Atomic absorption of arsenic in nitrogen (entrained air)-hydrogen

YAIR TALMI and CYRUS FELDMAN. 1975,13-34. Abstract | PDF ... Richard C. Chu , George P. Barron , and Paul A. W. Baumgarner. Analytical Chemistry 1972 ...
0 downloads 0 Views 709KB Size
Atomic Absorption of Arsenic in Nitrogen Air)-Hydrogen Flames Atsushi Ando,’ Masami Suzuki,2 Keiichiro Fuwa,8 and Bert L. Vallee Biophysics Research Laboratory, Department of Biological Chemistry, Harvard Medical School, and Dicision of Medical Biology, Peter Bent Brigham Hospital, Boston, Mass. Lines of As most suitable for atomic absorption are located below 2000 A, a region in which most conventional flames absorb radiation to a significant degree. Admixture of helium, argon, or nitrogen to air-hydrogen flames markedly reduces such flame absorption. A nitrogen (entrained airphydrogen flame proved most suitable for atomic absorption analysis of As. When used in conjunction with long-path-length Vycor cells, the sensitivity was 0.006 fig of As per ml (1% deflection). The effects of gas pressure, cell length, Flame conditions, cations, and anions on As absorption were investigated.

THE DETERMINATION of arsenic by atomic absorption spectrometry has presented dficulties, largely owing to the extremely high absorption of flame gases at wavelengths below 2000 A, the location of the most sensitive resonance lines of arsenic. Furthermore, it has been difficult to design hollow cathode tubes, adequate for analytical purposes, owing to the high volatility of the element, a technical problem which has been largely overcome now (1,2), Moreover, argon and nitrogen (entrained air)-hydrogen flames have been applied to atomic absorption analysis in the far-ultraviolet region, rendering flame absorption in this region feasible for analytical purposes (2-4). Hence it became possible to investigate and establish reproducible conditions for the determination of this element by atomic absorption spectroscopy. We have investigated the determination of arsenic by atomic absorption spectrometry, employing a long absorption cell (5). Using a new ring burner, mixtures of helium, nitrogen, and argon served as the medium for aspiration into the air-hydrogen flame. A number of parameters have been investigated which affect the determination of arsenic in this system, among them cation and anion interferences. EXPERIMENTAL

Apparatus. The absorption cell assembly for use in conjunction with the atomic absorption spectrometer is similar to that described previously (5). Light from the hollow cathode lamp was collimated by means of a spherical quartz lens (focal length 18 cm) passed through the absorption cell and ring burner and focused on the entrance slit on the monochromator.

On leave of absence from the Geological Survey of Japan. Present address, Department of Synthetic Chemistry, Faculty of Engineering, Nagoya University, Nagoya, Japan. 8 Present address, Department of Agricultural Chemistry, University of Tokyo, Tokyo, Japan. 1

2

(1) W. Slavin, “Atomic Absorption Spectroscopy,” Interscience, New York, 1968, pp. 80-1. (2) R. M. Dagnall, K. C. Thompson, and T. S. West, Talanfa,14, 551 (1967); Imperial College, London, personal communication

for arsenic data, 1967. (3) H.L. Kahn and J. E. Schallis, Perkin-Elmer Corp., Norwalk,

