Part I Fundamentals and Instrumentation 6 2 Excitation Sources and Atomising Systems This section contains reports of work concerned with atom formation and/or excit- ation relevant t o atomic emission absorption and fluorescence analysis. 2.1 ARCS AND SPARKS Although several alternative means of sample atomisation and excitation exist the large majority of practical analyses by emission spectroscopy use arc or spark excit- ation. These techniques are of course well-established but remain an area of consider- able research interest. The following pages reflect the diversity of the investigations recently reported and demonstrate how complex are the physics and chemistry of these sources and their interaction with analytical samples.Excitation processes in spark discharges (415) and in the d.c. arc plasma (424) have been studied. French workers (1066) have shown that all elements e.g. halogens N 0 Pb etc. in salts and oxides could be determined by means of a battery of low tension (1000 V) but powerful (1000 J in 200 ps) condensers. This source permitted work at atmospheric pressure in both the visible and ultraviolet spectral range. The conditions governing the arcing of samples in briquetted form using a Zeiss Q-24 prism spectrograph and ABR - 3 source have been examined (85). An inter- mittent a.c. arc combined with rotation of the sample ensured constant intensity of spectral lines provided that impulse duration impulse frequency and current were suitably chosen.The matrix effect with the a.c. arc was buffered by C powder even in the case of calcites fluorites and barytes which have in their matrix the spectro- chemically efficient cations Ca" and Ba* (642). An a.c. arc has been used (775) instead of the more usual spark excitation in a comparative study of intensity ratio and density ratio methods. Muntz (23) used a helium sheathed tubular electrode arc with ultrasonic nebulisation of sample solutions t o examine seven different bases (Ni Cu Al Mg C o Fe Ti) t o type the alloy from one set of standard solutions containing 7 Part I Fundamentals and Instrum en tation 18 elements.Arc excitation of fissile materials was compared with hollow cathode excitation as an alternative t o mass spectrometry (574). The trend in many industries is for fast analysis and spark sources of higher repetition rates have been developed. For various reasons the maximum repetition rate of these sources is about 400 sparks s-l. The source described by Schroeder and Strasheim (625) produced up t o 1000 sparks s-l and some analytical experiments indicated that analysis speed could be increased by a factor of 10-20 with no loss of reproducibility. Spark-off curves precision sensitivity self absorption and matrix effects were discussed in application t o the analysis of wear metals in lubricating oils (2 19 948)." M u l t i s ~ u r c e ~ ~ conditions which have been successful for many years in the steel industry are now provided at the high repetition rates of up t o 400 sparks s-l (420) and tables of discharge characteristics and precision data have been given (553 554 872 9 12). Factors influencing calibration drift were associated with the counter electrode (5 5 3). Modification of a commercial air interrupter high voltage spark source for produc- tion of high current discharges with wave forms ranging from fully oscillatory t o fully unidirectional was discussed (564 964 967). Line intensities in low energy neutral and first ion spectra were enhanced significantly without corresponding increase in background when the electrode was rotated above 1200 r.p.m.t o ensure reproducible spatial geometry. A simple device in which the cathode spot of a d.c. arc operating in an inert gas moved rapidly over the surface of a metal sample producing a fine aerosol was described (422 952). This aerosol can be transported in the inert gas up t o 20 m t o any suitable excitation device such as the stable low powered d.c. capillary arc chamber which has been designed for optical excitation of the aerosol (553 1006). Ignition difficulties which can occur when sparking highly alloyed steels were investi- gated with a special electronic controlled spark generator (624).Removal of the inter- element and metallurgical effects in the spectrographic analysis of high and low alloy steel was claimed f o r the low energy high power density micro spark method (804). The micro-spark method has also been used (1075) with a Cu-needle electrode t o examine the microhomogeneity of ingots prepared in moulds with directed cooling. A 21 amp d.c. plasma arc assembly designed t o exclude air was described for the determination of non-metals mixed with a suitable solvent. The wt. % of H 0 S and N was estimated from the diatomic spectral bands C - H 314.3 nm 0 - H 306.3 nm C - S 257.5 nm and C - N 388.3 nm. Results were given for N and S in crude oil (56 7 5 7).A study was made (626) of the distribution of particles in a d.c. arc of 8.5 A during the vaporisation of very small amounts of elements. The introduction of ammonium compounds into d.c. arc plasma was shown t o influence the apparent sensitivity of trace elements (84). Electrodes with double cavities were used to study the carrier effect of various salts and possible explanations of these effects were given. Morris and Worden (31) determined 24 trace elements in M o by a spectrographic method using a low current arc in Ar. The sample was packed in graphite and a small quantity of CO was produced when the oxide was reduced in the anode. This change in composition of the arc gas was essential for the excitation of trace impurities.The stability of a d.c. arc was improved by using a reduced pressure static argon atmosphere with a Ta sphere attached t o the. counter electrode. This led t o greater sensitivity and the technique was useful in trace analysis especially of Cr in blood hair and urine (257). A d.c. arc plasma burning in air at atmospheric pressure was used Part I Fundamentals and Instrumentation 8 (545) t o investigate the direct determination of trace elements in rare-earth matrices. The trace elements concentrated in the cathode region while the rare earth elements concentrated in the anode region. The cathode region was less dense and trace elements were determined independent of rare earth matrices.The electrodeposition of trace elements on mercury coated electrodes was used (584) t o determine Ag Bi Cd Co Cr Fe Ga Ge In Mo Ni Pb Sn T1 and Zn simultaneously by a high voltage spark discharge. This method of preconcentration avoided some causes of erratic results reduced contamination from reagents and gave satisfactory photographically recorded spectra for all elements except Pb. Pitz (238) measured the absolute intensity of the carbon arc in the region 190-250 nm by comparing it with the intensity of an Hg high pressure lamp and deuterium discharge lamps calibrated by means of synchrotron radiation. Once standardised the carbon arc was very reproducible.The spectrum