630 Analyst, October, 1966, Vol. 91, PP. 630-637 Improvement in Performance of a Simple Atomic Absorptiometer by Using Pre-heated Air and Town Gas BY R. A. G. RAWSON (Rothamsted Experimental Station, Harpenden, Hertfordshire) A simple atomic absorptiometer for the routine determination of mag- nesium, copper and zinc is described. A pre-heated air supply to the atomiser increased the response efficiency of the absorptiometer 16-fold. An adjustable slot in the burner enabled a wide range of fuel gases to be used. A horizontal monochromator slit gave improved light transmission and absorption. ALTHOUGH the conventional atomic-absorption method by Box and Walshl was adequate for determining magnesium, it was too insensitive for copper and zinc. Such insensitivity was probably caused by incomplete vaporisation of solids and was enhanced, for magnesium, by interfering elements that formed refractory compounds.Very hot flames, above 2000" C, generally improve the sensitivity, but these are costly, inconvenient to produce and have complex spectra in which there is more interference. Improved atomisation, which produced smaller solid particles and converted the solution to aerosol more efficiently, was a better alternative. This was achieved by pre-heating the air and gas supplies to the atomiser and spray chamber, respectively, and by using a newly designed burner. The need for such improvement was expressed by Allan2 and by Russell, Shelton and Walsh.3 INSTRUMENTATION APPARATUS- Source of emission-The conventional sealed-off hollow-cathode discharge tubes were made by Messrs.Hilger and Watts Ltd. Ancillary eq.uipment-The power pack for the hollow-cathode discharge tube and E.H.T. supplies for the photomultiplier valve (R.C.A. IP28) and the amplifier were as specified by Box and Wa1sh.l All electronic equipment was simple in design, cheap and reliable. Housing for bwner and lamp-The housing was made of 10-mm Asbestolite and the burner section was amply ventilated. The lamp and burner were in adjacent compartments separated by an Asbestolite partition with a hole of 25mm diameter in the appropriate - U .- - g : 2 8 0 5 36 38 40 41 40 3% 36 Extinction, per cent. Fig. 1. Town gas flame showing ad- vantage of horizontal monochromator slit over vertical slit with 2-mm burner slot widthRAWSON 631 position. A cowling and discharge duct above the burner is necessary in a laboratory with inadequate ventilation.Optics-The only optical system required in this unit was the monochromator, because the hollow-cathode lamp emits a near-parallel light beam. Focusing the beam, therefore, had little merit and did not warrant the quartz lenses that are needed for ultraviolet trans- mission. A horizontal monochromator slit was used to take advantage of the horizontal distribution of atomic vapour just above the blue cone of the flame (Fig. 1). With a vertical arrangement, a collimating slit placed after the flame improved the sensitivity by decreasing the unabsorbed emission transmitted in the upper part of the beam. After collimation, the smaller signal obtained from the detector had to be amplified more to obtain full-scale deflection.An optical bench was unnecessary because the components were rigidly constructed and easily aligned on the optical axis. E Fig. 2. Adjustable burner assembly. (Lettered parts of the apparatus are referred to in the text) The b.urn.er (the patent has been applied for)-This (Fig. 2) was designed for use with various fuel gases and was made from a stainless-steel tube of the following dimensions and standard wire gauge (s.w.g.) : main burner tube, A, and trunk tube, D: 25 mm over-all diameter (o.d.), 22 mm internal diameter (id.), 16"s.w.g.; tube, B: 22 mm o.d., 19 mm i.d., 16" s.w.g.; branch tubes, C: 19 mm o.d., 15 mm i.d., 14" s.w.g. The slot, 120 mm long, 3 mm deep, with width adjustable by arm, F, could be set between 0 and 2-25mm.The deep walls of the slot help laminar flow of the pre-mixed gases and also increase the quenching ability of the slot, thereby lessening the risk of "flashback." Narrow lips, E, were welded on the edges of the fixed tube, A, and the rotating tube, B, to maintain equal thickness in all positions, The inner walls of the slot were polished, straight and exactly parallel. The Y-shaped burner (each leg was equidistant from the centre and either end of the burner) was better than the T or "fish-tail" design, because these caused turbulent gas flow, and gases and solid particles would fractionate more. Another advantage was that each leg of the Y burner fed only a 60-mm length of slot. The main volume of gas carried by the trunk was equally split at the junction of the legs, ensuring even distribution to each half of the burner head. The long threaded rod, G, along the central axis of the burner head allowed the slot to be locked at any width (by using feeler gauges) and also acted as a baffle.Cooling the burner with water was unnecessary because, with pre-heated gases, a constant operating temperature was reached much more rapidly, contrary to previous observations by Clinton.