Sample Solution

Inert Gos

Ai I

T Hydrogen

0

IO mrn

c d

Figure 1. Burner assembly ARSENICHOLLOW CATHODE TUBES(Westinghouse Electric Corp., neon filled WL-22873 and WL-22836) were used at 20 ma. ABSORPTION CELLSconsisted either of Vycor tubing (Corning Glass Co., 743173, 96% silica glass) or Morganite recrystallized alumina (morganite, Triangle RR), each 91 cm long and with an inside diameter of 1 cm. Alumina tubing was employed in most experiments; Vycor tubing was substituted only when very high sensitivity was required. BURNER. The new platinum burner is shown in Figure 1. Hydrogen was fed through a portal supplying 24 small holes of the ring head. Air was delivered from a separate supply tube which entered into an outer mantle of the atomizer and, ultimately, the outer space of its concentric orifice. Sample solutions were aspirated pneumatically through the inner tube of the atomizer by means of helium, nitrogen, or argon to be delivered to its inner orifice or atomized into the airhydrogen flame. The flow rate of sample solutions was 2.0 i= 0.5 ml per minute at 15 psi (2.5 liters per minute) of nitrogen or argon, adjusted by changes of their gas pressures. The atomization of sample solutions with helium was inefficient. Commercial hydrogen, air, helium, nitrogen, and argon were employed throughout. The burner is centered at the entrance hole of the absorption cell, leaving a gap of 2 mm. Light from the hollow cathode lamp passes into the absorption tube through the space between the ring head and the atomizer. MONOCHROMATOR AND DETECTOR. A Zeiss M 4 Q I11 monochromator (Dispersion 8 A per mm at 2000 A cf/4.5), and a multiplier phototube (EM1 9526B) were used. The circuits for the high voltage power supply and amplifier have been described (5, 6). The output signal of the amplifier was displayed on a direct current microammeter with a range from 0 to 100 pa. Measurement of Temperature. The temperature inside the absorption cell was measured by both electrical and optical means. For temperatures below 1000 “C,a thermocouple probe (Pacific Transducer Corp. Model 326-CA), calibrated by a thermometer, was used. The line reversal method em-

Conn., personal communication, 1968.

(4) W. W. McGee and J. D. Winefordner, Anal. Chim.Acra, 37,

429 (1967).

(5) K. Fuwa and B. L. Vallee, ANAL.CHEM., 35, 942 (1963). 1974

28,175 (1956). (6) B. L. Vallee and M. Margoshes, ANAL.CHEM.,

ANALYTICAL CHEMISTRY, VOL. 4 1 , NO. 14, DECEMBER 1969

ploying the sodium D line serves for temperatures above 1000 “C (Leeds & Northrup 8622-C optical pyrometer and Wallace ocular spectroscope). Measurement of Absorption Intensities along Cell Path. A quartz tube (General Electric Co., Lamp Glass Department, Material 204) 91 cm long and with an inside diameter of 1 cm is placed perpendicularly across the axis of the light path. The light beam from the hollow cathode, measuring 5 mm in diameter, passes through the walls of the cell. Absorption is measured at right angles to the cell. Moving the absorption cell horizontally in front of the slit of the monochromator, the degree of atomic absorption was measured at any distance from the ring burner head while atomizing a solution of arsenic into the flame as it traverses the absorption cell (Figure 4). REAGENTS Arsenic Standards. A stock solution of 1000 pg of As per ml was prepared by first dissolving 0.132 gram of spectroscopically pure arsenic trioxide (Johnson Matthey Co.) in 2 ml of 1N sodium hydroxide. After slight acidification with hydrochloric acid, the solution was diluted to 100 ml. Dilute standard solutions were freshly prepared before use. For comparison, standard solutions of As were also prepared from sodium arsenate, Na2HAs04.7 H z 0 (Merck). Solutions of other metals were prepared from spectroscopically pure metals or salts (Johnson Matthey Go.). Organo-Arsenic Compounds. p-Aminobenzenearsonic acid and benzenearsonic acid containing 34.51 and 37.08% As, respectively (Eastman Organic Chemicals), served as certified standards and were dissolved in water. Water. Metal-free water was obtained by passing tap water through a mixed cation-anion exchange resin followed by glass distillation (7). PROCEDURE The microammeter which displays the output signal of the multiplier phototube was adjusted to zero after excluding light from the entrance slit of the monochromator. The signal from the hollow cathode tube, l o , was set to 80 on the microammeter while aspirating either water or the reference solution into the flame. Sample solutions were then aspirated and the deflection of the microammeter was recorded.

