirkbright, G. F., and Snook, R. D., Appl. Spectrosc., 1983, 21. 22. 23. 23a. 24. 25. 26. 27. 28. 29. 30. 31. - 37, 11. Kirkbright, G. F., Millard, D. L., and Snook, R. D., Spectrochim. Acta, Part B , 1983, 38, 649. Kirkbright, G. F., and Snook, R. D., Anal. Chem., 1979, 51, 1938. Dean, J. R., Ziang, L., and Snook, R. D., Analyst, 1985, 110, 1395. Ng, K. C. , and Caruso, J. A., Anal. Chim.Acta, 1982,143,209. Ng, K. C., and Caruso, J. A., Anal. Chem., 1983, 54, 1513. Ng, K. C., and Caruso, J. A., Analyst, 1983, 108, 476. Ng, K. C., and Caruso, J . A., Anal. Chem., 1983,55,2032. Gray, A. L., and Date, A. R., Analyst, 1983, 108, 1033. Crabi, G., Cavalli, P., Achilli, M., Rossi, G. , and Omenetto, N., At. Spectrosc., 1982, 3, 81. Casetta, B., Di Pasquale, G., and Soffientini, A., At. Spec- trosc., 1985, 6, 62. Swaidan, H. M., and Christian, G. D., Can. J . Spectrosc., 1983,28, 177.184 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 41a. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Hartenstein, S. D., Swaidan, H. M., and Christian, G. D., Analyst, 1983, 108, 1323.Swaidan, H. M., and Christian, G. D., Anal. Chem., 1984,56, 120. Aziz, A., Broekaert, J. A. C., and Leis, F., Spectrochim. Acta, Part B, 1982, 37,369. Matusiewicz, H., Fricke, F. L., and Barnes, R. M., J . Anal. At. Spectrom., 1986, 1, 203. Barnes, R. M., and Fodor, P., Spectrochim. Acta, Part B , 1983, 38, 1191. Matusiewicz, H., and Barnes, R. M., Spectrochim. Acta, Part B , 1984,39,891. Hull, D. R., and Horlick, G., Spectrochim. Acta, Part B , 1984, 39, 843. Matusiewicz, H., and Barnes, R. M., Appl. Spectrosc., 1984, 38, 745. Matusiewicz, H., and Barnes, R. M., Spectrochim. Acta, Part B , 1985,40,29. Matusiewicz, H., and Barnes, R. M., Spectrochim. Acta, Part B , 1985,40,41. Matusiewicz, H., Znt. Lab., 1983, 13, 24. Matusiewicz, H., and Barnes, R.M., Appl. Spectrosc., 1985, 39, 715. Blakemore, W. M., Casey, P. H., and Collie, W. R., Anal. Chem., 1984, 56, 1376. Hull, D. R., Wohlers, C. C., Schleicher, R. G., and Smith, S. B., paper presented at the Pittsburgh Conference and Exposition on Analytical Chemistry and Applied Spectro- scopy, New Orleans, LA, USA, February 25-March 1, 1985, paper No. 949. Hull, D. R., Wohlers, C. C., Schleicher, R. G., and Smith, S. B., Allied Analytical Systems, Andover, MA, USA, Technical Presentation No. T34, 1985. Kleinmann, I., and Svoboda, V., Anal. Chem., 1969,41,1029. Salin, E. D., and Horlick, G., Anal. Chem., 1979, 51, 2284. Sommer, D., and Ohls, K., Fresenius 2. Anal. Chem., 1980, 304,97. Ohls, K., and Sommer, D., in Barnes, R. M., Editor, “Developments in Atomic Plasma Spectrochemical Analysis,” Heyden, London, 1981, p.321. Kirkbright, G. F., and Walton, S . J., Analyst, 1982, 107, 276. Kirkbright, G. F., and Li-Xing, Z . , Analyst, 1982, 107, 617. Li-Xing, Z . , Kirkbright, G. F., Cope, M. J., and Watson, J. M., Appl. Spectrosc., 1983, 37, 250. Page, A. G., Madraswala, K. H., Godbole, S. V., Kulkarni, M. J., Mallapurkar, V. S . , and Joshi, B. D., Fresenius 2. Anal. Chem., 1983, 315, 38. Page, A. G., Godbole, S. V., Madraswala, K. H., Kulkarni, M. J., Mallapurkar, V. S., and Joshi, B. D., Spectrochim. Acta, Part B , 1984, 39, 551. Haraguchi, H., Abdullah, M., Hasegawa, T., Kurosawa, M., and Fuwa, K., Bull. Chem. SOC. Jpn., 1984, 57, 1839. Abdullah, M., Fuwa, K., and Haraguchi, H., Spectrochim. Acta, Part B , 1984, 39, 1129.Abdullah, M., and Haraguchi, H., Anal. Chem., 1985, 57, 2059. Salin, E. D., and Sing, R. L. A., Anal. Chem., 1984,56,2596. Keilsohn, J. P., Deutsch, R. D., and Hieftje, G. M., Appl. Spectrosc., 1983,37, 101. Ohls, K., and Sommer, D., Fresenius Z . Anal. Chem., 1979, 296, 241. Farnsworth, P. B., and Hieftje, G. M., Anal. Chem., 1983,55, 1414. Human, H. G. C., Scott, R. H., Oakes, A. R., and West, C. D., Analyst, 1976, 101, 265. Scott, R. H., Spectrochim. Acta, Part B , 1978,33, 123. Aziz, A., Broekaert, J. A. C., Laqua, K., and Leis, F., Spectrochim. Acta, Part B, 1984, 39, 1091. Jarrell-Ash ICAP Solid Sampling System Prospect No. D24, Waltham, MA, USA, 1982. Marks, J. Y., Fornwalt, D. E., and Yungk, R. E., Spectrochim. Acta, Part B, 1983, 38, 107. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. Dalager, P., Goulter, J., andTasker, D., Res. Develop., 1985, 27, 114. Applied Research Laboratories 3500 ICP AES Systems Con- ductive Solids Nebulizer Sample Introduction System Prospect, Sunland, CA, USA, 1985. Thompson, M., and Walsh, J. N., “A Handbook of Inductively Coupled Plasma Spectrometry,” Blackie, Glasgow, 1983, p. 176. Abercrombie, F. N., Silvester, M. D., Murray, A. D., and Barringer, A. R., in Barnes, R. M., Editor, “Applications of Inductively Coupled Plasmas to Emission Spectroscopy,” The Franklin Institute Press, Philadelphia, 1978, p. 121. Thompson, M., Goulter, J. E., and Sieper, F., Analyst, 1981, 106, 32. Thompson, M., Goulter, J. E., and Sieper, F., Jena Rev. 1981, 202. Thompson, M., and Hale, M., Areas Glaciat. Terr., 1984,225. Carr, J. W., and Horlick, G., Spectrochim. Acta, Part B , 1982, 37, 1. Kawaguchi, H., Xu, J., Tanaka, T., and Mizuike, A., Bunseki Kagaku, 1982, 31, E185. Ishizuka, T., and Uwamino, Y., Spectrochim. Acta, Part B , 1983, 38, 519. Gray, A. L., Analyst, 1985, 110, 551. Salin, E. D., and Habib, M. M., Anal. Chem., 1984,56,1186. Habib, M. M., and Salin, E. D., Anal. Chem., 1985,57,2055. Matusiewicz, H., Fish, J., and Malinski, T., Spectroscopy, submitted for publication. Matusiewicz, H., and Fricke, F. L., Anal. Chim. Acta, submitted for publication. Prack, E. R., and Bastiaans, G. J., Anal. Chem., 1983, 55, 1654. Koch, 0. G., LaFleur, P. D., Morrison, G. H., Jackwerth, E., Townshend, A., and Tolg, G., Pure Appl. Chem., 1982, 54, 1565. Omenetto, N., in Facchetti, S . , Editor, “Analytical Techniques for Heavy Metals in Biological Fluids,” Elsevier, Amsterdam, 1981, p. 155. Matusiewicz, H., Horvath, Z., and Barnes, R. M., Appl. Spectrosc., 1985,39, 558. Matusiewicz, H., and Barnes, R. M., Anal. Chem., 1985, 57, 406. Cope, M. J., Kirkbright, G. F., and Burr, P. M., Analyst, 1982, 107, 611. Mahanti, H. S . , and Barnes, R. M., Anal. Chem., 1983, 55, 403. Jing, L. S., and Barnes, R. M., Appl. Spectrosc., 1984,38,284. Ohls, K., Spectrochim. Acta, Part B, 1984, 39, 1105. Rica, C. C., Kirkbright, G. F., and Snook, R. D., At. Spectrosc., 1981, 2, 172. Rica, C. C., and Kirkbright, G. F., Sci. Total Environ., 1982, 22, 193. Barnett, N. W., Chen, L. S . , and Kirkbright, G. F., ’4nal. Chim. Acta, 1983, 149, 115. Tepperman, K., Finer, R., Donovan, S., Elder, R. C., Doi, J., Ratliff, D., and Ng, K., Science, 1984, 225, 430. Scherbatskoy, T., Matusiewicz, H., and Barnes, R. M., Nature (London), submitted for publication. Mahanti, H. S., and Barnes, R. M., Anal. Chim. Acta, 1983, 151, 409. Long, S. E., and Snook, R. D., At. Spectrosc., 1982, 3, 171. Matusiewicz, H., and Fricke, F. L., in “Biological Monitoring of Exposure to Metals,” John Wiley, submitted. Zil’bershtein, Kh. I., ZCP ZnJ Newsleft., 1985, 10, 964. Douglas, D. J., and Houk, R. S . , Prog. Anal. At. Spectrosc., 1985, in the press. Paper J5l39 Received September 23rd, I985 Accepted January 15th, I986
ISSN:0267-9477
DOI:10.1039/JA9860100171
出版商:RSC
年代:1986
数据来源: RSC
|
13. |
A predictive model of plasma matrix effects in inductively coupled plasma atomic emission spectrometry |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 185-193
Michael H. Ramsey,
Preview
|
PDF (1069KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 185 A Predictive Model of Plasma Matrix Effects in Inductively Coupled Plasma Atomic Emission Spectrometry* Michael H. Ramsey and Michael Thompson Applied Geochemistry Research Group, Department of Geology, Imperial College, London SW7 ZBP, UK Matrix effects in inductively coupled plasma atomic emission spectrometry (ICP-AES) due to metallic elements in solution are found to be largely due to plasma processes. The effects of such matrix elements on the sensitivities of analyte lines are evaluated. Most matrix elements give rise to considerably greater effects than do the frequently studied alkali metals, under operating conditions normally adopted for routine use. Analytes are affected to a degree dependent on the total excitation potential.Lines with high excitation potential are strongly affected. Different matrix elements produce widely varying effects on the sensitivity of a single line, depending on the energy required to dissociate and ionise the matrix. These observations are consistent with the hypothesis that the matrix effects are due to cooling of the plasma by the matrix. Such a model enables matrix effects to be quantitatively predicted for matrices at 0.05 M. Keywords; Inductively coupled plasma atomic emission spectrometry; matrix effects; excitation temperature; predictive model Our perception of matrix effects in inductively coupled plasma atomic emission spectrometry (ICP-AES) has changed as the method has developed. Until recently, matrix effects were relatively unknown, largely because of the quite proper application of matrix matching as a routine precaution.In circumstances where matrix matching is not possible (e.g. , widely variable samples or sample masses), it has become increasingly clear that matrix effects must be seriously considered in practical analysis. An understanding of matrix effects, however, is still rudimentary. No general means of predicting their magnitude is available, even on an empirical basis. Current trends in ICP-AES analysis towards the use of higher sample concentrations in the test solutions, together with the demand for greater accuracy, emphasise the need for such understanding. Matrix effects in ICP-AES stem from two sources, viz., (i) changes in nebuliser operation resulting from variations in the gross physical properties of the test solutions1 and (ii) changes in excitation conditions in the plasma itself.In a previous study2 we investigated matrix effects on a number of analytes caused by concomitant calcium, and found them to be located almost exclusively in the plasma. The magnitude of the effect depended strongly and directly on the total excitation poten- tial of the analyte line. Moreover, at normal compromise viewing heights (i.e. , under conditions that would influence practical analysis), the effects of calcium were much more severe than those of the alkali metals, the only elements that had previously been subjected to detailed investigation in relation to their matrix effe~ts.37~ The practical problems posed by the calcium effects in geochemical analysis2 drew attention to the almost complete lack of information on the plasma matrix effects of elements in general.Ideally, it should be possible to predict the extent to which any analyte line is affected by any matrix element. This paper describes work undertaken to remedy that shortfall. In this study the effects of 18 matrices on 16 analyte lines were tested experimentally. However , the broader perspec- tives opened by this large study enabled us to discern consistent patterns in the results that could be attributed to * Presented in part at FACSS, the twelfth annual meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Philadelphia, PA, USA, September 29th-October 4th, 1985. fundamental properties of the elements involved. The pattern of the effects on various analyte lines previously observed in a calcium matrix was also found in many other matrices.Thus, for a given matrix it became possible to postulate a small drop in excitation temperature AT to account for this pattern. Estimated values of AT gave a quantitatively consistent explanation of the sensitivity variations of all the analyte lines. Estimates of AT depended on a prior knowledge of the original plasma temperature T . Despite the large uncertainty in the value of T , values of AT can be much more accurately estimated. To account for the varying capacity of different matrix elements to produce matrix effects was more difficult. For example, there was no discernible relationship between the effect on an analyte and the ionisation potential of the matrix element.However, the total energy required to bring the matrix from ambient up to the temperature T (including the lattice energy and ionisation energy) was found to be linearly related to AT, suggesting a simple “cooling” mechanism. This enabled us to formulate for any matrix element - analyte line combination a predictive model, capable of providing a reliable quantitative estimate of the matrix effect. No spatially resolved information was sought in this study. Conditions of observation and operation were used that are common to most routine users of simultaneous ICP-AES. This enabled conclusions to be drawn that, although incom- plete in terms of the understanding of the plasma as a whole, can help identify and overcome problems of practical analysis.Experimental Equipment The details of the instrumentation and conditions are given in Table 1. An ARL 3520 sequential system was utilised only for the determination of the excitation temperature for the Ca atom lines given in Table 2. All multi-analyte measurements were made using the ARL 34000 simultaneous system and the lines used are given in Table 3. The statistical manipulation of the measurements was handled with the MINITAB com- puting system (Pennsylvania State University 1981.1). Materials All solutions were prepared in 1.0 M hydrochloric acid (analytical-reagent grade), with the exception of the lead186 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY , JUNE 1986, VOL. 1 Table 1. Description of ICP-AES instrumentation and operating conditions Instrument * Parameter 34000c 3520 Determination .. . . . . . . Computer . . . . . . . . . . Forwardpower/kW . . . . . . FrequencyIMHz . . . . . . Viewing height above load coil/mm Viewing window square of side / m m . . . . . . . . Torch type . . . . . . . . Gas flow-rated min-1: Coolant . . . . . . . . . . Auxiliary . . . . . . . . Injector(humidified) . . . . Spray chamber . . . . . . . . Nebuliser (concentric glass Meinhard) . . , . . . . . Solution uptake rate (unpumped)/ ml min-1 . . Uptake tube (poiyethy1ene);mm . . Nebulisertipwash/ml . . . . Pre-flus,h time, number of and timeofintegrations/s . . . . Simultaneous PDP 11/04 1.25 27.12 14 4 Fassel 12 0.4 1.01 Scott Double pass TR-30-3A 1.05 350 X 0.5 i.d. 0.5 20,3 x 5 Sequential PDP 11/23 1.20 27.12 15 4 Fassel 12 0.8 1 .oo Conical Single pass TR-30-3A 2.40 400 X 0.8 i.d.0.5 30,3 x 3 * Applied Research Laboratories 1-m vacuum spectrometer. Table 2. Nine calcium atom lines used for the determination of excitation temperature W aveleng th/nm 239.856 272.165 299.496 299.731 422.673 504.163 616.644 657.278 732.615 Excitation potentiallev 5.168 4.555 6.019 6.022 2.933 5.168 4.532 1.886 4.625 Relative sensitivity 32 5 50 69 12 841 62 26 1 94 Table 3. Information on analyte elements Element Wavelengthhm Ba . . . . Be . . . . Cd . . . . c o . . . . Cr . . . . c u . . . . La . . . . Li . . . . Mo . . . . Na . . , . Ni . . . . Pb . . . . Rb . . . . Sr . . . . v . . . . Zn . . . . 455.40 313.04 226.50 228.62 267.72 324.75 398.85 670.78 281.62 588.99 231.60 220.35 780.02 407.77 311.07 202.55 A tom/ion 1/11 I1 I1 I1 I1 I1 I I1 I I1 I I1 I1 I IT I1 I1 Order 1 2 2 3 3 2 2 1 2 1 2 2 1 1 2 2 Total excitation po tential/eV 7.93 13.28 14.46 13.70 12.91 3.82 9.12 1.85 13.25 2.10 14.02 14.78 1.59 8.73 11.07 15.51 matrix, for which 1.6 M nitric acid (analytical-reagent grade) was used.Stock solutions of matrix elements were prepared from high-purity metals or salts (Specpure, Johnson Matthey Chemicals) selected to minimise the addition of any other elements. Analyte elements were present at 5 pg ml-1, except for Be, Cd and Mo, which were at 0.5 pg ml-1. Procedure The relative sensitivity, as a measure of a matrix effect, is a key concept in this study. It is defined as the net intensity of an analyte in a solution containing a matrix element, relative to its intensity in the absence of the matrix element.Relative sensitivities for each matrix were measured by a separate sub-experiment designed to maximise the accuracy of the estimate. Attention to this detail is essential in order to avoid confusing inconsistencies. Matrix effects are difficult to measure with high accuracy, especially the smaller effects. The sub-experiments were conducted as follows. Four test solutions were run on the ICP-AES system under standard conditions but within the briefest possible time scale. The four solutions were as follows: (i) the blank matrix (1.0 M hydrochloric acid); (ii) a solution of the analytes in the blank matrix; (iii) a solution of the matrix element at a specified level in the blank matrix; and (iv) a solution containing both the analytes and the matrix element at the specified levels.If the response of a line at analyte concentration c in the nth solution was r,, then (r2 - rl)/c gave the reference net sensitivity and (r4 - Q)/C gave the net sensitivity in the solution of the matrix element. The relative sensitivity S in the solution of the matrix element was therefore (r4 - r3)/(r2 - r l ) . The term (r4 - r3) allowed for the compensation required for possible back- ground enhancements or line overlaps due to the matrix element, which would be identical in solutions (iii) and (iv). Conducting the study in the short-term sub-experiments minimised the effects of any possible drifts in sensitivity or background, a factor that easily distorts small differences in a series of relative sensitivity measurements.For analyte - matrix combinations presenting severe line overlaps or background enhancement, r3 >> (r4 - r3). In this study, such combinations were rare and, when encountered, the results were discounted as being subject to unacceptable experimen- tal error. Results and Discussion Mechanism of Matrix Effects Due to Calcium In a study of the matrix effects from calcium,2 a high correlation was found between the relative sensitivity of the analyte line in the presence of Ca and its total excitation potential (this includes ionisation energy for ion lines). The mechanism underlying this relationship was sought by con- sideration of the equation describing the net intensity I of the emission from an atom line? dA hvng 4nz I=- exp(-E/kT) .. . . where d is the depth of the source, A is the transition probability, h is Planck’s constant, v is the frequency of the transition, n is the number of atoms, g is the statistical weight of the excited state, 2 is the partition function, E is the excitation potential of the excited state, k is the Boltzmann constant and T is the atomic excitation temperature. Assuming initially that only the exponential term varies significantly with small changes of temperature, then equation (l), in its logarithmic form, differentiated with respect to temperature gives a E -lnZ=- aT k F ’ * ‘ * For a relatively small change in temperature AT, this becomes - * (2) (3) If the intensity of analyte emission is ZO in the blank matrix and I’ when calcium is present, then the intensity change caused by the calcium is given by AlnZ= 1nI’ - 1nP = In ($) = I n s .. (4)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 187 0.1 0.0 - .- c -0.1 > v) c a > .- 4- .- 0) -0.2 2 .- 4- - - r: -0.3 4 -0.4 I I I I I I I 0 4 8 12 16 20 0 4 8 12 16 -0.5 Total excitation potential of analyte/eV 0 Fig. 1. (a) Relative sensitivities of 14 analytes in 10000 yg ml-1 calcium plotted against the total excitation potential of each analyte line. These linear relationships are consistent with a fall in excitation temperature in the plasma of 130 k 14 K caused by the calcium. ( b ) Relative sensitivity data adjusted for the change in the numbers of atoms and ions caused by the change in temperature. The ion lines lie on a linear trend consistent with a fall in excitation temperature in the plasma of 94 k 17 K caused by the calcium.The atom lines, in parentheses, were not used in the regression. Element symbols are placed exactly over the experimental points 4 t O f /' ',*' 0 ,*' 0 S " i ' 0 I?' 1 I I I 0 1 2 3 4 E2 - El/eV Fig. 2. Determination of excitation temperature using pairs of calcium atom lines. The modified intensity ratio varies as a function of the difference in excitation potential with a slope yielding an excitation temperature of 7460 k 500 K where S is the relative sensitivity. Equation (3) now becomes AT l n S = E - k72 ' * ' . . . ( 5 ) The implication of this equation is that if the introduction of calcium to a plasma with excitation temperature T caused a drop in temperature AT, then there would be a linear relationship between the relative sensitivities of a range of analytes (Ins) and their corresponding excitation potentials.The relationship would apply equally to ion lines for which it can be derived by an analogous method. Hence a decrease in excitation temperature is consistent with the linear relationship between relative sensitivity and excitation potential previously published.2 The same data are replotted with the logarithm of the relative sensitivity in Fig. Inspection of equation (5) shows that the slope of the regression line shown in Fig. I(a) can be used to calculate AT, the decrease in excitation temperature required to produce the observed changes in the analyte sensitivities.The calcula- l ( 4 * tion, however, requires a value for the excitation temperature T , which was determined in a separate experiment. Determination of Excitation Temperature The method employed was that described by Boumans,5 using the intensities of the nine different calcium atom lines listed in Table 2. The sequential spectrometer (ARL 3520) was used to measure the relative intensities I of the nine lines, in their regions of linear response, and the values were corrected for the variations in photomultiplier response. Data for frequency (Y), statistical weight (g) and transition probability (A) were taken from Wiese et aZ.6 and used to calculate a modified intensity I, where For any pair of atom lines of excitation potentials El and E2, combining equations (1) and (6) gives Fig.2 shows the values of the modified intensity ratio against difference in excitation potential for the 36 pairs of atom lines measured. The slope of the regression line through the points gave an excitation temperature of 7460 K with an approximate standard deviation of 500 K. This value agrees well with other values of excitation temperature under similar operating conditions .7 Determination of Change in Excitation Temperature Using this value of excitation temperature (T), the change in excitation temperature (AT) can be calculated from the slope of the regression line shown in Fig. l(a), using equation (5). The value calculated was - 130 K with a standard deviation of 14 K, for analytes in a 1% mlV calcium solution.From comparison of this result with the uncertainties in the direct excitation temperature determination (k500 K), it was evident that (i) a direct measurement of T alone could not detect such small changes in excitation temperature, and (ii) the value of -130 K was dependent to a secondary degree on the uncertainty in the direct measurement T , and therefore188 JOURNAL OF -100 A 0 - I I 1 1 I I LNALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 slightly less certain than its small standard deviation would suggest. The initial assumption of a constant number of atoms (n) and ions with temperature was clearly an over simplification. The Saha equation can be used to calculate the degree of an ionisation aj of an element j , if local thermodynamic equi- librium (LTE) is assumed.From Boumansg we have a working version of the Saha equation: wherep, is electron pressure (atm), T the ionisation tempera- ture (K), Vii the ionisation potential (ev), Zu, the partition function for the atom of element j and 2, for the correspond- ing ion. Values of Vij, Zuj and 2, were interpolated from Drawin and Felenbok,g except for Mn, Be and La, for which no data for 2 were available at 7500 K. Extrapolation of data from Boumansg was considered justified as an approximation in these instances. .r I I i The ionisation temperature was taken to be equal to the experimentally determined excitation temperature by assum- ing LTE. Disparities between these temperatures have been reportedlo and direct measurement of ionisation temperature would clearly be necessary for more accurate estimation of a,.The value used for electron pressurep, (2.55 X 10-3 atm) was calculated from the electron density (n,) reported by Nojiri et al. ,lo measured under similar operating conditions, using the equation pe=nekT . . . . . . * * (9) For each emitting species the ratio was calculated between the number no of atoms or ions present at the initial temperature T (7500 K) and the number n’ at the lower temperature T - AT (7370 K). Equation (5) then becomes . . (10) The relative sensitivity data shown in Fig. l(a) when corrected for n’lno is replotted in Fig. l(b). The ion lines still give a linear trend after this correction, with a slope yielding a fall in excitation temperature of 94 k 17 K. The scatter of the atom lines of Li and Cu after the correction suggests that these simple assumptions do not fully describe the behaviour of atom lines.However, the over-all picture remains one of a drop in excitation temperature of ca. 100 K being a convincing explanation for the changes in sensitivity of the analytes in a 1% mlV calcium matrix. To estimate the decrease in excitation temperature at various calcium concentrations, a AT value was calculated for each analyte at each calcium concentration previously stu- died* using equation (10). These values, together with an average for each calcium concentration, are shown in Fig. 3. A small fall in temperature of ca. 20 K can be discerned for a calcium concentration as low as 100 pg ml-1. The trend appears to be slightly non-linear, curving away from the temperature axis.The Multi-matrix Domain Further support for the hypothesis of matrix effects reflecting a decrease in excitation temperature was sought by extending these techniques to a range of different matrix elements. The 18 matrix elements listed in Table 4 were studied for their effect on the 16 analyte lines given in Table 3. For the lead matrix only two analytes, Zn and Cd, were measured. Solutions of matrix elements were prepared at a concentration Table 4. Information on matrix elements (1) (2) (3) (4) ( 5 ) First Degree of Second Degree of “Total potential/eV ionisation/eV potentiallev ionisation/eV energy”/eV Matrix ionisation first ionisation second ionisation element A1 . . . . . . 5.98 Ba . . . . . . 5.21 B e . . . . .. 9.32 Ca . . . . . . 6.11 Cu . . . . . . 7.72 Fe . . . . . . 7.90 K . . . . . . 4.34 La . . . . . . 5.61 Li . . . . . . 5.39 Mg . . . . 7.64 Mn . . . . 7.43 N a . . . . . . 5.14 Ni . . . . . . 7.63 Pb . . . . . . 7.41 Rb . . . . 4.18 Sr . . . . . . 5.69 Ti . . . . . . 6.83 Zr . . . . . . 6.84 0.9487 0.9970 0.5372 0.9917 0.7320 0.8887 0.9965 0.9948 0.9908 0.9370 0.9352 0.9934 0.7859 0.9183 0.9973 0.9947 0.9789 0.9700 18.8 10.0 18.2 11.9 20.3 16.2 31.8 11.4 75.6 15.0 15.6 47.3 18.2 15.0 27.5 11.0 13.6 13.1 <0.0001 0.0382 <0.0001 0.0045 <o . 000 1 <0.0001 CO .0001 0.0243 CO .0001 CO. 000 1 <o .ooo 1 <o .ooo 1 <o. 000 1 <o .ooo 1 <o . 