72 Analyst, February, 1968, Vol. 93, pfi. 72-78 Molecular-emission Spectroscopy in Cool Flames Part 11.8 The Behaviour of Phosphorus-containing Compounds BY R. M. DAGNALL, K. C. THOMPSON AND T. S. WEST (Chemistry Department, Imperial College, London, S. W.7) A sensitive and selective molecular-emission method for the determina- tion of phosphorus is described, in which the intense green emission obtained by nebulising orthophosphoric acid solution into a cool, nitrogen - hydrogen diffusion flame is measured a t 528 mu. The limit of detection in aqueous solution under the conditions of measurement IS 0.1 p.p.m. of phosphorus. Most cations produce a depressive matrix effect that can be readily overcome by a preliminary ion-exchange separation. Fifty-fold molar excesses of acetic, hydrobromic, hydrochloric, nitric, oxalic, sulphuric and tartaric acids do not interfere.IN the previous paper in this series we reported on the application of cool, nitrogen - hydrogen diffusion flames to the determination of sulphur.1 The procedure was based on the blue emission from the S, species that is formed in such flames. Like sulphur, phosphorus has its principal atomic resonance lines in the far ultraviolet region of the spectrum (1775, 1783 and 1788 A) and, because of the almost total absorption by air, flame gases and quartz optics in this region, it is not possible to apply atomic-absorption or atomic-fluorescence spectro- scopy in their usual modes. Consequently, we have examined the determination of phos- phorus by molecular-emission spectroscopy in cool diffusion flames.The more usual methods of determining phosphorus by emission-spectroscopic techniques involve indirect procedures, such as the depressive effect of phosphate on the emission of calcium and magnesium.2 However, only a somewhat limited range of phosphate concen- trations can be examined by using a given concentration of calcium. In addition, the method is not very sensitive; by using a 100 p.p.m. calcium solution the estimated limit of detection is about 30 p.p.m. of phosphorus. A direct method, with a limit of detection of 3 p.p.m., has been described by Brite,s who measured the emission from the continuum at 540 mp obtained with an oxy-hydrogen total- consumption burner and organic solvents. Davis and co-workers4 also used an oxy-hydrogen turbulent-flow burner to obtain a limit of detection in aqueous solution of 8 p.p.m. at a slit width of 2.0 mm.Wide slits are necessary to gather sufficient energy from the continuum. Unfortunately, the use of organic solvents leads to high background emission, which must be subtracted from the phosphorus emission. Another method,b which is also concerned with the determination of phosphorus in organic compounds, involves the measurement of the emission from the H-P-0 species. In this instance, organic vapours leaving a gas- chromatographic column, with nitrogen as carrier gas, were mixed with a volume of oxygen to give the same nitrogen-to-oxygen ratio as occurs in air. This mixture was then burned, with hydrogen as fuel gas, in a flame-photometer burner.The use of narrow band-pass interference filters and a photomultiplier resulted in a very sensitive and selective detector for sulphur and phosphorus compounds. The limit of detection at 526mp is 0.0063 p.p.m. of phosphorus. However, it is applicable only to organophosphorus compounds and, in this instance, only to those which can be handled directly by gas-phase chromatography. * For details of Part I of this series, see reference list, p. 78. 0 SAC and the author.DAGNALL, THOMPSON AND WEST 73 The nitrogen-hydrogen diffusion flame has been found by us to offer considerable advantages as an atom reservoir in both atomic-absorption and atomic-fluorescence spectro- scopy.6~7 This arises because its background radiation is less than that of pre-mixed flames, and because its low temperature results in very low excitation of even the alkali elements.It also exhibits negligible absorption of radiation in the far ultraviolet region of the spectrum, so that increased signals may be obtained in atomic-absorption and atomic-fluorescence spectroscopy for elements such as arsenic.' The flame also has high reducing properties and limited quenching action. In addition, because of its low temperature and the limited