* Pre-heating the incoming gases to 328" K increased the volume of gases above the blue cone only slightly, because this increased the normal flame temperature (2073" K) by only 40" K. The wide burner slot gave complete freedom from clogging and decreased flame noise caused by pressure fluctuations, because the back pressure in the spray chamber beneath the burner was only about 1 mm of water.Atomiser-The atomiser from the E.E.L. flame photometer (Design Xo. 866150), which has an adjustable capillary to control the pressure drop across the nozzle, was modified. The capillary was removed, the orifice of the nozzle drilled with a No. 58 drill, and the capillary replaced. Spray clinmnber-The chamber used in the E.E.L. flame photometer was modified by drilling a hole, of 8 mm diameter, near the end cap to hold a 0" to 250" C thermometer. A water seal on the drain tube maintained constant pressure; it consisted of a glass U-tube with the delivery leg bent over so that its contents, when displaced, dripped steadily from this632 [Analyst, Vol. 91 leg. A glass ring (19 mm i.d.) with 6 holes on its inner surface (to focus the gas stream on the spray jet) was fitted on the gas inlet inside the end cap so that the gas surrounded the spray (Fig.3). This improved the mixing of gases and the heating of spray and lessened flame turbulence. RAWSON: IMPROVEMENT IN PERFORMANCE OF A SIMPLE ATOMIC /Gas ring Fig. 3. End cap of spray chamber Pre-heaters (the patent has been applied for)-Tubular brass pre-heaters (5 mm id., 8 mm 0.d. and 2.5 mm apart) in the air and gas lines passed through the flame 25 mm above the burner head, pre-heating the air and gas before entering the spray chamber. The sections of tubing from the flame to the spray chamber were insulated with asbestos lagging. Both heaters were connected to the spray chamber with brazed screw sockets and asbestos washers, because soldered connections broke down at such high temperatures.Stainless-steel tubes would be better than brass. Pre-heaters, spray chamber and burner were rigidly fixed together to ensure alignment when the assembly was moved. Monochromator-A diffraction-grating monochromator (Model D292), supplied by Hilger and Watts Ltd., was mounted with the entry slit horizontal. The monochromator is sensitive to heat and was shielded from the burner housing by asbestos. A short length of 12.5 mm i.d. copper tubing (its interior surface painted with camera black), fitted to the entry slit, decreased stray light entering between the burner housing and the monochromator. PhotomuZti$&er-The photomultiplier valve (R.C.A. IP28), designed to respond to wave- lengths between 2150 and 5500A, also operated satisfactorily at the zinc emission line (2138-5A).The position of the valve, in relation to the exit slit of the monochromator, was critical. The signal was strongest when the monochromator slit was parallel to the photo-cathode of the valve, i.e., at 90” to the valve base. The amplifier meter reading was most stable when the E.H.T. supply to the photomultiplier valve was about 860 volts and the amplification control did not exceed 30 per cent. of its total movement. Pump ~ O Y town gas-A small diaphragm pump (Charles Austin Pumps Ltd.; capacity 5 litres per minute at 5 p.s.i.) controlled by a needle valve and flow-meter was used to ensure a constant supply of town gas. Air-compressor-A compressor, supplying clean dry air continuously at about 25 litres per minute at 50 p.s.i., was controlled by a diaphragm reduction valve, needle valve and flow-meter.RESULTS AND DISCUSSION HOLLOW-CATHODE LAMPS- The hollow-cathode lamps were run at the lowest possible currents (e.g., 4 to 5 mA for To prevent clean-up” of the filler gas, lamps were run for 8 hours continuously, once every month, magnesium) for minimum spectral line half-width and maximum sensitivity. when not in use. 4 < CHOICE OF FUEL- Propane and methane proved to be unsuitable because their burning velocities in air were too small (40 cm per second) and caused their flames to “blow off’’ unless the streaming velocity was kept slow, particularly when water vapour was introduced. A slow streaming velocity produced a weak flame that was vulnerable to draught and obstruction.AlthoughOctober, 19661 ABSORPTIOMETER BY USING PRE-HEATED AIR AND TOWN GAS 633 these gases had large B.T.U. values and produced more total heat than town gas, the hottest part of their flames was cooler, so that they were less able to vaporise solids. Town