RESULTS Absorption Lines. Absorbance of a 1 pg per ml solution of arsenic, as reflected by absorption in the three resonance lines of the element located at 1890, 1937, and 1972 A are shown in Table I, as is the transmittance of radiation by the nitrogen (entrained air)-hydrogen flame path. The line at 1890 A has been reported to be the most sensitive one ( I ) . However, under our conditions the signal-noise ratio at this wavelength was both variable and unfavorable, owing to the absorption of water vapor. Detection of arsenic by using the line at 1937 A was not beset with similar difficulties and proved to be of greater practical value, yielding both the best stability and sensitivity, followed by that of the line at 1972

loo W

0

z a

75

1

EE

2 d I-

Helium

/

50

t-

Z

25 -

cc w

L

0 0

I

I

2

4

1

6

8

1

IO

HYDROGEN PRESSURE, psi

Figure 2. Effect of hydrogen pressure on absorption of radiation at As 1937 A by helium, nitrogen, and argon (entrained air)hydrogen flames when hydrogen pressure is varied Helium, nitrogen, and argon pressure 15 psi (2.5 I/min); air pressure 1 psi (0.2 l/min).

Cell. Morganite alumina, 1 cm i.d. X 91 cm long the air-hydrogen flame is unexpectedly high in long-path cells. At this wavelength and at hydrogen pressure of 6 psi (14 liters per minute), where maximal sensitivity for arsenic might be expected, the flame transmits only 1.5 to 3% of the total incident radiation, rendering this flame impractical for analytical purposes. However, as is apparent, the properties of helium, nitrogen, and argon (entrained air)-hydrogen flames overcome this problem. At all hydrogen pressures studied the transmittance in helium was highest, followed by nitrogen and argon in that order, increasing linearly with increasing hydrogen pressure (Figure 2). However, while the helium (entrained air)-hydrogen flame was most transparent at this wavelength, it was not stable. Furthermore, with this gas it was dficult to atomize sample solutions into the flame, perhaps because of some hydrodynamic properties of helium, limiting practical experimentation to nitrogen and argon (entrained air)-hydrogen. Oxygen is known to absorb radiation in the far ultraviolet region (9),suggesting the use of the lowest possible air-i.e., oxygen-flow rate. Flame absorption increases linearly with increasing air pressure, and the nitrogen (entrained air)(9) J. W. Robinson, “Atomic Absorption Spectroscopy,” Marcel Dekker, New York, 1966.

A.

Considerations in Selection of Flames. All flames and their combustion products absorb radiation, particularly at wavelengths below 2000 A, a phenomenon which is enhanced markedly in flames contained in long-path absorption cells. Down to 1950 A the air-hydrogen flame is m?re transparent than the air-acetylene flame, but at the 1937-A line the two flames are comparable (1,s). The absorption at 1937 A for (7) R. E. Thiers, “Trace Analysis,” J. H. Yoe and H. J. Koch, Jr.,

Eds., Wiley, New York, 1957, pp. 637-66. (8) J. E. Allan, Fourth Australian Spectroscopy Conference, August

1963.

Table I. Transmittance of Radiation and Absorbance Values (1 pg of As per MI) at Different Resonance Lines of As in a Nitrogen-(Entrained Air)-Hydrogen Flame Absorbance of Wavelength transmittance 1 pg/ml arsenic 1890 9.5 0.127 1937 0.217 42 1972 45 0.140 Nitrogen pressure 15 psi (2.5 l/min); hydrogen pressure 6 psi (14 limin); air pressure 1 psi (0.2 l/min) Cell. Morganite alumina, 1 cm id X 91 cm long

9

ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969

1975

1

0.3

8

I

Argon

0.2

z a m

U 0 UJ CD

a

0.1

I 0

4

2

8

6

IO

HYDROGEN PRESSURE, psi Figure 3. Effect of hydrogen pressure on absorption intensity of arsenic at 1937 A, in nitrogen and argon (entrained air)hydrogen flames

-

-

Nitrogen and argon pressure 15 psi (2.5 l/min); air pressure 1 psi (0.2 l/min); hydrogen pressure 2 10 psi (8 20 I/min). Sample flow rate* 2.0 ml/min. Cell. Morganite alumina 1 cm i d . X 91 cm long

hydrogen and argon (entrained air)-hydrogen flames remain stable even when the air pressure gauge reads 0 psi. In actuality, air is entrained from the bottom hole of the ring burner and sucked into the absorption cell, and this small quantity is sufficient to allow for combustion in this system. At this wavelength virtually no difference in transmittance between air pressures of 0 to 1 psi (0.2 liter per minute) could be observed. Of course, nitrogen and argon also absorb radiation in the far-ultraviolet region. Flame absorption increases linearly with an increase of these gases. At low both and high nitrogen or argon pressures, the flames were unstable, but they proved steady at 12 and 17 psi (2.0 lo 2.8 liters per minute).