was a continuum decreasing in intensity t o short wavelengths and with superimposed fine line structure. Spectral radiances of the order of l o 5 erg s-l cm-2 s f 1 nm-l were found. A special apparatus was designed (632) t o study the parameters affecting light intensity fluctua- tions in emission spectrometry. Conclusions were drawn about the use of hemi- spherical electrodes argon atmospheres and changes in polarity. The changes in the carbon arc under the influence of a stationary homogeneous magnetic field were studied by Leushacke and Nickel (365 366 55 1). When low fields of 1 0 G were applied the arc became mechanically unstable and started t o rotate twisting t o a helix with higher inductions.The increase in spectral line intensities was studied and was due basically t o an intensification of the cathode layer effect. The effects of added elements having low ionisation potential in a d.c. arc plasma burning in an inhomogeneous magnetic field were investigated by Georgijevic et a1 (550 552). The influence o n the electron density of the plasma was considerable with changes affecting behaviour and the ratio of spectral line intensity t o background. The addition of small concentrations could be advantageous for the determination of other elements with low ionisation potentials. The influence of lithium carbonate o n temper- ature distribution in an arc has also been investigated (629). To photograph the plasma produced by spark discharges Strasheim and Blum (623) used a rotating mirror ultra high speed camera with a minimum framing interval of 0.25 p s .The spectral radiation coming from different zones of the spark gap was recorded by means of a time resolving technique and the formation of cathode spots on different structures of sample surface was observed (623). Results for the radial temperature distribution in the free burning arc according t o Krinberg theory were compared with experimental determinations in Ar N and air (627 628). It was found that the convection transport of heat was not sufficient explanation for the radial temperature distribution found in the outer zones of the arc.The experimental and theoretical radial temperature distribution curves for arcs burn- ing in N Ar C 0 2 with added water vapour Li and I vapour were discussed (79 80 81 82 83 670). Dieudonne and Bril (1057) also studied the characteristics of atomic excitation in arcs in rare gas atmospheres. Energy levels higher than in air were reached favouring the appearance of high excitation potential lines. The photographic measurement of arc radii was described using an adjustable velocity? explosively driven double probe (697). The influence of pressures of up t o 40 atmospheres on a d.c. arc was reported and showed different characteristics for varying electrode materials (568). The background intensity was found t o be at least as good a reference as an internal 9 Part I Fundamentals and Instrumentation standard particularly because it was applicable for quite a wide spectral range in an examination of a gas stabilised 15 amp d.c.arc (566). Marinkovic and Vickers (33) described a stabilized d.c. arc for the production of atoms for AAS which should have great potential in emission spectroscopy. The arc was stable with a temperature similar t o that obtained in N2 0 - C2 H2 flames. Detec- tion limits for Al B Mg V and W were broadly comparable with those obtained using flames. A 3.4 m Jarrell-Ash Ebert spectrograph was used t o investigate high temperature reactions in three component systems (605).These reactions were interdependently connected with the evaporation of the elements and in the case of cathodic evapor- ation the burn-off of the material depended strongly on the atmosphere during burn- ing (604). Two papers discussed different techniques of measuring transport parameters in d.c. arc plasmas (608 962). The action of spectral carriers and their role in d.c. arcs was discussed and the influence of fluoride chloride and iodide carriers on the excitation of trace impurities in alumina was studied (76 77 78). Decker (69) showed that electrode temperatures in a d.c. arc were dependent on electrode dimensions and o n the electrical power in the arc system.With a fixed arc gap the cathode temperature was directly proportional t o the power and the log of the anode temperature was directly proportional t o the log of the power. As the electrode temperature appeared t o be independent of both the atmosphere used and the nature of the sample it should be possible t o vary the electrode temperature t o suit the type of sample being analysed. Curry and Cooley described (374) a compact gas jet operated at 12 amps d.c. with a reduction in background emission and a shorter burn time when compared with earlier designs by Helz or Margoshes. For the direct spectrographic determination of highly volatile fluorides such as B and Si present as impurities in UF4 it was necessary (603) t o suppress the U excitation during the d.c.arc exposure and t o improve sensitivity. With samples mixed with alumina the limit of detection did not meet specifications but using zinc oxide at a concentration of 35% B and Si spectra were obtained free from most interference from U lines enabling their determination with good sensitivity and reproducibility. A specially shaped electrode was used (564) t o collect the material evaporated from a given microsector of mineral by a laser beam. The electrode was then subject t o a separate excitation in certain cases a d.c. arc in a magnetic field or in an impulse discharge. When applying this laser microanalysis technique t o the detection of impurities in galena there was an increase above the expected intensity of a great number of spectral lines.The intensification factor for CuI 323.8 nm was 4.65 with the background unchanged and thus improved the signal t o noise ratio. Investigations were conducted into the nature of selective electrical erosion of different alloys in spark and arc sources used in spectrochemical analysis. In a d.c. arc the erosion corresponded directly t o the melting points in Ti - Ni Mg - Al and Mg - Zn binary alloys and volatilisation was mainly a thermal process (71 0). The decisive role of thermal conductivity o n the erosion mechanism was shown using an a.c. arc with Cu Ni Cu - Mn Cu - Zn and Cd - Zn alloys (702).The dependence - of electrode erosion on electrode composition in an arc discharge was much stronger for the anode than the cathode and it was indicated that the transport rate of elec- trode material into the discharge caused part of a third element effect described (703). Erosion by spark discharges was shown t o have both thermal and electromechanical Part I Fundamentals and Instrumen tation 10 properties and hard brittle materials with high melting points underwent stronger spark erosion than malleable materials with low melting points (709 710). A linear relationship was shown between t h e spectral intensities and t h e