Atomization of Molybdenum in Air-Acetylene and Nitrous Oxide-Acetylene Flames R. E. Sturgeon and C. L. Chakrabarti” Department of Chemistry, Carleton University, Ottawa, Ontario, Canada K 7.S 586
The effects of flame stoichiometry, nature of the metal complex or compound, and the solvent on the atomic absorption signal of molybdenum have been investigated. Spatial distributions (profiles) of molybdenum free-atom populations are presented. These profiles were obtained by using various molybdenum compounds or complexes dissolved in aqueous or organic solvents and nebulizing them into unshielded, air-acetylene and nitrous oxide-acetylene flames of various stoichiometries. The highest free-atom population density (maximum absorbance) was localized in a rather small area of the flame and was obtained only in a highly reducing nitrous oxide-acetylene flame by using organomolybdenum compounds. However, the chemical nature of the organometallic compounds is also important in determining the sensitivity. Equally important was the particular compound or complex-solvent combination employed. This latter aspect was expressed in terms of specific “enrichment factors” for the compound or complex-solvent systems studied.
pends only on the temperature as well as on the kind of species involved. The dissociation of a diatomic molecule, MX, can be represented as follows: MX=M+X
K, =
pm PMX
where M is a metal atom, X represents C1, 0, OH, H , C, etc., K , is the equilibrium constant and p represents the partial pressure of the components (in atmospheres). Equation 1 can be rearranged to give
&=Kg
(3)
PX Le., the ratio of the free and combined atoms of the metal depends solely on the equilibrium constant and the partial pressure of the atom X. The presence of gaseous monoxides in carbon-rich flames may be accounted for by various equilibria such as: PMX
M+O=MO The remarkable enhancement in sensitivity exhibited by many elements in highly reducing flames has been documented in the literature (1-8). In addition to the particular flame (whether air-CzH2 or N20-C2H2), flame stoichiometry and the chemical nature of the analyte-compound (whether simple or complex salt, oxygen-bonded or organometallic) often play a decisive role in the formation of atoms and in determining the degree of atomization (9-12). The enhancement in sensitivity given by high-temperature, highly-reducing flames is particularly remarkable for the analyte which forms involatile solids and/or stable gaseous oxides. Molybdenum forms three stable gaseous oxides having dissociation energies, in kcal mole-’ as follows ( 1 3 ) : Moo3 (411 f 7 ) , Moo2 (162 f 10) and MOO (116 & 15). Hence, it is expected that for molybdenum, the particular flame, the flame stoichiometry, and the analyte compound or complex-solvent system will have large effects on the degree of atomization. This study was undertaken to determine the effect of these variables on the degree of atomization of molybdenum. Rasmuson et al. ( 1 4 ) have suggested that nitrous oxideacetylene flames having N20/C2H2 flow ratios less than 2 contain more reactant carbon than the N20 can oxidize to CO, and, accordingly, it is logical to identify these flames as carbon-rich. In this paper, a carbon-rich flame is defined as a flame in which the C/O ratio 21. The above two definitions describe the same C/O ratio for a carbon-rich flame but the definition based upon C/O ratio is operationally simpler in the present case where the organic solvents introduced into the flame function as supplementary fuels and may contribute both C and 0 to the flame gas composition. In a flame having a particular stoichiometry and temperature, the stability of gaseous oxides often becomes the limiting factor in the production of free atoms. The equilibrium constant for the dissociation of gaseous oxides de-
(2)
M M
+ OH
MO
(4)
+H
+ H20 + M O + H2
(5) (6)
The equilibrium partial pressures reported by Rasmuson et al. (14) for a nitrous oxide-acetylene flame having a N20/C2H2 flow ratio of about 1.9 are: 0 = 1 X 10-5 atm atm, Le., p o is about four orders of and OH = 1 X magnitude greater than P O H . A t the much lower flow rates used in this study ( < l )the , partial pressures of both 0 and OH are likely to be even smaller. Therefore, formation of only oxides need be considered, that of hydroxides can be ignored. Since in the carbon-rich flame used, the partial pressure of 0 is extremely small (calculated to be 4.2 X l o p 9 atm for a N20/C2H2 flow ratio of 1.65-ref. ( 4 ) ) ,formation of only monoxide need be considered, and that of higher oxides ignored (15, p 129). Considering only Equation 4 and the dissociation of MO, one can express the partial pressure, P M , of free metal atoms: (7) The partial pressure of free metal atoms may be expressed in terms of their number density, N, through the relation p~ = N k T , where k = Boltzmann constant and T is the absolute temperature. Thus, from Equation 7, N , the number of free metal atoms, may be expressed as follows