The sensitivity of arsenic detection as a function of hydrogen pressure is shown in Figure 3. In both argon and nitrogen absorbance was greatest at about 6 psi (14 liters per minute) of hydrogen and below 4 psi (0.65 liter per minute) of air. To compensate for the increased absorption at this wavelength in argon (entrained air)-hydrogen over that in nitrogen (entrained air)-hydrogen flames, using the same slit width, a high gain sensitivity was employed for the former, which resulted in minimal differences of sensitivities discernible in these two flames. The distribution of arsenic absorption intensities along the cell path is shown in Figure 4. The integrated areas under the curves correspond to the total absorbance of radiation in the entire cell. At 6 psi (14 liters per minute) of hydrogen pressure, the sensitivity is optimal, in good agreement with the results of Figure 3. Absorption markedly decreases beyond 91 cm, obviating the use of longer path cells. Table I1 shows the effect of spectral slit width on arsenic absorption. From the above experimental data, working conditions can be summarized which yield the highest sensitivity of detection of arsenic while holding flame absorption in a long cell path to the lowest level as follows: Air pressure Hydrogen pressure Nitrogen and argon pressure Length of cell Slit width

0-1 psi (0-0.2 ljmin) 6 psi (14 ljmin) 12-16 psi (2-2.8 ljmin) 91 cm 0 . 1 mm (spectral slit width 1.2A)

Admission of a small amount of air to the flame, in addition to that entrained by the burner, changes the flame conditions only minimally, but has the practical advantage of facilitating ignition of the flame. Flame Temperature. The temperatures along the absorption cell have been measured in nitrogen and argon (entrained air)-hydrogen flames while aspirating distilled water. The temperatures of both flames were nearly constant between 280' and 240 "Calong almost the entire length of the cell, except near the head of the burner. Three centimeters from the head of the burner the flame temperature was 1400 "C

.. 0

20

40

CELL

60

80

100

LENGTH, c m

Figure 4. Atomic absorption of arsenic along the path, in nitrogen and argon (entrained air)-hydrogen flames at 1937 A Nitrogen and argon pressure 15 psi (2.5 l/min); air pressure 1 psi (0.2 l/min); hydrogen pressure 4,6,8,10 psi (11,14,17,20 l/min) Cell. Quartz, 1 cm i d . X 91 cm long 1976

0

ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969

Table 11. Effect of Spectral Slit Width on As Absorption" Relative absorbance Mechanical Spectral slit Absorbance of (max. value slit width, mm width, 8, 1 pg Asjml = loo) 0.6 0.217 100 0.05 1.2 0.217 100 0.1 0.195 90 0.25 3 0.154 71 0.5 6 0. I30 60 1 .O 12 0.098 45 2.0 24 a As 1937 A. Experimental conditions, same as Table I.

0.7 0.6

w 0.5 0

z a 0.4 Q

cn

m 4

Table 111. Arsenic Sensitivity Using Nitrogen-(Entrained Air)Hydrogen Flame in Absorption Cell Absorbance cella Material Length, Wavetength, Sensitivity, cm A pg/m1/1 Z Vycor 91 1937 0.006 91 1972 0.01 Morganite 91 1937 0.02 60 1937 0.023 Nitrogen pressure 15 psi (2.5 l/min); hydrogen pressure 6 psi (14 l/min); air pressure 1 psi (0.2 l/min). a 1 cm. i.d.