amount of erosion loss of Cu Mn Si and Ti by spark discharge from various Fe alloys and a n A1 alloy (339 3 44).2.2 PLASMAS The characteristics and applications of plasmas generated in an inert gas stream have been studied b y several groups. Of t h e more basic work t h e apparatus used b y Mullen et a1 (753) t o measure high temperature plasma reactions was interesting. In this t h e R F induction plasma produced by means of a 100 kW 1 MHz R F generator was passed through a reaction channel t o react with supersonically injected target gas molecules. The reaction products were then monitored b y quadrupole mass spcctro- metry. Kleinmann and Cajko (75) investigated spectrophysical characteristics (spatial distribution of radiation shape of spectral lines temperature of particles electron density continuum intensity) of a high frequency (50 MHz) low power (270 W ) discharge in Ar.The discharge was shown not t o b e in thermodynamic equilibrium. Isagawa and Niki (350) studied t h e distribution of spectral line intensity and continuum in t h e plasma j e t along t h e flame axis measuring t h e Fe 371.99 nm Mg 285.21 n m and Mg 280.27 nm lines. They concluded that line t o continuum ratio was maximised b y cooling the plasma b y means of a "cooling device" rather than by allowing it t o cool by spouting itself i n t o t h e atmosphere. Barnes and Schleicher (610) reported t h e beginning of a study aimed at developing a description of t h e physical and chemical interactions in an induction plasma discharge and their influence o n spectrochemical processes.Two other studies of plasma temperatures have been noted (705 740). Measuring t h e temperature i n a 55A DC plasma jet b y t h e two-line method (Zn 307.2 nm and Zn 307.6 nm) it was found (705) that at and above 1 6 mm the temperature was n o t influenced b y t h e effective ionisation potential of t h e samples. Comparisons have been made (740) between temperature measurements of argon induction plasmas and calculations based o n t h e Elenbaas-Heller equation. Agreement was good for the central portion of t h e plasma column while discrepancies nearer t h e walls were attributed t o over-estimation of gas temperatures when Saha equilibrium was assumed.A patent has been granted (242) covering a high frequency discharge plasma torch generator with a single A1 electrode. The electrode is watercooled and use of A1 at t h e t i p of t h e electrode was claimed t o preclude erosion even a t high temperatures. Analytical applications of plasmas have increased and commercial instruments using these as t h e excitation source are now available (see Section 6.2). Greenfield (774) has recently reviewed his considerable experience in this field. Plasma jets using graphite electrodes microwave torches capacitively- and inductively-coupled plasma torches and electrodeless inductively-coupled torches were described and t h e advantages of the latter type stressed.Power generators incorporating Hartley-type oscillators were stated t o remove mismatching effects caused b y introduction of aerosols into plasmas. Ultrasonic nebulisation was advocated for introduction of liquids and fluidised-bed devices f o r solids. This technique has been applied (1 1 1 ) t o t h e analysis of powders using a modified Radyne SC15 plasma generator. Fassel and co-workers have made significant contributions in particular by demon- strating t h e selectivity and sensitivity of this technique when used with suitable optical equipment. Recent reviews (224 9 3 1 ) reveal that they favour use of a n induction- 11 Part I Fundamentals and Instrumentation coupled plasma operated at 40-50 MHz a n d 2 kW making analytical measurements in t h e tail flame (T-3000" K) rather t h a n in t h e toroidal plasma (T-900O0K).While.it has been claimed (846 8 9 1 ) t h a t a n Ar plasma has been developed t o t h e stage a t which it is available as a routine analytical tool t h e experience of some workers has been less favourable. A 2 kW 6 0 MHz induction-coupled plasma for example showed rather irreproducible behaviour (454) which made operation difficult. Robin ( 4 4 4 ) reported results obtained with a STEL (France) 4-6 MHz generator which confirmed t h a t while plasmas have advantages in emission spectro- scopy t h e y are inferior to flames when used in t h e absorption mode chiefly because t h e high background emission tends t o saturate t h e detector.Demonstrations of t h e applicability of plasmas t o practical analyses have been provided b y Vecsernyes and Zaray (327) Heemstra (56) Gebhardt and Horn (609) and Kleinmann (1032). The last a u t h o r noted t h e severe matrix effects of elements with low ionisation energies in a high frequency low voltage plasma. Vecsernyes (61 1 ) has also discussed t h e analytical significance of nine operating parameters in t h e design and operation of a laboratory-built plasma. He commented particularly o n t h e experimental difficulties associated with direct analysis of powder samples.Finally t w o examples of t h e use of low-power microwave-generated plasmas should be noted. Fallgatter e t a1 ( 3 8 ) used a 1 0 0 W 2450 MHz generator t o produce a plasma with temperature -5000" K. The system however was n o t in thermodynamic equilibrium and also suffered from memory effects. Aldous et a1 (829) used a similar system equipped with an EM1 9 6 0 1 B photomultiplier and simple amplifier-detector system t o detect t h e transient emission signals produced from -0.1 pl of sample introduced b y evaporation from a Pt o r W loop heated b y electrical resistance. Both of these investigations were restricted t o volatile elements in aqueous solutions.It has recently been announced t h a t ARL are to produce a n emission spectrometer based upon a low-power microwave plasma which it is claimed will provide empirical formulae of eluent fractions from a gas chromatographic separation of organic mixtures. I t will cover C H N 0 P S and halogens. 2 . 3 GLOW DISCHARGE LAMPS Although t h e most important application of glow discharge lamps in atomic spectro- scopy is as t h e light source i n atomic absorption o r fluorescence t h e y can be used as t h e excitation source f o r analytical emission spectroscopy. Their advantages in this mode include high signal-to-noise ratio relatively few inter-element interferences and t h e requirement of only a few mg of sample.The development of this application has been largely attributable t o Grimm". During t h e past year h e has described their application (421 595) particularly to t h e determination of t h e gases N H and 0 in metals such as Ni alloys. The sample was placed in a cavity diameter 2 0 mm and formed t h e cathode. The lamp contained 3 t o r r of Ar ran a t 0.5 to 1 kV and 1 5 0 mA and a 60 second exposure was used. T h o r n t o n ( 4 2 3 ) described t h e determination of trace metals i n Ni alloys placing 0.01 g of sample i n t h e hollow-cathode in a n He atmosphere. Further practical applications of t h e glow lamp have included t h e analysis of special steels ( 5 5 5 ) using a technique i n which t h e line and background emission intensities were measured alternately b y t h e .same photomultiplier b y means of an oscillating * W.Grimm Naturw. 