gas, which had a burning velocity of over 100cm per second because of its high hydrogen content, was used at a moderate flow-rate to give a firm flame free from interference. Town gas is not an ideal gas to use, because its composition varies. The only factors that affect reproducibility are its B.T.U. value, burning velocity and specific gravity, although suppliers of town gas rigidly control these factors for economic reasons. With hydrogen as a fuel (burning velocity in air 270 cm per second) the burner slot must be closed to 0.50 mm to prevent “flash back.” This increases back pressure in the spray chamber, increases the streaming velocity and narrows the flame, which gives unsatisfactory results with the horizontal monochromator slit.Increasing the streaming velocity increases flame turbulence and extends the flame front of the pre-mixed cone, thereby increasing the surface area of the reaction zone. This elongates the cone and, therefore, the contours of atomic vapour density, which then do not match the monochromator slit and so sensitivity is lost. FLAME TEMPERATURE- The flame temperature was increased by raising the carbon monoxide content of town gas from 20 to 50 per cent., but the hotter flame had no effect on magnesium absorption.Much hotter flames obtained with acetylene vaporised refractory compounds, but improved atomisation, which made the particles of such compounds smaller, was a better method. INTERFERING ABSORPTION LINES When zinc was determined a t 2138A, the flow of town gas through the burner, before ignition, caused complete extinction of the source beam. This was caused, partly by carbon monoxide in the town gas, which has a strong absorption line at this wavelength, but mostly by hydrogen sulphide which decomposes endothermically at this wavelength. The radiation responsible for this decomposition ranges5 in wavelength from 1800 to 2300 A. Accord- ing to the theory of the structure of flames, about half of the carbon monoxide and all of the hydrogen sulphide is oxidised in the pre-mixed cone on ignition. This was confirmed by a fall in extinction to between 40 and 50 per cent, on ignition.The amplifier gain was increased to restore 100 per cent. transmission, but this greatly decreased the sensitivity for zinc. Acetylene also had a strong absorption line at 2138 A that persisted after ignition, particularly in a fuel-rich flame, because of carbon monoxide in the interconal gases produced by combustion. Carbon dioxide, nitrogen, oxygen, methane and propane did not absorb at this wavelength. ATOMISATION- Increasing the diameter of the air nozzle of the atomiser lessened the air pressure behind the nozzle, which slowed heat transfer to the capillary. This permitted a hotter air supply to be used without vaporising the solution within the capillary, which would retard uptake of solution and cause the spray chamber to over-heat.The air was hot enough to warm the test solution as it passed through the capillary and to supply the latent heat required to vaporise it at the nozzle. Increasing the diameter of the nozzle also slowed the air passing out of it and allowed the vapour pressure of the solution to be increased without vaporising it within the capillary. ADJUSTMENT OF THE ATOMISER AND PRE-HEATERS- Great sensitivity was attained economically by feeding a small volume of hot air at low pressure, about 25 p.s.i., to the atomiser, but this gave a weak flame susceptible t o draught and pressure variation. A firm flame was, therefore, preferred with an air pressure of 30 to 35 p s i .A t this pressure, with the appropriate amount of town gas, the pre- heaters were at a red heat over about 2 inches of their length. The atomiser then took up solution rapidly and the spray chamber became correspondingly cold. The capillary of the atomiser was slowly adjusted until uptake was only about 2.5 ml per minute and then left to operate on distilled water for about 15 minutes to allow the spray chamber to reach constant temperature. The capillary was finally adjusted by rotating it in an anti-clockwise direction, about 5” at a time, until the run-off from the spray chamber stopped, indicating 100 per cent. atomisation.634 [Analyst, Vol. 91 To obtain the correct final balance, the air pressure to the atomiser was altered by 1 p s i at a time and the gas flow adjusted.This altered the heat output to the pre-heaters and made it possible to adjust the temperature in the spray chamber to 54" C After each run, the flame was extinguished and the spray chamber flushed out with dis- tilled water by passing cold air through the atomiser for 5 to 10 minutes to remove solids trapped by the spray chamber. The use of a secondary air supply, which did not pass through the atomiser and could therefore be heated without