supply of oxygen to these flames, it permits and promotes the existence of species not normally observed in usual pre-mixed flames, e.g., S,.1 The emission from phosphorus-containing compounds is a further instance of this phenomenon. In view of the voluminous literature available concerning organophosphorus compounds, we have concentrated on inorganic applications in this study.THE NITROGEN - HYDROGEN DIFFUSION FLAME- The burner head used was the standard 1.8 x 76cm air-acetylene emission head supplied with the Unicam SPSOOA flame spectrometer. The burner head is perforated near one end by a 1-cm square pattern of 13 holes. However, any other suitable emission burner head could be used in its place. Nitrogen was used as the nebulising gas (at 15 p.s.i.), with the conventional SPSOOA nebulising system. Hydrogen was introduced, as usual, at the bottom of the burner, at a pressure somewhat above that necessary to prevent the flame from lifting off the burner. The actual pressure of hydrogen was found to be uncritical. The principal characteristics of this flame are that it is colourIess and cool (temperature ranges from 280" to 850" C, depending upon position in the flame),l and gives weak OH band emission (about forty times less than a pre-mixed air-hydrogen flame) and only weak sodium emission (emission from a 2 p.p.m.sodium solution at 689mp is about fifty times less than in an air - hydrogen flame .) The green phosphorus emission, which is at a maximum at 528 mp, is easily obtained by spraying a solution of orthophosphoric acid. As with sulphur, it was observed that the signal was drastically decreased by the addition of small amounts of air or nitrous oxide through a third jet in the base of the burner.* Temperature measurements show that this is due to the increase in flame temperature obtained by the addition of oxygen.The green colour was noted to be particularly prominent in the cool inner regions of the diffusion flame (about 350" C), and was most marked about 1 cm above the burner top (280" C). A nitrogen - hydrogen total-consumption burner gave negligible emission, probably because the turbulent nature of the flame causes considerable entrainment of air, and hence gives rise to a much hotter flame than that obtained with the simple diffusion flame. The separated air - hydrogen flame, previously designed to aid the breakdown of sulphates and so give a more uniform sulphur emission,l was also investigated. The experimental arrange- ment was exactly as used for the sulphur studies, Le., a pre-mixed air - hydrogen flame with a cooled borosilicate glass or quartz sheath, but it was found not to be particularly advantageous in this instance.The green colour persisted a long time after nebulisation had been terminated. This is thought to be caused by condensation of phosphoric acid or phos- phorus pentoxide on to the walls of the glass sheath and subsequent slow evaporation. The emission occurred mainly around the cool walls of the sheath, just above the burner (about 350" C), and was only slightly more intense (about 20 per cent.) than in the normal diffusion flame. IDENTITY OF NEBULISED SPECIES- The green emission was most marked in the nitrogen - hydrogen diffusion flame on nebu- lising orthophosphoric acid. Metallic phosphates gave only a weak signal from lack of dissociation in the relatively cool flame (see Interference studies).A spectral scan of the emission obtained while spraying a 1.2 x 1 0 - 2 ~ aqueous solution of orthophosphoric acid, with a slit width of 0.03 mm, is shown in Fig. 1. The air - hydrogen flame yielded an exactly similar spectrum, but was at least five times less intense, and there was much more of a continuum. Also, the comparatively large background emission of the air - hydrogen detracts from the sensitivity towards dilute orthophosphoric acid solutions. In aU instances emission corresponds only to the H-P-0 species.6 EXPERIMENTAL74 DAGNALL, THOMPSON AND WEST : MOLECULAR-EMISSION [Analyst, Vol. 93 8o t Wavelength, mp Fig. 1. Emission spectrum of H-P-0 species obtained by nebulising a 1.2 x 10-ZM solution of orthophosphoric acid into the nitrogen - hydrogen diffusion flame The intensity of the emission at 528 mp in the diffusion flame exhibits a linear response with orthophosphoric acid solutions over the range 