1967,22 586. W. Grimm Spectrochim. Acta 1968,23B 443. Part I Fundamentals and Instrumentation 12 quartz disc; determination of non-metallic Contaminants 0 S P and halogens (8 14) and the determination of U235 in natural U (352). In the last analytical measurement was made at 424.4124 nm using the 424.4372 nm U238 line as standard - the Jarrell- Ash 3.4 m Ebert-type grating spectrograph having the dispersion necessary t o separate these lines completely.The U sample t o be analysed was deposited into the hollow- cathode cavity electrolytically and the discharge was carried by a flowing Ar stream. Other descriptions of the design and application of such lamps for emission analysis have been given by Johnson and Gram (865 866 949) Dogan et a1 (681) Ropert (596 1072) Boumans (592) Jager and Butler (216 594) Moal and Brossier (597) and Dieudonne and Bril (593) - in most instances using glow discharge sources manu- factured by R.S.V. GmbH. Three further papers are noteworthy. In the first (598) design parameters were optimised and the stability of the discharge was shown t o be critically dependent upon the distance between the electrodes.The use of this lamp f o r determination of P and also for the continuous analysis of solutions was con- sidered. A simple demountable water-cooled graphite hollow-cathode system for solution analysis was described by Prakash and Harrison (1 06). Finally a twin hollow- cathode system was described (599) which permitted excitation of sample and standard material under identical conditions. Hollow cathode discharges have also been used as sputtering devices for the non- flame atomic absorption analysis of alloys (54 169 7 11). Atomic fluorescence gave (169) limits of detection for Ni Cr and Cu in Fe base alloys of the order of 10 pg/g. Atomic absorption investigations showed that precision was poorer than that obtain- able with a flame (5- 15%) (54) and that analysis time was long - varying from 15-20 min (71 1) t o 3-4 min (54).From the summary above it will be seen that this is an area in which there is considerable activity. A more complete understanding of the cathodic sputtering processes is being built-up and the stability of hollow-cathode discharges - an internal standard need not be used - has been pointed out. It will be interesting t o note whether present research efforts result in the adoption of this technique for routine emission analysis in the near future. 2.4 LASERS Laser microprobe analysis is now an established technique in emission spectroscopy and various aspects of its application have been described (560).Webb and Webb (314) for example used a Jarrell-Ash Q-switched ruby laser with a Hilger medium quartz spectrograph for semi-quantitative analysis. Microprobe analysis has been applied t o studying minerals (564) and the welding seams of steels (563). It has been shown (62) that staining of a metal sample surface during analysis was prevented by prior application of a thin coating of collodion. The laser microprobe has been used (706) for atomic absorption analysis. Here the samples were held at 45O t o the laser beam and the optical axis of a spectrograph. The integrated absorption signal from high alloy steel samples was measured.Two studies of laser microprobe analysis in atmospheres other than air have appeared. With a Carl-Zeiss Jena LMA-1 laser Petrakiev and Dimitrov (5 62) concluded that in N2 O2 and Ar atmospheres spectral line intensities and signal-to-background ratios increased by up t o 1 5-fold resulting in significant improvements in reproduc- ibility and sensitivity. Treytl et a1 (406) however investigated the use of N 2 0 2 Ar Ne and vacuum and concluded that special atmospheres were of no benefit in general work. Line-to-background ratio showed no systematic changes with change of energy 13 Part I Fundamentals and Instrumentation between 1.2 3.6 and 8 mJ but was best in vacuum a t 1.2 mJ.This group also reported (255) that spectral emission of metallic elements viz.Ag and Mg was suppressed in organic matrices such as albumin and serum owing t o reduced sample vaporisation. Treytl et a1 have also (40) described a method of maximising the signal-to- background ratio from a Q-spoiled laser by photoelectric time differentiation of the emission spectra. With a similar aim Scott and Strasheim (559 971) used a high- repetition rate Nd laser with a time-resolved direct-reading detection system and achieved reproducibility which permitted use of the technique for quantitative analysis of metal samples.More fundamental studies have been made of the interaction of a laser beam with a sample. Tlalka (561 ) reviewed various theories of the mechanism of this interaction. Allen (741) probed a laser-produced plasma with an electron beam and obtained the target surface temperature as a function of time and the mass vaporisation rate. Finally Piepmeier et a1 (870 968 1008) have also studied the processes occurring when a Qswitched Nd laser atomised a metallic Cu sample. At atmospheric pressure the crater diameter (90 p m ) and sample consumption (35 ng) were relatively constant with variation in power from 10 t o 100 mJ. At 1 torr however both increased with increasing power and spectral changes also occurred.2.5 FLAMES 2.5.1 Flame Types The dominant position of premixed air/C2 H2 flames for atomic absorption analysis remains unaltered. These flames have gained acceptance as sources for atomic emission and on shielding with an inert gas have proved attractive for fluorescence applications. The current state of the art was clearly defined by Koirtyohann (447) in his plenary lecture in Paris. He pointed out that the flame/burner/nebuliser system is still the most troublesome item of today's atomic absorption instruments. We still use inefficient pneumatic nebulisers which require large sample volumes; high precision seems t o take place t o sensitivity in most instrument designs; working curves inexplic- ably change during a series of analyses; we have no coherent theory of chemical interference for many elements; a non-quenching interference-free flame is still required for atomic fluorescence; solid sample analysis is not part of current practice; and there are still many elements which are determined only poorly by atomic absorption - boron being an example.The premixed 0 2 / H 2 flame has received little attention as an analytical source. A recent study (453 926) showed that despite a high maximum temperature (2950°K) detection limits were worse by factors of 4 t o 10 than those obtained in NzO/C2H2 flames. Solute volatilisation interferences were serious particularly in fuel-rich flames and equilibrium calculations indicated that vapour-phase metal