000 1 0.0154 0.0008 0.0005 5.67 5.58 5.01 6.11 5.65 7.02 4.32 5.86 5.34 7.16 6.95 5.11 6.00 6.81 4.17 5.83 6.70 6.64 (6) “Dis- sociation energy ”/eV 3.17 5.84 3.73 5.17 0.69 2.91 2.98 8.51 3.43 3.24 2.97 2.91 2.79 0.80 2.84 5.36 4.58 6.53 (7) “Matrix energy demand”/eV 8.85 11.42 8.73 11.28 6.35 9.93 7.30 14.37 8.77 10.40 9.92 8.02 8.78 7.60 7.01 11.19 11.28 13.17 (8) Concentra- tion/ yg ml-1 1350 6870 45 1 2000 3180 2790 1960 6950 347 1220 2750 1150 2940 10360 4270 4380 2400 4560 (9) Relative sensitivity of Zn 0.955 0.818 0.916 0.816 0.876 0.992 0.704 0.958 0.892 0.877 0.957 0.915 0.962 0.985 0.844 0.773 0.768 -JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL.1 4 6 189 I La , 8 8 10 0.7 I 4 5 6 7 8 Ti Zr C N > > Lc 0.9 c. .- .- c .- In Q, Q, .- 4- - a 0.8 a ’ * 1 .o C N w- 0.9 > > In C c. .- .- 4- .- s Q, .- 4- 5 0.8 CT La , Rb (6 Pb Na Li A1 Be Ni 0.7 Mg MnFe I , 1 ‘La Sr Ca Ti Zr First ionisation potential of matrix elementlev 1 .o RbK ( Cl Li Na Al 1 Pb iij 0.8 !t Ti Sr Ca Ba Zr 1 .o C 0 > > In C : 0.9 +- .- .- c .- s Q, .- c 0.8 U rr7 V, + a2V2 matrix elementlev \ pb‘N\a k; \ \ BiNi \ \ Mg \ MnFy \ \ \ Ti \ Zr \ \ \ \ \ Fig.4. Matrix effect on a single analyte, zinc, expressed as relative sensitivity, for a range of 17 matrix elements at 0.05 M expressed as a function of various properties of the matrix element. (a) First ionisation potential. No over-all trend is evident ( r = -0.14) although elements from Groups I and I1 of the Periodic Table lie on their own “pseudo trends” (broken lines). (b) “Total ionisation energ ” including first and second ionisation potentials and degrees of ionisation.Again, no over-all trend is evident ( r = -0.47). z) “Dissociation energy” derived from lattice energy without cation ionisation potentials. A discernible trend is evident but with a large amount of variability ( r = -0.86). (d) “Matrix energy demand” (MED) as summation of “total ionisation energy” and “dissociation energy” gives a linear relationship with a high degree of correlation ( r = -0.97). MED can therefore be used as a predictive property of 0.05 M. The actual concentrations used are given in column 8 in Table 4. In order to simplify data presentation, only one analyte line, Zn, is initially considered in the multi-matrix domain. The relative sensitivities, measured by the method described, are given in column 9 in Table 4. In this context calcium can be seen to cause larger matrix effects than most elements.Some elements, notably lanthanum, titanium and zirconium, produced greater effects than calcium. A primary objective of this work was to identify which fundamental properties of the matrix elements determined the magnitude of the matrix effect and hence enable useful predictions to be made. A property previously suggested to relate to the matrix effects of elements from Group I of the Periodic Table is the first ionisation potential of the matrix element. Fig. 4(a) shows the relative sensitivity data for Zn given in column 9 in Table 4 plotted against the first ionisation potential of the matrix element (column 1). (The value for the copper matrix was omitted owing to uncertainties caused by a large spectral overlap .) No functional relationship is evident (correlation coefficient r = -0.14), although points for both Group I and I1 elements fall misleadingly on their own “trends,” but in opposite directions.This illustrates the limitations of generalisations derived from studies encompass- ing only a few analytes, particularly when restricted to a single group of the Periodic Table. From a consideration of degrees of ionisation (calculated from the Saha equation, and given in columns 2 and 4 in Table 4), it is evident that at 7500 I( certain matrix elements are only partially first ionised ( e . g . , Be 54%), while others are significantly second ionised ( e . g . , Ba 4%). The total “ioni- sation energy,” VTOT, given in column 5 in Table 4 is thus more appropriate than the first ionisation potential, and was given by VTOT = aiVi + a2V2 where V is appropriate ionisation potential and a the corresponding degree of ionisation.Fig. 4(b) shows the same data (Zn relative sensitivity) plotted again this function VTOT190 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 5. Regression coefficients of analyte relative sensitivity against matrix energy demand (MED) Slope/eV-l x lo2 Intercept Analyte line Coefficient Ba . . . . -1.056 Be . . . . -2.328 Cd . . . . -2.877 CO . . . . -2.437 Cr . . . . -2.504 Cu . . . . +0.428 La . . . . -1.281 Li . . . . +1.064 Na . . . . +0.985 Ni . . . . -2.790 Pb . . , , -1.707 Rb . . . . +1.431 Sr . . . . -1.733 Zn . . . . -3.996 MO . . . . -2.456 V . . . . . .-1.874 Standard error 0.284 0.208 0.284 0.248 0.240 0.180 0.561 0.199 0.498 0.197 0.315 0.331 0.245 0.347 0.195 0.261 Coefficient 1.057 1.129 1.176 1.158 1.170 0.935 1.052 0.868 1.166 0.883 1.179 1.095 0.839 1.102 1.122 1.278 Standard error 0.018 0.021 0.028 0.024 0.024 0.018 0.053 0.020 0.049 0.020 0.031 0.033 0.025 0.034 0.019 0.026 1.1 I 1 0.7 Y Zn I I I I 1 8 10 12 14 16 ”Matrix energy demand”leV = MED Fig. 5. Variation in matrix effects on ten representative analytes as from a range of matrix elements expressed as a function of their MED. Regression coefficients are given in Table 5 for each matrix element. This again proved inadequate as a functional relationship ( r = - 0.47). Considering the matrix element to cause removal of energy from the plasma, a further possible energy requirement was the lattice energy of the species which entered the plasma after desolvation (e.g., A1C13 in HC1).Lattice energy (Un), however, is defined in relation to fully ionised products (e.g., Ca2+ or Al3+), which is clearly inappropriate in the ICP except for univalent ions. A modified lattice energy with the sum of the ionisation potentials subtracted, U, - X Vi,was therefore considered. This “dissociation energy” for each matrice element (column 6 in Table 4) is plotted against the Zn relative sensitivity data in Fig. 4(c). This relationship has a significant correlation ( r = -0.86) and showed a discernable trend, with La being now correctly extreme in both values. The next logical step was to combine the “dissociation energy” and the “ionisation energy,” VToT, in a total “matrix energy demand” (MED), M .This is given by n i = l M = un - c vi + a1 v1 + a2 v2 . . (11) i = l where Un is the lattice energy of the matrix element in the valency state n, as the chloride (e.g., for A1 in HC1 this would be the lattice energy of A1C13 expressed in eV). V, is the ith ionisation potential (in eV) summed to the maximum value n. The values of all these terms are given in column 7 in Table 4 for each matrix element. The values of MED for the matrix elements are plotted against the same Zn relative sensitivity data in Fig. 4(d). The trend is now clearly linear ( r = -0.97). Regression statistics are given in Table 5. The scatter of points can be attributed to three sources: (i) experimental error in the relative sensitivity estimated as +1%; (ii) uncertainties in the MED data, for example in the value used for ionisation temperature to calculate the degree of ionisation; and (iii) deviation from the model.Considering now the multi-analyte data, a consistent pattern emerges. The behaviour of the zinc line as an analyte in various matrices was typical of the complete range of analytes studied, although the other analytes were affected, as expected, to a smaller extent, dependent on their total excitation potentials. Each analyte line tested showed a similar linear relationship to that of zinc against the MED of the matrix element. Regression data are given in Table 5. All the standard errors on the slopes of the individual regression lines are similar, indicating equally small scatter of points about the line (although the slopes of a few are not significantly different from zero, e.g., Cu).The regression lines of ten representative analytes are shown in Fig. 5. Lines for the other analytes were excluded to improve the clarity of the diagram. The atom lines increase in relative sensitivity with increasing MED, the opposite of all the ion lines. This fact qualitatively supports the hypothesis of decreasing excitation temperature with matrix element addition, as the fraction of the element present as atoms would therefore increase whereas the population of ions should decrease according to the Saha equation. The net change in relative sensitivity is therefore determined by a balance between changing degree of ionisation and changing emission intensity.Matrix elements with values of MED less than 9 eV (e.g., Li, Na, K, Rb, Be, Ni, Cu, A1 and Pb) can be seen to produce no matrix effect exceeding +lo%, for any analyte, under the operating conditions used, The relationship between analyte relative sensitivities for one value of MED (i.e., Ca) has already been shown to be a function of the analyte line excitation potential (Fig. 1). This “pattern” is clearly seen in Fig. 5 , extending to other matrix elements with higher values of MED. The pattern becomes obscured at lower values of MED. A more systematic expression of the relationship should be exposed by plotting for each analyte the slope of the regression line (from Table 5 ) against their appropriate excitation potential. This is shown in Fig.6(a). There is clearly a single linear trend incorporating both the enhancement of the atom lines and the increasing suppression of the ion lines. (Only the Pb value lies far from the line which may indicate a suspect excitation potential value.) The regression statistics for the line are slope -0.00312 2 0.00012 and intercept 0.0161 k 0.0013. A visual inspection of Fig. 5 suggested that there also might be a systematic trend in the intercepts of the regression lines. These are plotted in Fig. 6(b) again, against the analyte line excitation potential. The least-squares regression line again gives a significant result: slope 0.0244 k 0.0013 eV-1 and intercept 0.835 k 0.014. The scatter of points, however, is greater than in Fig. 6(a) as the typical error bars indicate.These two regression equations can be combined to produce a single equation that predicts the relative sensitivity (Sim) of any analyte i line of excitation potential E, (eV) in any matrix element rn of matrix energy demand M,,, (eV) at a concentration of 0.05 M byJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 191 0.02 c I 2 0.00 e - B 2 a a Lc Q) v) $ -0.02 - a) \ \ \ \ \ \ \ Ba\\ T Sr \\ Pb La V \ 1 N i‘ cd\ \ \ \ I I I Zn, ‘\ g 1.2 1 / / -0.04 Total excitation potential of analyte/eV Fig. 6. potential of the analyte line. Both trends are linear and enable a predictive model of matrix effects to be constructed (a) Slopes and ( b ) intercepts of the regression lines shown in Fig. 5 (and listed in Table 5 ) as a function of the total excitation Matrix elements at 0.05 M RbPbNa Al Ca 1 .oo 0.95 \ ; 2p*T , l / l .o o ~ 1.05- ‘6 8 10 12 14 “Matrix energy demand“/eV Fig. 7. Contour map of predicted matrix effects generated with e uation (12) annotated with selected analyte and matrix elements. TXe contour lines are for values of equal relative sensitivity. To predict a relative sensitivity of any analyte - matrix combination, locate the co-ordinates of the excitation potential of the analyte line and the MED of the matrix element. Biases of >lo% occur only in the top right-hand corner of the diagram Sj, = M , (0.0161 - 0.00312 Ei) + 0.0244 Ei + 0.835 (12) The implications of this relationship can be best perceived by the graphical representation of equation (12) shown in Fig. 7.The two axes give the energies associated with the plasma processes, the abscissa showing the “matrix energy demand” for the matrix and the ordinate showing the excitation potential for the analyte. The contour lines are for constant values of predicted relative sensitivity (i.e. , matrix effect) from equation (12). Several general observations can be made from Fig. 7. A large area of the diagram (about 50%) has an expected bias of less than 5%. This confirms the general observation that ICP-AES is largely free from serious matrix effects. For example, no analyte will be seriously biased by a sodium matrix of 0.05 M. Further, there is only a relatively small area 0.8 I I I 1 I (ca. 25%) of the figure that predicts large (>lo%) matrix effects. These are restricted to analytes lines of high excitation potential (>8 eV) in matrices of large MED (>lo eV).A suggested use of this diagram is to identify matrix - analyte combinations that will require matrix matching or some form of bias correction.2 However, an obvious extension of this work would be to characterise the effect of different concentrations of matrix elements on equation (12). With this extra dimension it should be possible to calculate corrections for matrix effects from fundamental properties of the analyte and matrix elements over a range of matrix concentration. In order to make an initial appraisal to the likely performance of such a correction procedure, equation (12) was used to generate predicted relative sensitivities for each of the matrix - analyte line combinations previously measured.A graphical comparison of the predicted and experimental values is shown in Fig. 8. Of an original 258 measurements, 13 data points have been omitted where the experimental error was high owing to large translational interferences (back- ground or line overlap). The regression of experimental on predicted relative sensitivity gave a slope of 0.98 f 0.03, which is not significantly different from 1.0. This validates the predictive power of the model represented by equation (12). The standard deviation of the residuals was 0.025, which quantifies the average uncertainty of the proposed correction technique. For practical purposes the correction procedure would therefore reduce the bias derived from serious matrix effects to a few per cent.The accuracy of the correction should improve with a more refined model, based on more precise measurements of relative sensitivity and calculation of MED values. Mechanism of Plasma Matrix Effects For a given matrix element, the matrix energy demand is seemingly associated with a charcteristic fall in the excitation temperature, AT, as determined by equation (10). With equimolar matrices (0.05 M) there is a relationship between AT and the MED illustrated in Fig. 9. The regression line shown in Fig. 9 had the coefficients: slope 10.46 k 0.58 K eV-1 and intercept 74.30 t 5.80 K. The standard deviation of the value of AT calculated from this equation was estimated by simulation. It is also a function of MED but ranges from 7 to 11 K. The zinc analyte line sensitivity alone was employed to determine AT because of the relatively small uncertainties.However, all analytes gaveJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 I /@ I 1 I 0.8 0.9 1 .o 1.1 u. I 0.7 Predicted relative sensitivity Fig. 8. Closeness of fit of the predictive model and the experimental relative sensitivities from which it was generated. A method of matrix effect correction based on the current model would be expected to reduce most matrix effects (-30 to +5%) to within a few per cent \ \ -100 6 8 10 12 14 ” Matrix energy de m a n d “ /eV Fig. 9. Calculated fall in excitation temperature caused by a range of matrix elements using the zinc analyte as a thermometer. The trend is clearly linear and supports the hypothesis that matrix elements remove energy from the plasma in proportion to their MED linear relationships between their relative sensitivity and MED of the matrix element, and therefore they would all give linear relationships between AT and MED over this range.The smaller the slope coefficient of the regression, however, the greater is the uncertainty in the value of AT. The qualitative interpretation of this relationship is that the higher the MED of the matrix element, the more energy is taken from the plasma and the greater is the consequent drop in excitation temperature. The strong similarity between the effects on analyte sensitivities of the addition of calcium and a reduction of forward power2 lends further support to this hypo thesis. The similarity reported11 between the multi-analyte response to variations in forward power, injector gas flow-rate or air temperature can now be extended to include the effect of all matrix elements with MED greater than 10 eV.A common mechanism of variation in plasma excitation tem- perature would seem probable. This would also explain the successful extension of the “parameter related internal stan- dard method” (PRISM), initially devised for improving precision11 in removing bias due to matrix effects.2 It is not clear whether the empirical connection between MED and AT, which provides a predictive model of matrix effects, can be directly extended into a causal relationship. There seems to be a discrepancy between the values of the power absorbed by the matrix, and that released by the plasma gas in cooling by AT.The power absorption associated with the MED can be calculated, given a figure for the efficiency of solution nebulisation and transfer to the plasma. A power of 0.010 W is consumed when a 0.05 M Ca solution sprayed at 1% efficiency at a rate of 1.05 ml min-1 (the Ca MED = 11.28 eV or 1083 kJ mol-l). The power loss associated with the temperature drop AT = 43.7 K in 1 1 min-1 argon is much greater, however (at least ~ 5 0 ) . Conclusions A predictive model for matrix effects in ICP-AES was constructed that describes the relative sensitivity to be expected for any combination of analyte element line and matrix element at 0.05 M. The model is based on the total excitation potential of the analyte line and the “matrix energy demand” (MED) of the matrix element.The MED is calculated from fundamental properties of the matrix element and plasma. The theoretical basis for the model is based on the hypothesis that matrix effects are a symptom of a reduction in excitation temperature of the plasma caused by the matrix element. The decreases in excitation temperature measured were in the range 0-100 K. The hypothesis is consistent with all the experimental results for all matrices. Large matrix effects (>lo%) are found only where matrix elements with values of MED greater than 10 eV cause significant reductions in excitation temperature (>30 K). In that case only analyte lines with high values of excitation potential (>8 eV) will exhibit the large reduction in sensitivity. Conversely, most analyte - matrix combinations relatively free from matrix effects in ICP-AES can be readily identified. The rationale for the correction of matrix effects using the parameter-related internal standard method is explained by the model.An extension of the model to variable concentrations of matrix elements should enable a method to be devised for the routine correction of matrix effects. At present, however, the model cannot account quantitatively for the observed form of the relationship between matrix effect and matrix concentra- tion. Many matrices seem to exhibit the general “exponential decay” shape characterised for calcium.2 The multi-element domain, both multi-analyte and multi- matrix, has been shown to give an extra dimension to the study of matrix effects in ICP-AES. Whilst broadly confirming findings of spatial studies, multi-element patterns reveal important trends that would be impossible to perceive from data sets on limited numbers of analyte lines or matrix elements. The authors express their thanks to Mr. S. J. Walton of Applied Research Laboratories, Luton, for assistance with the use of the sequential spectrometer. References 1. 2. 3. 4. Greenfield, S . , McGeachin, H. McD., and Smith, P. B . , Anal. Chim. Acta, 1976,84, 67. Thompson, M., and Ramsey, M. H., Analyst, 1985,110,1413. Faires, L. M., Apel, C. T., and Niemczyk, T. M., Appl. Spectrosc., 1983, 37, 558. Blades, M. W., and Horlick, G., Spectrochim. Acta, Part B, 1981, 36, 881.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 193 5. 6. Boumans, P. W. J. M., “The Theory of Spectrochemical Excitation,” Hilger and Watts, London, 1966, p. 103. Wiese, W. L., Smith, M. W., and Miles, B. M., “Atomic Transition Probabilities,” Volume 2, US Department of Commerce, National Standard Reference Data System NRDS- NBS-22, National Bureau of Standards, Washington, DC, 1969. 7 . Alder, J. F., Bombleka, R. M., and Kirkbright, G. F., Spectrochim. Acta, Part B , 1980, 35, 163. 8. Boumans, P. W. J. M., “The Theory of Spectrochemical Excitation,” Hilger and Watts, London, 1966, p. 164. 9. 10. Drawin, H. W., and Felenbok, P., “Data for Plasmas in Local Thermodynamic Equilibrium,” Gauthier-Villars, Paris, 1965. Nojiri, Y., Tanabe, K., Uchida, H., Haraguchi, H., Fuwa, K., and Winefordner, J. D., Spectrochim. Acta, Part B , 1983, 38, 61. Ramsey, M. H., and Thompson, M., Analyst, 1985, 110,519. 11. Paper J566 Received November 26th, 1985 Accepted January 22nd, 1986
ISSN:0267-9477
DOI:10.1039/JA9860100185
出版商:RSC
年代:1986
数据来源: RSC
|
14. |
Studies of a low-noise laminar flow torch for inductively coupled plasma atomic emission spectrometry. Part 1. Fundamental characteristics |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 195-201
John Davies,
Preview
|
PDF (764KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 195 Studies of a Low-noise Laminar Flow Torch for Inductively Coupled Plasma Atomic Emission Spectrometry Part 1. Fundamental Characteristics John Davies Trace Analysis Laboratory, Department of Chemistry, Imperial College of Science and Technology, London SW7 2AY, UK Richard D. Snook Chelsea Instruments Ltd., 5 Epirus Road, London SW6 7UR, UK The fundamental properties of the laminar flow torch (LFT) designed in our laboratory are described. Noise power spectra are presented which show that the principle source of noise in a conventional tangential flow torch (TFT) has been removed in the laminar flow torch. Spatial profiles of Ca atom and ion emission intensities are observed to be displaced towards the induction coil in comparison with those in a conventional tangential flow torch. Electron densities measured in the LFT are lower than those in the TFT but Ca ion to atom ratios are higher in the LFT and hence ionisation temperatures are comparable to those in the TFT.Detection limits in the LFT are therefore superior by an order of magnitude compared with the TFT. Keywords: Inductively coupled plasma; laminar flow torch; noise One of the earliest differences of opinion that arose over the torch design of the inductively coupled plasma (ICP) was whether tangential flow of the coolant and plasma gases resulted in a more stable discharge and superior analytical performance than laminar flow. Reed,lJ in the design of his early torches employed tangential flow of gases as a means of recirculating some of the plasma to maintain the discharge.This is commonly known as vortex stabilisation. Greenfield et al.,3 in their original paper used the same method and such vortex stabilisation was thought to be essential for operation of the discharge. Tangential flow is an excellent method of establishing a vortex by creating a reduction of pressure at the tip of the injector tube causing some of the ionised gas to move countercurrent to the main gas flow. Wendt and F a ~ s e l , ~ however, who independently of Greenfield and co-workers reported a similar use of a radiofrequency discharge , described the use of a laminar flow torch where there is no recirculation of gases. Moreover they claimed that a vortex flow possesses more turbulence and decreases rather than increases the stability of the discharge.They created laminar flow by positioning a screen at the base of the torch where the coolant gas enters the base. It became evident,5 however, that there was little to choose between tangential and laminar flow. Despite this fact, commercial torches are nearly all designed with tangential gas inlets and, indeed, few workers4.6-9 have documented the use or study of a laminar flow torch. In a previous paper10 we described a laminar flow torch viewed in an axial (end-on) configuration which showed decreased noise to background levels and consequent superior detection limits compared with the conventional TFT. The reduction in noise was attributed to removal of turbulence in the injector channel caused by the tangentially flowing plasma and coolant gases.In this paper we describe the fundamental characteristics of the LFT and plasma viewed in the more conventional vertical (side-on) configuration and show that spatial profiles of ion and atom emission intensities, electron densities, ionisation and excitation temperatures and ana- lytical utility are comparable to those in the TFT. Detection limits in the LFT are superior, however, because of the reduction in source flicker noise in the LFT. Experimental Laminar Flow Torch Fig. 1 shows a schematic representation of the LFT designed in our laboratory. The plasma and coolant gases enter the torch base via inlets normal to the base rather than tangential as in the TFT. The coolant gas enters the torch base through the normal gas inlet and proceeds around a gallery from which it traverses axially up the torch base through a series of 21 2 cm I' - I1 II II Induction coil- Injector tube Ib IIII d'!, I1 II ii Ii Fig.1. Schematic representation of the laminar flow torch196 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 1. Argon gas data for laminar flow torch Flared portion Non-flared portion Parameter Coolant annuli Plasma annuli v h 3 s - ' . . . . . . 2.2 x 10-4 0.7 x 10-5 a h 2 . . . . . . . . 2.9 x 10-5 5.0 x 10-5 rJm . . . . . . . . 9.5 x 10-3 8.5 X 10-3 rilm . . . . . . . . 9.0 x 10-3 3.0 x 10-3 0.353 K . . . . . . . . 0.947 Nm . . . . . . . . Re . . . . . . . . - - 604 6 Coolant annuli 2.2 x 10-4 1.3 x 10-4 9.5 x 10-3 6.0 x 10-3 0.632 935 - Plasma annuli 21 holes 5.0 x 10-5 - 0.7 x 10-5 5.0 x 10-3 3.0 x 10-3 8.0 x 10-7 - - - 0.6 43 <1 - 2.0 x 10-3 Table 2.Operating conditions ICP instrumentation Parameter A: IPC-Monospek B: HF 1500 - Spex 1 1 R.f.power/kW . . . . Photomultiplier tube Entrancelexit slit voltage/kV . . . . 1 1.4 widths/pm . . . . . . 