0.2 to 500 p.p.m.of phosphorus. Larger amounts of phosphorus were not examined. The limit of detection (signal-to-noise ratio = 1) was 0.1 p.p.m. of phosphorus, with a slit width of 0-17 mm. The use of filters and a more sensitive photomultiplier (E.M.I. 9601 B was used in these studies) would give considerably lower limits. There is also little doubt that the use of a fast monochromator or a simple filter system to isolate the desired radiation around 528 mp would improve the sensitivity considerably. The emission at 528mp was not very dependent upon the hydrogen pressure or the height of measurement in the flame, because it extends uniformly over most of the inner flame regions. Measurements were optimised with the hydrogen pressure somewhat above that necessary to prevent flame lift-off, and with the top of the burner 3 cm below the bottom of the monochromator slit.INTERFERENCE STUDIES- Resfionse towards various phosphorus-containing compounds-Table I shows the emission signals obtained for 2 x M solutions of various phosphorus-containing compounds usually encountered. The readings were obtained by taking the peak (at 528 mp)-to-trough (at 515mp) height from a recorded spectrum. This was considered to be more accurate than an emission signal reading at 528 mp because it eliminates any errors caused by a slight sodium continuum at 528 mp. The peak-to-trough height was found to vary almost linearly with orthophosphoric acid concentration.TABLE I EMISSION INTENSITIES OF VARIOUS PHOSPHORUS COMPOUNDS IN THE NITROGEN - HYDROGEN DIFFUSION FLAME Compound, 2 x 1 0 - 3 ~ Orthophosphoric acid . . .. .. .. Sodium dihydrogen orthophosphate . . .. Disodium hydrogen orthophosphate . . .. Disodium hydrogen phosphite . . .. .. Sodium pyrophosphate . . .. .. .. Calcium hydrogen orthophosphate. . .. .. Emission reading, peak-to-trough height 48 16 6 5.6 6 12February, 19681 SPECTROSCOPY I N COOL FLAMES. PART I1 76 All readings were obtained with a slit width of 0-05mm and a gain of 3-10, The sodium-to-phosphorus ratio has a critical effect on the emission signal; as the ratio increases so the emission signal decreases. The addition of hydrochloric acid (concentrations up to M) to the solution of &sodium hydrogen orthophosphate, or an increased hydrogen flow to the diffusion flame, did not produce any increased response.TABLE I1 EFFECT OF SOME EXTRANEOUS IONS ON THE H-P-0 EMISSION IN THE DIFFUSION FLAME Interference Aluminium chloride . . .. Ammonium chloride . . .. Ammonium iron (I1 I) sulphate Ammonium iron(I1) sulphate . . Cadmium chloride . . .. Calcium chloride .. .. Cobalt(I1) chloride . . .. Copper(I1) chloride . . .. Lead nitrate . . .. .. Lithium chloride .. .. Sodium chloride .. .. Magnesium chloride . . .. Concentration, M (5-6 x 1 0 - 3 ~ with respect to H3P0,) .. 1 x 10-2 .. .. .. .. .. * . .. .. .. .. .. 1 x 10-8 1 x 10-8 5 x 10-8 1 x 10-2 1 x 10-8 5 x 10-2 1 x 10-2 3 x 10-2 3 x 10-2 1 x 10-2 5 x 10-2 1 x 10-8 3 x 10-2 1 x 10-2 1 x 10-8 2 x 10-3 Emission reading, peak-to-trough height 54 0 55 22 17 21 21 13 4 22 > 100 32 23 18 2 6 3 2 Efect of extraweom ions-Interference studies (Table 11) were carried out with a 5.6 x 1 0 - 3 ~ solution of orthophosphoric acid and optimised diffusion flarne conditions.Following the addition of the extraneous ion, all of the solutions were made 0 . 2 ~ with respect to hydrochloric acid. As before, all of the readings are peak (at 528 mp)-to-trough (at 515 mp) heights to eliminate any light leaks in the monochromator, etc. In this instance the slit width was 0.03 mm and the gain was 3-9. The only increased response is caused by copper and arises from the strong Cu-Cl and Cu-H band emission in the same region as the phosphorus emission.It would appear from Table I1 that the degree of the depressive effects is not related to the volatility or stability of the various metal phosphate species produced in the flame. However, there is some correlation between these effects and the ease with which the species may be reduced by hydrogen. This may be illustrated with reference to lead orthophosphate [Pb,(PO,),], which has a higher melting-point than either lithium orthophosphate (Li,PO,) or sodium pyrophosphate (Na,P,O,). Also, iron(II1) forms a more stable phosphate species than sodium or lithium but grves less