monoxide formation was extensive.The premixed N20/C2,H2 flame has now been generally accepted as useful for atomic emission*. A list of detection limits has been published (266 1010) for 68 elements and the use of gas chromatography integrators for evaluation of spectral scans has been reported (1042). The use of the N2 shielded N 2 0 / C 2 H 2 flame supported on a slot-burner gave u p t o 3 times improvement in detection limit over the unshielded flame (1042) for A1 396.2 Re 346.1 and Re 448.9 nm. * E. E. Pickett and S. R. Koirtyohann Spectvochirn Acta 1968,23B 235. Part I Fundamentals and Instrumen tation 14 Methyl acetylene appears t o be a less satisfactory alternative t o acetylene for use in premixed flames (664).Flames based on the use of liquid fuels (e.g. hexane acetone) and air (100 260 490) may have some application in the design of portable instruments for field use. Bailey and Rankin introduced fuel into a conventional burner chamber by means of a second nebuliser. Performance was similar t o that using gaseous alkane fuels. 2.5.2 Burners and Nebulisers The important parameters in burner design which affect flame flashback have been reviewed by a number of authors (101 618 849 902 1009). Using these criteria it is possible t o design burners which will not flash back under any conditions of gas flow.Such burners would have plate separations of 0.3 mm for N 2 0 / C 2 H 2 and 0.5 mm for air/C2H2 if a conventional slot design were adopted but would be of limited analy- tical use. Capillary burners have been designed (618) which are safe usable and have convenient configurations. The minimum total gas flow required t o maintain a stoichi- ometric flame on several commercial burners has been calculated (61 8 849 902 1009). One important observation was that the measured burning velocity of the stoichiometric N2 O/Cz H2 flame (61 8 ) + + C2 Kz 3 N 2 0 = 2CO H 2 0 3Nz C / O = 0.667 was found t o be 285 cm/sec as against the only other value of 160 cm/sec*.An improved N2 0 / C 2 H2 burner (304 820) of gold-plated copper construction featured water cooling and a slot width of 0.64 mm tapered t o 0.48 mm at the base. The burner warmed u p rapidly operating at a head temperature of 60°C. Increases in sensitivity of 2.0 t o 2.6 times were claimed over the Techtron AB 50 burner which it superficially resembled. Burners with independent sample aerosol introduction have been designed (478) for use with ultrasonic nebulisers. Separation of the nebulisation and combustion gas supplies resulted in constant nebuliser performance irrespective of flame variations. Ultrasonic nebulisers have not found widespread use in atomic spectroscopy of their poor stability.Owen (881) has experimented with various designs and reported (479) a direct impi\ngement device which gave reproducible performance comparable with that of a pneumatic nebuliser. The sample solution was pumped at 0.25 t o 3 ml/min across the vibrating face of a piezo-electric disc protected by a glass coating. Vibration at 1.4 MHz produced a fine aerosol with a uniform drop size and efficiencies of up to 50%. An improved spray chamber design by Nicolas (19) is now commercially available on Techtron instruments. The chamber has a constriction about one-third of the distance from the inlet. Fuel gas is introduced at the nebuliser end of the chamber at a low velocity point.The constriction aids mixing of fuel and oxidant gases removes large droplets and successfully damps out shock waves which are propagated in conventional spray chambers. The efficiency of nebuliser chambers can be increased by the use of infra-red heating as in some types of burner available from Beckman Instruments. Some recent experiences with heated chambers have been reported by Uny et a1 (70 132) who * W. G. Parker and H. G. Wolfhard Proc. 4 t h Int. Symp. Combustion 420 ( 1 9 5 2 ) andProc. 5th Int. Symp. Combustion 7 1 8 (1954). 15 Part I Fundamentals and Instrumentation claim a n increase i n sensitivity of 1 0 t o 30 times depending o n t h e element and b y Rawson (465) who has constructed a stainless steel unit having similar performance.While correction for background i n flame emission has been compensated b y repeti- tive optical scanning (641) a n older alternative approach is t h a t of modulated intro- duction of t h e sample. I n t h e past this has involved t h e use of periodic sample injection via a rotating t a p at a b o u t 50 Hz. Difficulties were usually encountered as t h e modulation technique also caused flame background fluctuations. In t h e most recent reports (35 639) t h e sample introduction was only partially modulated. A latex t u b e leading t o a t o t a l consumption burner which supported a premixed nitrous oxide/ acetylene flame was sandwiched between t h e plates of a piezoelectric bimorph.Modulation was produced b y compression of t h e tubing wall. A dual channel device was also described where sample and blank were aspirated i n t o t h e same burner through separate tubes and modulation was 1 80° out-of-phase thus maintaining a constant liquid flow-rat e. The noise spectra of premixed N2 0 / C 2 H2 flames supported o n a sprayer burner* were measured (638) over t h e frequency range lov3 t o 0.5 Hz. Below 0.1 Hz t h e noise was mainly inversely proportional t o t h e frequency (l/f noise). Between 0.1 Hz and 0.5 Hz "white" noise predominated. Noise arising from t h e nebuliser ( b o t h l/f and white) made a significant contribution t o t h e t o t a l noise of t h e analytical signal when a sample solution was aspirated i n t o t h e flame.The noise arising from Swan band emission ( 7 7 1 ) in a n air/C2H2 flame was shown t o cause interference o n several F e and Cr lines in AAS using a commercially available instrument. 2.5.3 Discrete Sampling Devices Several discrete sampling techniques f o r use with flames are currently available. Usually t h e sample solution is measured i n t o a boat o r coil and after evaporation to dryness inserted i n t o t h e flame. A Ta sampling boat similar t o t h a t developed b y Kahn et a l t has been used f o r atomic fluorescence (482) and improved by t h e use of electrical pre-heating (8 92).Pre-evaporation is quicker if prior extraction into an organic solvent is carried o u t . T h e Delves sampling cup$ is now available from Perkin Elmer and has been described recently in their literature (92 292). A boat of pyrolytic graphite was used (492) for t h e introduction of A1 solutions ( u p t o 1Opl) i n t o a N2 0 / C 2 H2 flame. The evaporation time was 3-5 sec although Cd was evaporated in less than 1 sec in an air/C2H2 flame. Several easily vaporised elements (e.g. Cd Zn Pb and Ag) may be determined b y placing a d r o p of solution o n a Pt loop which is t h e n inserted i n t o t h e f l a m e s . A similar device has been developed (685) for use with an air/C2H2 flame f o r use i n t h e determination of Zn in A1 and Cu alloys b y atomic absorption.Harris and Newton (1 0 4 5 ) have injected dithizone extracts containing Cu directly into t h e capillary of a conventional pneumatic nebuliser. As little as 30 pg Cu could b e detected i n 1 0 0 1.11 a n d it is possible t h a t under these conditions t h e nebuliser could have a higher efficiency t h a n with continuous operation. Best results were obtained with a n oscilloscope readout. The direct analysis of solid samples has commanded little attention f r o m flame spectroscopists until recently. Introduction of a powdered sample i n t o t h e unburned * V.G. Mossotti and M. Duggan Appl. Optics 1968 7 1325. t H. L. Kahn G. E. Peterson and J. E. Schallis At. Absorp. Newsl. 