35 40 Entrance slit heightlmm 3 3 Argon coolant gad1 min-l 13 12 Argon plasma gas11 min- 0.4 1 .o Argon injector gad1 min-l 0.6 0.4 holes arranged circumferentially around the torch base. The torch base, fabricated from brass, is split into two parts to facilitate machining of the gallery. The criterion for the maintenance of the laminar flow is that the Reynolds number, Re is less than 2300.11 For an annulus12 Re is given by the relationship V .2(1 - K)T, Re = a.a v.1 R e = - where v = average gas velocity, a = kinematic viscosity of argon (1.25 x 10-5 m2 s-l), a = area of annulus, K = ri/ro, ro = radius of outer tube, rl = radius of inner tube and 1 = diameter of tube. Table 1 lists the values of the Reynolds numbers achieved in the LFT along with the appropriate data. and for a tubell a Instrumentation The high-frequency generator used in this work (except that described in Reduction in Noise, see later) was a conventional crystal controlled generator [International Plasma Corpora- tion (IPC) Model 120-271 operating at 27.12 MHz. Power from the generator was transferred via a coaxial cable (RG8/U) to a capacitively coupled matching network and water cooled 1.5 turn induction coil constructed from 6 mm outer diameter copper tubing with an over-all coil internal diameter of 29 mm.A single-channel monochromator (Rank Hilger Monospek 1000) equipped with an EM1 6256B photomultiplier tube was used for signal detection. Radiation from the discharge was focused as a 1 : 1 image on to the entrance slit of the monochromator using a 35 mm diameter silica lens with a focal length of 180 mm. Signal registration was achieved with a fast response potentiome tric chart recorder (Servoscribe RE 541.20). Sample introduction into the ICP was facilitated with a concentric glass nebuliser (Meinhard Associates Model T-230 A3) combined with a Scott-type double-pass spray chamber. The plasma was operated under the conditions shown in Table 2 unless otherwise stated. Throughout this work the dimensions of the torch base and silica tubing of the TFT were the same as those in the LFT.The configuration factor13 of the LFT and TFT was 0.95. Both torches utilised an extended coolant tube in order to reduce molecular band emission and air entrainment as previously reported14 (extending 40 mm beyond the induction coil). The noise power spectrum analyser was a Solatron 1200 Digital Signal Processor. Signals from the PMT were processed by this unit and presented as spectra of decibels, dB, (where dB = -20 logloV) versus frequency on a Hewlett-Packard 7470A X - Y plotter. For the noise power measurements a Plasma Therm HF 1500 RF generator and matching network were employed with a Spex 1-m monochromator. The operating conditions chosen for these studies are shown in Table 2, column B.Reagents All reagents used were of AnalaR grade (BDH Chemicals). De-ionised distilled water was used throughout the experi- ments. Operating Procedure The operating procedure and conditions used with the LFT are the same as those used with the TFT. Plasma ignition was achieved with plasma and coolant gas flow-rates of 0.4 and 13 1 min-1, respectively, and after ignition the aerosol was introduced. The introduction of the aerosol was found to be much easier with the LFT. Introduction of the aerosol into the TFT always had to be carried out with care and somewhat slowly, and in some instances extinguished the plasma even though the aerosol chamber had been sufficiently flushed with argon to remove any air. With the LFT, the aerosol could be introduced immediately after ignition without such precautions and has never extinguished the plasma.Results and Discussion Reduction in Noise The reduction in noise achieved by supporting a plasma in the LFT rather than in a TFT has been described in an earlier paper.10 The reduction in noise and consequent superior detection limits were attributed to establishing laminar flow of the gases rather than recirculation caused by vortex stabilisa- tion as so with conventional TFTs. Although the effects of this assumption could be observed the reduction in noise was not quantified. In this study we recorded noise power spectra to determine the frequencies at which the principle noise components are manifest in the LFT and TFT. For these measurements a 10 p.p.m.solution of Ca was nebulised into the ICP under the conditions shown in Table 2, column B. The Ca I1 393.366-nm line was monitored at a viewing height of 25 mm above the load coil (ALC). The signal generated at the PMT was fed into the Solatron 1200 and the output of the processor (the Fourier transform of the signal multiplied by its complex conjugate), the noise power spectrum, was displayed on the X - Y plotter. Fig. 2 shows the noise power spectra obtained from the LFT and TFT. For the TFT [Fig. 2(a)] there is a principle noiseJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 1 0 ( b ) 0) cn .- 50 i -100 197 40 ( a ) 117 0 component at 117 Hz, corresponding to a peak power of 2.1 dB Hz-1, which is believed to correspond to the modulation of the optical signal caused by the rotation of the plasma and coolant gases, which in turn causes non-centrosymmetric rotation of the injector channel.This phenomena was described by Belchamber and Horlick,lS although owing to the different design of torch and different operating para- meters they reported principle frequencies at 200-300 Hz. In our study, significant frequencies of lower power (area under the peak) in the TFT are evident at 59 and 630 Hz. It is difficult at this stage to assign the source of these components. The peak at 50 Hz is an instrumental artefact corresponding to the mains electrical frequency used in the UK. The white noise level (base line) for the TFT has a dB level of -23 dB equivalent to an average voltage of 71 mV. In the LFT a more stable plasma was sustained as is evident from Fig.2(b). Here we see that under the same experimental conditions there are no significant frequency components in the noise power spectrum other than the 50 Hz artefact. Moreover, the white noise level, caused by pseudo-random noise in the plasma is reduced to -42 dB or 7.9 mV. We believe that the combination of the removal of the component at 117 Hz and the reduction of the white noise by an order of magnitude is responsible for the improvement in detection limits by more than one order of magnitude reported in our previous paperlo and later in this paper. Size of the Plasma Discharge The most apparent physical difference between the LFT and the TFT observed was the size of the plasma discharges sustained in the two torches. The LFT plasma discharge was approximately 40% smaller than the discharge sustained in the TFT, extending for only 25 mm above the load coil in comparison with 40 mm in the TFT.The LFT discharge was extremely stable with no visible turbulence or flicker ob- served. Spatial Profiles The spatial profiles of calcium atom and ion emission intensities, electron densities and temperatures were measured in the LET and TFT for comparison. For all of these I I 1 1 1 5 10 15 20 25 Viewing height ALC/mrn Fig. 3. Effect of varying operating parameters on the Ca I signal emission intensity in the LFT. (a) Effect of injector gas flow-rate: A. 1.0; B, 0.8; and C , 0.6 1 min-1. (b) Effect of r.f. power: D, 1.2; E, 1.0; and F. 0.8 kW measurements the operating conditions were the same as those in Table 2, column A, except that the slit height was reduce to 2 mm to provide adequate vertical spatial resolution.Atomic and ionic emission intensities It is well known that changes in injector gas flow-rate and r.f. power have severe effects on the vertical spatial profile of soft line emission ( e . g . , Ca I) in the ICP and a much lesser effect on hard line emission (e.g. , Ca II).16J7 Consequently, we investigated the dependence of the emission profiles of Ca I and Ca I1 (422.673 and 393.366 nm, respectively) on theseJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 198 35 30 25 20 15 to CI .- 70 .- UJ .- E 6C 50 a, .- - a, CT 4c 3c 2c I ( , I 1 I 5 10 15 20 25 Viewing height ALCImm Fig. 4. Effect of varying operating parameters on the Ca I1 signal emission intensity in the LFT.(a) Effect of injector gas flow-rate: A, 1.0; B, 0.8; and C, 0.61 min-1. ( b ) Effect of r.f. power: D, 1.2; E, 1.0; and F, 0.8 kW parameters in the LFT. Figs. 3 and 4 show the effects of changing the injector gas flow-rate and r.f. power on Ca I and Ca 11, respectively. Fig. 3(a) shows that increasing the injector flow-rate from 0.6 through 0.8 to 1.0 1 min-l moves the position of Ca I peak emission intensity from 3 to 12 mm ALC. The reasons for this are two-fold. Firstly, the increased gas flow cools the energy input region such that the analyte takes a longer vertical distance to be desolvated and, secondly, more solvent is introduced into the plasma requiring more energy for complete desolvation.For Ca I1 the effect is also observed. At an injector gas flow-rate of 1.0 1 min-1 the position of peak emission moves to a higher viewing height but is also reduced in intensity. These results concur well with those of Blades and Horlickl6 who studied the vertical spatial emission profiles in a tangential vortex stabilised torch. An increase in the r.f. power also shifts the Ca I emission profile [Fig. 3(b)] to lower viewing heights but no such shift is observed for the Ca I1 emission profile [Fig. 4(b)]. In both instances, however, an increase in r.f. power increases the net signal intensity. Again this is in accordance with the work of Blades and Horlick.16 A more interesting observation can be made when we compare the effect of r.f. power on these profiles in the LFT plasma with those observed in the TFT plasma.Fig. 5 shows clearly that for all r.f. powers the net emission intensity of Ca I is lower in the LFT than in the TFT and that the peak emission 25 20 15 to t4 .z 10 L- h S .- ; 5 m i CI UJ C a, .- t4 .- C .: 70 to .- : a, m .- CI - 50 4c 30 2( 1 c a ) I ' I , 5 10 15 20 25 Viewing height ALC/mm Fig. 5. Comparison of the signal emission intensity in the LFT (broken line) and the TFT (solid line). Effect of r.f. power: A, 1.2,; B, 1.0; C, 0.8 kW. (a) Ca 1422.673 nm; and (b) Ca I1 393.366 nm intensities occur some 5 mm lower in the LFT under the same operating conditions. For Ca 11, however, the net emission intensity is higher in the LFT than in the TFT. Again the position of peak emission intensity is ca.5 mm lower in the LFT. The relative positions of peak intensities can be correlated with the background intensity and the electron density. Electron density The electron density, n,, was measured by determining the Stark half-width, AS1/2, of the broadened HP Balmer line at 486.133 nm, which can be related to n, by the relationship n, C(n,, T)AS1/23/2. The Doppler broadening component, AD, was calculated by assuming a temperature of 6000 K and using the relation AD = 7.16 x l o - 7 A . m where A = wavelength of the Doppler broadened line, T = temperature and M = atomic mass. The Stark half-width was finally obtained from the measured half-width after deconvolution of the Doppler and instrumental half-widths. The procedure is well documented in the work of Alder et al.18 Fig. 6 shows the spatial profile of electron densities for the LFT and the TFT. It is evident from the profiles that electron densities in the LFT are lower than those in the TFT by ca. 30%. Huska and Clump6 have observed that the electricalJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986. VOL. 1 199 70 60 I I I I I 5 10 15 20 25 Viewing height ALCImm 7 \ - - - c-- 6 ,/- \ - I I I I - I Fig. 6. Electron densities determined in A, the LFT-ICP and B, the TFT-ICP. For operating conditions see Table 2 0 .- +-I 2 2 ; 40- n 2 30- 0) in 5 0 - I3 0) Y 0 +-I - .- 20 10 Table 3. Fe I emission line data - - Fe I wavelengthhm 368.222 368.411 370.108 371.993 372.438 372.762 373.486 373.713 0 .- 4- z 2 -5z 0) - 4 ; - 3 : L2 0 c. P) in .- - 2 - 1 Excitation energy of the emitting upper Log oscillator factor level/cm-' (&nfrn.") 55754 0.28 49135 -0.28 51192 -0.01 26876 -0.43 45221 -0.70 34547 -0.52 33695 0.31 27 167 0.57 , 1 5 10 15 20 25 5 000 Viewing height ALCImm Fig.7. Ionisation and excitation temperatures determined in the LFT-ICP (broken line) and the TFT-ICP (solid line). For operating conditions see Table 2 coupling efficiency was lower when coaxial flow was used instead of vortex flow for stabilisation of the discharge. It may be therefore that the electrical coupling efficiency in the LFT is lower than that in the T M causing the electron densities to be that much lower. However, Miller and Aye1119 found in similar experiments that coupling efficiencies were not depen- dent on whether coaxial or vortex flow was utilised, in direct contrast to the results of Huska and Clump.The most important observation to be made from Figs. 5 and 6 is that the values of electron densities at the viewing heights where maximum emission intensity occurs for Ca I and Ca I1 are the same for both torches, i.e., Ca I1 emission peaks at an electron density of 1.1 x 1021 electrons m-3 and Ca I emission peaks at an electron density of 1.7 x 1021 electrons m-3 in both torches. Thus the difference in position of peak emission intensities in the LFT and T M are adequately explained by interpretation of the electron densities at these heights. The knowledge of electron density is therefore important when comparing the spatial emission profiles of emitting species in i I / I I I I I I L I I 1 5 10 15 20 25 Viewing height ALCImm Fig.8. Com arison of the signal to back round ratio of the Ca I 422.673 nm &) and the Ca I1 393.366 nm fB) lines in the TFT (solid line) and LFT (broken line); 1 p.p.m. of Ca 1500 v) C 3 > 4- .- F +-I .- g 1000 i I \ 5 10 15 20 25 Fig. 9. Comparison of the magnitude of background emission intensity at 422.673 nm in A, the LFT-ICP and €3, the TFT-ICp the plasma. Indeed Raaijmakers and co-workers20321 have shown that the state of an ICP can be adequately described by a measured electron density distribution, this measurement being independent of local thermal equilibrium (LTE). We have shown in this paper that maximum emission intensities occur at the same electron densities in both types of torches. It would appear therefore that any comparison between plasma performances and analytical utilities should include measurement and interpretation of electron densities.Ionisation and excitation temperatures The ionisation temperature, Tion,, was determined from the measurement of the relative emission intensities of the Ca I and Ca I1 lines and using the combined Saha - Boltzmann relations hip : Viewing height ALCImmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 200 2 m 0 X F . .- c I .- s E l 0 rn C En c - i7j / B \ P 0 X 0 7 . .- c I .- s 0 0 m C c3) m c - .- 5 10 15 20 Viewing height ALC/mm Fi . 10. Comparison of signal to noise ratios in the LFT (A) and TFT (Bg) for Ca I1 393.366 nm, 1 p.p.m. solution in water Table 4. Detection limits in aqueous media Detection limit, p.p.b.Literature Element Wavelengthinm TFT LFT values24 4 3 A1 Ba Ca Cr c u La Mn Na Ni Re Si Mg - . . . . . . . . . . * . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328.068 396.152 455.403 393.366 283.563 324.754 333.749 279.553 257.610 589.955 352.454 358.015 288.158 5.0 0.6 0.13 5.0 21 13 - 42 20 - - - - 0.5 0.83 0.05 0.010 0.5 0.4 0.7 1.8 0.09 0.97 2.6 0.7 1.6 4.7 0.87 0.13 4.7 3.6 6.7 0.93 19 20 19 30 81 18 where nca II = number density of the singly charged ion, nca I = number density of the neutral atom, rn = electronic mass, k = Boltzmann’s constant, h = Planck’s constant, 2 = partition function, Eion. = ionisation energy, 6E = lowering of ionisation energy and T = temperature. The excitation temperature, T,,,., was determined using iron as a thermometric species and applying the Saha - Boltzmann equation. A plot of log(ln,,h3/gJm,J verszu En, the energy of the upper level n, where g is the statistical weight, f the oscillator strength and I the intensity of the transition between the upper and lower energy levels n and m, gives, under thermal equilibrium conditions, a straight line graph with a gradient equal to (-2.303kTe,,,)-1. Eight Fe I lines were chosen (Table 3) whose transition probabilities are known from data by Bridges and Kornblinth,2* with un- certainties in the data estimated to be ? 10%. Fig. 7 shows the graph of Tion, and T,,,. in the LFT and the TIT from which it can be seen that temperatures are comparable in both torches. Maximum temperatures in the LFT occur lower in the discharge compared with those in the TFT.The Ca I1 to Ca I ratio in the LFT is ca. 40% higher than the same ratio in the TFT, which results in a higher value for Tion. being observed in the LFT even though the electron density is lower. Signal to Background and Signal to Noise Ratios An assessment of the merit of any analytical technique is given by the signal to background (SBR) and signal to noise (SNR) ratios. Fig. 8 shows a comparison of the SBRs for Ca I and Ca I1 emission intensity from which it is observed that the SBRs in LFT are higher for the Ca I1 but lower for the Ca I line in comparison with the TFT. The lower SBR for Ca I is a result of a fall in intensity of this emission line and not a result of an increasing background as can be seen from the absolute values of the background for both torches (Fig.9). The magnitude of the background at 1&12 mm in the TFT (where Ca I emission reaches a maximum) is of the same order as that at 4-6 mm in the LFT (where Ca I emission also reaches a maximum). Such correlation is also observed for the Ca I1 line. Signal to noise ratios (Fig. 10) are at least one order of magnitude higher in the LFT than in the TFT for the reasons discussed under Reduction in Noise, which results in superior detection limits. Detection Limits The detection limit of an element has been defined23 as that concentration of analyte which produces an output signal twice the root mean square of the background noise. Using this definition, detection limits were obtained for several elements by extrapolation for both the LFT and the TFT.The operating conditions of Table 2, column A, were used and compromise viewing heights of 18 and 12 mm were used in the TFT and LFT, respectively. In Table 4 these results are shown and compared with literature values.24 It is evident that detection limits are an order of magnitude lower in the LFT compared with the TFT and the literature values. This is partly due to an increase in the SNR (already discussed) and partly to an increase in the analyte absolute signal intensity. Dynamic Range The dynamic range of emission intensity with respect to concentration for Ag, Ba and Ca were measured at their wavelengths of 328.068, 455.403 and 393.366 nm, respec- tively. Linear calibrations of 5.5, 5.5 and 6.5 orders of magnitude were obtained, with curvature of the calibrations occurring at 100,50 and 20 p.p.m., respectively. These are an order of magnitude better than those obtainable with our TFT plasma.This would be expected as improvements in the linear range are made at the lower levels (i.e., improvements in detection limit) leaving the upper limit of linearity unchanged. Conclusions The LFT described here and in the previous paperlo shows similar characteristics to the conventional TFT. Temperatures are similar although electron densities are somewhat lower in the L M . A possible reason for this difference could be that the electrical coupling efficiency of the LFT is lower than that in the TFT. As a consequence, spatial profiles of atom and ion line emission intensities are observed to be shifted lower in the LIT.However, the behaviour of these profiles with respect to power is observed to be normal, i.e., the Ca I spatial emission profile shifts to lower viewing heights in the plasma as power is increased while no such shift is observed for the Ca I1 spatial emission profile. The analytical utility of the LFT has been shown to be superior to the TM in that the linear dynamic range is better while the reduction in noise makes the LFT superior in terms of detection limits. The major noise contribution in the LM-ICP is now due to the nebuliser and if this noise can be reduced, the LFT technique will approach detector noise limits. We acknowledge support for J. D. by the S.E.R.C. and Chelsea Instruments Ltd.under the C.A.S.E. Studentship Scheme. We also thank Dr. D. E. M. Spillane of the DIAS, UMIST, Manchester, for the use of the noise power spectrum analyser and Plasma Therm HF 1500 ICP.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 201 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Reed, T. B., J. Appl. Phys., 1961, 32, 821. Reed, T. B. , J. Appl. Phys., 1963, 34, 2266. Greenfield, S., Jones, I. L., and Berry, C. T., Analyst, 1964, 89, 713. Wendt, R. H., and Fassel, V. A., Anal. Chem., 1965,37,920. Greenfield, S . , Jones, I. L., McGeachin, H. M. C. D., and Smith, P. B., Anal. Chim. Acta, 1975, 74, 225. Huska, P. A., and Clump, C. W., Znd. Eng. Chem. Proc. Des. Dev., 1967, 6, 238. Wendt, R. H., and Fassel, V. A., Anal. Chem., 1966,38,337. Truitt, D., and Robinson, J . W., Anal. Chim. Acta, 1970, 49, 401. Truitt, D., and Robinson, J. W., Anal. Chim. Acta, 1970, 51, 61. Davies, J., and Snook, R. D., Analyst, 1985, 110, 887. Kirkbright, G. F., and Sargent, M., “Atomic Absorption and Fluorescence Spectroscopy,” Academic Press, New York, 1974. Bird, R. B., Stewart, W. E., and Lightfoot, E. N., “Transport Phenomena,” Wiley, New York, 1960. Allemand, C. D., and Barnes, R. M., Appl. Spectrosc., 1977, 31, 434. Davies, J., Dean, J. R., and Snook, R. D., Analyst, 1985,110, 535. Belchamber, R. M., and Horlick, G., Spectrochim. Acta, Part B , 1982, 37, 17. 16. 17. 18. 19. 20. 21. 22. 23. 24. Blades, M. W., and Horlick, G., Spectrochim. Acta, Part B , 1981,36, 861. Boumans, P. W. J. M., and DeBoer, F. J . , Spectrochim. Acta, Part B , 1972, 27, 391. Alder, J. F., Bombelka, R. M., and Kirkbright, G. F . , Spectrochim. Acta, Part B , 1980, 35, 163. Miller, R. C., and Ayen, R. J., Ind. Eng. Chem. Proc. Des. Dev., 1969, 8, 370. Raaijmakers, I. J . M. M., Boumans, P. W. J. M., Van Der Sijde, B., and Schram, D. C., Spectrochim. Acta, Part B , 1983, 38, 697. Schram, D. C., Raaijmakers, I. J. M. M., Van Der Sijde, B., Schenkelaars, H. J. W., and Boumans, P. W. J. M., Spectro- chim. Acta, Part B , 1983, 38, 1545. Bridges, J. M., and Kornblinth, R. L., Astrophys. I., 1974,192, 793. Willard, H. H., Merritt, L. L., Jr., Dean, J. A., and Settle, F. A., Jr., “Instrumental Methods of Analysis,” Sixth Edition, Van Nostrand, New York, 1981. Boumans, P. W. J. M., Spectrochim. Acta, Part B , 1981, 36, 169. Paper J5l.51 Received November 30th, 1985 Accepted January 24th, 1986
ISSN:0267-9477
DOI:10.1039/JA9860100195
出版商:RSC
年代:1986
数据来源: RSC
|
15. |
Modification of a commercial electrothermal vaporisation system for inductively coupled plasma spectrometry: evaluation and matrix effects |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 203-209
Henryk Matusiewicz,
Preview
|
PDF (885KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 203 Modification of a Commercial Electrothermal Vaporisation Svstem for Inductively Coupled Plasma Spectrometry: Evaluation and Matrix Effects Henryk Matusiewicz” and Fred L. Fricke Elemental Analysis Research Center, US Food and Drug Administration, I 14 I Central Parkwa y, Cincinnati, OH 45202, USA Ramon M. Barnes Department of Chemistry, GRC Towers, University of Massachusetts, Amherst, MA 0 1003-0035, USA The development and analytical utility of an electrothermal vaporisation technique (Em) employing a Perkin-Elmer HGA-500 graphite furnace for sample introduction into the inductively coupled plasma (ICP) are assessed. Evaluation of a novel configuration with custom graphite cuvettes is described. Operational characteristics, including the effect of the length of the transport tube to the ICP torch, vaporisation temperature, carrier argon flow-rate, observation height above the coil and plasma power, are investigated.Detection limits for Be, Cd, Co, Cu, Fe, Mn, Ni, Pb and Zn ranged from 0.6 to 20 ng ml-1 with precision varying from 1.2 to 8.5% RSD at 300 ng ml-1 (3 PI). The effects of major biological matrix constituents (Car Fe, K, Mg, Na and P) on the determination of trace elements by EN-ICP is also investigated. Significant enhancement and/or suppression of the analyte emission is observed. Keywords: Electrothermal sample introduction system; inductively coupled plasma; matrix effects In recent years emphasis on improving electrothermal vap- orisation (ETV) devices for introducing samples into an inductively coupled plasma (ICP) has been directed towards increasing the efficiency of sample transport into the ICP and lowering detection limits.Many versions of ETV devices have been explored for use with the ICP. Current designs for the introduction of micro-volume liquid samples have been reviewed. Several researchers have modified Perkin-Elmer graphite furnaces ( e . g . , HGA-74, HGA-500 and HGA-2000) for sample introduction into an ICP or microwave induced plasma (MIP) Aziz et al. used an HGA-74 graphite furnace with an aerosol transport system in which sample aerosol travelled through a 30-cm long glass tube to the base of an ICP or MIP tor~h.6~7 Crabi et aL8 modified an HGA-500 graphite furnace with a L’vov platform to introduce samples into an ICP.Recently Christian and co-workers+ll adapted an HGA-2000 graphite furnace to transport sample aerosol through 20 cm of tubing to a spray chamber connected to an ICP torch. This system was used for single and simultaneous multi-element analyses. Some of the problems associated with the ETV-ICP technique include matrix effects,5J712-17 pulse effects (or “piston” effect) ,8718 analyte deposition on chamber walls or in the transfer line (transport efficiency),”8~13.19.20 optimisation for simultaneous multi-element determinationlOJlJlJ2 (this will result in compromise parameters with higher detection limits than in the single-element mode) and formation of refractory carbides. 19 The ultimate scope and application of ETV-ICP will depend on its freedom from matrix interferences.Early reports, in general, described furnace vaporisation of samples into the ICP as being virtually free from matrix interferences.14 However, recent work has moderated this early optimism and demonstrated that matrix effects exist.6,8,12J-17 This had * Author to whom correspondence should be addressed. On leave, as an International Visiting Scientist, from Technical University of Poznan, Department of Analytical Chemistry, 60-965 Poznan, Poland. been predicted by Fassel23 who observed: “The problem associated with variable or incomplete sample vaporisation and effects due to changes in matrix composition and substrate condition on the vaporisation process would be expected to remain .” However, few interference studies have been published. Aziz et a1.6 found large matrix effects from biological samples (orchard leaves, bovine liver and serum) on the analyte emission.Ng and Carus012 observed both enhan- cement and depression on the analyte emission with synthetic ocean water. They reported a 190, 200, 40 and 20% enhancement of signal for 1 yg ml-l of Au, Cd, Li and Zn, respectively, and a 70 and 40% suppression of signal for 1 pg ml-1 of As and Sn, respectively. Crabi et al.8concluded that background correction was necessary for aluminium in the presence on varying amounts of sodium chloride and, indirectly, calcium. Hull and Horlickls observed no matrix effect from excess of Na (as NaC1) in the Pb emission signal. Millard et al. 13 observed enhancement of Cd and As analytical signals in the presence of concomitant elements such as Se.Kitazume16 observed that signals of B, Ge and P were suppressed with an increase in alkali metal concentration. Also when K or Na were present in the solution, 1&15% decreases in Ge and 30-50% decreases in B and P intensities were observed. Kirkbright and Snookl7 noticed that the detection limit for Cd was better in the presence of uranium than in its absence. This improvement in detection limits is due to the formation of unique aerosol crystals, which enhance the analyte transport efficiency between the vaporisation chamber and the injector tip. This paper describes a modification of an HGA-500 furnace that allows vertical mounting of the graphite tube, insertion of a graphite cuvette and direct connection to the base of an ICP torch. A variety of custom-made graphite cuvettes were evaluated.Optimum and compromise simultaneous multi- element conditions were established for Be, Cd, Co, Cu, Mn, Pb and Zn determinations. The effect of major biological matrix constituents (Na, K, Ca, Mg, Fe and P) on the emission signal, appearance time and peak shape of trace elements using this ETV-ICP technique were studied. Parameters were varied to minimise matrix effects from major matrix constituents.204 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 1. Experimental ETV-ICP compromise operating conditions Jarrell-Ash Model 1160 Plasma Atomcomp- Forward power1W . . . . . . . . 