interference. These effects could be explained if cool diffusion flames, unlike normal pre-mixed flames, are assumed to produce atoms from metal salt solutions principally by reduction, and not by thermal breakdown.In a normal pre-mixed flame no interference from compound form- ation is observed with sodium, because its salts are thermally unstable at temperatures of about 2000" C. Phosphate compounds of elements such as cadmium, cobalt, lead and iron are much more likely to be reduced by hydrogen at temperatures of about 400" C1 than aluminium, lithium, sodium and magnesium compounds. The depressive effect increases with increasing concentrations of sodium and lithium. This appears to be a mass-action effect, because increasing the metal chloride concentration must increase the free metal concentration in the flame and, as the chlorides would be expected to be thermally less stable than the corresponding phosphates, the dissociation of the metal phosphate would be depressed.The effect is not so marked with increasing concentrations of cadmium, lead and iron, suggesting that the breakdown process is not entirely thermal but a reduction process that is not very dependent on metal-ion concentration.76 DAGNALL, THOMPSON AND WEST : MOLECULAR-EMISSION [A fldyst, VOl. 93 The effect of varying concentrations of sodium, lithium and calcium on a solution 5 x M with respect to orthophosphoric acid and 0.2 M with respect to hydrochloric acid is shown in Fig. 2. It can be seen that the sodium and calcium curves show a pronounced change of slope at a metal concentration equivalent to that of the phosphorus, corresponding to the formation of sodium dihydrogen orthophosphate (NaH,PO,) and calcium hydrogen orthophosphate (CaHPO,).Lithium does not show such a change until the metal con- centration is twice that of phosphorus, Le., corresponding to the formation of dilithium hydrogen orthophosphate (Li,HPO,) . The curve then levels off until the metal-to-phosphorus ratio (3: 1) corresponds to the fonnation of lithium orthophosphate (Li,PO,). A gradual decrease is then observed, with further increase in lithium concentration. Thus it would appear that the lithium dihydrogen orthophosphate is not as stable in the diffusion flame as the dilithium hydrogen orthophosphate. The following reaction presumably occurs- BLiH,PO, -+ Li,HPO, + H,PO,. It is a well known fact that lithium salts differ in behaviour from the corresponding sodium and potassium salts.Metal-to-phosphorus ratio Fig, 2. Effect of sodium, lithium and calcium on the phosphorus emission : curve A, varying lithium concentrations; curve B, vary- ing sodium concentrations; curve C, varying calcium concentrations At low calcium-to-phosphorus ratios (below 06), the calcium graph is linear and when extrapolated gives a ratio of 1 : 1. As this ratio increases above 0.5, the curve becomes rounded and extrapolation tends to give a calcium-to-phosphorus ratio of 1-5:l. Thus it would appear that in solutions containing a 2-fold, or more, molar excess of orthophosphoric acid over calcium, calcium hydrogen orthophosphate (CaHPO,) is formed, but in solutions containing greater amounts of calcium, calcium orthophosphate [Ca,(PO,),] is formed. However, acetic, hydrobromic, hydrochloric, nitric, oxalic, sulphuric and tartaric acids, when present in a 50-fold molar excess, produced no variation in response equal to, or greater than, +5 per cent.February, 19681 SPECTROSCOPY I N COOL FLAMES.PART I1 77 The presence of a 100-fold molar excess of the above acids, except for sulphuric acid, produced about a 10 per cent. decrease in signal because of the increased viscosity and attendant depression of nebulisation. Strong solutions of sulphuric acid produced an increase of signal caused by the intense blue S, emission. The presence of 10 per cent. v/v of organic solvents miscible with water produced with all about an 85 per cent. decrease in signal. This was almost identical with the effects that they produced With the S2 emission.The reason is not known at present, but the cause is most probably the quenching action of CH, CHO radicals, etc. Temperature measurements proved that the decrease in signal is not caused by any increase of temperature. Removal of cationic interferertces-It was considered that cationic interferences could