If the experimental conditions are such that the absorbance is directly proportional to N, then
Thus, for metals which form gaseous monoxides in the ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
677
flame, the absorbance may be expected to vary directly with the partial pressure of the gaseous monoxides and inversely with that of the atomic oxygen-a highly oxygendeficient environment being a requirement for the existence of free atoms of those elements which form stable gaseous monoxides. The interconal zone of highly-reduced flames provides such an environment where relatively high concentrations of carbon-containing species, as well as atomic and molecular hydrogen, drastically reduce the oxygen concentration (16). The possibility of metal monoxide formation is thus greatly reduced and, in addition, reduction of the oxide is highly favored thermodynamically. The reaction MO + C CO M (where C = carbon or carbon containing species) is highly exothermic as the heat of formation of CO (256.7 kcal mol-’) is greater than that of the most stable metal monoxide (less than 200 kcal mol-I). Around 2500 K, carbon is thermodynamically capable of reducing all metal monoxides (17, 18). I t is therefore evident that the role of a flame as a reservoir of metal atoms is a function of its reducing environment as well as its temperature, and a flame cannot be regarded simply as a thermal dissociation medium (19-22). The events of (a) evaporation of the solvent from aerosol droplets, (b) vaporization of residual salt particles or their decomposition or reaction products, and (c) the dissociation of these residual species, each require a finite time. All of these processes are kinetically controlled, and some may not reach completion. The position in the flame where a maximum concentration of free atoms occurs reflects the balance between the increase of free atom population with height (time) and the decrease of this population by other processes. This position is therefore dependent upon the rise velocity of the flame, the size distribution of the aerosol droplets, the rate of evaporation, vaporization, dissociation, and decrease in the concentration of free metal atoms through recombination, diffusion, and dilution. Since under proper experimental conditions, absorbance is directly proportional to the concentration of the absorbing species, the effect of the above processes on the spatial distribution of free analyte atoms in the flame can be studied by the use of flame profiles (23, 24). These profiles are absorbance contour maps showing the concentration distribution in two dimensions in the flame. The ordinate and abscissa give the coordinates of the point in the flame in a plane a t right angles to the optical axis. Each contour line, passing through points of equal absorbance, is plotted with respect to both the height from the burner top (the vertical traverse), and the horizontal displacement from the center of the flame with the slot of the burner aligned along the optical axis (the horizontal traverse). The contours show the absorbance increases from the outermost to the innermost contour line with the maximum in the center. Only relative, not absolute concentrations of atomic species are represented in such profiles, but this does not detract from their usefulness in this study.
-
+
EXPERIMENTAL Apparatus. A Jarrell-Ash, model 82-500 atomic absorption and emission spectrophotometer with a 0.50-m focal length Ebert grating monochromator (ruled area 52 X 52 mm, 1180 linedmm, blazed a t 500.0 nm) was used. Fixed entrance and exit slits of 25 p corresponding to a calculated spectral band pass of 0.04 n m (first order) were used. T h e spectrophotometer was fitted with a Varian Techtron laminar-flow burner assembly, a Beckman 10-in. recorder (model 1005) and a Westinghouse single-element hollow-cathode molybdenum lamp. Varian Techtron burner heads (6-cm slot NzO-CzHg and 10-cm slot air-CzHz, model A B 51) supported the unshielded flames. A metal plate containing a circular aperture of 1-mm diameter was rigidly clamped t o the optical rail between t h e burner and the convex lens nearest to the monochromator. This 678
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4 , APRIL 1976