1000 Reflected power/W . . . . . . . . <lo Argon outer gas flow-rate11 min-l .. 18 rate11 min-1 . . . . . . . . . . 0.8 Argon intermediate gas flow- Argon carrier gas flow-rate11 min-l . . 1.65 Plasma torch . . . . . . . . . . Standard Jarrell-Ash, modified: orifice diameter 1.5 mm, straight injection tube Direct reader- Magnification of the entrance slit . . 3.6X HeightofthesWmm . . . . . . 3 Entranceslitwidth1pm . . . . . . 25 Exit slit width/pm . . . . . . . . 50 Electrothermal vaporiser argon flow-rates- Carrier . . . . . . . . . . . . Variable Outer11 min-1 . . . . . . . . . . 3.0 Perkin-Elmer HGA-50& Cycle TemperaturePC* Ramp/s Hold/s Drying . . . . . . . . 80 10 10 Intermediate heating . . . . 200 5 5 Vaporisation . . . . . . Variable 1 4 130 10 10 150 10 10 300 5 5 Ashing . . . . . . . . None Heatingrate .. . . . . >1000"C~-~ Cleaningcycle . . . . . . 280OaCfor4sifnecessary * Instrument setting. Experimental Instrumentation The Jarrell-Ash Model 1160 Plasma Atomcomp (Jarrell-Ash Division, Waltham, MA) ICP polychromator capable of monitoring 36 channels simultaneously used in this work has been described in detail previously.24 Operating parameters are listed in Table 1. The original interface between the PDP 11/34 and ICP data acquisition electronics was replaced by an 8085-based Intelligent Controller (Jarrell-Ash part number 004523). One-half second before the start of the furnace vaporisation cycle, the computer was activated, and sample exposure times of 0.1 s for a total duration of 6.3 s were used to calculate peak-area emission intensities. Operating parameters for the Perkin-Elmer HGA-500 graphite furnace (Perkin-Elmer, Norwalk, CT) also are listed in Table 1.Reagents Distilled, de-ionised water (18 M a ) was utilised throughout (Millipore Milli-Q water purification system). Calibration standards and solutions were prepared by appropriate serial dilution of 1000 or 10000 yg ml-1 ICP reference standards (Spex Industries, Metuchen, NJ). A solution of 100 ng ml-1 each of Be, Cd, Co, Cu, Mn, Pb and Zn was prepared in 2% nitric acid. Major matrix constituent solutions were made ranging from 25 to 1000 pg ml-1 of Ca, K, Mg, Na and P and 25 to 400 pg ml-1 of Fe (as nitrates). Also, each of the matrix solutions was prepared to contain 100 ng ml-1 of the trace elements. 8 7 t Argon carrier gas inlet port 2.0 Cross section Fig.1. Schematic diagram of the HGA-500 graphite furnace configuration used in the electrothermal vaporisation technique showing coupling with the ICP torch. All dimensions are in mm. (a) The system configuration: 1, graphite tube; 2, graphite rod; 3, graphite contact cylinders; 4, ceramic rings; 5 PTFE tubing (length 30 mm, i.d. 5.5 mm); 6, atomiser workheads; 7, PTFE seals; 8, quartz ball joint (1215, length 10 mm, i.d. 5.5 mm); and 9, window assembly. Pneumatic cylinder outer argon gas entrance and exit orts are not shown. Diameter of hole in a graphite tube is 3.Ix) mm. Tube dimensions are 28 x 5.8 (id.) mm, with wall thickness of 1.1 mm. ( b ) Details of a graphite rod Modified Electrothermal Vaporisation System The HGA-500 furnace was mounted vertically at the base of the ICP torch and is illustrated in Fig.l(a). In order to load samples into the furnace, in this arrangement, the furnace was re-designed to accept graphite cuvettes. After evaluation of several configurations, a pyrolytically coated graphite cuvette with a crater towards one end was selected [Fig. l(b)]. This end of the cuvette was also shaped to the contours of the graphite tube to ensure good electrical contact when inserted into the tube. The furnace assembly together with its temperature sensor was disconnected from its conventional adjustable mount for convenience. The sample introduction hole of the original, pyrolytically coated graphite tube (Perkin-Elmer No. B0109-322) was enlarged with a drill, or a slot was cut in order to insert the graphite cuvette.The two holes in the bottom (right side) graphite contact cylinder of the HGA-500 were plugged with graphite plugs and a new hole o r slot was aligned with the hole or slot in the graphite tube. Ceramic rings were placed between the two graphite contact cylinders and between the cylinders and window assemblies to form a gas-tight seal. The two quartz windows were replaced with two PTFE tubes to connect the argon carrier gas to theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 205 ETV and the ETV to the ICP. The PTFE tubes were held firmly in place against the ceramic rings by O-rings contained in the original window assemblies in order to prevent analyte loss. The inner diameters of the tubes and ceramic rings were selected to match the inner diameter of the graphite tube (5.8 mm).The upper PTFE tube was connected to a quartz tube with a PTFE seal, and the quartz tube was connected to the ICP torch with a ball and socket joint (12/5). The distance between the graphite cuvette and the injection tube was approximately 17 cm. The argon carrier gas was connected to the bottom PTFE tube and was controlled with an electronic mass flow controller (Tylan Model FC-260). The ICP outer gas flow was also controlled (Tylan Model FC-261). The graphite tube and contact cylinders were cooled by a separate argon source, which entered the original holes in the graphite contact cylinders and flowed into the space between the graphite tube and the contact cylinders. In addition, each electrode block clamp was independently water cooled.Results and Discussion Parameter Selection The ICP outer and intermediate argon gas flow-rates, the furnace drying temperature and time, the heating rate, the outer argon flow around the graphite tube and solution injection volume for each cuvette design were kept constant for all elements investigated, based upon earlier res~lts.2~4J1~22 The effect of graphite cuvette design, length of the transport tube from the graphite cuvette to the ICP, carrier argon flow-rates, vaporisation temperature, viewing height and radiofrequency power were examined. The analy- tical operating parameters for standard solutions were establi- shed to achieve the best compromise detection limits for several elements in order to permit simultaneous multi- element determination. Graphite Cuvettes A number of factors must be considered in the configuration of a cuvette for electrothermal vaporisation in the HGA-500 graphite furnace. These factors limit the cuvette dimensions, shape and material.The major difficulty with graphite cuvettes is the formation of refractory elemental carbides. Carbide formation can be reduced with pure pyrolytic graphite cuvettes.25 To obtain high sensitivity a cuvette that can accommodate as large a sample volume as possible is desirable. However, the cuvette mass should be minimised to ensure fast heating rate and rapid vaporisation of each analyte. Several configurations were examined, all of which were fabricated by shaping graphite rods and platforms. The absolute detection limits (pg) were essentially the same for all designs; however, the concentration detection limits (ng ml- l) are limited by the crater size formed in the cuvette.A 51-11 sample was injected into the graphite rod with a crater towards one end. The rod with a crater that had a capacity of 15 pl was used for most of the studies. Aging of the graphite cuvette and tube was not apparent until after 100 firings, when the tube and crater surface were noticeably eroded. Graphite cuvettes with a crater were fabricated from a 3.05 mm spectrographic rod (Bay Carbon, Inc., grade BCI 100 MD and/or Ultra Carbon ultra “F” purity grade U-7) and then pyrolytically coated by the manufacturer. Graphite platforms with a crater were made from plates of anisotropic pyrolytic graphite (Super-Temp Operations, Santa Fe Springs, CA).Their thicknesses varied from 1.0 to 3.0 mm, and they were typically 35 mm long by 3.5 mm wide. The craters, 0.5-2.5 mm deep, 4.0 mm long and 3.0 mm wide, held from 6.0 to 30.0 1-11. The cuvettes were hand sanded to fit snugly in the hole or slot in the graphite tube. The platforms were unsuitable for use in the HGA-500, because the sample tended to spread to the edge of the platforms and then wet the tube wall. Also, to machine a rectangular slot into the tube to accommodate a platform was more difficult than to drill a round hole in the rod. Sample Transport Distance and Argon Carrier Flow-rate The main effects of increasing transport distance are a decrease in emission signal, apparently due to loss of aerosol, increase in the time between starting the electrothermal vaporisation cycle and arrival of the aerosol at the analytical zone in the ICP, a decrease in instantaneous noise and an increase in peak broadening.8~18.20.26~27 Crabi et a1.8 and Gunn et a1.18 recommended a minimum connecting tube length, which was a main consideration in our arrangement.The effect of transport distance on the analytical perfor- mance of the present system was studied, and results are compiled in Fig. 2. The maximum signal was observed at the shortest distance and tends to a minimum at greater than 27 cm. The analyte appearance time and peak shape did not change at any of the transport distances measured (17,22,27 and 32 cm). In the present design, a distance of 17 cm from the graphite cuvette to the load coil was the shortest practical length for convenient access.An optimum flow-rate range of 1.3-2.0 1 min-1 was observed based upon signal to background ratios (SBR) for the elements investigated, and a compromise flow-rate of 1.65 1 min-1 was used. Furnace Heating Rate and Temperatures Based upon earlier work2.8 in which a fast heating rate was applied, the fastest ramp time available for the atomisation cycle was chosen. In this instance, the rate of sample introduction is independent of the carrier gas flow-rate and is determined mainly by the rate of applied temperature ramp. Three drying stages (see Table 1) were used for smooth drying without losses by sputtering and to avoid extinguishing the plasma. As experiments were conducted with pure aqueous or slightly acidified solutions, an ashing step was not necessary.However, to ensure complete drying an intermediate heating 300 v) D C 0 $ 200 v) v) C 3 c 8 2 . ([I Y ([I 100 4- a, z 0 17 22 27 32 Distancekrn 3000 v) U 2000 $ In v) C 3 c 00 2! 1000 g . lu x lu c z 0 Fig. 2. Effect of the distance from the graphite cuvette to the load coil on the emission signal for A1 308.2, Cd 228.8, Mn 257.6, Pb 220.3 and Zn 213.8 nm. Aliquots of 3 pl of 300 ng ml-1 of Cd, Mn, Pb and Zn and 1000 ng ml-1 of A1 solutions. Carrier flow-rate 1.65 1 min-l206 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 2. Comparison of detection limits obtained with the graphite tube - rod-crater configuration (3 ~ 1 ) and ETV-ICP* - simultaneous multi-element analysis systems Detection limit? Present work (area) Reference 11$ (peak height) Element Ag .. . . A1 . . . . Ba . . . . Be . . . . Ca . . . . Cd . . . . c o . . . . Cr . . . . c u . . . . Fe . . . . Mn . . . . Ni . . . . Pb . . . . Sb . . . . Si . . . . Zn . . . . Wavelengthlnm - . . - . . - . . . . 313.0 . . 228.811 . . 228.6 . . 324.7 . . 259.9 . . 257.6 . . 231.611 . . 220.3 - . . - . . - . . - . . . . 213.811 ng ml-1 - - - 0.6 3 4 13 4 0.7 20 12 - - - - 2 Pg - - - 1.8 9 12 39 12 60 36 - - 2.1 - - 6 ng ml-l 10 170 150 30 20 20 5. 10 10 3 50 50 250 50 7 - Pgll 100 1700 1500 300 200 200 50 100 100 30 500 500 2500 500 70 - * Vaporisation temperature 2800 "C for 4 s, carrier flow-rate 1.65 1 min-l. With a Perkin-Elmer HGA-500 and Jarrell-Ash Model 1160 Plasma ? The limit of detection corresponds to the concentration giving a signal equivalent to three times the standard deviation of the background.$ Detection limits (3a), 10-pl samples, Ag 320.0, A1 308.2, Ba 493.4, Ca 396.8, Cd 228.8, Co 228.6, Cr 205.5, Cu 324.7, Fe 259.9, Mn 257.6, Ni 231.6, Pb 220.3, Sb 217.5, Si 251.6 and Zn 213.8 nm. Vaporisation temperature 2400 "C for 10 s, carrier flow-rate 1.0-2.5 1 min-1. With a Perkin-Elmer HGA-2000 and Jarrell-Ash Model 955 Plasma Atomcomp; pyrolytically-coated graphite tube. 0 Samples 5 p1; Ag 328.07, Cr 267.72, Cu 324.75 and Mn 257.61 nm. Carrier flow-rate 0.8 1 min-I. With a Varian CRA-90 and Jarrell-Ash Model 965 Plasma Atomcomp; tantalum boat. 7 Our calculation based on reported sample volume used. 11 Used in the second order on the polychromator.Atomcomp. step was applied; Be, Co, Cu and Mn required a vaporisation temperature of 2800 "C to obtain the best signal to background ratios. A higher temperature could not be applied as the graphite cuvettes and tubes deteriorated, and carbon depo- sited in the transport tube and torch. The temperatures were not measured with an optical pyrometer but were instrumental settings and should be regarded as approximate values. Plasma Viewing Height and R.f. Power The viewing height was varied from 2 to 16 mm in 2-mm increments above the load coil. The argon flow-rate was fixed at 1.65 1 min-1. An optimum height range of 4-10 mm was observed for the elements investigated, and a compromise height of 8 mm was used. The best SBR for Be was at 4 mm; Cu and Fe at 6 mm; Cd, Co, Ni, Pb and Zn at 8 mm; and Mn at 10 mm.These heights are considerably below the observation height (15-25 mm) commonly used with nebuliser introduc- tion systems, probably because the electrothermal atomiser delivers a dry aerosol to the plasma. The r.f. power was varied from 600 to 1500 W in 100-W increments. The argon flow-rate was fixed at 1.65 1 min-1 and the viewing height maintained at 8 mm above the load coil. The best SBR for Be was at 1300 W; Ni at 1100 W; Fe, Pb and Zn at 1000 W; Cd and Mn at 900 W; Co at 800 W; and Cu at 700 W. A compromise r.f. power of 1000 W was used, in agreement with previous w ~ r k . ~ - ~ ~ J ~ J ~ , ~ ~ Analytical Performance Detection limits were determined by using 3 p1 of a solution containing 300 ng ml-1 of analyte.The detection limit was defined as the concentratjon required to produce a net emission intensity equivalent to three times the standard deviation of the background. The signal was defined as the integral area (summing each 0.1-s signal under the total peak) of the peak minus the integral area of the background over the same measurement time. The detection limits for several elements are presented as both mass and concentration in Table 2. The results are compared only with those obtained using the ETV-ICP in the simultaneous multi-element mode and the same model of polychromator.l1322 In most instances improved detection limits were obtained with the present method, possibly as a result of better aerosol transport resulting from the minimised furnace torch distance.Detec- tion limits using single-element determination and various spectrometers have been given by Matusiewicz and Barnes.28 In these studies (3 pl of 300 ng ml-1) the precision, for ten replicate determinations, typically ranged from 1.2 to 8.5% RSD. Precision of the background at the analytical wavelength ranged from 1.3 to 3.4% RSD. The linear range for this system is 3-4 orders of magnitude for the elements studied. However, concentrations in excess of 10 pg ml-1 for all elements require extensive cleaning of the system and several repeated firings. In agreement with previous findings,28 upon initiation of the vaporisation cycle the background change is not noticeable and is without a negative peak. This may be a consequence of the small volume of argon heated in the graphite tube, as the internal volume of the tube is approximately 1 ml.The volatile elements (Cd, Pb and Zn) appear slightly before the more refractory elements (Be and Co) in agreement with other reports.11,2*,22 Details are given under Individual Element Characteristics. The emission peaks are sharp and end with a slight tail. The range of the arrival times is 1.6-2.5 s, and the total peak widths at the base are 1-2.5 s.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 207 ~~ Table 3. Effects* of major elements on 100 ng ml-1 (500 pg) of analyte obtained with the ETV-ICP technique? in 2% nitric acid Concentration Cd228.89 . Co228.6 . Cu324.7 . Mn257.6 . Pb220.3 . Zn213.89 . 100 200 250 400 1000 4000 25 100 200 250 400 1000 4000 25 100 200 250 400 1000 4000 25 100 200 25 0 400 1000 4000 25 100 200 250 400 1000 4000 25 100 200 250 400 1000 4000 25 100 200 250 400 1000 4000 of major element/ Element and line/nm pg ml-1 Be313.0 .. . . . . 25 . . . . . . . . . . . . . . . . . . . . I . . . . I Signal enhancement and/or suppression, Yo K -2 -3 ND$ -4 ND - 15 ND +31 + 43 ND + 37 ND + 14 ND + 10 +6 ND +7 ND -3 ND + 24 + 18 ND + 14 ND +8 ND + 63 + 46 ND + 54 ND + 40 ND + 33 + 29 ND + 25 ND + 24 ND + 43 + 22 ND +10 ND +7 ND Na + 13 +3 ND +8 ND + 15 + 13 + 72 + 66 ND + 65 ND + 29 + 17 +2 -3 ND +5 ND 0 - 13 +2 0 ND -4 ND - 13 -41 + 57 + 77 ND + 142 ND +116 + 95 0 +3 ND 0 ND -1 - 40 + 79 + 78 ND +61 ND + 39 + 10 Ca +21 + 12 ND -5 ND -5 ND + 13 +2 ND -4 ND -6 ND + 25 + 24 ND +11 ND +9 ND - 12 -2 ND +3 ND - 14 ND + 22 + 46 ND + 48 ND + 62 ND +2 +5 ND +6 ND + 23 ND + 29 + 22 ND +21 ND - 17 ND Mg +2 -7 ND - 25 ND -61 ND +9 + 26 ND +139 ND + 145 ND +4 -1 ND -5 ND - 10 ND - 24 - 9 ND - 12 ND - 13 ND +6 + 13 ND + 26 ND + 104 ND -6 -3 ND -3 ND - 14 ND + 28 + 24 ND + 28 ND + 47 ND P -1 + 16 ND +11 ND +9 ND + 37 + 55 ND +81 ND + 126 ND -6 + 12 ND 0 ND - 18 ND -5 -18 ND - 30 ND - 43 ND -8 +9 ND +11 ND +21 ND + 64 +53 ND +21 ND + 10 ND +3 -1 ND 0 ND + 18 ND Fe +6 +3 +1 ND 0 ND ND -11 -7 -2 ND +1 ND ND -6 - 12 - 19 ND - 28 ND ND ND ND ND ND ND ND ND +4 + 10 +21 ND + 36 ND ND - 15 - 36 - 46 ND - 62 ND ND +7 -4 +1 ND + 24 ND ND Cx - CHNO~ * Signal enhancement or suppression (%) = analyte emission intensity in 2% nitric acid.-t ETV-ICP operating conditions: carrier flow-rate 1.65 1 min-1, r.f. power 1000 W, vaporisation temperature 2800 "C, height above coil 8 mm and injection volume 5 p1.$ ND = not determined. 9 Used in the second order on the polychromator. X 100 where C , = net analyte emission intensity in matrix x; and CHNOl = net CHN03 Major Matrix (Inorganic Salts) Effects The effects of adding various amounts of Ca, Fe, K, Mg, Na and P on the trace element (Be, Cd, Co, Cu, Mn, Pb and Zn) signals are summarised in Table 3 for compromise operating conditions established without major concomitants. Varied and significant, in some instances severe, enhancements and/or suppressions were observed with all major concomi- tants. Generally, signals were enhanced when the alkali or alkaline earth elements were present.These effects are not due to spectral interferences, stray light or a background change. This was demonstrated by vaporising a blank (2% HN03) solution and the concomitant elements and measuring at and near the analytical line peak wavelength (20.03 and k0.06 nm). This indicated that the effects are probably a combination of physical thermal vaporisation effects occurring208 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 during ETV, transport effects and changes in the plasma characteristics.5 In general, background correction is desir- able but not necessary in every instance. The viewing height was varied from 2 to 16 mm above the load coil with a constant r.f. power of 1000 Win an attempt to minimise the matrix effects. The data indicate that above or below 8 mm the effects remained the same or became worse.The power was varied from 600 to 1500 W with a constant viewing height of 8 mm. The matrix effects can be reduced by changing the power for each element in a particular matrix element, however no suitable compromise power could be found to eliminate the matrix effects simultaneously. Individual Element Characteristics Beryllium The presence of Ca, K and Na did not change the beryllium signal appearance time, which was 2.5 f 0.3 s. However, 1000 pg ml-1 of Mg increased the appearance time by 0.6 s, 1000 pg ml-1 of P and 400 pg ml-1 of Fe decreased the time by 0.6 and 0.3 s, respectively. At a Be concentration of 100 ng ml-l, 300 times the detection limit, background correction for each matrix was necessary.The Be 313.04-nm line is located between two unknown peaks (313.03 and 316.06 nm), probably due to the OH molecule.29 The Na and Mg did not affect the peak width; K at 25-1000 pg ml-1 and Fe at 25-400 pg ml-1 broadened the peak by 0.3 and 0.5 s, respectively; P increased the peak width from 0.1 to 0.8 s as the concentration of P was increased from 25 to 1000 pg ml-1; Ca narrowed the peak by 0.5 s. The increase in Be appearance time in the presence of Fe is probably due to interactions in the graphite cuve tte. Cadmium The cadmium signal appearance time was 1.6 k 0.3 s and was not changed by the Na and Mg matrix. The presence of Ca, K and P at 25-1000 pg ml-1 and Fe at 25-400 pg ml-1 increased the appearance time by 0.2,0.3,0.8 and 0.2 s, respectively.At a Cd concentration of 100 ng ml-1, 50 times the detection limit, no background correction for each matrix was neces- sary. Ca, K and Na did not affect the peak width. The peak width was increased by Mg at 100 pg ml-1 by 0.4 s, at 250 pg ml-1 by 0.6 s and at 1000 pg ml-1 by 0.7 s; 25 pg ml-l of P increased the peak width by 0.1 s, 100 pg ml-1 by 0.3 s, 250 pg ml:l by 0.6 s and 1000 pg ml-1 by 0.8 s; 100 and 200 pg ml-l of Fe increased the peak width by 0.2 s and 400 pg ml-l by 0.6 s. The increase in Cd appearance time is due to the formation of cadmium phosphate. Cobalt The cobalt signal appearance time was 2.4 k 0.3 s. The presence of Ca, K, Mg, Na and P did not change the signal appearance time. Fe at 25-400 pg ml-l increased the appearance time by 0.5 s.At a Co concentration of 100 ng ml-1, 40 times the detection limit, no background correction was necessary for K, Na, Mg, P and Fe. Ca produced a background enhancement; K, Ca and Fe did not affect the peak width; Na at a concentration of 25-1000 pg ml-1 narrowed the peak by 0.3 s; P at a concentration of 25-1000 pg ml-1 broadened the peak by 0.2 s; 1000 pg ml-l of Mg broadened the peak by 0.3 s. The increase in Co appearance time in the presence of Fe is probably due to interactions in the graphite cuvette. Copper The copper signal appearance time was 2.2 k 0.3 s. The presence of Fe, Mg, Na and P did not change the appearance time. Ca and K (1000 pg ml-1 of each) increased the arrival time by 0.3 s; at a Cu concentration of 100 ng ml-1, 12 times the detection limit, no background correction was necessary for Fe, Mg, Na and P; Ca and K produced a background enhancement; K and P did not affect the peak width; Na at a concentration of 25-1000 pg ml-1 narrowed the peak by 0.3 s; Ca at 1000 yg ml-1 narrowed the peak by 0.4 s; Mg at 1000 pg ml-1 broadened the peak by 0.3 s.Manganese The manganese signal appearance time was 2.2 k 0.3 s. The presence of Fe, K, Mg, Na and P did not change the appearance time, but Ca increased the time by 0.3 s. At an Mn concentration of 100 ng ml-1, 240 times the detection limit, Ca, Fe, K, Mg, Na and P all caused background enhancement; Na did not affect the peak width; K, Mg and P each at 1000 pg ml-1 broadened the peak by 0.9,0.3 and 0.4 s, respectively; Ca at a concentration of 25-1000 pg ml-1 narrowed the peak by 0.5 s; 100 and 200 yg ml-1 of Fe broadened the peak by 0.5 s and 400 pg ml-1 by 0.7 s.Lead The lead signal appearance time was 1.7 k 0.3 s. The presence of K and Na did not change the appearance time. The presence of Ca, Mg and P at 25-1000 pg ml-1 increased the time by 0.3, 0.7 and 0.7 s, respectively; Fe at 25-400 pg ml-1 increased the time from 0.1 to 1.0 s; at a Pb concentration of 100 ng ml-1, 14 times the detection limit, no background correction was necessary for Fe, K, Mg, Na and P; Ca caused a background enhancement; 1000 pg ml-1 of K broadened the peak by 0.4 s; Ca from 100 to 1000 pg ml-1 broadened the peak by 0.4 s; Mg, Na and P at concentrations of from 25 to 1000 pg ml-1 narrowed the peak by 0.4, 0.7 and 0.3 s, respectively; Fe at a concentration of 400 pg ml-1 narrowed the peak by 0.6 s.The increase in Pb time in the presence of P is due to the formation of lead phosphate. The increase in Pb appearance time in the presence of Mg and Fe is probably due to the interactions in the graphite cuvette. There is no effect on the Pb intensity from the addition of Na up to a ratio of 10 000 : 1 , which is consistent with Hull and Horlick's study. l5 However, with the ratio of 40 000 : 1 , a 40% decrease in the Pb intensity was observed. Zinc The zinc signal appearance time was 1.7 k 0.3 s. The presence of K, Na and P did not change the time; Ca and Fe each increased the appearance time by 0.3 s; Mg also increased the time from 0.2 to 0.8 s as the concentration of Mg was increased from 25 to 1000 yg ml-1; at a Zn concentration of 100 ng ml-1, 83 times the detection limit, no background correction for each matrix was necessary; Fe,'K, Mg, Na and P did not affect the peak width; Ca at a concentration of from 25 to 1000 pg ml-1 broadened the peak by 0.2 s.The increase in Zn appearance time in the presence of Mg is probably due to interactions in the graphite cuvette. Conclusions The experimental results summarised above suggest that the present furnace design could be useful in the development of ETV-ICP. Generally, the furnace should have a small internal volume and have the option for using some type of platform. The transport distance from the furnace to the ICP should be as short as possible. In most instances a lower observation zone compared with wet aerosol introduction via pneumatic nebulisation is optimum for the determination of trace elements. A number of matrix effects were observed with this ETV-ICP system, and two parameters, viewing height and power, were investigated in an attempt to eliminate or minimise the effects.We were not successful in minimising the effects, and further work is required to elucidate and minimise them. Experiments are needed to differentiate betweenJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 209 furnace vaporisation/atomisation, transport and plasma effects. Meanwhile, standard addition or matrix matched standards and samples are required for ETV-ICP analyses with this arrangement. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.13. Matusiewicz, H., and Barnes, R. M., Appl. Spectrosc., 1984, 38, 745. Matusiewicz, H., and Barnes, R. M., Spectrochim. Actu, Part B , 1984, 39, 891. Browner, R. F., and Boorn, A. W., Anal. Chem., 1984, 56, 786A and 875A. Ng, K. C., and Caruso, J. A., Appl. Spectrosc., 1985,39, 719. Long, S. E., Snook, R. D., and Browner, R. F., Spectrochim. Acta, Part B , 1985, 40, 553. Aziz, A . , Broekaert, J. A. C., and Leis, F., Spectrochim. Actu, Part B , 1982,37, 369. Aziz, A., Broekaert, J. A. C., and Leis, F., Spectrochim. Acta, Part B , 1982, 37, 381. Crabi, G., Cavalli, P., Achilli, M., Rossi, G . , and Omenetto, N . , At. Spectrosc., 1982, 3, 81. Swaidan, H. M., and Christian, G. D., Can. J . Spectrosc., 1983, 28, 177. Hertenstein, S. D., Swaidan, H. M., and Christian, G. D., Analyst, 1983, 108, 1323. Swaidan, H. M., and Christian, G. D., Anal. Chem., 1984,56, 120. Ng, K. C., and Caruso, J. A., Anal. Chem., 1983, 54, 1513. Millard, D. L., Shan, H. C., and Kirkbright, G. F., Analyst, 1980, 105, 502. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. Rica, C. C., Kirkbright, G. F., and Snook, R. D., At. Spectrosc., 1981, 2, 172. Hull, D. R., and Horlick, G., Spectrochim. Acta, Part B , 1984, 39, 843. Kitazume, E., Anal. Chem., 1983, 55, 802. Kirkbright, G. F., and Snook, R. D., Appl. Spectrosc., 1983, 3 7 , l l . Gunn, A. M., Millard, D. L., and Kirkbright, G. F., Analyst, 1978, 103, 1066. Kirkbright, G. F., and Snook, R. D., Anal. Chem., 1979, 51, 1938. Kirkbright, G. F., Millard, D. L., and Snook, R. D., Spectrochim. Acta, Part B , 1983, 38, 649. Blakemore, W. M., Casey, P. H., and Collie, W. R., Anal. Chem., 1984, 56, 1376. Tikkanen, M. W., and Niemczyk, T. M., Anal. Chem., 1984, 56, 1997. Fassel, V. A., Am. SOC. Test. Muter., Spec. Tech. Publ. No. 618, 1977, p. 22. Wolnik, K. A., Kuennen, R. W., and Fricke, F. L., in Barnes, R. M., Editor, “Developments in Atomic Plasma Spectro- chemical Analysis,” Heyden, Philadelphia, 1981, p. 685. Slavin, W., and Manning, D. C., Spectrochim. Acta, Part B , 1982,37, 955. Kirkbright, G. F., Pure Appl. Chem., 1982, 54, 769. Kirkbright, G. F., and Snook, R. D., Appl. Spectrosc., 1983, 37, 11. Matusiewicz, H., and Barnes, R. M., Spectrochim. Acta, Part B , 1985, 40, 29. Schramel, P., and Xu, L.-Q., Anal. Chem., 1982, 54, 1333. Paper JAY1 Received September 27th, 1985 Acccepted January 15th, 1986