1968 7 35. $ H. T. Delves Analyst 1970 95 431. 0 G. M. Katz Anal. Biochem. 1968,26 381. Part I Fundamentals and Instrumen tation 16 gas mixture of a flame has been described (533) for Pd on a number of catalyst supports. Aspiration of a suspension of stannous oxide stannic oxide or stannic sul- phide has been used (261) for the determination of Sn. In both instances the atomic absorption response was dependent on the nature of the sample. Direct analysis of electrically conducting samples has been achieved (1007) by an apparatus in which the sample forms the cathode of a low current d.c.arc. The cathode spot moves t o give uniform sampling and the aerosol produced is carried in a stream of Ar t o a N2O/C2H2 flame. Coefficients of variation of 2.6% have been obtained for Cr Mn and Ni in steels by atomic emission. Other devices which could possibly be of use for solid sample nebulisation are described in the section dealing with non-flame atomisation. 2.5.4 Theoretical Studies An accurate measurement of flame temperature is essential in understanding the spectroscopic equilibrium and kinetic processes involving species contained in the flame. Reif et a1 (450 61 4 932) have emphasised the point that spectroscopic tenfper- ature measurements on conventional analytical flames may be in error by several hundreds of degrees from the average o r weighted average temperature.The uncertainty arises from the non-isothermal nature of all flames used for analysis and Abel inversion techniques must be used t o convert experimental lateral temperature profiles t o true radial profiles. Illustrative calculations were based on temperature and concentration profiles generated by equations similar t o those used by Sasaki". Two of the most easily applied spectroscopic temperature measurement techniques the two-line method and the slope method depend o n accurate transition probabilities for Fe lines and a proper choice of lines. Reif et a1 (451 615) determined relative transition probabilities for 4 3 Fe lines relative t o the Fe 373.487 nm line.The work was carried out in an isothermal region of a premixed NzO/CzHz flame using the reversal temperature of the Fe 371.994 nm line t o determine the flame temperature with an experimental error of +1O0K. Durie and Smith (172) found that chemilumine- scence increased the measured Na D line reversal temperatures of premixed propane/ O2 /N2 flames when small amounts of S O 2 NO or Ar were introduced. + + + ("S) Na H SO2 -+ (2P) Na HSOz + + (2 S ) Na H 4- NO + ( 2 P) Na HNO Flame temperature measurements based on an atomic fluorescence technique have been proposed (183 485).Direct-line fluorescence of T1 at 535.0 nm is excited using a continuum source at the resonance wavelength of 377.6 nm. Resonance fluorescence at 377.6 nm is then excited by 535.0 nm radiation from the thermally populated metastable P!42 level. The effects of flame temperature on the atomic fluorescence of Pb at the direct-line 405.8 nm and the resonance line of 283.3 nm have received comment from Bailey (1 12) and thermally-assisted fluorescence of In and Ga has been discussed by Omenetto and Rossi (25). Calculations of the degree of thermal ionisation of metal additives in flames assum- ing the Saha equilibrium have been tabulated for 48 elements at three flame temper- atures ( 3 78) using the assumptions of earlier authors?.Although the calculation was * Y. Sasaki Jap. J. Appl. Phys. 1966,S 439. t D. @. Manning and L. Capacho-Delgado Anal; Chim A d a 1966,36 312. 17 Part I Fundamentals and Instrumentation admittedly approximate it should serve as a useful guide t o practising analysts. Signi- ficant deviations have however been observed (61 7 939) between calculated and experimentally determined degrees of ionisation. The differences are known from earlier work t o arise from either slow recombination of ions or chemi-ionisation. Ionic absorption measurements (619) have been suggested as a possible but unlikely complement t o atomic absorption for analysis.The measurement of the free a t o m fraction (p) of metal additives in flames continues t o undergo developments in experimental technique. Ag has been shown t o have a /3 = 1 .O in H2 / 0 2 /Ar flames and can therefore be used as a reference element for a relative absorption technique (677 907) using a continuum source and iso- thermal flame. The relative measurement approach obviates the necessity for determin- ing nebuliser efficiency and spectral response of the detection system. Accurate values for oscillator strengths were required in common with many other methods for deter- mining 0. Willis (72) using an absolute technique noted that the value of p approached 1 .O for Na Cu Ag and Au in a premixed air/C2 H2 flame.Errors by earlier workers were shown t o result from neglect of hyperfine structure. A method for measuring p factors which was not dependent on a knowledge of oscillator strengths (307) used relative measurements of atomic and metal monoxide emission spectra in a series of flames having the same temperature but different fuel-to-oxidant ratios. Nitrogen was used t o reduce the temperature of the hotter flames t o that of the more fuel-rich flames studied. The method required that the metal and its monoxide be the only species present. It was applied t o the rare-earth elements. A detailed investigation (665 754) of the behaviour of Cr in H2/02/N2 flames showed that comparable concentrations of Cry CrO C r 0 2 and HCr03 were present In the gas phase in all flames studied (1 800-2700" K).Calculations of free atom fractions based on tabulated data have been used t o explain equilibrium atom-formation processes in N2 O / C 2 Hz flames ( 3 12,903). Com- puterised calculations ( 3 12) have enabled quantitative predictions t o be made on the suitability of unusual flames for atomic spectroscopy e.g. 02/C2H2 02/C2N2 N 2 0 / H 2 NO/C2H2 N20/C3H8. Although some of the tabulated data for metal compounds is of limited accuracy this approach can only serve to eliminate much of the speculation associated with this area of flame chemistry.While these calculations showed that the degree of metal atom formation could be adequately described by an equilibrium model it is difficult t o appreciate the contrary views of other workers (754) who have adopted a much less rigorous