0.3 0.2

0.1

0.1

0.25

0.5

Arsenic,

0.75

1.0

p g h l

Figure 5. Working curves for arsenic I. 1937 6 ; Vycor, 1 sin i.d. X 91 cm long 11. 1972 A; Vycor, 1 cm i.d. X 91 cm long 111. 1937 A; morganite alumina, 1cm i.d. X 91 em long IV. 1937 A; morganite alumina, 1 cm i.d. X 60 cm long

as measured by the optical reversal method, decreasing sharply to become constant at the level cited along the path. Absorption Cell. The effect of different types of cells on arsenic absorption is shown in Figure 5. Sensitivity was greatest in Vycor tubing, probably due to reflections of radiation from the inner surface of the cell (5); but as the inner surface of the tube becomes progressively coated with salts sprayed into the flame, the sensitivity changes, requiring frequent cleaning of this type of tubing. Morganite alumina tubes yielded less sensitivity but more stable signals. Sensitivity is a function of length of the absorption cell, as demonstrated by morganite cells, 60 and 91 cm in length, respectively. U p to the latter value sensitivity remains a function of cell length but longer cells do not increase sensitivity further. Sensitivity and Repeatability. Sensitivities of arsenic as a function of different cell lengths and wavelengths are shown in Table 111. With 91-cm-length Vycor and morganite tubing, the sensitivity values were 0.006 and 0.02 pg of As per ml, respectively, and the range for effective quantitative work was between 0.1 and 1.0 pg of As per ml of solution. Five replicate determinations of three different arsenic concentrations were performed to ascertain the repeatability of the method (Table IV). The coefficients of variation varied from 2 to 6z3 the latter at the lowest concentration tested. Arsenite and arsenate solutions containing the same arsenic concentration resulted in the same absorption. Accuracy. The accuracy for organo-arsenic standards was ascertained by analysis of organo-arsenic compounds; 99 and 100% of the known arsenic content of p-aminoben-

Table IV. Repeatability of Arsenic Standards Used for Flame Absorption Working Curve As, pg/ml HCl, M log ZojZ & S.DUa c. V.,% 0.10 ... 0.061 f 0.003 5 .,. 0.356 f 0.012 3 0.50 1.o ... 0.713 =!= 0.016 2 0.1 0.053 =!= 0.003 6 0.10 0.228 & 0.004 3 0.50 'L 0.541 i 0.013 2 1.o Nitrogen pressure 15 psi (2.5 I/min); hydrogen pressure 6 psi (14 limin); air pressure 1 psi (0.2 l/min) Cell. Vycor, 1 em id. X 91 cm long a Five determinations, As 1937 A.

zenearsonic acid and of benzenearsonic acid, respectively, were found. Interferences. Table V shows the effect of acids, buffers, and salts on arsenic absorption. All acids tested a: well as ammonia affected absorption of arsenic at 1937 A. The absorption due to ammonia was highest, probably because of the absorption by both OH and NH1. Among all of the acids studied hydrochloric acid had the least effect on absorption at this wavelength, rendering it a suitable diluent. Interferences by various metals were tested also. Magnesium, calcium, aluminum, nickel, cobalt, and manganese reduce arsenic absorption (Table VI), attributable perhaps to the formation of arsenides in the flames. To test this hypothesis absorption intensity changes of magnesium, calcium, nickel, and cobalt were measured in the presence of arsenous acid, and the consequent decrease of absorbance of these metals tends to confirm the formation of arsenides in the flame as the basis of their effects. The effect of iron on arsenic absorption was complicated: Increasing concentrations of iron first increased and finally decreased arsenic absorption. The measurement of arsenic thus suffers from multiple interferences; acid strength, concentration of salts, and elements such as nickel, cobalt, and aluminum all can affect it. Hence, careful control is required in the preparation of both standards and sample solutions.

ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969

1977

Table V. Effect of Acids, Buffers, and Salts on Arsenic Absorption” As, pglml 1 .O

Added M HNOa ’ ‘ 0 . 1 0.2 HCIOd 0.1 0.2 0.1 HC1

0.5

Absorbance 0.217 0.169 0.186 0.182 0.255 0.182 0. I67 0.126 0.382 0.597 0.118 0.079 0.121 0.125 0.118

Recovery,

HOAC‘ ’0.085 Tartaric acid 0.05 H.204 0.02 EDTA 0.001 Tris 0.005 0.155 0.01 0.074 0.5W 0.111 TCA a Flame condition same as Table IV. Cell. Morganite alumina, 1 cm i.d. X 91 cm long.