ISSN:0267-9477
DOI:10.1039/JA9860100203
出版商:RSC
年代:1986
数据来源: RSC
|
16. |
Application of inductively coupled plasma mass spectrometry (ICP-MS) for trace metal determination in foods |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 211-219
Seumas Munro,
Preview
|
PDF (931KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 211 Application of Inductively Coupled Plasma Mass Spectrometry (ICP-MS) for Trace Metal Determination in Foods Seumas Munro and Les Ebdon Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth, Devon PL4 8AA, UK David J. McWeeny Ministry of Agriculture Fisheries and Food, Food Science Laboratory, Queen Street, Norwich NR2 4SX, UK The application of inductively coupled plasma mass spectrometry (ICP-MS) for trace and ultra-trace metal determination in foodstuffs has been investigated. The ICP-MS technique was compared with electrothermal atomisation atomic absorption spectrometry (ETA-AAS) for Cd and Pb in three biological standard materials. Results for Mo and V by ICP-MS were compared with the certified values.The ETA-AAS gave results that were closest to the certified values, with good precision. The ICP-MS results were encouraging and compared well for Cd and Pb, with both ETA-AAS and the certified results. The precision was acceptable at the higher element concentrations but degraded considerably at the lower levels. An investigation into the causes of the poor precision revealed molecular interferences and instrument hardware were partially to blame. Noticeable improvements in accuracy and dramatic improvements in precision were then obtained. Molecular ion interferences over the 51-81 a.m.u. mass range were investigated using several chemical matrices (HN03, HCI, H2S04, H3P04, H202 and NH3) and the results confirmed that dilute HN03 is the most suitable acid matrix.Data are presented on the problems of the determination of As and V in the presence of 51CIO+ and 75ArCI+ and the possibilities for data correction explored. Keywords: Inductively coupled plasma mass spectrometry; trace metal determination; food analysis; molecular ion interferences Increased awareness of the importance of trace elements in foods and their speciation has created a demand for yet more sensitive determinations. Often it is required to determine Cd, Mo, Pb and V at the sub-microgram per gram level. Of the available analytical techniques in most food science labora- tories only electrothermal atomisation atomic absorption spectrometry (ETA-AAS) offers this capability. This method is relatively slow and is not without problems.Thus the closely at different levels of chloride concentration to deter- mine if correction for the 51C10+ and 75ArC1+ ions could be made using WlO+ and 77ArC1+ and at what concentration chloride becomes a problem. Experimental Instrumentation development of inductively coupled plasma mass spectro- metry'-5 by Gray and Fassel and other workers has been greeted with interest in the field of food analysis. These advances were recently reviewed by two other significant contributors, Douglas and Houk.6 Commercial ICP-MS instrumentation has been available only since early 1984. Consequently few papers have been published concerning precision and stability in the analysis of real samples using ICP-MS. Early publications were con- cerned mainly with advancing the technique, and only limited data were reported on possible isotope overlap caused by molecular ions, and much of this concerned the now disused boundary layer sampling techniques (Table 1).Several of the molecular ions reported have been a function of the nature of the plasma source (e.g., capillary arc plasma1 now of historical interest only in comparison with ICP*-5), the mode of ion sampling (boundary layerl-4 in comparison with continuum flow ion extraction5 and the distance of the sampling aperture from the load coils). This paper presents analytical results for Cd, Mo, Pb and V determined in biological standard materials by ICP-MS, those for Cd and Pb are compared with those obtained by ETA-AAS. The ICP-MS results were encouraging. However, an investigation was instigated to determine the possible causes of poor precision; this indicated several molecular ion interferences and some hardware problems.The molecular ions were investigated critically over the 51-81 a.m.u. mass range using several chemical matrices often utilised for food analysis (H2SO4, H3P04, HC1, HN03, H202 and NH3). Two elements, As and V, were examined more The ICP-MS results were obtained using a Plasmaquad (VG Isotopes, Ion Path, Road Three, Winsford, Cheshire, UK). Full instrument details are given in Table 2. The ETA-AAS results were obtained using an HGA-500 furnace fitted to a 3030 spectrometer (Perkin-Elmer Ltd., Post Office Lane, Beaconsfield, Buckinghamshire, UK) fitted with deuterium arc background correction. Operating conditions were those used routinely in this laboratory (Food Science Laboratory, MAFF) for food analysis.Chemicals and Reagents Three standard biological materials were analysed: oyster tissue (NBS SRM 1566, National Bureau of Standards, Washington, DC, USA), bovine liver (NBS SRM 1577a) and a laboratory collaborative trial curly kale material. All reagents used were of Aristar (BDH Chemicals, Ltd., Poole, Dorset, UK) or equivalent quality and all solutions were prepared with de-ionised, distilled water. Methodology Determination of total Cd, Mo, Pb and V in biological standard materials Samples (0.50 g dry mass) of oyster tissue, bovine liver and curly kale were wet digested with concentrated HN03 (5 cm3) and concentrated H2S04 (2.5 cm3) in pre-cleaned boiling tubes on a programmable aluminium heating block.The samples were completely charred at 180 "C and the remaining212 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 1. Cited major molecules and their probable identity A.m.u. Molecular ions 17 MO+*.t; NH?+* 18 NH,+*,$;OH,+*,t 19 OH3+*,t,S 20 AT2+*,$ 31 NOH+$ Much reduced 32 O2+*9t3$ insize§ 30 NO+*,t)$ 36 NH4+ .H,O$; 36Ar+* 37 40 40Ar+ * 9 t 41 NArH+*,t; Na.+H20* 44 N20+$; C02+$; Sr2+* 45 N,OH+$ 48 NO+.H20$ 55 OH3+ .2H20$ 68-70 Baz+*,T 76 ASH+! 80 40Ar2*7t9§77 81 40Ar2H+*)t,§,7 91 AsO+*,W 92 AsOH+T 103 ArCu+ll 104 SrO+t 105 SrOH+t; YO+*,?; ArCu 11 106 YOH+t 154 BaO+t 155 BaOH+t 254 UO+t,§ 270 U02+t7§ * Houk et a1.2: ICP-MS, boundary layer sampling, sampling t Date and Gray? ICP-MS, boundary layer sampling, sampling $ Grayl: capillary arc plasma mass spectrometry (CAP-MS), 0 Gray and Dates: ICP-MS, continuum layer sampling, sampling 7 Date and Gray,: ICP-MS, boundary layer sampling, sampling 1) Cantle et aL7: plasma discharge source mass spectrometry.29 N,H+t I 33 O2H+$ H30+. H20 * ,t ,$; 36ArH+ * aperture 50 pm. aperture 70 pm. boundary layer sampling, sampling aperture 75 pm. aperture 0.4-0.5 mm, no desolvation. aperture 70 pm. Table 2. Instrument details for inductively coupled plasma mass spectrometer Instrument Plasmaquad (VG Isotopes, UK) Plasma system- R.f. generator . , . . 27.12 MHz, 0-1500 W RMS carrier wave power Torch system . . . . BTP 1500 ICP, 3-turn water cooled coil and quartz torch Gascontrols . . . . . . Afinegascontroller(Negrettiand Zambra) modulates the incoming Ar supply to the rotameter controlled auxilliary, nebuliser and plasma gas flows Nebuliser .. . . . . Jarrell-Ash cross-flow nebuliser pumped at 1.2 cm3 min-1, PTFE high-solids nebuliser (P.S. Analytical, Arthur House, Far North Building, Cray Avenue, Orpington, Kent, UK) pumped at 1.2 cm3 min-l spray chamber with a coarse glass sinter in the drain line Spray chamber . . . . Double by pass (glass), water-cooled Ion sampling- Sampling cone . . . . Titanium nitride coated Ni cone Skimmer cone . . . . Nickel (aperture 1 mm) Load coil - sampling cone Multiplier . . . . . . Continuous dynode electron multiplier (aperture 0.75 mm) distance . . . . . . 12mm (Model 401, Detector Technology Inc., USA) USA) Multi-channel analyser . . Tracor Northern, N 7200 (Wisconsin, Data system- System .. . . . . . . For all analyses: Forward r.f. power . . Coolantgasflow . . . . Auxiliarygasflow . . . . Nebulisergasflow . . . . Nebuliser pressure . . 28000 CPU, 160 kbyte RAM. 1 I Mbytes hard disc 640 kbyte, quad density floppy discs (Olivetti) resolution, mass calibration and counts over the mass range of interest prior to analysis. A minimum 1-h warm-up time was set before analysis The system was optimised for: 1400 W 12 1 min-* 11 min-1 0.45 1 min-1 1.38 bar (2Op.s.i.) organic matter and HN03 were removed by the addition of 20 vol. H202 (5 cm3) to the cooled digests, which were then reheated to 220 "C and the water boiled off. The digests were diluted to 100 cm3 to contain 0.1% Mg(N03)2, 0.5% NH4H2P04 (both as matrix modifier for ETA-AAS) and 1 pg cm-3 of both Bi and Y (as nitrates as internal standards for the ICP-MS).The-digests were analysed for Cd and Pb by ETA-AAS. The Cd, Mo, Pb and V were determined by ICP-MS according to the instrument parameters (Table 2) and scan details [Table 3(a))] shown. The digests were unfiltered so the PTFE solids nebuliser, which is unaffected by particu- late material, was used. Results were calculated on the basis of the mean of triplicate unspiked sample digest solutions, corrected for the recoveries of duplicate spiked sample digest solutions. The spiked digests contained 2, 0.5, 2.5 and 1 pg of V, Mo, Cd and Pb, respectively. Three unspiked and two spiked digest blanks were digested at the same time. Sample solutions For the molecular interference studies described, the matrices were prepared as 0.4, 0.8, 1.2 and 2.0% V/V solutions of the following stock solutions: de-ionised, distilled water (dH20) ; de-ionised, distilled, Millipore filtered water (mH20) ; H2SO4, Aristar 98% m/V, 1.84 g (3111-3 (BDH Chemicals Ltd.); HC1, Ultrar 36% mlm, 1.18 g cm-3 (Hopkin & Williams); HN03, Primar 70% mlV, 1.42 g cm-3 (Fisons); H3P04, Aristar, 85% mlV, 1.7 g cm-3 (BDH Chemicals Ltd.); NH3, Pronalys, 33% mlm, 0.88 g cm-3 (May & Baker); and H202, Aristar, 100 vol.30% mlV, 1.099 g cm-3 (BDH Chemicals Ltd.). Internal standards and standard solutions were prepared from analyti- cal-reagent grade 1000 pg cm-3 stock solutions when required. Dilute HCl was used as the source of C1 ions for the interference studies on As and V.(The As203 1000 pg cm-3 stock standard solution was stabilised in 20% mlm HCI solution, which was equivalent to 2.3 p.p.m. of C1 in a 10 p.p. b. As solution, corrections were made accordingly.) The As standard solutions and blanks were prepared in 1% HN03. All solutions were diluted with de-ionised, distilled H20 and sufficient Y and/or Bi was added as internal standard. Scans were performed as listed [Table 3 ( b ) and (c)]. The data for low level C1 concentration effects on As and V were replicated ten times at each concentration level. The Jarrell-Ash crossflow nebuliser was used throughout these interference experiments.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 213 Table 3. Scan details Table 5. Molecular and interfering ions observed during matrix interference studies.Where isotope masses are not given, the major isotope is assumed Parameter V, Mo and Cd Pb (a) V , Mo, Cd and Pb determinations in biological standard materials- Mass range/a.m.u. . . . . . . 49.5-1 16.5 204.5-215 Sweeps . . . . . . . . . . 240 480 Channels . . . . . . . . . . 1024 512 Dwell timelys . . . . . . . . 500 500 Total run time/s . . . . . . . . 122 122 Internal standard (1 yg g-1) . . 89Y 209Bi A.m.u. Probable identity 47 PO+ A.m.u. 67 Probable identity H34S02+ 35C102+ 48 POH+ 32s160 NO+H20 68 69 49 ~ ~ S O H + ( b ) Matrix interference studies- Mass range/a.m.u. . . . . . . 49.5-210.5 Sweeps . . . . . . . . . . 120 Channels . . . . . . . . . . 2048 Dwell time/ys . . . . . . . . 500 Total run time/s .. . . . . . . 122 Internal standard (1 pg g-1) . . 209Bi ( c ) The effect of [Cl] on V and As results- Mass range/a.m.u. . . . . . . 49.5-90.5 Sweeps . . . . . . . . . . 480 Channels . . . . . . . . . . 512 Dwell time/ps . . . . . . . . 500 Total run time/s . . . . . . , . 122 Internalstandard (0.1 ygg-1) . . SOY 70 71 37C102Hi 50 34SO+ 51 35C10+ ~_~soH+ 72 73 75 53 37C10+ aAr35C1+ 54 40ArN+ 56 40ArO+ 37C10H + 76 77 57 41ArOH+ Table 4. Comparison of results for trace element determination in biological standards between ETA-AAS and ICP-MS. Values in parentheses are non-certified results 78 80 58,60 Ni from Ni cones 40Ar4"Ar+ HP03+ * 40Ar40A1-N + * H2P03+ '5AsO+ P202+* 62 P2+ 81 Levels found/pg g-l 63 P02+ 63,65 Cu from Cu cones 64 HP02+ 32s02+ 32S2+ W u H + 91 94 Sample ETA-AAS ICP-MS Certified result Vanadium- Oyster tissue Bovine liver Curly kale (NBS1566). .. . (NBS1577a) . . (from laboratory collaborative trial) 98 100 102 1.96 k 0.63 2.3 t 0.1 0.20 ? 0.06 - 65 H32S02+ 103 105 106 205 66 WUH+ ,34S02+ Molybdenum- Bovineliver . . . . - Oystertissue . . . . - Curlykale . . . . - * Combination unproven 0.16 -t 0.02 (S0.2) 2.56 k 0.02 15.7 k 0.5 - 3.5 k 0.5 levels, i.e., Pb in bovine liver. The variable results for V in oyster tissue were probably due to molecular interference of 51ClO+ caused by the 1% C1 present in oyster tissue. No correction has been made for this. The causes of the poor stability and precision of the ICP-MS results (based on several months analyses) are now attributed to: (1) sampling cones; (2) temperature variations in (i) quadrupole and (ii) plasma argon supply and nebuliser spray chamber; and (3) molecular ion interferences, matrix and plasma derived.Cadmium- Oystertissue . . . . 3.4k0.08 Bovineliver . . . . 0.44 k 0.09 Curlykale . . . . 0.17 k 0.05 3.5 t- 0.5 0.47 t- 0.15 0.14 t- 0.04 3.5 ? 0.4 0.44 t 0.06 (0.21 5 0.04) Lead- Oystertissue . . . . 0.51 ? 0.03 Bovine liver . . . . 0.16 k 0.03 Curlykale . . . . 4.6 t 0.32 0.50 k 0.19 0.1 k 0.1 4.82 t- 0.35 0.48 k 0.04 0.135 k 0.015 (4.8 5 0.7) Sampling cones Food chemically wet digested for total elements are routinely prepared in this laboratory to have a final concentration of 10% H2S04. This rapidly destroyed copper sampling cones (0.4-0.6 mm aperture) in about 8 h and prevented stable ion sampling.The lifespan of the copper sampling cones was increased to 30 h using a total digest acid concentration of 2.5% H2S04, but this increased the digestion time. The 30-h period was still considered insufficient for the production of a stable ion sampling system. Small aperture (0.4 mm) nickel cones were available as an alternative, but these were rapidly destroyed by 2.5% HzS04. The aperture rapidly degraded in symmetry and closed after 8 h of operation. Similar observations were made with dilute Results and Discussions Determination of Total Cd, Mo, Pb and V in Biological Standard Materials The ETA-AAS results for Cd and Pb in the digests compared well with the certified results (Table 4). The precision decreased at low element concentrations, e.g., Cd in curly kale and Pb in bovine liver, however, these results, based on direct injection, could probably be improved by solvent extraction.In all instances the ETA-AAS analysis was more precise than the ICP-MS analysis for Cd and Pb. The ICP-MS results were all within the range of the accepted values with the exception of Mo in bovine liver. The precision however was only acceptable at the higher element concentrations and it degraded considerably at the lower214 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 H3P04 solutions. Both the copper and nickel cones sputtered in the region immediately behind the sampling aperture (Gray and Dates) further reducing their lifespan. Recently, titanium nitride coated cones have been tested, the narrow aperture versions (0.4 mm) experienced similar problems to the nickel with 2.5% H2S04.However, the enlarged (0.75-mm aperture) treated nickel cone has, as with the larger aperture copper cones (0.8 mm), given improved sensitivity and reduced both background counts and the tendency to block with salt solutions. The coated nickel cones are now favoured as opposed to copper as they run hotter and produce a smaller percentage of molecular ions. In continuum ion extraction with large apertures the only boundary layer ions seen are MO+ and possibly MOH+, which bleed in from the edges of the aperture. The most probable source for the other molecular ions is the relatively dense region in the conical aperture and skimmer apices where oxygen atoms in particular, but also chlorine, probably adsorb on the metal surfaces and are released by ion bombardment either as molecules or as atoms that subsequently react in the gas phase.Cone material has to be chosen according to the analysis required, and both the skimmer and the sampling cone should be of the same material (Gray and Dates). Temperature variation Quadrupole. Early hardware failures within the region of the r.f. output to the quadrupole indicated that better temperature control of the instrumental cabinet was essential. Modifications were made to improve instrument cooling and these have reduced drift. To stabilise the quadrupole tem- perature a warm-up time of at least 30 min was necessary in the mass range of interest. Plasma Ar supply and the nebuliser spray chamber. The heat absorbed in the gas control box as the plasma r.f.generator operates was considerable, but modifications by the manufac- turer, involving venting of the generator to the outside, have much improved this problem. The nebuliser spray chamber is situated within the torch box and held by an aluminium former. The spray chamber temperature was monitored over 90 min of analysis and found to attain a maximum temperature of 47 "C, which was not reached until well after the 1-h warm-up period. The possibility that this high temperature could increase the water vapour loading of the plasma with its subsequent cooling was considered. To test this hypothesis a 1 pg g-1 standard solution of Y was continuously nebulised over 90 min, and the I I I I 142 0 20 40 60 80 Tim eim in Fig.1. Graphs of A, Y+ (l-pg gT*); B, OH3+ (19); C, * 6 0 ' 8 0 + (34); and D, 02H3 (37) counts against time; and E, uncooled spray chamber temperature against time I I I I I I 1 ( C) I I I I I , I I I (9) I I I 50 55 60 65 70 75 80 A.m.u. Fig. 2. Effect of 2% matrix solutions on 50-84 a.m.u.: the major ions. Coun& full scale, 1600; 1 yg ggl Bi = 6 X lo5 counts. (a) dH20; ( b ) mH20; ( c ) H,O,; ( d ) NH,; ( e ) HN03; (f) HCI; (g) H2S0,; and ( h ) WO-2JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986. VOL. 1 215 , I I 1 I I 1 I I I I I ( d) I I I I 1 I 1 I I I I I 50 55 60 65 70 75 80 A.m.u. Fig. 3. Effect of 2% matrix solutions on 50-84 a.m.u.: the minor ions. Counts full scale, 320. (a) dH,O; (b) mH,O; ( c ) H,O,; (d) NH,; (e) HNO,; U, HCl; (g) H,SO,; and (h) H3P04 sample was repetatively analysed for masses 89(Y+), 19(OH3+), 34('60180+) and 37(O2H3+) to see if the water loading (at 37 a.m.u.) increased with nebuliser temperature.The results in Fig. 1 were inconclusive. The downward drift of 89 was closely mapped by that of mass 34 but less so that of mass 37 and not at all by mass 19. Therefore the increase in nebuliser temperature was matched by a decrease in ion counts on 89, 34 and 37. As the experiment was inconclusive the spray chamber temperature was reduced to a constant by water cooling to 24 "C (attained 45 min after plasma ignition), as a stable rather than a drifting spray chamber temperature was considered desirable. When all these modifications were incorporated, dramatic improvements in precision were obtained.When internal standards were used these hardware modifications improved the precision for food digest analysis for: cadmium from 29% relative to 3% relative at the 3.5 pg 8-1 level, molybdenum from 12% relative to 5.7% at the 0.2 yg g-1 level, lead from 7% relative to 1.3% at the 5 pg g-1 level and for vanadium from 30% relative to 14% at the 0.2 yg g-1 level. When the analyses ,reported in Table 4 were repeated noticeable improvements in precision and some improvement in accuracy were obtained, e.g., cadmium 3.4 If: 0.1 yg 8-1 in oyster tissue and 0.47 -t 0.03 pg g-1 in bovine liver; molybdenum 0.23 k 0.04 pg g-1 in oyster tissue and 3.4 +_ 0.09 pg g-1 in bovine liver; lead 0.38 k 0.08 yg 8-1 in oyster tissue, 4.87 k 0.19 pg g-1 in curly kale and 0.075 k 0.011 pg g-1 in bovine liver; and vanadium 2.02 & 0.07 pg g-1 in oyster tissue and 0.21 k 0.04 pg g-1 in bovine liver.Greater improvements in precision might be obtained by decreasing the dilution factor, presently 200, but this was needed to reduce the sulphuric acid concentration to 2.5% of concentrated. Molecular ion interference, matrix and plasma derived Postcone and plasma molecular ion behaviour is complex and the extent of the isotope overlap in the 51-81 a.m.u. mass range is considerable, even some of the higher mass units are not exempt (Table 5 ) . The major and minor molecular ions have different characteristics according to the chemical matrix (Figs. 2 and 3). Major molecular ion peaks were observed at 54(ArN+), 56(Ar0+), at 76,78 and 80 a.m.u.(ArAr+) and at 58 and 60 atomic ion interfaces from the nickel cone occur. These peaks overlap isotopes of Cr, Fe, Ni and Se, respec- tively, and are particularly severe on Fe and Se, overlapping two isotopes, accounting for 77 and 73%, respectively, of the relative isotopic abundance. Without chloride addition, C10+ peaks at 51 and 53 a.m.u. were observed from the chlorine content of the food samples. Ammonia solution affected ArN+ formation to a much greater extent than HN03 [Figs. 4(a) and 5(b)]. The ArN+ formation was favoured in the presence of the more volatile and efficiently nebulised NH3. Dilute HN03, H202 and NH3 spectra were similar to those obtained with H20, with the exception of the ArN+ behaviour mentioned above. There- fore of all the matrices examined dilute HN03 would be the most suitable for elemental analysis within this mass range. Increased plasma loading with C1 ions affected both C1 associated and non-associated molecular ion peaks [Figs.4(6) and 5(a)]. As counts increased on the C1 associated masses (Table 5) so those of 56ArO+ ,76ArAr+, 78ArAr+ and 80ArAr+ decreased [Figs. 4(b), 5(c and d ) ] . This effect was only observed with C1 ions and the data indicated 51 + s3C10+ was favoured to 56ArO+, and 75+77ArCl+ to ArAr+. Further C10+ formation increased at a greater rate than that of ArCI+. Gray and Date5 reported ArAr+ formation to be the possible product of polymerisation reactions within the expansion chamber, between the plasma and the quadrupole. In this experiment the larger collisional cross-section of C1 may favour C10+ formation over ArO+, ArCI+ and ArAr+ within this zone.Therefore, molecular ion interference on 51V or 75As cannot be directly corrected using masses 75,77 or 51 and 53, respectively. Isotope overlap of C1 molecular ions occurred mainly on V and As but also on Cr, Zn, Ga and Ge.216 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 E F G r / D I 1 I 1 50 55 60 65 70 75 80 50 55 60 65 70 75 80 A.m.u. A.m.u. 1 1 I I I I , I I I I I I I I 55 60 65 70 75 80 50 55 60 65 70 75 80 A.m.u. A.m.u. 50 Fig. 4. Effect of NH3 concentration on 50-84 a.m.u.; counts full scale, 3200; 1 pg g-1 Bi = 6 X 105 counts; A 0.4, B 0.8, C 1.2 and D 2.0% NH3. ( b ) Effect of HC1 concentration on 50-84 a.m.u.; counts full scale, 3200; E 0.4, F 0.8, G 1.2 and H 2.0% HC1. ( c ) Effect of H,S04 concentration on 5&84 a.m.u.; counts full scale, 3200, I 0.4, J 0.8, K 1.2 and L 2.0% H,S04.(d) Effect of H3P04 concentration on 50-84 a.m.u. ; counts full scale, 3200; M 0.4, N 0.8, 0 1.2 and P 2.0% H,PO, (a)217 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 4 - .: 3 E 0 a CI g 2'1 E 1 - ( a ) several molecular ions were still formed [Fig. 4(d)] at 63 (PO,+), 64 (HP02+), 71 (ArP+) and 81 a.m.u. (H2P03+), which would cause isotope overlap on isotopes of Ni, Cu, Zn, Three mass units in the 51-81 a.m.u. size remained unhindered by molecular ion isotope overlap, 55 (Mn), 59 (Co) and 74 (Ge). These isotopes of Mn and Co are both naturally 100% abundant. ( b ) 8 h (a) - f i ~ s - - r--r-lr 6 !