approach. Evidence for non-equilibrium processes in low pressure flames has been presented (448) and both chemiluminescence and chemiatomisation phenomena observed. Excitation of metal atoms in hydrogen flames occurs (390) in the recombination of H' and OH' radicals. In hydrocarbon flames alternative mechanisms can give excitation t o energies of up t o 9.0 eV. The role of particulate chromium oxide in radical recombination reactions has been studied (665 754) in H2 /02 /N2 flames.The Doppler half-width (AX*) of an atom in a flame is easily calculated if the flame temperature is known. The Lorentz (collisional) half-width (AXL ) cannot be calculated accurately because of uncertainties in collisional cross-section and molecular concentrations of colliding species. A convenient means of determining ALL is via the a-parameter a = (ln2)1/2(AXL/AX,) and this can be obtained from experimental curves of growth*. The growth curves may be based on emission or absorption measurements. * C. van Trigt T. Hollander and C. T. J. Alkemade J. Quant. Spectry. Rad. Transfer 1965 5 813. Part I Fundamentals and Instrumentation 18 A method based on atomic fluorescence measurements by Winefordner et a1 (678) using H2 / 0 2 /Ar flames gave results which were systematically greater than those measured by emission.The advantages of the fluorescence method were discussed. Experimental growth curves in atomic fluorescence using continuum or line source excitation were in good agreement with theoretical predictions (7 1 ) for Mg 285.2 nm in an isothermal flame of well-defined geometry. Practical analytical curves cannot be determined under such favourable conditions. Svoboda and Winefordner (48 9) have discussed departures from ideality and their effects on the shapes of atomic fluore- scence curves.Van Gelder (88) has made a similar study of atomic absorption curves and cites self-absorption in the light source imperfect light collimation through the flame hyperfine structure and especially measurement of non-resonance radiation by the detector as the main causes of premature curvature. Atomic line profiles and related topics were the subject of an excellent review by de Galan (445). Measure- ments of the atomic absorption coefficient of the sodium D1 line have been made (254) using magnetic scanning of a vapour cell at 100 t o 180°C. Atomic line profiles from light sources and flames have been measured experimentally (637) using a piezo- electric Fabry-Perot interferometer. The self-absorption caused by cooler outer regions of a flame used for atomic emission will affect the measured line profile.This has been studied in detail (616) using a two flame atomic absorption technique. One flame acting as the light source represented the inner isothermal portion of the theoretical flame while the second flame represented the cooler outer regions of the theoretical model where self-absorption occurs. The quantitative theory of flame equilibrium processes is now quite well developed and fair agreement with experiment can be obtained. In contrast the study of chemical interferences is predominantly a kinetic problem as well as a problem of identifying the involatile species involved. Classical approaches involving measurement of degrees of interference at varying burner heights with differing excesses of inter- ferent and with varying flame conditions continue t o be made for aqueous solutions (65 441 458 459 460 1053 1061 1064) and for oils (1021).A more direct approach t o the problem of chemical interference is that adopted by Hieftje (461 925 990) who has developed the single droplet technique introduced by himself and Malmstadt". Using this method droplets of a uniform diameter are introduced singly into a tall well-defined flame and the rates of desolvation volatilisation and dissoci- ation measured as a function of distance travelled. The determination of aerosol size distribution using spectral attenuation measurements and based on Mie scattering theory has been reported (1017).Page and Newman (640) used a Langmuir probe t o measure the volatilisation and diffusion of aerosol droplets of a 1 M KC1 solution. Each droplet evaporated t o give a small 'filament' of vapour about 3 p m in diameter separated from its neighbours by a mean of 500 y m or 170 diameters. 2.6 NON-FLAME CELLS 1971 saw a profusion of publications describing the use of non-flame atom cells for atomic absorption and fluorescence spectroscopy. The high absolute sensitivity obtain- able permits the analysis of very small samples and the low emissive backgrounds assist performance in atomic fluorescence. Furthermore the devices are usually small and may be totally enclosed t o allow analysis of radioactive samples or t o extend the wavelength range t o the vacuum ultra-violet.Kirkbright (436) published an excellent review with 65 references of non-flame devices and their historical development. * G. M. Hieftje and H. V. Malmstadt Anal. Chem 1968,40 1860. 19 Part I Fundamentals and I n s t r u m e n t a t i o n 2.6.1 Tube Furnaces and Filaments A number of instrument companies have made designs commercially available. The Perkin-Elmer HGA-70 is based o n t h e Massman t u b e furnace and features a program- mable power supply. It has been described i n several application reports (1 5 166 1 7 1 189 384 397 512 522 8 4 4 918).Varian-Techtron market a carbon rod device based o n t h e West carbon filament and also their so-called "mini-Massman" which is interchangeable with t h e carbon rod. Both are used with a programmable power supply and have been described with applications in a number of reports (1 13 1 14 116 175 209 2 5 9 412 4 7 1 650 8 4 1 851 8 7 6 8 8 6 945 1043 1044). Instru- mentation Laboratory Inc. have developed a n electrically heated Ta strip with a V- shaped indentation to receive t h e sample. The strip is operated in a n Ar atmosphere and has been described in detail (271 480 570 845 8 4 8 8 7 7 935 984). Most of t h e reports o n these commercially available devices were somewhat repetitive and journal publications usually covered t h e symposium material also.Other instrument com- panies known t o b e actively evaluating non-flame a t o m cells include Jarrell-Ash Rank Precision Industries Shandon Southern Instruments (469) and Carl Zeiss (571 ). In addition several workers have reported t h e use of either unspecified o r home-built devices based o n accepted designs of t u b e (Massman) furnaces (218 51 5 5 3 4 ) o r o f filament atomisers (393 41 7 470). A t u b e furnace f o r continuous operation (2 1 1 ) employed a n ultrasonic nebuliser t o introduce liquid samples. Solid samples were introduced using graphite cups (145 2 1 1). Simultaneous correction for molecular absorption and light scattering inter- ference was made.Direct measurements of atmospheric pollutants have been made using a heated cylindrical absorption tube packed with carbon rods through which t h e air was drawn continuously*. Lifetime of