aperture, used throughout this study, selected t h e area of the emerging beam of t h e incident radiation over which the absorbance, emission, or t h e temperature was measured. Temperature profiles were obtained with a Varian Techtron model AA-4 spectrophotometer using t h e same burner systems as above, and a 1000-watt xenon-arc continuum source (model 976C-1, Canrad Precision Ind. Inc., Hanovia L a m p Division). These profiles could not be obtained with t h e Jarrell-Ash spectrophotometer system because of the enclosed nature of its housing for t h e atomizerburner. T h e flame was traversed horizontally and vertically, relative t o the selected beam of the incident radiation, by rotating t h e micrometer screws on the burner mounting. A Varian Techtron model AA-5 atomic absorption spectrophotometer with t h e burner unit replaced with a Carbon Rod Atomizer, model 63 (Varian Techtron Pty. Ltd.), was employed for t h e solvent extraction studies. T e s t solutions of 5-pl volume were pipetted with an Eppendorf syringe, fitted with disposable plastic tips. A tungsten continuum source (G.E. 30 AT 24/13, 3.5 V, uv spectrum lamp) was used for the correction of absorption due t o molecular bands and scattering of t h e primary source radiation. Reagents. Reagents used were of analytical reagent grade: ammonium molybdate (Anachemia, 81-83%), molybdenum hexacarbonyl (Alpha Inorganics), molybdenum cyclopentadienyl tricarbonyl dimer (Alpha Inorganics), t o be denoted [MoCp(CO)s]n,molybdenum trioxide (Allied Chemical 99.5%), cupferron (Mallinckrodt Chemicals), and indium tris(2-ethyl hexanoate) (Eastman Kodak Co.). Other reagents included chemically pure acids, bases, “Spectrograde” organic solvents, and freshly distilled, deionized water. T h e cupferron was purified and recrystallized according t o t h e procedure outlined by Kemp (25). Molybdenum cyclopentadienyl tricarbonyl was purified as prescribed by Eisch and King (26), and the percentage of Mo was determined by neutron activation analysis t o be 95 f 3% (using Moo3 as a standard). T h e hexacarbonyl compound was purified by passing it through a column packed with activated silica gel (grade 923, Davison Chemical) and eluting it with ethyl ether. T h e percentage of molybdenum in t h e recrystallized product was determined by neutron activation analysis t o be 96 f 3%. Procedure. T h e effect of varying the Mo compound or complex, the solvent, the flame type, and stoichiometry on the sensitivity and the absorbance profile of molybdenum was studied. Using a modified version of the computer program developed by Chakrabarti e t al. (27), the absorbance profiles were plotted with a computer-plotter, thus eliminating t h e laborious and time-consuming hand-plotting technique. Following the method of Zaugg (28), a m monium phosphomolybdate was extracted into each of the following organic solvents: methyl isobutyl ketone (MIBK), n-butyl acetate, and 2-octanol. T h e concentration of molybdenum in these organic extracts was determined with the Carbon Rod Atomizer 63 and also with t h e Jarrell-Ash apparatus with an air-acetylene flame. Solutions containing 1000 l g l m l of molybdenum as t h e hexacarbonyl were prepared in MIBK, n-butyl acetate, and 2-octanol. Solutions containing 1000 pg/ml of molybdenum as molybdenum cyclopentadienyl tricarbonyl could be prepared only in MIBK and n-butyl acetate. In all cases, the test solutions were used immediately after they had been prepared, by appropriate dilution. I t was observed t h a t [MoCp(CO)3]2was unstable and underwent photodecomposition on exposure t o fluorescent light. Molybdenum was also extracted as a cupferrate complex from a cold aqueous solution of M o o s adjusted to a p H of 1. T h e percentage extraction was determined by carrying out successive extractions on t h e same sample until t h e n e t absorbance obtained for t h e last extract was less than 0.0044. T h e percentage of Mo removed in t h e first extraction (single extraction was used throughout this study) was calculated from the ratio of the absorbance of the first t o the sum total of all extracts. T h e amount of molybdenum left in t h e aqueous phase was also checked. Aqueous solutions containing 1000 pg/ml of Mo were prepared by dissolving Moos in a 40% solution of aqueous ammonia. Absorbance profiles in an air-acetylene flame for each of these compound (or complex)-solvent systems were plotted using optimal experimental conditoons. A nitrous oxide-acetylene flame was used only with solutions of Mo(CO)6, [MoCp(CO)s]*,and MoOx. Temperature profiles were plotted for t h e air-acetylene and nitrous oxide-acetylene flames using the Varian Techtron model AA-4 spectrophotometer. Both flames were fed with a solution of either indium tris(2-ethyl hexanoate) in MIBK or an aqueous solution of indium nitrate. T h e temperature was determined by t h e two-line absorption method of Browner and Winefordner (29).
Table I. Sensitivities of Compound- or Complex-Solvent Systems Air-acetylene flame0
C,H, flow rate, Compound/complex
Ammoniumphosphomoly bdate Mo( co ) 4 [MOCP(CO),1 2 Mo-cupferrate Ammoniumphosphomolybdate Mo(CO), [MoCp(CO), 1 Mo-cupferrate Ammoniumphosphomoly bdate Mo( CO 1 6 Mo-cupferrate MOO, a Air flow rate = 7 . 5 l./min; cate analyses.