$ Q, 3 , Ga and Br.1 E 4 - z 2 - LT P c 6 I 3 2 4l c I v) c 3 V) 8 - s 6 - m' - 4 - c 0 X m E 2 - a E F - mH20 - dH20 0 0.4 0.8 1.2 1.6 2.0 Matrix concentration, o/o 1 , I 0 0.4 0.8 1.2 1.6 2.0 Matrix concentration, o/o I , I I t 0 0.4 0.8 1.2 1.6 2.0 Matrix concentration, O/O 0 X 7 m $ Matrix concentration, o/o Fig. 5. H,O,; E, HN03; and F, H3P04 Effect of matrix concentration on counts for: (a) 51; ( b ) 54; ( c ) 56; and (d) 80 a.rn.u. A, HCl; B, H,SO,; C, NH3; D, 0 L=TGz3 20 40 60 80 100 0 -/ I I I I 2 4 6 8 1 0 CI concentration/pg cm-3 CI concentration/mg cm-3 Fig. 6 . Effect of low [(a) and (b)] C1 concentrations and high C1 concentrations on the C10+ [(a) and ( c ) ] and ArCl+ [(b) and (d)] ion counts in the presence of 10 p.p.b. of As (75 a.rn.u.).A, 51; B, 53; C, 75 and D, 77 a.m.u. The main molecular ions of H2SO4 were at 50 (34SO+), 64 (32S02+), 65 (H32S02+) and 66 a.m.u. (34S02+), which overlap the isotopes of Ti, Cu and Zn [Fig. 4(c)]. Nickel ion counts increased with increased concentration of H2S04 due to erosion of the sample cone. Both H2S04 and H3P04 solutions had higher surface tension and viscosity characteris- tics compared with the other matrices and the reduced peak intensity observed with H3P04 is indicative of reduced nebulisation efficiency. With the resulting hotter plasma LOW and high concentrations of C1 show similar effects on masses 51, 53, 75 and 77 a.m.u. [Fig. 6 (a and b ) ] . However, correction for C1 on either 51 or 75 a.m.u. (V and As) depend on the relative response ratio of peaks at 35 and 37 a.m.u.being constant (3.065). This was not so at low C1 concen- trations (Fig. 7) because of the undue contribution of other background counts at such low concentrations and was not observed until over 25 pg g-1 of C1 were present, at which level [Cl] was already enhancing counts.218 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Ti me/rn i n Fig. 8. Graph of ratio of standardisation (of Au) plotted against time. Nebuliser, Jarrel-Ash cross flow pumped at 0.8 cm3 min-1 In this simplified situation correction for counts of 75ArC1+ on 10 p.p.b. of As was possible to within 5% RSD by the following calculation: (Counts on 75 - blank counts on 75) - [(counts on 77 - blank counts on 77) x 3.0651 = counts due to As At higher C1 concentrations correction proved more diffi- cult, both at mass 51 and 75.At these concentrations only one analysis was performed for each sample. However, the declining counts on 56 and 80 may provide a means of assessing the C1 contribution, but further work is required before a satisfactory expression is obtained. It has been suggested that competitive surface absorption between 0 and C1 probably occurs but this would vary with surface conditions inside the expansion stage. The detection limits of As degraded with increasing C1 content (without correction), and above 25 yg 8-1 of C1,51V and 75As isotopes are enhanced by 10% for every 3.3 and 6 pg g-1 of C1, respectively. In complex matrices correction for isotope overlap will be difficult, and the detection limits on predominantly monoisotopic elements, or on elements with- out overlap “free” isotopes will deteriorate.We have exam- ined simple matrices but it is obviously important in real analysis to matrix match standards, dilute where possible to reduce some molecular ion species and, where appropriate, use standard additions. The determination of trace elements in pure materials may considerably increase isotope overlap problems due to MO+, MOH+ and Mz+. The extent and ratios to which these species form may not be a constant and may be affected by the solution matrix. Advantages of internal standards In the unmodified instrument, drifts were observed in the nebuliser, ion sampling and quadrupole regions of the ICP-MS and these were alleviated during an analysis sequence by the use of a suitable internal standard.These drifts were remarkably reduced by the modifications described above, and might be further reduced by an improved laboratory environment. The raw counts for V, Mo, Cd and Pb (1 yg cm-3 in dilute HN03) were monitored continuously over 2 h after a 30-min warm-up period and drifts of 3.7% for 51V, 3.9% 95M0, 3.8% 98M0, 3.9% 111Cd, 3.7% 114Cd, 4.3% 206Pb and 4.4% 208Pb were observed. These were further improved using 89Y (1 yg cm-3) as internal standard to 0.87, 0.99, 1.0, 1.4, 1.0, 1.3 and 1.3%, respectively. The ideal internal standard is a monoisotopic element providing minimal mole- cular interference in the mass region of interest. It should have a relatively rapid wash-out time. Both S9Y and 209Bi have been used in the present work at the 0.1 and 1 yg g-1 concentration levels.It is necessary to keep the proportion of Y2+ low if Y is to be a suitable internal standard but this has never been found to be a problem in our work. Fig. 8 (where time is plotted against the ratio of standardisation , i. e . , the ratio by which the counts are multiplied to give a constant internal standard count, using Au, throughout the analysis) illustrates the necessity for use of an internal standard, where changes in the viscosity, surface tension and aerosol characteristics of the pumped analyte standard [Fig. 8(b)] and sample matrices [Fig. 8(a)] are different. This causes changes in the temperature of the plasma and consequently the degree of ionisation of the analyte aerosol.Therefore, if unfiltered samples are to be nebulised, an internal standard is a necessity if analyte sensitivity is insufficient to allow sample dilution. Recently two or more internal standards have been recommended for ICP-OES (Lorber,g Ramsey and Thompson9) to determine non-random fluctuations in analytical channels and have proved sufficiently successful in reducing drift to as low as 0.1%. Conclusion This preliminary study of ICP-MS for the analysis of food indicates that a thorough investigation of the incidence of molecular ion interference that may arise from the matrix at concentration ranges present in the sample is necessary. Dilute nitric acid is clearly the preferred matrix for elemental analysis in the 51-81 a.1n.u. range. Correction for some molecular ion overlaps mathematica1l.y is possible. Two or more internal standards in the mass range of interest can improve both precision and accuracy. When real samples are to be analysed a high solids nebuliser is recommended, as is the pumping of solutions to the nebuliser. Careful tempera- ture control of the instrument is important including cooling of the nebuliser spray chamber. The choice of cone materials is critical. In practice ICP-MS is not without problems but itsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 219 excellent sensitivity will continue to suggest that it has a vital role to play in trace element analyses of foods. The authors gratefully acknowledge the help of Malcolm Baxter of the Food Science Laboratory, Norwich for ETA- AAS analysis of the food digest. We also acknowledge a grant from the Ministry of Agriculture, Fisheries and Food, which made this work possible. References 1. 2. Gray, A. L., Analyst, 1975, 100, 289. Houk, R. S., Fassel, V. A., Flesch, G. D., Svec, H. J., Gray, A. L., and Taylor, C. E., Anal. Chem., 1980, 52,2283. 3. Date, A. R., and Gray, A. L., Analyst, 1981, 106, 1255. 4. Date, A. R., and Gray, A. L., Spectrochim. Acta, Part B , 1983, 38, 29. 5. Gray, A. L., and Date, A. R., Analyst, 1983, 108, 1033. 6. Douglas, D. J., and Houk, R. S . , Prog. Anal. A t . Spectrosc., 1985, 8, 1. 7. Cantle, J. E., Hall, E. F., Shaw, C. J., andTurner, P. J., Int. J. Mass. Spectrom. Ion Phys., 1983, 46, 11. 8. Lorber, A., Anal. Chem., 1984, 56, 1404. 9. Ramsey, M. H., and Thompson, M., Analyst, 1984,109,1625. Paper J5155 Received November 18th, 1985 Accepted February 18th, 1986
ISSN:0267-9477
DOI:10.1039/JA9860100211
出版商:RSC
年代:1986
数据来源: RSC
|
17. |
Determination of dissolved inorganic antimony(V) and antimony(III) species in natural waters by hydride generation atomic absorption spectrometry |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 221-225
S. C. Apte,
Preview
|
PDF (596KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 22 1 Determination of Dissolved Inorganic Antimony(V) and Antimony(ll1) Species in Natural Waters by Hydride Generation Atomic Absorption Spectrometry S. C. Apte* and A. G. Howardt Department of Chemistry, The University, Southampton, Hampshire SO9 5NH, UK A method is described for the determination of dissolved antimony(ll1) and antimony(\/) species in natural waters. The technique is based on the rapid evolution of stibine from the sample solution by reduction with sodium tetrahydroborate(lll), cryogenic pre-concentration and atomisation in an electrically heated quartz tube furnace mounted in the light path of an atomic absorption spectrometer. Differentiation of antimony(ll1) and -(V) species is achieved by exploiting the pH dependence of the tetrahydroborate(ll1) reduction step.The procedure is applicable over a working range of 0-300 ng 1-1 of antimony and permits a sample throughput of ca. 7 analyses per hour. Detection limits are ca. 10 ng 1-1 for total antimony and ca. 1 ng I-' for antimony(ll1). Potentially interfering species have been investigated but, at levels expected in natural waters, did not influence the analytical results obtained using the method. Keywords: Antimony determination; cryogenic pie-concentration; speciation; environmental analysis; h ydride generation atomic absorption spectrometry Studies of the chemistry and cycling of antimony in the aquatic environment require sensitive analytical techniques capable of determining analyte concentrations typically in the range 1-300 ng 1-1.In addition to determining total element concentration, it is also desirable to gain information on the chemical form of the element (speciation), as this will affect reactivity, toxicity and the mode of assimilation by living organisms. For metalloids, an important aspect of inorganic speciation is oxidation state. Whilst arsenic and antimony are predicted on thermodynamic grounds to exist in their pentav- alent oxidation states in natural waters, the lower oxidation states [e.g., As(II1) and Sb(III)] have been detected and are believed to result, at least in part, from biologically mediated reduction processes. 1-4 Few methods suitable for ultratrace antimony determina- tion have been reported in the literature. Colorimetric methods based on the formation of ion-pair complexes between antimony(II1) and a cationic dye such as Rhodamine B5 or Crystal Violet6 have detection limits of ca. 10 ng 1-1 of antimony, but they require lengthy sample preparation and pre-concentration procedures, which limit sample throughput to typically one per hour.' The more commonly employed analytical approach involves hydride generation interfaced with some form of atomic spectroscopy, popularly atomic absorption spectro- metry (AAS).Such methods have been widely used to determine the metalloids (including antimony) in a variety of matrices, typically in the pg 1-1 concentration range. Hydride methods have the advantage of being simple, rapid and relatively interference free. In order to achieve the sub-pg 1-1 detection limits required for natural water analysis, it is necessary to employ a pre-concentration step.This can be conveniently achieved by cryogenic pre-concentration of the evolved hydride.8.9 An additional advantage offered by hydride generation is that by careful control of r e d u c t i ~ n ~ , ~ - ~ l information on analyte oxidation states may be gained. Few workers, however, have extended the capabilities of the hydride technique to low-level antimony speciation analysis. This paper describes a relatively simple, yet highly sensitive, method for the determination of antimony speciation in water, '' Present address; Water Research Centre, PO Box 16. I-Ienley t To whom correspondence should be addressed. Road, Medmenham, Buckinghamshire, UK.which uses oxidation state information obtained by exploiting the pH dependency of the hydride generation reactions. A sample throughput of better than seven samples per hour has been achieved as a result of efficient hydride generator design. This compares with a sample throughput of ca. four per hour reported for other low-level hydride methods of comparable sensitivity. A detailed study of the effects of potentially interfering substances is also presented. Experimental Reagents and Glassware All chemicals were of analytical-reagent grade unless other- wise stated. Glass- and plasticware were cleaned by soaking in dilute (1 + 9) nitric acid and were rinsed with distilled water before use. Stock 1000 mg 1-1 antimony standard solutions were prepared from antimony potassium tartrate dissolved in distilled water and potassium antimonate (laboratory-reagent grade) dissolved in dilute (1 + 9) hydrochloric acid.The concentration of the latter standard was checked against the antimony(II1) solution by flame AAS. Lower concentration standards were prepared by dilution on the day of use. Although it has been reported that on occasion dilute antimony(II1) standards undergo rapid oxidation,' this was not evident over a period of some 5 h. Laboratory-reagent grade sodium tetrahydroborate(II1) was used to prepare unstabilised solutions in distilled water (2 and 4% rnlV). These solutions were sufficiently stable to be employed during the working day. The sodium tetrahydroborate(II1) solution was the major contributor to the experimental blank and antimony levels were unacceptably high in batches of this reagent obtained from several sources.It is therefore recommended that when setting up the method, several batches of the reagent be screened to select suitable material. Alternatively it may be necessary to employ one of the tetrahydroborate(II1) purification procedures suggested by Aggett and Aspelllo or Andreae and Byrd.12 The citric acid solution used for pH adjustment was prepared as a 2 M solution in distilled water. Potassium iodide solution (1 M) was made up in distilled water. Concentrated hydrochloric acid was purified before use by sub-boiling distillation.222 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Apparatus The apparatus consists of three distinct stages: a hydride generator (Fig.l), a cryogenic trap and an atomisation - AAS detection system. The last two stages have been described in detail in a previous publicationl3 and only a brief description will therefore be given here. Hydrides are generated from solution in a purpose-built generator (Fig. l), consisting of a cylindrical chamber with a medium porosity frit located at its base. A side port, capped with a silicone rubber septum (replaced after every 20-30 determinations), is used for the introduction of tetrahydro- borate(II1) solution into the sample by syringe. The generation chamber is connected to the rest of the apparatus via a ground-glass joint. When in use, the sample is held in the chamber by a combination of upward gas pressure and surface tension.Following generation, the hydride-containing gas stream is dried by passage through sodium hydroxide pellets (renewed every 30 determinations) located in the head of the chamber, and is then cryogenically pre-concentrated. This concentration step is carried out in a glass U-tube packed with glass beads (40 mesh, acid washed and trimethylchlorosilane treated) cooled with liquid nitrogen. When trapping is complete, the liquid nitrogen Dewar is removed and the trap is allowed to warm to room temperature. The volatilised hydride is swept into an electrically heated quartz tube, having a thermally shielded inlet arm, aligned in the optical path of a Varian AA 175 AB AAS fitted with an EM1 9783 B photomultiplier tube giving extended response in the UV region.Background correction was employed throughout the reported work. The resulting spectrometer output is recorded on a Tekman TE 200 y - t chart recorder. The following spectrometer operating conditions are employed: source , hollow-cathode lamp; 5 mA operating current; wavelength 217.6 nm; and spectral band pass 0.2 nm. Recommended Procedures The sample is split into two aliquots: antimony(II1) is selectively determined in one aliquot and total antimony is determined in the other. The antimony(V) concentration can be determined by calculating the difference between the two values. Full details of reagent additions employed during these steps are given in Table 1. p -To cold trap Sodium Id? Ground-alass hydroxide pellets Silicone rubber --A 1 septum 3 %3 cm I 15cm '11 2cm I eTap t Nitrogen carrier gas Fig.1. Hydride generation flask Determination of total antimony With the carrier gas flow-rate set to approximately 50 ml min-1 the sample aliquot (usually 25 or 50 ml) is pipetted into the reactor. Acidity is adjusted to a hydrogen ion concentration of approximately 1 M by the addition of concentrated hydrochloric acid and potassium iodide solution (1 ml) is then added in order to ensure complete reduction of antimony(V) to antimony(II1). The reactor is connected to the cryogenic trap assembly and the nitrogen flow-rate is readjusted to 200 ml min-1. After the solution has been purged for 60 s, the trap is allowed to cool in liquid nitrogen for 60 s prior to injection of sodium tetrahydro- borate(II1) solution via the side port.The injection is performed using an all-glass syringe having a stainless-steel needle, at a rate such that 5 ml of solution are added in approximately 15 s. After a further 60 s, the liquid nitrogen is removed and the trap allowed to warm to room temperature releasing the stibine into the atomiser. The resulting spec- trometer output is recorded. Between determinations, condensed water is removed from the trap by warming in hot water for ca. 3 min while the hydride generator is disconnected from the system and thoroughly rinsed with distilled water. Determination of antimony(I2I) The procedure described above is employed but with citric acid solution being used to adjust the pH (see Table 1) and with the omission of the concentrated hydrochloric acid and potassium iodide solution.Results and Discussion Hydride Generation Under optimum conditions the hydride generation reaction proceeds rapidly and can be considered as being essentially instantaneous. In such a system it is important to select a hydride generator that efficiently mixes the sample and reagents, and provides rapid, efficient purging of evolved hydrides from the sample solution. In our initial work a generator consisting of a chamber and a fritted bubbler, similar to those described in other low level methods,l was employed. Using this system stripping times of ca. 5 min were required to purge evolved hydrides from solution. A long purge time was undesirable as it greatly increased sample analysis time and prolonged contact of the hydride with the Table 1.Details of reagent volumes and concentrations Total inorganic antimony determination- Samplevolume/ml . . . . 25 Concentrated hydrochloric Potassium iodide solution Sodium tetrahydroborate(II1) acidlml . . . . . . . . 2.5 (1 M)/ml . . . . . . . . 1 .o (2%mlV)lml . . . . . . 5 Antimony(ZZf) determinaiion- Samplevolumelml . . . . 25 or 50 Citricacid(2M)/ml . . . . 1 or2 Sodium tetrahydroborate(II1) solutionlml . . . . . . 5 (2"/0 miV) o r 5 (4% miV) Conditions common to both methods- Carrier gas flow-ratelml min-1 200 Trappingtimels . . . . . . 60JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 223 Q) c m sample solution increasing the chances of stibine hydrolysis. The reported hydride generator (Fig. 1) purges the sample much more efficiently and requires a stripping time of only 60 s.' '-Sb(lll) Reduction Conditions The principle of pH selective reduction of metalloid species was first reported in conjunction with the determination of arsenic(II1) and arsenic(V) species in waters,9JO and later for the determination of antimony(II1) and antimony(V) species. 1 ~ 1 Whilst no comprehensive explanation of the principles behind pH selective reduction has been cited in the literature, it is believed that the effect may exploit variations in the rate of the tetrahydroborate(II1) reduction of antimony(V) to antimony(III), as suggested by Aggett and Aspell.10 At high pH values this reaction proceeds extremely slowly compared with the reduction of antimony(II1) to stibine. In this work, the basis for the pH selective determination of antimony oxidation states was confirmed by monitoring the signal response of antimony(II1) and antimony(V) standards over a range of sample pH acidity. Initial sample pH was adjusted by the addition of acid or citrate buffer.The results (Fig. 2) showed that the antimony(II1) signal response was 0.12 I / , , a a 3 5 PH 0 HCVM Fig. 2. Variation of absorbance with hydride generation pH Table 2. Performance data Linear calibration range- Totalantimonyhg . . . . 0-5 Antimony(III)/ng . . . . 0-10 Sample throughput- Rate (both methods)/h-I . . 7 Antimony determination- Detection limithg 1-1 Precision, 70 Total antimony Antimony( 111) 25-ml sample volume . . 10 1.9 25-ml sample volume . . 2 4 50-ml sample volume . . 1 6.9 relatively unaffected over a wide pH range.With antimony(V) standards however, hydride yield was effectively zero above pH 2, in agreement with previously published work.1.11 Based on these findings, differentiation between oxidation states was achieved by determining inorganic antimony(II1) at pH 2.2 (adjusted by addition of citric acid solution) and the deter- mination of total inorganic antimony at a sample acidity of ca. 1 M in hydrochloric acid. The selectivity of the antimony( 111) analytical method was further confirmed by the analysis of solutions containing 10 ng of antimony(II1) in the presence of a ten-fold excess of antimony(V). Average recovery (triplicate determination) was 101 -t 1%. The addition of potassium iodide solution was found to be necessary to obtain equivalent signal responses from anti- mony(V) and antimony(II1) standards in acidic media.With- out potassium iodide addition, the signal response of anti- mony(V) was no better than 60% of that obtained with antimony(II1) standards. This is also in agreement with the findings of other workers.1.11 Under the selected pH conditions for speciation analysis, parameters such as borohydride addition, trapping time and carrier gas flow-rate were optimised for 25- and 50-ml sample volumes. These optimised analysis conditions are summarised in Table 1. Cryogenic Pre-concentration The cold-trap design and glass-bead packing used were those previously reported for use in arsenic determination. 13 Drying of the gas stream prior to cryogenic pre-concentration was found to be necessary in order to prevent ice build-up in the trap, which resulted in eventual blockage.Both calcium chloride and sodium hydroxide pellets located at the head of the generation chamber were effective as drying agents and adsorbed negligible amounts of stibine. The latter desiccant was favoured only on the grounds of proven effectiveness as a drying agent during arsenic determination. 13 Early development work on the trapping step was hampered by the appearance of several spurious peaks that eluted after the stibine peak. These peaks were present irrespective of whether the AAS was operated with, or without, background correction. A similar effect has been reported by Andreae and Byrd during low level hydride determination of tin species.12 These peaks were not however related to stibine peak-height response and diminished in intensity during subsequent determinations in a run.Attempts at characterising these extra peaks by mass spectrometry were unsuccessful, showing only the presence of stibine and carbon dioxide. When cryogenic pre-concentration was complete, the trap was allowed to warm to room temperature unassisted by additional heating. This was sufficient to cause the elution of a sharp peak that showed little sign of tailing. The elution of the pre-concentrated hydride into the atom cell was relatively Table 3. Interference study. Figures in parentheses are o/o signal response, otherwise no signal depression observed; sample, 200 ng 1 - 1 of antimony Concentration of substance added Sb(II1) determination Total Sb determination 1OOOp.p.m.(5000000-foldexcess) 10p.p.m. (50000-fold excess) . . Sz-, Cr(III), Fe(III), Fe(II), Na(I), Br-, F-, NO3-, S042-, K(I), Cl- Mo(VI), AI(III), Bi(III), Ca(II), Co(II), Mn(II), M m ) , a m , Srm) 3 VW) 1 p.p.m. ( 5 000-fold excess) . . . . Se(VI), Cd(fI), Cu(II), Ni(II), Te(IV) 100 pgl-1(500-foId excess) . . , . Cr(VI), Ag(I), As(III), As(V), Se(IV), NO2-, Pb(II), Sn(I1) (82%) 50 pg l-l(250-foId excess) . . . . - Br-, K(I), S042-, C1-, Na(1) S2-, Fe(III), Fe(II), Mn(VII), Se(VI), AI(III), Cd(II), Ca(II), Co(II), cu(II), Mn(II), Mg(II),Hg(II), Sr(II), V(V), F-, NO3- Cr(III), Cr(VI), Ag(I), Bi(III), Te(1V) As(III), Pb(II), As(V) (86%), NOz-, Mo(VI) (85%) Sn(1V) (80%)224 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 insensitive to fluctuations in laboratory temperature during analytical runs.Atomisation The quartz tube atomiser (Fig. 3) was heated by means of a purpose-built electric furnace. This consisted of two “coffin- shaped” halves made from alumina cement into which hand-wound wire elements had been embedded. Power supply to the furnace was controlled by means of a variable transformer. The furnace was operated at the optimum temperature range for atomisation, 800-1000 “C. After extended use, the performance of the quartz tube atomiser deteriorated, giving rise to noise and poor precision. As a result, the atomiser assembly was replaced about every 3 months and furnace temperature was re-optimised accord- ingly. 2 c m . I Borosilicate glass 2 mm i.d. I + i.d. ! I AAS light path Analytical Performance Data Performance data is summarised in Table 2.The calibration graphs were linear from 0 to 5 ng for total antimony and 0 to 10 ng for antimony(II1). Within batch precision was evaluated by replicate analyses of ten, 5-ng standards (Table 2). Detection limits based on three times the standard deviation of the blank were 10 ng 1-1 for total antimony and 1 ng 1-1 for antimony(II1). These values do not represent the ultimate performance of the system as they are largely controlled by blank contributions, which could be improved upon by enhanced reagent clean-up procedures. Interferences The effect of other substances on the hydride generation methods was assessed by adding a known excess (by mass) of a foreign ion to a distilled water solution containing 200 ng 1-1 of antimony.Results of the study are presented in Table 3. No substance tested interfered at levels that would be realistically expected in relatively unpolluted natural waters. Possible bias during sea water analysis was assessed by a spiking recovery experiment using sea water (Table 4). Excellent recoveries were obtained using the total antimony method, but some difficulty was experienced in obtaining good recoveries of antimony(II1) added to sea water. This was thought to arise from oxidation of the added antimony(II1) spike to anti- mony(V). Satisfactory results were obtained after first purging the sea water with nitrogen in order to remove dissolved gases. This was not regarded as a serious interference, but a general statement of the instability of this species at high concentra- tions in oxygenated sea water. Sample Storage and Analysis of Real Samples The described methods have been used to determine anti- mony speciation in a number of natural waters and some typical results are given in Table 5.Antimony(II1) has always been a minor species in the waters analysed. For speciation analysis, sample storage is an important problem and often a neglected factor. Ideally, samples should be filtered (ca. 0.45 pm cut-off) immediately after collection, using filters chosen to give negligible antimony adsorption. In our experience glass-fibre filters are suitable for this purpose. Priority must then be given to performing the antimony(II1) determination as rapidly as possible (tentatively within 2 h), as it is this species that is most susceptible to decay.If delays are unavoidable prior to analysis, then storage of filtered samples by deep freezing, as recommended by Andreae,s is advised. Fig. 3. tube atomisation cell (a) Glass bead stibine trap; and (b) construction of the quartz Table 4. Results of spiking recovery experiments on sea water Amount added n Recovery, ‘70 Total antimony determination- 5 ng Sb(II1) . . . . . . . . . . 3 100.0 * 5.5 5ngSb(V) . . . . . . . . . . 3 98.0 k 6.3 10 ng Sb(II1) . . . . . . . . 3 98.7 * 3.5 Antimony(IIZ) determination- Conclusions The method described provides a simple, sensitive and rapid method of determining inorganic antimony speciation in natural waters. It has been proven effective in the routine analysis of natural water samples.Interferences have been studied in detail and found not to pose a problem at levels encountered in natural waters. This work was supported by a studentship (to S.C.A.) from the Science and Engineering Research Council. Table 5. Analysis of some real water samples Sample location Date Salinity, YO Total Sbhg 1-1 Sb(III)/ng 1-1 . . . . 168 2 Loch Ewe, North West Scotland 26.3.83 - 9.4.83 - 145 3 Tamar River Estuary, England . . . . 18.7.83 <2 352 10 16 616 <1 33 382 5JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 225 References 1. Andreae, M. O., Asmode, J. F., and Van7 Dack, L., Anal. Chem., 1981, 53, 1766. 2. Andreae, M. O., Anal. Chem., 1977, 49, 820. 3. Howard, A. G., Arbab-Zavar, M. H., and Apte, S . , Mar. Chem., 1982, 11, 493. 4. Howard, A. G., Arbab-Zavar, M. H., and Apte, S., Estuarine, Coastal Shelf Sci., 1984, 19, 493. 5. Portman, J. E., and Riley, J. P., Anal. Chim. Acta, 1964, 31, 509. 6. Abu-Hilal, A. H., and Riley, J. P., Anal. Chim. Acta, 1981, 131, 175. 7. Standing Committee of Analysts, (to Review Standard Methods for Quality Control of the Water Cycle), “Antimony in Effluents, and Raw, Potable and Seawaters by Spectro- photometry using Crystal Violet 1982 Version: Tentative Method,” Department of the Environment, National Water Council, HMSO, London, 1982. 8. Andreae, M. O., in, Grasshoff, K., and Ehrhardt, M., Editors, “Methods of Seawater Analysis,” Verlag-Chemie, Berlin, 1983, p. 225 Braman, R. S., Johnson, D. L., Foreback, C. C., Ammons, J. M., and Bricker, J. L., Anal. Chem., 1977, 49,621. Aggett, J., and Aspell, A. C., Analyst, 1976, 101, 341. Yamamoto, M., Urata, K., and Yamamoto, Y., Spectrochim. Acta, Part B , 1981, 36, 671. Andreae, M. O., and Byrd, J. T., Anal. Chim. A m , 1984,156, 147. Howard, A. G., and Arbab-Zavar, M. H., Analyst, 1981, 106, 213. 9. 10. 11. 12. 13. Paper J5l.59 Received December 12th) 1985 Accepted January 2 7th) I986