t h e rods might be expected t o be limited because of oxidation by t h e flowing air (834 9 8 9 ) . An inductively heated furnace incorporating a 6 kW Radyne power unit gave a maximum temperature of 21 70" K ( 4 8 ) ; a n improved version gave a maximum temper- ature of 2670°K (148). Soviet scientists (484 495 572 715 722 7 2 6 ) have developed a graphite cell which could be used f o r t h e determination of solid samples (about 2 0 mg mixed with an equal weight o f graphite powder).The crucible was clamped between graphite rods in a water cooled holder and heated b y a current < 4 0 0 A supplied from t h e secon- dary winding of a step-down transformer. A pulse of 5 sec duration gave a heating rate of 1800"K/sec and a final temperature of 2570°K. The device was used for b o t h atomic absorption and fluorescence analysis. A graphite cell f o r t h e measurement of vapour pressures and heats of vaporisation has been constructed b y Nickolaev e t a1 (525) and t h e use of King furnaces and heat pipe ovens described by Tomkins ( 4 7 3 ) t . A King furnace enclosed in a glove box has been used ( 5 3 9 ) f o r t h e atomic absorption determination of T e and other radioactive elements.Relevant t o t h e use of non-flame sampling was a paper b y Antonetti et a1 (5 1 8 ) who discussed t h e micromanipulative techniques necessary when volumes of 1-11 were t o be analysed. An automated system was constructed ( 9 0 1 ) for AAS and AFS where the instru- ment generated a d r o p of reproducible size placed it o n a Pt loop and vaporised t h e * H. P. Loftin et al Spectry. Lett 1970,3 161. t c.f. F. S. Tompkins and B. Ercoli App. Optics 1967 6 1299. Part I Fundamentals and Instrumentation 20 sample by passing a 6.3 V a.c. The measurement device was gated through the Pt loop and the total integrated signal converted t o digital output.About 150 analyses per hour could be achieved. 2.6.2 Cells for Mercury Determination Although flameless atomic absorption was used as long ago as 1939 by Woodson* who determined Hg in air the widely accepted cold-vapour absorption technique involving SnC12 reduction was introduced in 1964 by Poluektov et a1.F and developed by Hatch and OttS. Manning has published a thorough review of the topics§. Despite the number of publications very little new instrumentation has been described. This is no doubt because of the ease with which reliable determinations may be carried out at low levels using inexpensive modifications of commercial atomic absorption spectrometers. Nearly all manufacturers of atomic absorption instruments provide conversion kits for Hg determination.Furthermore heated devices such as tube furnaces and filaments have been used for this purpose. Several instruments specific- ally designed for Hg analysis are listed in Table A. Several reports (370 41 1 524) have described instrument adaptations where Hg is obtained by strongly heating the sample is adsorbed on Pt (524) or Au (370,411) and is subsequently liberated by heating and swept into an absorption cell. Automated digestion and measurement systems have been constructed (1 57 202 409 5 17) and the use of a Wickbold apparatus for sample preparation particularly with organic solutions has been described (504)q. Although Hg vapour has usually been detected by atomic absorption techniques the use of atomic fluorescence offers some additional advantages including elimination of spurious absorption signals by condensed species.In the system of Thompson and Wildy (764) there was no optical cell or recirculation system. Hg vapour escaped from the end of a 10 nm id glass tube at the optical centre of the fluorescence system. A d.c. glow discharge and atomic emission has also been used (408) for the detection of evolved Hg. An enclosed absorption cell has been described (28) which features a side-arm containing an Hg reservoir. An analytical calibration curve for the cell could be con- structed by placing the side-arm in a variety of cooling mixtures.The apparatus was used for the determination of Hg in lunar rocks (549). 2.7 OTHER EXCITATION AND ATOMISING SYSTEMS Two interesting papers have been noted which fall into none of the above categories. In the work described by Hirokawa (860) the analytical sample was excited by ion bombardment the ions being produced by a low voltage impulse discharge through N z . This indirect method was useful for analysing both metals and non-conducting materials such as oxides. In the latter instances mixing with a conducting material e.g. carbon was unnecessary. Analytical emission lines monitored included Ti I 319.19 nm Ti I1 309.19 nm C I1 250.91 nm Fe 1309.81 nm Si 1243.52 nm and B I 249.68 nm.Rousselet et a1 (468) described a system where the sample (volume * T. T. Woodson Rev. Sci Ins. 1939,10 308. N. S. Poluektov R. A. Vitkun and Y. V. Zelyukova Zh. Anal. Khim 1964,19 873. 5 W. R. Hatch and W. L. Ott,AnaZyt. Chem. 1968,40 2085. 8 D. C. Manning,At. Absorp. Newsl. 1970 9 97. 1 See ASTM Method D 2785-70 or Institute of Petroleum Method IP 243/69T for description of Wickbold apparatus. mg a filters testing.and Two liquids.gives operate gases 0.0002 .Mains on catalyst SnC1 feature to 0-0.2 1-1 an testing.through and discharged.battery ranges silver circulator.mg Batteries down atmosphere are operated scales with hot air two a for current use m-3.gases two in to air Mains mg determinations on for atmosphere for batteries on built-in by mainly rechargeable 0.0002-0.5 circulator.for /day.m-3 analysis are rechargeable direct used AAS air 0-0.7 Description with when be m-' mg Ni/Cd gas designed for injected Carried lb.liquids and off detector 6 mg and detector also 0.01 volt and Built-in m-3 turn to 12 sample Can to instrument furnace.liquid m-3 mg continuously.p1 0.0025 FLAMELESS Weight portable instrument instrument m3.portable is down system.for mg 50 BY use.beam mg 0-0.2 hours cut-out Drift 7% reduction hours a photometer.Hand-held Measures 0-1.0 8 Double Designed 0.002-5 operated.Laboratory reduction Hand-held ranges for safety Laboratory Typically in to mP3.MERCURY OF address DETERMINATION THE and Division) Lane U.S.A.Supplier's 11 Ltd England.England FOR Inc. Instruments Heath Heath Co.Division Division 15238 Research Crown AB 60153 (Coleman Ltd Bucks.The The Ins.Drive Pa Surrey Bromma Ill.Cheshire Industries Cheshire Mond Mond 11 Instruments House Berks.Elmer Elmer Dept Dept Alpha Incentive 11074 D D Ltd Ltd INSTRUMENTS & & Baccharach AMBAC 625 Pittsburgh U.S.A.Shandon-Southern Camberley IRD Development Box S-161 Sweden.Wallac Crown Theale Perkin Maywood Perkin Beaconsfield ICI R Runcorn ICI R Runcorn A TABLE MV-2 Sniffer Name Probe Vapour MAS-SO Meter Mercury Mercury W J HGM-2300 Coleman Total Mercury Part I Fu ndam en tals and Ins tru m en ta tion 22 10-100 nl) was deposited in a W cup and bombarded under vacuum by electrons accelerated by a 2 k V cm-l electric field.High voltage stability and vacuum quality were important criteria for successful operation.