limit, achieved complete atomisation with ease but did not improve the sensitivity appreciably. Probably it did not contribute to the disintegration of the spray but evaporated the larger droplets that otherwise would have fallen out to the drain.Solid aerosol particles from such large droplets would not be easily vaporised by the flame. With pre-heated air, the hot droplets of spray were most probably split instantly into minute fragments on meeting the incoming air at about 200" C. The very fine particles of solid aerosol resulting were easily vaporised by the flame, giving a high density of atomic vapour. This physical change of state is possible only when the necessary latent heat of vaporisation is readily available. RAWSON: IMPROVEMENT IN PERFORMANCE OF A SIMPLE ATOMIC 1" C. USE O F ORGANIC SOLVENT- The use of an organic solvent in the test solution was said to increase sensitivity greatly under normal conditions, as reported by Allan.2 However, with pre-heating, 20 per cent. v/v of ethanol or propanol had no significant effect.Also, extracting and concentrating with an immiscible organic solvent was not necessary with copper because the sensitivity was adequate. ATOMISATION EFFICIENCY- With this system atomisation was efficient (Table I) when the spray chamber was kept at a constant temperature of 54" C during continuous uptake of distilled water or test solutions. TABLE I PERFORMANCE CHARACTERISTICS OF COMMERCIAL INSTRUMENTS AND OUR APPARATUS Unicam flame spectrophotometer SP900 r----A------, E.E.L. flame Our Cool flame Hot flame photometer instrument Air pressure, p.s.i. .. . . .. 22 29 12 34 Rate of uptake, ml in 10 minutes . . 39 43 24 21 Atomisation, efficiency per cent. . . 7 - 7 7 13.5 100 Fall-out to drain . . .. .. . . 36 40 21 ni 1 IMPROVED SENSITIVITY- An accurate comparison of this system with, and without, pre-heating is not possible because altering any one factor inevitably affects others.However, removing the pre- heaters, without manually altering anything else, decreased sensitivity to one-sixth, despite the solution being taken up faster (Fig. 4). This fraction becomes one-sixteenth when allow- ance is made for the increase in the rate of solution uptake (Table 11). TABLE I1 EFFICIENCY OF CONVERSION TO ATOMIC VAPOUR Soh tion, Percentage Volume p.p.m..of Uptake, of solution atomised, Optical magnesium nil per minute atomised ml per minute density A. With pre-heaters . . 0.065 1-78 100 1.78 0.152 B. Without prc-heaters 0-400 4-76 8 0.38 0.152 C . Ratio A/B . . . . 6.150 2.67 - 4-68 - To obtain the same optical density without, as with, pre-heating, 6.15 times the con- centration of magnesium and 2.67 times the amount of test solution were required, Despite the sensitivity increases of 6.15-fold with pre-heating, the volume of solution atomised in- creased only 4.68-fold.This figure does not allow for about 45 per cent. of solid aerosol trapped by the spray chamber. So there was a vast improvement in converting the solution to atomic vapour. Without pre-heating, only 27 per cent. of solid was trapped.October, 19661 ABSORPTIOMETER BY USING PRE-HEATED AIR AND TOWN GAS 635 Magnesium, p.p.m. Fig. 4. Absorption by magnesium atoms, with and without yre-heated air and gas supplies, in the presencc of 300 p.p.m. of strontium as strontium chloride a\. With pre-heaters: ratc of uptake = 1.78 ml per minute; spray chamber temperature = 55” C ; and atomisation efficiency = 100 per cent.B. Without pre-heaters: rate of uptake = 4.76 ml per minute; spray chamber temperature = 15” C; and atomisation efficiency = 8 per cent. THE FLAME- In normal operation (burner slot width 2 mm), the flame was 125 mm long, 11 mm wide and had a blue cone about 6 mm high when aspirating solution. Air, at room temperature and pressure, was supplied at 21.6 litres per minute. The volume of town gas used varied according to the type of flame required, but was usually about 5-4 litres per minute. Its average composition was hydrogen, 49 per cent. v/v; carbon monoxide, 20 per cent. v/v; methane, 10 per cent. v/v; nitrogen, 7 per cent. v/v; carbon dioxide, 5 per cent.v/v; un- specified hydrocarbons, 5 per cent. v/v; ethane, 4 per cent. v/v; and oxygen, 0.6 per cent. v/v: traces of sulphur compounds were always present. To find the vertical distribution of “ground-state” atoms in the flame, the monochromator slit was collimated to 2 x 0.20 mm and the extinction noted as the burner was lowered by 1 mm at a time (Fig. 1). This indicated the sensitivity that might be expected from a 0.2 x 7-mm vertical monochromator slit. The distribution of “ground-state” atoms in a cross-section of the flame was examined by superimposing a 1-mm vertical slit on the 0-2-mm horizontal slit. This showed that the intensity of the source beam fell rapidly