ISSN:0267-9477
DOI:10.1039/JA9860100221
出版商:RSC
年代:1986
数据来源: RSC
|
18. |
Application of a simplified model for atom formation by a tungsten-strip heater in atomic absorption spectrometry |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 227-230
Susumu Nakamura,
Preview
|
PDF (431KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 227 Application of a Simplified Model for Atom Formation by a Tungsten-strip Heater in Atomic Absorption Spectrometry Susumu Nakamura National Chemical Laboratory for Industry, Yatabe, Tsukuba, lbaraki 305, Japan A model has been proposed for the mechanism involved in atom formation in electrothermal atomisation atomic absorption spectrometry. To simplify the investigation, a tungsten-strip heater was used as the atomiser. The model is based on the assumption that the rate of atom formation, given by a simple Arrhenius type expression, is a function of the number of atoms of the sample on the heater and the heater temperature. Signal measurements were made using a fast-response electronic system developed to remove signal distortion.Experimental results obtained for Ag, Cu and Fe demonstrated the validity of the proposed model. Activation energies of the model are also given for these analytes. Keywords: Atom formation model; tungsten-strip heater; electrothermal atomisation atomic absorption spectrometry Information about the mechanism of atom formation in electrothermal atomisers is essential for achieving improve- ments in the instrumentation and analytical characteristics of electrothermal atomic absorption spectrometry. L'vov repor- ted' the first investigation on atomisation mechanisms. Johnson et al. ,2 Fuller,3.4 Torsi and co-workers,5-7 Sturgeon et af.,s van den Broek and de Galan,9 Smets,1° Holcombe and Rayson,ll Men'shikov et af.12 and Chungl3 have since proposed models to describe the mechanism of atom forma- tion idon electrothermal atomisers. Most of these studies concerned graphite furnaces.However, as Men'shikov et al. 12 indicated, the experimental vaporisation curves consist of two or more overlapping curves as the result of the superposition of several processes with different activation energies. To explain such phenomena, a complicated expression is required. In addition Smets'o has pointed out that the kinetic behaviour would differ markedly from the ideal evaporation curve owing to the porosity of the graphite wall. Some researchers14.15 used a Ta-lined graphite tube furnace or a Ta-tube furnace. According to Chakrabarti et al.,Ih however, a simple theoretical model for the atomisation of analytes with tube-type furnaces has some limitations, which stem largely from the simplifying assumptions and over- simplified treatment of the atom loss processes.Such limita- tions do not seriously impair its usefulness in offering a theoretical framework for understanding the effect of various experimental parameters on the analytical sensitivity. In this study, an attempt is made at applying a model to atom formation with a W-strip heater. The model, which can be used to explain the shape of absorption signals, is basically the same as those proposed by Fuller,3.4 Smets,lO Holcombe and Rayson" and Men'shikov et al.12 with respect to the utilisation of kinetic theory. However, by using the heater, it is possible to avoid such problems as the porosity of the material and the complex phenomena occurring in the tube furnaces.As the result, the atom formation model can be simplified and the physical meaning of the parameters used in the model equation more clearly explained. Experimental Instrumentation A Seiko Instrument & Electronics Ltd. (Tokyo) Model SAS-725 atomic absorption spectrometer equipped with a W-strip heater (Fig. 1) (power supply: Seiko Instrument & Electronics Ltd. Model SAS-714) with a deuterium back- ground corrector was used for absorbance measurements. With a commercial system it is impossible to get a signal that truly reflects the fast phenomenon of atomisation, because the response time of the system is of the same order of magnitude as the total width of the signal. Hence a fast-response system was used in this work (response time ca.10 p) in which the output of the photomultiplier was fed through a pre-amplifier and an analogue to digital converter into a computer. The details of the system have been described previously.17~18 Measurements of the heater temperature were performed with a Chino Works Ltd. (Tokyo) Model IR-P pyrometer, the electronic system of which had been modified to provide a fast response (26 ps). Temperatures of the surface of the heater were measured using the pyrometer (900 nm, E = 0.38). The precision of the temperature measurements was ca. k15 K. The accuracy of the temperatures indicated was ascertained by a method based on the melting-points of some metals, and it was demonstrated that the furnace temperature can be obtained, using the pyrometer, to within +15 K in the range 1000-2500 K.19 An NEC PC-8801 microcomputer was used for the simultaneous acquisition of absorbance and tempera- ture data.The data acquisition was performed with a 7.52-ms sampling interval corresponding to twice the chopping cycle (266 Hz = 3.76 ms) of the atomic absorption spectrometer. I I Fig. 1. W-strip heater, dimensions in mm228 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 3 - 2 - I 5 5- t -1 I I 1 I 6 8 10 12 Temperature/K x 10' Fig. 2. Thermogravimetric analyses of Ag, Cu and Fe samples; A, AgC104 to Ag; B, Cu(C104),.7H20 to Cu; and C, Fe(C104)3.6H2 to Fe. Arrows indicate the temperature corresponding to the ashing temperatures for the respective analyses by atomic absorption spectrometry 1 .o 0 Fig.3.. Signal profiles with differential heating rates used for obtaining plots of ln(No - N) versus time. Heating rate: (a) rl, step 16; and ( b ) r5, step 20 Time/s I 1 -5 . ID +o 4 0 6 7 8 9 10 -10 IITIK-' x 10-4 Fig. 5. Plots of In k and absorbance of Ag versus 1/T. Initial heating rate of the heater: 0, 8.4; 0, 8.5; 0, 8.6; A, 8.7; and H,, 8.9 K ms-l Reagents and Procedures All chemicals used were of analytical-reagent grade. Stock solutions were prepared by dissolving the pure metals in perchloric acid. This acid was chosen for the dissolutions in order to avoid oxidation of W and the formation of chlorides of the analytes, which may sublime during ashing. Test solutions were prepared by appropriate dilution to 1 mg ml-1 stock solutions just prior to their use.The acid concentration of the test solutions was 0.00001%. A 5- or 10-p.1 volume of the 'test solution (0.1 pg ml-1) was dropped on to the furnace with an Eppendorf pipette. Light from a hollow-cathode lamp and a D2 lamp passes along an optical pathway 2 mm above the centre of the heater. Samples were dried, ashed and atomised in the presence of H2 (0.5 1 min-1) - Ar ( 5 1 min-1) purge gas. The temperature of the central portion of the heater was monitored; this central portion (3 X 5 mm) is the analytical zone on to which the samples are dropped. Model for Atom Formation In this study, a model is considered that is based on the following assumptions about atom formation: (i) the number of atoms present in the optical pathway per unit time is proportional to the rate of atom formation on the surface; and (ii) the rate of atom formation is proportional to the number of atoms present as a thin layer on the furnace.According to the above assumptions, the rate of change in the number of analyte atoms on the heater per unit time (dNldt) is given by the equation dNldt= k ( N 0 - N ) . . . . . . (1) where No is the total number of analyte atoms on the furnace and N the number of analyte atoms vaporised during time t. As the number of atoms in the optical pathway is proportional to the signal, dNldt, No and N are given by the absorbance signal obtained with a 7.52-ms sampling interval, the accumu- lated absorbance signal during atomisation [i.e., No = Jom(dN/dt)dt] and the accumulated absorbance signal during a given time t [i.e., N = J$dN/dt)dt)], respectively.The rate D constant k can be calculated from dNldt, N and No by equation (1). I 1 I I I Tim e/s 0.3 0.4 0.5 Fig. 4. Plots of In (No - N) versus time. Temperature of the heater: A, 1500; B, 1600; C, 1700; D, 1800; and E, 1900 K Assuming that k is related to the temperature of the heater by expression (2), k = A exp(B/T) . . . . * (2) where A and B are constants, substituting expression (2) into equation (l), the following differential equation is obtainedJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 229 I X X I Table 1. Activation energies (kcal mol- 1) obtained using the proposed model Initial heating Ag rate of Step heater/K ms-1 Below m.p. Above m.p. c u Fe 16 ( T I ) 8.4 55 11 46 58 18 (4 8.6 70 16 44 56 19 (r4) 8.7 67 16 41 58 20 (5) 8.9 62 k 4* 14 k 2* 41 k 4* 66 k 9* 17 (r2) 8.5 66 14 40 59 * n = 10.x x X X X X X 0) C m - 1.02 2 2 X- x I , " X I X X X X X X IITIK-1 x 10-4 Fig. 6. Plots of In k and absorbance of Cu versus 1/T. Initial heating rate: 0 8.4; 0, 8.5; 0, 8.6; A, 8.7; and ., 8.9 K ms-1 X 0 X X I X X X X X X X I x I/T/K-1 X Fig. 7. Plots of In k and absorbance of Fe versus 1/T. Initial heating rate: 0. 8.4; 0, 8.5; 0, 8.6; A, 8.7; and ., 8.9 K ms-1 dN/dt = A(No - N ) exp (BIT) Expressing B as B = -E,/R gives equation (3), dNldt = A(No - N ) exp (-E,/RT) . . (3) where E, corresponds to the activation energy of the atomisation process of the metal on the W heater. I BO I I U !J u I I I I t 1 1 0 0.1 0.2 0.3 0.4 0.5 Mass of analytehg cm-2 Fig.8. below the melting-point; and B, above the melting-point Plots of activation energy of Ag versus mass of the analyte: A, Results and Discussion Chemical State Changes in the chemical state of the elements due to the ashing procedure were investigated by thermal gravimetric analysis [ Rigaku Industrial Corporation (Ohsaka) Model 8089A2 thermal gravimeter; heating rate 10 K min-1, sample mass 10 mg]. The atmospheric conditions are the same as are used for the ashing stage of the atomic absorption spec- trometry (a mixture of 0.5 1 min-1 of H2 and 5 1 min-1 of Ar). Ashing temperatures for the elements tested were 820 K for Ag and 1200 K for Cu and Fe. The results of the thermal gravimetric analyses (Fig. 2) demonstrated that all the analytes (perchlorates) had changed into the metallic state below the ashing temperature.There- fore only atom formation from the metallic state during atomisation of the analytes on the W heater will be considered here. Atom Formation Integrating equation (1), we obtain (dN/(No - N) = (kdt At the constant temperature (k = constant), this becomes ln(No - N) = - kt + C The time required for the heater to reach a certain tempera- ture varies with the heating rate. The time ( t ) and ln(No - N> at a specific temperature were obtained by measuring the absorbance signals with different heating rates (Fig. 3). As shown in Fig. 4, the relationship between ln(No - N) and t yields straight lines for the different temperatures for all parts of the Cu signal profile. This demonstrates that the atom formation process can be expressed by equation (1).For Ag (328 nm) atomisation, In k determined from equation (1) and the absorbance signal are plotted against the reciprocal temperature (Fig. 5 ) . The absorbance data give a230 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 temperature range in which the signal can be observed under the measurement conditions used. Clearly, fairly good agree- ment is observed for the k values for the different heating rates (steps 16-20). It can also be seen that the data yielded straight lines in the temperature regions above and below the melting-point of Ag. The data for Cu (325 nm) and Fe (248 nm) in Figs. 6 and 7 also yielded straight lines. The activation energies calculated from the data given in Figs.5-7 are given in Table 1. This shows that the activation energy for each analyte was within experimental error in spite of the different heating rates of the heater. There is a linear relationship between the activation energy and heat of vaporisation. Fig. 8 shows the effect of the mass of analyte (Ag) on the activation energy at temperatures below the melting-point of Ag. The activation energy increases with increasing mass, approaching the value of the heat of sublimation (69 kcal mol-1).20 Above 0.4 ng cm-2 (2.3 X 1012 cm-2), the experimental activation energy of Ag below the melting-point is the same as that for the sublimation of bulk Ag.20 With an increase in initial mass, Ag may form islands in the temperature region below the melting-point. Arthur and Cho21 proposed a kinetic model in which two-dimensional nucleation of mobile adsorbed atoms occurs upon adsorption, while loss of atoms from the edges of disc nuclei is the rate-controlling step for desorption.Their model implies that the bonding of metal atoms to the basal plane of the graphite is weak and non-localised, with adsorption occurring only when two-dimensional nucleation permits metal - metal bonding. Good agreement is observed between the value obtained for Cu in the present work (41 k 4 kcal mol-1) and Arthur and Cho’s result (47 kcal mol-1). It is probable that the analytes on the heater form islands between the melting-points of the Ag, Cu and Fe and a monolayer above the melting-points. This study was undertaken in order to obtain a model for the phenomena involved in the formation of atoms by a W-strip heater.The experimental results clearly show that atom formation can described in terms of simple kinetic theory. As this work confirmed that the use of a W-strip heater allows us to obtain activation energies with a good degree of accuracy, it will thus be interesting to use this heater to understand chemical interferences; such a study is being pursued. The author thanks Drs. Masaaki Kubota and Yoshinori Kobayashi for many useful discussions. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17 18. 19. 20. 21. References L’vov, B. V., “Atomic Absorption Spectrochemical Analysis,” Adam Hilger, London, 1970, p. 204. Johnson, D. J., Sharp, B. L., West, T. S., and Dagnell, M. R., Anal. Chem., 1975,47, 1234.Fuller, C. W., Analyst, 1974, 99, 736. Fuller, C. W., Analyst, 1975, 100, 229. Paveri-Fontana, S. L., Tessari, G., and Torsi, G., Anal. Chem., 1974,46, 1032. Torsi, G., and Tessari, G., Anal. Chem., 1975,47, 839. Tessari, G., and Torsi, G., Anal. Chem., 1975, 47, 842. Sturgeon, R. E., Chakrabarti, C. L., and Langford, C. H., Anal. Chem., 1976,48, 1792. van den Broek, W. M. G. T., and de Galan, L., Anal. Chem., 1977,49,2176. Smets, B., Spectrochim. Acta, Part B , 1980, 35, 33. Holcombe, J. A., and Rayson, G. D., Prog. Anal. At. Spectrosc., 1983, 6, 225. Men’shikov, V. I., S. E. Vorob’eva, S. E., and Tsykhanskii, V. D., 2. Anal. Khim., 1984, 39, 591. Chung, C. H., Anal. Chem., 1984,56,2714. Gregoire, D. C., and Chakrabarti, C. L., Spectrochim. Acta, Part B , 1982, 37, 611. Suzuki, M., and Ohta, K., Research report, No. 40, Asahi Glass Co. Ltd. (Asahi Garasu Kogyo Gijutsu Shoreika Kenkyu Houkoku), 1982, p. 219. Chakrabarti, C. L., Chang, S. B., Lawson, S. R., and Wong, M., Spectrochim. Acta, Part B , 1983, 38, 1287. Nakamura, S., Kobayashi, Y., and Kubota, M., Bunseki Kagaku, 1985,34, 682. Nakamura, S . , and Kubota, M., Bunseki Kagaku, 1986,35,61. Nakamura, S., Kobayashi, Y., and Kubota, M., Spectrochim. Acta, in the press. Moore, W. J., “Physical Chemistry,” 3rd Edition, Prentice- Hall, Englewood Cliffs, NJ, 1962, p. 59. Arthur, J. R., and Cho, A. Y., Surf Sci., 1973, 36, 641. Paper J.5146 Received October 22nd, 1985 Accepted January 21st, 1986
ISSN:0267-9477
DOI:10.1039/JA9860100227
出版商:RSC
年代:1986
数据来源: RSC
|
19. |
Comparative study of the sputtering process in the conventional and microwave-coupled hollow-cathode discharge |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 231-235
Sergio Caroli,
Preview
|
PDF (1383KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 23 1 Comparative Study of the Sputtering Process in the Conventional and Microwave-coupled Hollow-cathode Discharge Sergio Caroli," Oreste Senofonte, Nicola Violante, Oriano Falasca and Achille Marconi lstituto Superiore di Sanita, Viale Regina Elena 299, 00161 Rome, Italy Mauro Barbieri lnstituto di Medicina del Lavoro, Universita Cattolica del Sacro Cuore, Largo Francesco Vito 1, 00167 Rome, Italy The ablation rate of sample atoms in conventional d.c. hollow-cathode discharge was compared with that observed when a microwave field is superimposed on the discharge. On the basis of mass variation as well as of data from an electron microscope, it could be ascertained that the effect on a material of microwave irradiation decreases with increasing melting-point.The materials used in this study were zinc, copper, molybdenum and graphite. Thus, the enhanced emission intensity observed in the microwave-coupled hollow-cathode source is, at least for low-melting specimens, due to the combined effect of more efficient excitation and higher density of the atomic cloud within the cathode cavity. Keywords: Hollow-cathode discharge; microwave-coupled hollow-cathode discharge; sputtering mechanism In recent years the potential usefulness of the microwave- coupled hollow-cathode discharge (MW-HCD) has been investigated by a number of workers.1-7 This novel radiation source in atomic spectrometry combines the advantages of both types of excitation mechanisms and at the same time allows some of their shortcomings to be overcome.At present there is sufficient experimental evidence to support the view that MW-HCD gives rise to signal to background ratios (SBRs) that are definitely better for many elements than those of the conventional d.c. operating mode, usually being higher by one order of magnitude. On the other hand, the theoretical interpretation of the phenomena underlying the working mechanism of this compo- site source is still far from completely clear. However, the increase in electron density, paralleled by an analogous increase in the population of carrier gas metastables, at least with argon as the carrier gas, seems more likely to be a consequence of the additional energy supplied to the plasma. Whether this process is accompanied by a change in the ablation rate of the cathodic material is still open to debate.Preliminary data obtained in our laboratory showed that MW superimposition would not affect this to any great extent, thus further corroborating the assumption that their influence is mainly exerted on the gaseous phase. The lack of sufficient information on this aspect prompted us to undertake a systematic investigation on the possible effects of MW irradiation on the sputtering processes for a number of representative cathode materials. Experimental The vacuum spectrograph, MW-HCD tube and ancillary apparatus have been described in detail previously. 1~ Their characteristics are summarised in Table 1. The procedure followed in this study requires the machining of hollow cathodes from the bulk materials, followed by preliminary mechanical cleaning, rinsing three times in hexane and sputtering for 5 min in order to remove all impurities and adsorbed gases.Each cathode was sub- sequently accurately weighed and stored in a desiccator until measurements were started. For each element two series of discharges were performed, one in the conventional d.c. hollow-cathode mode and the other with superimposition of an MW field, as detailed in Table 2. The duration of each single discharge was 30 min, followed by a pause of the same duration in the desiccator and determination of the mass before initiating a subsequent cycle, for a total of six cycles. The whole set of measurements was replicated three times and the values were averaged. It is well known that under certain conditions of current intensity and gas pressure, such as those adopted in this study, the cathodic material is mainly sputtered from the bottom and lower half of the electrode inner wall whereas redeposition occurs in the upper tract. This results in the shaping of the cavity, after a sufficient length of time, like an hour-glass.It is one of the properties of the hollow-cathode discharge, as a consequence of its special geometry, that most of the volatilised material is confined within the cavity with a small probability of escaping. The process of mass loss takes place, however, at a rate that is still measurable and therefore we have assumed that it is directly related to the extent of atomisation. In order to appraise better the variation of mass with time, hollow cathodes with detachable bottoms (see Fig.1) were prepared. Losses or increases in the mass of the two components could thus be measured independently of one another. Table 1. Instrumentation Spectrograph . . 1 m/800 RSV Prazisionsmessgerate, Paschen- Runge mounting, equipped with two concave gratings of 1200 and 2400 grooves mm-1, respectively, spectral range 120-600 nm, entrance slit width 30 pm o.d., 6 mm i.d. and 15 mm deep cathodes, continuous flow of inert gas (either Ar or He) Vacuum gauge . . Generators . . HVG 2,25-500 mA current-stabilised unit, RSV Prazisionsmessgerate ; Bosch microwave unit with 20-200 W nominal output power and frequency 2450 MHz HCD tube . . Water-cooled demountable lamp, with 8 mm Thermotron TM lU2, Leybold-Heraeus Scanning electron microscope .. Cambridge Stereoscan 180 at 20 kV accelerating voltage and 250 pA beam current, 45" tilt angle; magnification from 13 to 1200X * To whom correspondence should be addressed.232 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 2. Operating conditions Transition Element point/K Zn . . . . . . . . 629.7$ Cu . . . . . . . . 1356$ Mo . . . . . . . . 28834 C . . . . . . . . 38239 * c = Continuous, p = pulsed. t Effective power within the lamp ca. 30 W. $ Melting-point. 9 Sublimation-point. Applied current intensity/mA* 100 (PI 100 (PI 300 (c) 300 (c) 300 (PI 300 (PI 300 (c) 300 (c) 300 (c) 300 (c) Resulting voltage range/V 290-3 10 28CL293 400-4 10 38CL390 30CL305 296300 25CL255 24CL250 30%400 37CL390 Carrier gas pressure/Pa 426 (Ar) 426 (Ar) 426 (Ar) 426 (Ar) 1064 (He) 1064 (He) 426 (Ar) 426 ( Ar) 426 (Ar) 426 (Ar) Microwave nominal power/Wt - 200 - 200 - 200 - 200 - 200 Fig.1. adopted Cross-sectional view of the hollow-cathode arrangement Table 3. Mass variation of bottom and wall of hollow cathodes as a function of time Mass variatioxdmg Element time/min Bottom Wall Discharge Zn (without MW) Zn(withMW) . . Cu (without MW) Cu(withMW) . . Mo (without MW) Mo(withMW) . . C (without MW) C(withMW) . . . . 120 . . 120 . . 120 . . 120 . . 120 . . 120 . . 120 . . 120 -0.5 +0.6 +6.8 +4.8 -4.9 -5.3 -1.6 - 1.2 -0.6 -0.9 -26.3 -26.6 -20.4 -11.6 -2.4 -2.2 Table 4. Ratio of mass variations as a function of the logarithm of the transition point (TP) of the cathode material Element RB * Rwt Log(TP/K) Zn .. . . . . . . -1.2 +1.5 2.84 Cu . . . . . . . . +0.71 +1.01 3.13 Mo . . . . . . . . +1.08 +0.57 3.46 C . . . . . . . . +0.75 +0.9 3.58 * Mass variation ratio for cathode bottom. t Mass variation ratio for cathode wall. 1 mm H Fig. 2. Zn surface sputtered without microwaves. Over-all view 1 mrn H Fig. 3. Zn surface sputtered with microwaves. Over-all view Four elements were selected as the cathode material on the basis of their well differentiated sputtering properties, namely zinc, copper, molybdenum and graphite. In addition, the detachable cathode bottoms, used during the different series of discharges, with and without MW irradiation, were inspected by scanning electron microscopy (SEM) in order to detect morphological changes possibly induced by the experimental conditions. Results and Discussion The data reported in Table 3 do not allow an obvious trend to be identified.On the other hand, if the ratio R of the mass variation with microwave superposition to that withoutJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 233 300 pm w Fig. 4. Zn surface sputtered with microwaves. Bottom centre 30 pm - Fig. 5. Zn surface sputtered without microwaves. Bottom centre 30 pm H Fig. 6. Zn surface sputtered with microwaves. Bottom centre 300 pm H Fig. 8. Cu surface sputtered with microwaves. Bottom centre 100 pm +--? Fig. 9. Cu surface sputtered without microwaves. Bottom edge 100 pm H Fig. 10. Cu surface sputtered with microwaves. Bottom edge 300 pm 300 pm H H Fig.7. Cu surface sputtered without microwaves. Bottom centre Fig. 11. MO surface sputtered without microwaves. Bottom centre234 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 300 prn H 1 rnm H Fig. 12. Mo surface sputtered with microwaves. Bottom centre Fig. 16. Graphite surface sputtered with microwaves. Over-all view 20 prn H Fig. 13. Mo surface sputtered without microwaves. Bottom edge 10 prn H Fig. 17. centre Graphite surface sputtered without microwaves. Bottom 100 prn H Fig. 14. Mo surface sputtered with microwaves. Bottom edge 10 pm H Fig. 18. Graphite surface sputtered with microwaves. Bottom centre coupling for both the bottom (B) and wall (W) of the cathodes are considered (Table 4), then a certain dependence on the transition points (TP) of the samples investigated emerges.This is described by the following equations: RB = 0.2710gTP + 3.16 ( Y = 0.84) . . . . (1) Rw = -0.7410g TP + 3.99 (Y = 0.85) . . . . (2) Although this cannot be considered as anything more than a qualitative trend owing to the complex process of the ablation and redeposition of material within a hollow cathode and the diffusion of the atomic vapour from the electrode bore, it confirms that the thermal effects associated with the MW of sample into the discharge zone. Moreover, if the correlation 1 mrn M F.ig. 15. Graphite surface sputtered without microwaves. Over-all superim??osition can significantly contribute to the transport viewJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE Fig. 19. EDXA of the centre of the bottom surface of Zn sputtered in the hollow-cathode conventional mode Lff Fig.20. EDXA of the centre of the bottom surface of Zn sputtered in the microwave-coupled hollow-cathode mode coefficients for the equations (1) and (2) are recalculated without considering graphite, the values improve consider- ably, passing from 0.84 to 0.92 and from 0.85 to 1. This is fully understandable, as graphite is a non-metal with a very high sublimation point (3550 “C). Its sputtering rate can therefore be only slightly affected by the thermal enhancement caused by the MW field. Morphological changes in the cathode bottom surface provide further support for this interpretation, as a series of photographs taken with an electron microscope clearly showed. Some of the effects are illustrated in Figs.2-18. With Zn, the element with the lowest transition point of the group investigated (Figs. 2-6), it is apparent that when the conven- tional HCD is applied, the bottom surface is homogeneo-usly sputtered even after 2 h of discharge, the grooves cayed by the sample machining still being readily visible. Microwave irradiation causes a much more irregular surface, with a well pronounced central zone where the effect of overheating can be recognised. On the other hand, no appreciable differences could be observed at the bottom edge, where redeposition mainly occurs. Thus, for low-melting metals, the contribution 1986, VOL. 1 235 of thermal volatilisation in the MW-HCD operation mode cannot be overlooked even in water-cooled systems. Copper does not show such evident signs of surface alteration when the discharge is subjected to MW irradiation, in keeping with its higher melting-point.Details of the surface attack under both excitation modes are shown in Figs. 7-10. This trend is more marked for molybdenum, as illustrated in Figs. 11-14, for which element only very faint differences may be observed, if any. As far as graphite is concerned, the surfaces under both treatments appear identical even at very high magnifications (Figs. 15-18). Energy-dispersive X-ray analyses (EDXA) were also car- ried out to check whether the morphological changes observed in some instances for the samples investigated were accom- panied by other phenomena at the microscopic level, such as surface enrichment of impurities. This was never the case, however, as is apparent from two of the spectra reported as examples (Figs.19 and 20). Although these findings are only preliminary and lead to little more than a hypothesis, the assumption seems reason- able that MW superimposition on HCD does not lead simply to a more efficient excitation of sample atoms in the vapour phase, but can also affect the rate at which they enter the discharge zone. Also, it seems justifiable to state that the interaction is probably rather complex, as the volatilisation mechanism turns out to be a combination of both a thermal process and physical sputtering, with the contribution of the thermal process being inversely related to the transition point. Moreover, diffusion and back-deposition of cathode material may be significantly affected by the altered working tempera- ture of the supporting gas in a manner that is difficult to forecast. At present, therefore, it can be assumed that the observed increase in emission intensity of analyte atoms can under certain circumstances be ascribed to the joint effect of higher density of sample atoms and better excitation in a more energetic environment associated with the increased electron density in the boosted discharge. Further experiments are in progress to shed more light on these aspects. 1. 2. 3. 4. 5 . 6. 7 . References Caroli, S., Alimonti, A., and Petrucci, F., Anal. Chim. Acta, 1982, 136, 269. Caroli, S. , Alimonti, A. , and Petrucci, F., Trends Anal. Chem., 1982, 1, 368. Caroli, S., Petrucci, F., and Alimonti, A., Can. J . Specfrosc., 1983, 28, 156. Caroli, S., Senofonte, O., Violante, N., Petrucci, F., and Alimonti, A., Spectrochim. Acta, Part B , 1984, 39, 1425. Caroli, S . , Petrucci, F., Alimonti, A., and Zhray, Gy., Spectrosc. Lett., 1985, 18, 609. Caroli, S . , Alimonti, A., and Petrucci, F., in Caroli, S . , Editor, “Improved Hollow Cathode Lamps for Atomic Spectroscopy,” Ellis Honvood, Chichester, 1985, p. 13. Human, H. G. C., and Butler, L. R. P., Spectrochim. Acta, Part B , 1970, 25, 647. Paper 5.514 7 Received October 18th, 1985 Accepted December 30th, 1985