Table VI. Effect of Diverse Metals on Arsenic Absorption” Diverse metals added, Recovery, As, pg/ml m/ml Absorbance 7i 0.5 0.118 100 Na ’30 0.118 100 K 77 0.125 106 Li 100 0.125 106 Ba 10 0.118 100 Ca 0.5 0.093 79 Mg 1 . 8 0.107 91 Hg 50 0.118 100 Pb 79 0.116 98 Cu 50 0.118 100 c u 120 0.087 74 Zn 49 0.118 100 Mn 10 0.100 85 co 2 0.080 68 Ni 2 0.044 37 A1 1.6 0.040 34 Cr 10 0.130 110 Fe 3.2 0.142 120 8 0.173 147 0.185 157 16 24 0.165 139 55 0.163 138 V 40 0. 118 100 a Flame condition same as Table IV. Cell. Morganite alumina 1 em id. X 91 cm long

DISCUSSION

Because of difficulties in spectral detection described, methods for the determination of arsenic by atomic absorption were unavailable until 1963, when Allan (8) found strong absorption of all three resonance lines in the region below 2000 A and Willis (10) described a procedure for arsenic determination by atomic absorption spectrometry. Earlier work utilized an air (oxygen)-acetylene flame with sensitivities of about 3 to 5 pg per ml at 1937 8, and 1 pg per ml at 1890 8, P,lI). (10) J. B. Willis, “Methods of Biochemical Analysis,” D. Glick, Ed,, Vol. XI, Interscience, New York, 1963. (11) W. T. Elwell and J. A. F. Gidley, “Atomic Absorption Spectrophotometry,” 2nd ed., Pergamon Press, New York, 1966. 1978

e

Absorbance (for added materials only)

100

78 86 84 118 84 77 58

176 275

(FINO; ’ ’ 0.090) (HNOa

(HClO4 (HClOe (HCI (HCI (HCl (NH40H (NHdOH

0.140 0.052) 0.1 11) 0.011)

0.017) 0.022) 0.196) 0.372)

100

67 103 106 100 131 63 94

Non Non Non Non Non (&SO4 Non

0.010)

Improvements in the design of hollow cathodes ( 1 ) employing a mixture of arsenic, gold, and lead in proper proportion resulted in high light emission, relative long life spans, and sensitivities with air-acetylene flames of 1,2, and 3 pg per ml at 1890, 1937, and 1972 A, respectively. Recently, a sensitivity value of about 1.5 pg per ml at 1972 A was reported with an oxyacetylene flame (12). Massmann (13) investigated the use of a heated graphite tube, employing an argon chamber as an absorption cell at 1890 and 1972 A, detecting as little as 0.1 pg per ml. Using a propa?e-butane burner, sensitivity of 10 pg per ml of arsenic at 1972 A has been reported (14). Dagnall, Thompson, and West (2) employed a microwaveexcited electrodeless discharge tube as a spectral source for arsenic determination. When compared with hollow cathode tubes, very high spectral intensities were obtained at the resonance lines. They also applied a nitrogen (entrained air)hydrogen flame of low temperature and low flame absorption with resulting improvement in limits of detection in the far ultraviolet region. Using a triple slot burner, the sensitivity was increased to 0.75 pg per ml. Similarly an argon (entrained air)-hydrogen flame for arsenic determination has been investigated (3). In the far-ultraviolet region, flame absorption decreases, becoming about one fourth that of the air-acetylene flame at 1937 A and resulting in a sensitivity of 0.21 pg per ml at this wavelength. Recently, a highly sensitive sampling boat technique for atomic absorption spectrometry has been reported with a limit of detection for arsenic of 0.02 pg per ml(Z5). In the present study the introduction of long-path-length cells has greatly increased the sensitivity for arsenic. Using 91-cm length Vycor and morganite tubing and a nitrogen (entrained air)-hydrogen flame, sensitivities were 0.006 and 0.02 pg of As per ml. Sensitivities with an argon (entrained air)hydrogen flame were similar.