as the vertical slit was moved away from its centre, and the amplifier gain was increased to restore full- scale deflection.This method proved unsatisfactory because the response curve of the amplifier had then to be taken into account. Instead, the instrument was set at full-scale deflection when the vertical slit was at the centre of the source beam. Then, without altering the amplifier gain, the extinction was expressed as the percentage of the incident light as the vertical slit was moved. ADVANTAGES OF THE HORIZONTAL MONOCHROMATOR SLIT- The monochromator slit gave better results when horizontal, with a broad flame, than when vertical, with broad or narrow flames. Fig. 1 shows that the density of atomic vapour decreases faster in the vertical direction than in the horizontal, giving the horizontal slit a distinct advantage. Collimating the horizontal slit to 4 or 5mm increased sensitivity slightly but was of little advantage because the increased amplifier gain required made the636 [A~zaZyst, Vol.91 meter less stable. Light transmission was a t a maximum because it was not necessary to collimate the source beam to the same extent as with a vertical slit. This was advantageous because the light transmitted by the diffraction grating monochromator was only one-third of that transmitted by a Unicam SP500 prism instrument under the same conditions. If less light was transmitted, full-scale deflection of the meter was obtained by passing more current through the hollow-cathode lamp (this broadened the spectral line-width of the source), by increasing the width of the monochromator slit (causing a loss of resolution), increasing the E.H.T.to the photomultiplier, or by increasing the amplifier gain (both of which caused unstable readings on the meter). RAWSON: IMPROVEMENT IN PERFORMANCE OF A SIMPLE ATOMIC 5 p.p.m. Fig. 5. Calibration lines for zinc, copper, manganese and calcium A = Zinc Monochromator slit = 0.30 mm Amplifier gain = 20 per cent. Lamp current = 10 mA Background noise = &On25 per cent. Flame emission = 0 per cent. Flame extinction = 42 per cent. Analytical limit (5 per cent. deflection) = 0.05 p.p.m. C = Manganese Monochromator slit = 0.2 mm Amplifier gain = 20 per cent. Lamp current = 10 mA Background noise = +Om25 per cent. Flame emission = 3 per cent. Analytical limit (5 per cent. deflection) = 0.25 p.p.m. B = Copper Monochromator slit = 0.10 mm Amplifier gain = 20 per cent.Lamp current = 5 mA Background noise = 0-75 per cent. over-all Flame emission = 1 per cent. Flame extinction = 3 per cent. Analytical limit (5 per cent. deflection) =0.125 p.p.m. D = Calcium (with 500 p.p.m. of lanthanum) Monochromator slit = 0.2 mm Amplifier gain = 23 per cent. Lamp current = 8 mA Background noise = f0.3 per cent. Flame emission = < 1 per cent. Analytical limit (5 per cent. deflection) = 0.25 p.p.m. COPPER, ZINC, MANGANESE AND CALCIUIIG- Fig. 5 shows calibration lines for copper, zinc, manganese and calcium. aids, e.g., optical system or scale expansion unit, were used at any time. No mechanical INTERFERENCE- Phosphorus, and particularly aluminium, interfered with the determination of mag- nesium, but all interference was eliminated by adding strontium as strontium chloride toOctober, 19661 637 the test solutions and standards to give 300 p.p.m. of strontium. The rich colour in the flame from strontium was also a good indicator of contamination. Other elements in concentrations commonly found in soil extracts and plant digests did not interfere when determining copper and zinc. Sensitivity for calcium was adequate but interferences were not examined. CONCLUSION ABSORPTIOMETER BY USING PRE-HEATED AIR AND TOWN GAS The instrument described is so improved in sensitivity that many elements can be directly determined without preliminary concentration. Some of the main requirements in atomic-absorption technique are : producing a very fine solid aerosol; converting solution to aerosol efficiently ; having the smallest practicable volume of interconal gases ; and having a long light path through the flame. Further detailed investigation is necessary to curtail the effects caused by the formation of refractory compounds. Devices that would make the particles of solid aerosol still smaller and so diminish chemical and physical interferences are needed. I thank Mr. B. Edwards for help with the burner design. REFERENCES 1. 2. 3. 4. Clinton, 0. E., Ibid., 1960, 16, 985. 5. Box, G. F., and Walsh, A., Spectrochiwz. Acta, 1960, 16, 255. Allan, J. E., Ibid., 1962, 18, 605. Russel, B. J., Shelton, J. P., and Walsh, A., Ibid., 1957, 8, 317. Mellor, J. W., “A Comprehensive Treatise on Inorganic and Theoretical Chemistry,” Longmans, Received January 7th, 1966 Green and Co., London, New York and Toronto, 1930, Volume 10, p. 128.