ISSN:0267-9477
DOI:10.1039/JA9860100231
出版商:RSC
年代:1986
数据来源: RSC
|
20. |
Laser-excited atomic fluorescence spectrometry as a practical analytical method. Part I. Design of a graphite tube atomiser for the determination of trace amounts of lead |
|
Journal of Analytical Atomic Spectrometry,
Volume 1,
Issue 3,
1986,
Page 237-241
Klaus Dittrich,
Preview
|
PDF (655KB)
|
|
摘要:
JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 237 Laser-excited Atomic Fluorescence Spectrometry as a Practical Analytical Method Part 1. Design of a Graphite Tube Atomiser for the Determination of Trace Amounts of Lead Klaus Dittrich and Hans-Joachim Stark Sektion Chemie, Karl-Marx-Universitat, WB Analytik, Talstr. 35, DDR-7010, Leipzig, GDR Tube atomiser designs for the laser-excited atomic fluorescence spectrometric (LAFS) technique were investigated. Commercially available cuvettes of the Beckman 1268 type and Perkin-Elmer HGA-500 - EA 3 type that had been modified for use with LAFS were used. The fluorescence was measured within the tube. These tube atomisers were tested and compared with a home-made rod atomiser by measuring the fluorescence of lead. The detection limits, sensitivities and reproducibility obtained by tube atomisation were greatly improved compared with rod atomisation.The influence of some metal nitrates and sodium sulphate on the fluorescence intensity of lead was investigated. With tube atomisation the interferences were reduced. Keywords: Laser-excited atomic fluorescence spectrometry; lead determination; graphite tube atomiser; trace analysis It has been shown that laser-excited atomic fluorescence spectrometry (LAFS) is potentially a very sensitive and selective analytical technique. 1-4 Detection limits at the femtogram level have been obtained (in most instances by extrapolation) using pulsed dye-laser excitation. Falks has shown theoretically that the detection limits in LAFS can be 108 times better than those obtained with AFS using conventional light sources, such as hollow-cathode or electrodeless discharge lamps.Such low detection limits have not so far been obtained in practice. We can conclude therefore that LAFS is the most sensitive method of optical atomic spectroscopy. However, the sensitivity of the method depends not only on the laser power, but also on the efficiency of atom production. Using flames for atomisation, the result is a very high atom dilution in the dynamic flame plasma, leading to poor detection limits. Hence it is better to employ electrothermal atomisers, which provide higher free atom concentrations in the vapour phase. Some workers have used graphite rod atomisers, because of the convenience of the optical geometry for implementing the fluorescence tech- nique.1-4.6.7 As shown in a previous paper,7 we obtained good sensitivities for some metal ions using a carbon rod atomiser and conventional light sources (laboratory-made EDLs) .Such a system does, however, have some disadvantages, principally the strongly negative correlation between plasma temperature and the distance from the top of the carbon rod. This means that after a short time the free atoms combine to form molecules and condensate. This leads to poor excessive vapour-phase interferences and poor analytical perfor- mance.2,6 Recently, Human4 used a Mini-Massman graphite- tube atomiser for AFS measurements, but the tube was used only as an atom generator, the LAFS being measured outside the tube. In this work we have attempted to combine the established analytical advantages of tube atomisers with the potential sensitivity of the LAFS technique.Experimental Apparatus A schematic diagram of the equipment used is shown in Fig. 1. Because the details are important, a description of some of the component parts will be given. Nitrogen laser The primary light source was a pulsed nitrogen laser operating at 337.13 nm (Model No. IGT 300, Central Institute of Scientific Instruments, Academy of Science, Berlin, GDR). The pulse frequency could be varied between 0.3 and 25 Hz and the pulse duration was 8 ns. Pulse frequency and pulse energy are inversely related and, taking account of the atomisation period, an optimum frequency of 6-7 Hz was chosen. The laser power supply voltage was variable between 14 and 20 kV but, in the interests of laser lifetime, an operating voltage of 15 kV was selected, which gave a pulse energy of 2 mJ and a power of 250 kW. Dye laser A tunable dye laser (Model No.FLGR-2, Carl Zeiss, Jena, GDR) of the Hansch type was used to excite the Pb direct-line fluorescence. The laser was equipped with a 2600 line mm-1 grating and with appropriate dyes could be tuned over the i? 1 7 IP Fig. 1. Schematic diagram of the atomic fluorescence spectrometer with laser excitation. 1, Power supply for pulsed nitrogen laser; 2, pulsed nitrogen laser; 3, mirror for the light triggering the measure- ment; 4, dye laser with generation of secondary harmonic (SHG); 5 , filter for visible part of laser light; 6, power supply for atomiser; 7, atomiser (carbon tube or carbon rod type); 8, condenser lens; 9, monochromator for fluorescent light; 10, photomultiplier (PM) tube; 11, power supply for PM tube; 12, power supply for trigger element; 13, trigger element for boxcar integrator; 14, boxcar integrator; 15, X - Y recorder; 16, pyroelectric joule meter for laser light; 17, power su ply for joule meter; 18, oscilloscope for joule meter; 19, mirror for regection of laser light; 20, thermostat; 21, nitrogen cylinder; and 22, argon cylinder238 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL.1 range 400-900 nm. A coumarine type dye (C462) was used to produce output in the range 525-590 nm with a maximum at 560 nm, yielding 107 pJ with a pulse duration of 6 ns. The line half-width was 40 pm.KDP crystals were used to frequency- double the laser first harmonic, achieving a 5-10% conversion efficiency with a line width of 20 pm and the pulse duration of 4 ns. The power in the second harmonic is sufficient to saturate only a few spectral transitions. Energy meter For controlling and measuring the pulse energy of the dye laser, a pyroelectric joule meter was used. The registration of the pulse was carried out with an oscilloscope. The maximum sensitivity of the system was 10 mV ClJ-1 and 0.1 p.J can be measured with this device. Detection system The fluoresced light was focused by a quartz lens (f = 40 mm; diameter = 38 mm, aperture = 1 : 1.05) on to the entrance slit of an SPM 1 monochromator (Carl Zeiss) with an aperture of 1 : 6.7. Depending on the prisms used (glass, quartz or sodium chloride) different wavelength regions can be utilised.For our purpose (direct-line fluorescence of lead at 405 nm), the glass prism has the best resolution (about 5 nm mm-l). The maximum slit width of 3 mm was used, giving a spectral band width of about 15 nm. This type of monochromator was suitable for measuring atomic fluorescence. The slit height was optimum for all atomisation systems at 1 mm (rod, Beckman tube and EA 3 tube). The intensity of the light was measured with a type M 11 FVC 520 photomultiplier (VEB Werk fur Fernsehelektronik, Berlin, GDR). The cathode of the photomultiplier is sensitive between 185 and 1200 nm. High voltages of up to 1.5 kV can be used. The optimum signal to background ratio was obtained at 1.3 kV.To improve the signal to noise ratio, the output pulse from the PMT was filtered and stretched from 80 to 800 ns. The signal was measured with a BCI 280 boxcar integrator (ZWG, AdW der DDR, Berlin, GDR). Triggering of this instrument by part of the nitrogen laser light was effected by a photodetector. The signal delay was corrected. The minimum gate duration of the integrator was 10 ns, but in following a prolonged signal a gate duration of 400 ns was optimal. Signals were recorded with an X - Y recorder (Model 620.01, Messgeratewerk Schlotheim, GDR). The best sensi- tivity of the recorder in relation to noise problems was 100 mV cm-1. Atomiser systems A laboratory-constructed device described previously7 was used as the graphite rod atomiser.The design of the atomiser is shown in Fig. 2 and the graphite rods used are shown in Fig. 3. The lower hole guarantees a high electrical resistance and a rapid heating rate of about 400 K s-1. The maximum final temperature is 2800°C. The top of the rod was pyrolytically coated with graphite. As base material, graphite of TO quality (VEB Elektrokohle, Berlin, GDR) was used. Two different modified commercial systems were used as graphite tube atomisers. 1. Type 1268 (Beckman Instruments, USA). The design of the modified system and of the tube is shown in Fig. 4. The hole for entry of the exciting laser beam was varied in distance from the central dosage point to establish the best position. The tube length was also varied and a length of 50 mm was found to be best compared with the original length of 68 mm.2. Modified HGA type (Perkin-Elmer, USA)/EA system (Carl Zeiss) (Fig. 5). In this instance the maximum distance between the centre of the tube and the centre of the entrance hole is 2 mm. In the outer graphite ring a third hole was bored. 20 mrn P I 5 I 6’ A- Fig. 2. Carbon rod atomiser. 1, Sample holder; 2, contacts for sample holder (graphite); 3, contacts for power supply (brass) (5, water cooled); 4, spring for holding the graphite rod; and 6, argon inlet @ 4.0 @ 6.0 Fig. 3. Schematic diagram of graphite rods used for AFS. All dimensions in millimetres. Upper de ression (diameter 4 mm) for the sample solution; lower depression 6iarneter 2.5 mm) for enhance- ment of the electrical resistance and faster heating The graphite tube in the modified Perkin-Elmer atomiser has two holes.However, it has an outer graphite ring to shield the graphite tube from air. This ring normally has two holes, one for sample introduction and the other for temperature measurement. In our modified atomiser we have no ternpera- ture control system. For geometric reasons we used one hole (originally for temperature measurement) for sample intro- duction, the second hole (originally for sample introduction) as the laser inlet and a new third hole as the laser outlet (Fig. Table 1 gives some parameters of the atomisers. In all instances pyrolytically coated tubes were used. The first results showed that the best type of tube atomiser was the HGA/EA 3 type, which was used in most instances. 6). Procedure A 10-20 1.11 volume of the analyte was placed on top of the rod or into the tube. Using programmed heating, the sample was dried, ashed and atomised, the laser beginning to operate in the ash phase.The system was excited at 283.307 nm and the direct-line atomic fluorescence at 405.782 nm was measured. Reagents Lead(II) stock solution, 0.2 mg ml-1 in 1 M nitric acid (Suprapur). The solution was diluted with 0.01 M nitric acid.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 239 3 1 Results and Discussion u 1 6 . - 15 8 9 1011 12 13 5 14 6 Fig. 4. Modified Beckman tube atomiser (type 1268). 1, Contacts for power supply; 2, inlet holes for outer argon shielding gas stream; 3, inlet and outlet for water cooling; 4, screws and springs for holding tube; 5 , tube holder (brass); 6, cone for direct tube holding (copper); 7, graphite tube; 8, gasket; 9, copper tube for shielding; 10, holes for entrance of laser light through the wall; 11, holes for entrance of laser light through 10 and outlet for outer argon shielding stream; 12, hole for sample introduction; 13 and 14, isolators; 15, screws for mechanical stabilisation; 16, quartz windows; and 17, inlet for inner argon gas stream 11 5 6 10 6 5 4 3 2 1 17 18 19 20 21 5 15 Fig.5. Modified HGAiEA 3 atomiser. Left. side view; right, front view. 1, Screw and 2, spring for mechanical holding; 3, isolator; 4 and 11, tube holding system (brass); 5 , inlets and outlets for water cooling; 6 and 10, inlet holes for outer argon shielding gas stream; 7 and 9, graphite shielding rings; 8, hole for sample introduction; 12 and 13, window holders; 14, quartz windows; 15 and 16, inlet holes for inner argon shielding gas stream; 17, graphite tube (normal dimensions); 18 and 19, asymmetric entrance hole for exciting laser light in the graphite tube and in the graphite ring (7); and 20 and 21, holder and screw for moving 4 Laser Beam Fig.6. Principle of the modified atomiser Znterferents. For studying matrix interferences the following salts of analytical-reagent grade were used: NaN03, KN03, Al(N03)3 and Na2S04. RbN03, Cd(N03)2, AgN03, CU(NO~)Z, Mg(N03)2, Optimisation of the Geometry of Excitation and of the Gas System Depending on the type of atomiser there are different parameters that can influence the fluorescence signal inten- sity.For the carbon rod atomiser these are (a) height of the excitation zone above the top of the rod, (b) diameter of the beam in relation to the top surface of the rod, (c) gas stream velocity and (d) heating rate. A minimum distance from the surface of the top of the rod gave the best results because at this point the atom concentra- tion is highest. At greater distances the plasma temperature is lower and thus the atom concentration is smaller owing to condensation and dilution processes. The geometry of the laser beam is about 0.5 x 3 mm (height X width). Depending on the distance from the dye laser, the laser beam diverges. The best results were obtained with a width of 6 mm (identical with the rod diameter) because with such a small distance from the top the atom-containing plasma is cylindrical.The argon flow-rate was varied between 40 and 200 1 h-1 and no influence was observed. We used a flow-rate of 80 1 h-1 as previously.7 The maximum heating rate gave the best fluorescence signals (peak height). We found no back- ground, because the lead evaporation was very fast, and no stray light, because we used direct-line fluorescence. For carbon tube atomisers similar and other parameters exist: (a) excitation geometry and fluorescence light measure- ment; (b) position of the hole in the tube for laser excitation; (c) gas flow-rate; (d) dimensions of the carbon tube; and (e) heating rate of the tube. With tube atomisers two measurement geometries are possible: excitation perpendicular to the tube axis and measurement along the tube axis, and the opposite conditions.In both instances equal sensitivities were obtained. In spite of this, we used excitation perpendicular to the axis and measured the atomic fluorescence along the tube axis. The reason was that additional holes for laser beam excitation of the appropriate size can be formed and the diameter can be very small. This means that there are only small losses of atoms through the tube holes, because the atomiser tube is more efficient in a closed form. For the energy absorption, a transition with high oscillator strength was used, and strong absorption in all parts of the atom cloud was therefore obtained. Excitation along the tube axis leads to large pre-filter effects, which results in a loss of fluorescence intensity.Excitation perpendicular to the axis does not give pre-filter effects. Because in most instances direct-line fluores- cence was used, for practical purposes no post-filter effects along the tube axis were observed. The position of the access port for the exciting beam can be in the centre or displaced from the centre. With the large tube (Beckman type) a distance of 9 mm from the centre was chosen (hole diameter = 4 mm). For lead the central position gave only 50% of the sensitivity obtained with the other position, because in the central position there are atom losses as a result of convection and diffusion processes. With small tubes (HGA/EA 3 type) the same tendency was found. However, the distance between the central sampling port and the hole for laser excitation can be varied only up to 2 mm.The dimensions of the hole are height 1.5 mm and width 4 mm. The inner gas stream was directed to the monochromator (the holes for the laser beam also lie in this direction). The gas flow-rate was optimised to 20 1 h-1. Up to 5 1 h-1 no changes were observed but at higher flow-rates smaller signals were obtained. With the small atomiser (HGA/EA 3 type) measurements were carried out with and without quartz windows at the end of the tube axis. There are three possibilities, as follows. (a) Both quartz windows are in place. Using a gas stream from both sides of the tube the influence of the gas flow-rate is240 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 Table 1. Optimum conditions of some atomisers in LAFS Type of atomiser Final temperature] "C Rod atomiser .. . . . . . . . . . . . . . . . . 2800 Modified 1268 type (Beckman) . . . . . . . . . . . . 3000 1268 type (Beckman) . . . . . . . . . . . . . . . . 3000 HGA/EA 3 type . . . . . . . . . . . . . . . . 3000 Inner Heating Tube Tube tube rate/ diameter/ length/ volume/ K s-1 mm mm mm3 400 300 8.6 68 3950 800 8.6 50 2900 6000 5.5 28 665 - - - Heated mass of graphite/ mg 300 8000 6000 1000 Table 2. Results of the determination of trace amounts of lead by LAFS Reciprocal sensitivity Detection Type of per 100 mVI Improvement limit (3a)l Improvement atomiser fg factor f g factor Carbon rod 10300 - 1450 - Carbon tube (HGA/EA 3) . . . . 118 87 68 21 . . . . . . . . . . Amount of lead/pg Fig. 7. Calibration graphs for trace determination of lead with (1) carbon rod and (2) carbon tube (HGA/EA 3 type) atomisation.(a) Logarithmic transformation; and ( b ) direct signal versus concentra- tion dependence very high (Fig. 5). The lifetime of the atoms in the tube will be shortened by closed windows and a high velocity of the gas streams. However, in the gas-stop regime during atomisation the most sensitive results were obtained. (b) Only the opposite position (in relation to the mono- chromator) is closed. The sensitivity obtained with this system was good [about 90% of the sensitivity in (a) under gas-stop conditions]. The gas flow-rate in the direction of the mono- chromator has no significant influence on the fluorescence intensity up to 5 1 h-1. Higher gas flow-rates decrease the peak-height fluorescence signal because of rapid losses of atoms in the fluorescence zone by convection.(c) Without windows present the sensitivity was poor because atmospheric N2 and O2 diffused into the atomiser, causing severe quenching of the fluorescence. The maximum heating rate (see Table 1) is advantageous and the best analytical results are obtained with the HGA/EA 3 type atomiser. Summarising the investigations, it can be concluded that the following tube atomiser design is the best for the determina- tion of lead: tubes of the HGA/EA 3 type, pyrolytically coated with carbon; two slits of length 4 mm and height 1.5 mm for the exciting laser beam; distance between the centre of the tube and the centre of the slits 2 mm in the direction of the monochromator; both ends of the cuvette closed by quartz windows; gas stop during the atomisation step; and measure- ment of fluorescence radiation along the tube axis. Comparison of Sensitivities of Carbon Rod and Carbon Tube Atomisers The dependences of the atomic fluorescence signal intensities on lead concentration in solutions were measured under comparable excitation conditions (excitation energy 8 pJ) and atomisation parameters.Two types of calibration graph were obtained: a linear dependence of signal intensity on lead concentration [Fig. 7(b)] and a logarithmic dependence [Fig. 7(a)]. Fig. 7(b) shows that a real linear relationship between fluorescence intensity and concentration exists only for small parts of the calibration graph. Therefore, it is better to use the logarithmic function for a wide concentration range.It can be seen from Fig. 7(b) that the sensitivity of the determination of lead in the carbon tube atomiser is higher but the curvature of the calibration graph is greater for tubes than for rods. The reason may be pre-filter effects in the more concentrated plasma or recording problems. For the recording system used, 30 was equivalent to 100 mV and therefore to facilitate comparison of sensitivities, recipro- cal sensitivities relative to 100 mV are given in Table 2. Measurements were carried out at lead concentrations up to 2 pg per 10 yl with the tube atomiser and up to 50 pg per 10 1-11 with the rod atomiser. Although the sensitivity for the tube was better than that for for the rod, the absolute noise level was 3.5 times lower for the rod system.The detection limits, calculated from these values are shown in Table 2. The reasons for the lower noise with the rod atomiser are the plasma dilution and the greater possibility of avoiding stray and continuum light. It can be concluded from these results that the carbon tube atomiser is a better atom reservoir for LAFS than the carbon rod atomiser. The reasons for this conclusion with respect to the carbon tube atomiser are as follows: higher heating rate; more homogeneous temperature in the plasma; higher atomi- sation efficiency; longer lifetime; and higher concentration of atoms in the plasma (smaller plasma volume). Comparison of Reproducibilities of Carbon Rod and Carbon Tube Atomisers To compare the reproducibilities of the two atomisers, 80 measurements on each were carried out, the relative standard deviation (RSD) being determined five times each on sixteen different volumes (lead concentration 200 pg per 10 yl).For the carbon rod the RSDs were 17.2, 14.6,9.8,14.6 and 15.6% (mean 15%) and for the HGA/EA 3 carbon tube the RSDs were 3.1, 3.7, 3.0, 8.7 and 12.6% (mean 6%). In contrast to the results at low concentration (see the previous section), the reproducibility at higher concentration is better for the carbon tube, owing to the more homogeneous temperature of the generated plasma and of the longer lifetime of the excited atoms in this plasma in comparison with the open rod system. Interferences The influence of some metal nitrates and sodium sulphate on the fluorescence intensity of lead was investigated.It was found that the use of nitrate media resulted in negativeJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, JUNE 1986, VOL. 1 241 interferences, probably caused by quenching of the fluor- escence by NO, molecules. It was found that thermally stable nitrates, such as sodium and potassium nitrate, cause a stronger signal depression compared with less stable nitrates, such as aluminium and cadmium nitrate. When using graphite rods, interference from nitrates was greater than that from carbonates. Conclusions It has been shown that the sensitivity of LAFS measurements of lead is higher using a carbon tube atomiser instead of a carbon rod atomiser, owing to the more homogeneous temperature in tubes and the greater atom concentration. In electrothermal atomisation atomic absorption spec- trometry nitrate media are preferred in order to avoid vapour-phase interferences from metal chlorides. However, it was noted in this study that negative interferences were obtained from the nitrate anion and it is believed that these were caused by quenching of the fluorescence by NO, molecules. More details on this and other interferences will be given in a subsequent paper. References 1. Weeks, S. J., Haraguchi, H., and Winefordner, J. D., Anal. Chem., 1978,50, 360. 2. Bolshov, R. A., Zybin, A. V., and Smirenkina, I. I., Spectrochim. Acta, Part B, 1981, 36, 1143. 3. Tilch, J., Paetzold, H.-J., Falk, H., and Schmidt, K. P., “Analytiktreffen 1982, Atomspektroskopie, Neubrandenburg 8-12.11.82, Kurzreferateband,” DV No. 55. 4. Human, H. G. C., Omenetto, N., Cavalli, P., and Rossi, G., Spectrochim. Acta, Part B, 1984,39, 1344. 5. Falk, H., Progr. Anal. At. Spectrosc., 1980,3, 181. 6 . Omenetto, N., and Human, H. G. C., Spectrochim. Acta, Part B, 1984,39, 1333. 7. Dittrich, K., Wennrich, R., and Mothes, W., Chem. Anal. (Warsaw), 1977,22, 1053. Paper J.5112 Received July lst, 1985 Accepted November 30th, 1985
ISSN:0267-9477
DOI:10.1039/JA9860100237
出版商:RSC
年代:1986
数据来源: RSC
|
|