generally be overcome by removal on a cation-exchange resin, in its hydrogen form. According to Samuelson,Q phosphate can be quantitatively separated on such a resin from the following cations: Li+, Na+, K+, NH,+, Rb+, Cs+, Mg2+, Caz+, Sr2+, Ba2+, Znw, MnZ+, Co2+, Ni2+, Cd2+, AP+ and Fes+. Alternatively, it should also be possible to remove phosphate on an anion-exchange resin and then regenerate it as phosphoric acid. By using such a method it might be possible to obtain a concentration step.Some preliminary measurements have been carried out with the cation-exchange resin Zeo-Karb 225. It was found that aluminium, calcium, iron and magnesium could be removed quantita- tively simply by shaking a few grams of the washed Zeo-Karb resin, in the hydrogen form, with about 30ml of solution containing orthophosphoric acid plus a 10-fold molar excess of the interfering ion. The interference caused by sodium and potassium was reduced by about 95 per cent., but this is presumably because Zeo-Karb 225 is not an efficient resin for univalent cations and neither is the batch-method of exchange. There is no reason to assume that most other cationic interferences cannot be removed by such methods. The separated air - hydrogen flame was also investigated, but it did not show any advantage in minimising interferences.In fact, the results tended to be inferior because of the higher temperature and the increased emission signals from easily excited elements. Elements and compounds with emission bands and lines close to 628 mp gave large apparent increases in signal. PREPARATION OF CALIBRATION GRAPH APPARATUS- A Unicam SP9OOA flame-emission /at omic-absorption spec t ropho t ometer , fitted with standard air - acetylene (rectangular) burner head and E.M.I. 9601B photomultiplier, was used. Fuel gas-Hydrogen, from a cylinder. Diluent gas-Nitrogen, from a cylinder. REAGENTS- M-Prepare by diluting AnalaR “syrupy” orthophosphoric acid (90 per cent. w/w), standardising and subsequently adjusting the solution to exactly lo-* M.Orthophosphoric acid, 1 ml of M orthophosphoric acid = 310.2 pg of phosphorus. PROCEDURE- Calibration graph for 3.1 to 31 P.9.m. of phosphorus-Transfer, by pipette, 1 to 10-ml aliquots of 1 0 - s ~ orthophosphoric acid into a series of 100-ml calibrated flasks and dilute to volume with distilled water. Nebulise the solutions into the nitrogen - hydrogen diffusion flame under the following experimental conditions : nitrogen pressure, 15 p.s.i. ; hydrogen flow-rate, about 15 cm on ,the dibutyl phthalate filled manometer; top of burner head, 1 cm below the bottom of the monochromator slit; slit width, 0-06mm; gain, 3.10; and band- width, 3. Measure the emission in the usual way at 528 mp and plot signal against concen- tration of phosphorus. Other concentration ranges (between 0-3 and 310 p.p.m.) can be prepared by suitable dilution.78 DAGNALL, THOMPSON AND WEST Linear calibration graphs over the ranges 3-1 to 31 p.p.m. and 31 to 620 p.p.m. of phosphorus have also been prepared in a similar manner, but in the presence of 0.1 M con- centrations of hydrochloric acid. In these instances, slit widths of 0.06 mm and 0.02 mm, and gains of 3-10 and 3.5, respectively, were used. We are grateful to I.C.I. Ltd. for the award of a grant to one of us (T.S.W.) for the purchase of the flame spectrometer used in this study, and to the Science Research Council for the award of a Research Studentship to K.C.T. Thanks are also given to Mr. R. 0. Walker of this Department for assistance with some of the experimental work. REFERENCES 1. Dagnall, R. M., Thompson, K. C., and West, T. S., Analyst, 1967, 92, 506. 2. Dippel, W. A., Bricker, C . E., and Furman, N. H., Analyt. Chem., 1954, 26, 553. 3. Brite, D. W., Ibid., 1955, 27, 1815. 4. Davis, A., Dinan, F. J., Lobbett, E. J., Chazin, J. D., and Tufts, L. E., Ibid., 1964, 36, 1066. 5. Brody, S. S., and Chaney, J. E., J. Gas Chromat., 1966, 4, 42. 6. Dagnall, R. M., Thompson, K. C., and West, T. S., Talanta, 1967,14, 1467. 7.--- , Ibid., in the press. 8. Maclkon, R., Analyst, 1964, 89, 745. 9. Samuelson, O., “Ion Exchangers in Analytical Chemistry,” John Wiley & Sons Inc., New York: NOTE-Reference 1 is to Part I of this series. Received August 22nd, 1967 Almquist & Wiksell, Stockholm; Chapman & Hall Ltd., London, 1953, p. 24,