(12) K. E. Smith and C. W. Frank, Appl. Spectry., 22, 765 (1968). (13) H. Massmann, 2.Anal. C/zem., 225, 203 (1967). (14) A. S. Bazhov,Zaaudsk.Lab., 33, (9), 1096 (1967). (15) H. L. Kahn and G. E. Peterson, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, March 1968.

ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969

The helium, nitrogen, and argon (entrained air)-hydrogen flames are colorless, transparent, and invisible inside the absorption cell, and are fairly stable at the proper ratio of gas mixtures. The flame absorption in the far-ultraviolet region is decreased substantially by adding a n inert gas to the primary air-hydrogen mixture. Inert gases do not participate directly in the flame reaction, and this decreased absorption of the flames seems to be caused primarily by dilution but depends on the nature of the inert gas added. Flame absorption decreases from helium, nitrogen, to argon (entrained air)hydrogen, the helium (entrained air)-hydrogen flame being the most transparent at this wavelength. In our experiments difficulty was encountered atomizing the sample solution with helium, forcing us to abandon this gas, though further studies along these lines are indicated. With nitrogen and argon (entrained air)-hydrogen flames, the sensitivity for arsenic was nearly equal, though the nitrogen (entrained air)-hydrogen flame is more transparent than the argon (entrained air)-hydrogen flame. We have chosen nitrogen, however, because of its lower cost and better transparency, allowing the choice of narrow slits. Hydrogen pressure also affects flame absorption markedly. In the helium, nitrogen, and argon (entrained air)-hydrogen flames, flame absorption linearly decreased with increasing hydrogen pressures, probably through the rapid decay of OH in the presence of excess hydrogen (26, 17) which has a relatively low absorption at this wavelength. The presence of inert gases also decreases flame temperature and flame velocity (18). Dagnall et al. (19) measured the temperature of the nitrogen (entrained air)-hydrogen flame using a probe in air while aspirating distilled water, and from 1 to 6 cm above the burner head relatively low temperature (16) A. 6.Gaydon, “The Spectroscopy of Flames,” Wiley, New York, 1957, p. 83. (17) I. Rubeska and B. Moldan, Analysf, 93,148 (1968). (18) F. Pungor, “Flame Photometry Theory,” Van Nostrand, London, 1967, p. 43. (19) R. M. Dagnall, K. C . Thompson, and T. S. West, Analyst, 92, 506 (1967).

readings from 280 to 480” C were recorded in the center part of the flame. In the present study, the temperatures of nitrogen and argon (entrained air)-hydrogen flames were low-Le., from 280’ to 240” C-over almost the entire path length of the 91-cm absorption cell. Only a small quantity of air is entrained from the bottom hole, and the air supply tube of the burner constitutes a diluted, oxygen-poor combustion system with substantially decreased thermal energy. Maximum sensitivity for arsenic is achieved at relatively high hydrogen pressures-i.e., 6 psi (14 liters per minjproviding a strong reducing medium and preventing oxide formation of arsenic. But even in this very strong reducing environment, the flame itself remains an oxidation system, and dissociated neutral atoms of arsenic combine with oxygen to form white arsenic oxide deposits on the inside wall of the cell. While Vycor tubing yields higher sensitivity, such arsenic oxide deposits seriously alter the reflectivity of the inner wall. Though the sensitivity which can be achieved with morganite tubing is intrinsically less, it is free from interference of wall deposits, making such cells preferable for routine analysis of arsenic. Some acids, ammonia, and metallic elements seriously depress arsenic detection. The influence of ammonia is attributed to absorption interference of the NHs band spectrum. In contrast, the interference of many metallic elements is attributed to the formation of arsenides in the flames. Low temperatures are thought to give rise to significant concentration of undissociated arsenides, seriously depressing absorption for some elements (Ni, Co, Al, etc.), and this effect is remarkably high. Higher flame temperatures are desirable to destroy undissociated arsenides. Since higher flame temperatures, however, render the flame gases more opaque, care must be taken either to balance these variables to eliminate interferences as much as possible or to cancel them by means of proper working standards.

RECEIVED for review May 9, 1969. Accepted September 15, 1969. Work supported by Grants-in-Aid HE-07927 and GM-15003 from the National Institutes of Health, Department of Health, Education, and Welfare.

ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969

0

1979