Solvent
Sensitivityb,
Nitrous oxide-acetylene flame0 Sensitivityb, p g / m l / l % absorption
C,H, flow rate, I./min
I./min
p g / m l / l % absorption
MIBK MIBK MIBK MIBK
1.85 1.85 1.85 1.85
0.54 I0.03 0.27 = 0.01 0.48 t 0.02 0.53 i 0.03
4.50 4.50
not used in this study 0.052 z 0.002 0.100 i 0.004 not used in this study
n-Butyl acetate n-Butyl acetate n-Butyl acetate n-Butyl acetate
2.30 2.30 2.30 2.30
0.45 i 0.04 0.22 i 0.01 0.52 i 0 . 0 2 0.54 i 0.03
4.50 4.50
not used in this study 0.055 L 0.002 0.130 c 0.004 not used in this study
2.70 2-Octanol 2.10 2-Octanol 2.70 2-Octanol Aqueous ammoniia 2.70 N,O flow rate = 4.5 l./min. b The
not used in this study 0.40 2 0.03 5.20 0.062 i 0.002 0.11 i 0.01 not used in this study 0.66 i 0.03 6.50 0.55 0.01 0.63 0.01 i variation is the standard deviation calculated from 9 repli+_
+_
Table 11. Molybdenum Compound or Complex-Solvent Enrichment Factors Enrichment factor Analyte
Q
Solvent
.4i-C,HZa
AmmoniumMIBK phosphomoly bdate MIBK Mo(CO), MIBK [MOCP(CO),I * MIBK Mo-cupferrate Ammoniumn-Butyl acetate phosphomol y bdate n-Butyl acetate Mo(CO), n-Butyl acetate EMoCP(CO), I , n-Butyl acetate Mo-cupferrate Ammonium2-Octan01 phosphomoly bdate 2-Octanol Mo(CO), Mo-cupferrate 2-Octanol MOO, Aqueous ammonia Flow rates of the fuel and the oxidant as reported in Table I, and those
Sensitivities for each of the above systems were determined using the Ma 313.2-nm resonance line and the optimum experimental conditions of lamp current, flame stoichiometry, and flame coordinates. In addition to the percentage consumption of the solvents, their aspiration rates were also measured. The absorbance given by the test solution entering the flame and that given by the arrested portion of the test solution draining from the nebulizerburner-chamber were measured, and the gain or loss of the analyte from the test solution inside the nebulizer-burner-chamber was noted. Since under the flame conditions employed in these studies, which are similar to those used by West et al. (301,the degree of ionization is small (0.3%),no ionization suppressant was used, nor was any correction made for ionization. Absorbance profiles are most conveniently constructed using the experimental arrangement described earlier. When a wide beam of light is employed, one obtains an average of the absorbance over the cross-sectional area of the beam. As a result, an absorbance value somewhat less than the maximum absorbance is obtained. This type of measurement is referred to as a “large-area” absorbance technique. In the case of most elements, by sampling only the smaller regions containing the highest free atom population with the use of the aperture described earlier, a higher absorbance is obtained. This is termed the “small-area’’absorbance technique. Differences in the absorbance observed with these two techniques are especially marked for those atomic population distributions which show a small region of high absorbance.
RESULTS AND DISCUSSION In agreement with Rann and Hambly (23) and Chakrabarti e t al. (241, the “small-area” absorbance technique for
N,0-C2H2a
1.6 1.0 1.3 1.4
not used in this study 1.2 1.2 not used in this study
1.7
not used in this study 1.o
1.1 1.8 1.1
1.1
not used in this study
2.9 not used in this study 3.4 2.5 not used in this study 1.0 0.87 0.94 of the solvents as reported in Table IV.
measuring the molybdenum free atom population in the flames gave greater absorbance values than those given by the “large-area” absorbance technique; for example, a 20% increase in t h e absorbance values was obtained for the Mo(CO)6 and [MoCp(CO)s]zsolutions. T h e sensitivities obtained for the various compound/ complex-solvent systems are presented in Table I. T h e flow rates of acetylene and oxidant gases employed provided flames with C/O ratios >1. If the contributions of solvents (given later) as additional fuels are also taken into consideration, t h e flames were highly carbon-rich even after making allowances for the contribution of the ambie n t air to their oxygen content. The following generalizations may be drawn from Table I. a) Compared to the sensitivities in the air-acetylene flames, the sensitivities in the nitrous oxide-acetylene flames are higher by factors of 4-5. b) T h e sensitivities are generally higher with t h e organomolybdenum compounds than the inorganic compounds or compounds containing metal-oxygen bonds. c) T h e sensitivities obtained for a n aqueous solution of Moo3 and ammonium phosphomolybdate and the molybdenum cupferrate in MIBK are similar if allowances are made for the solvent effects. All these inorganic compounds or complexes contain metal-oxygen bonds. d ) T h e most sensitive system studied is Mo(CO)F, in ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
679
Table 111. Normalized Solvent Effects
C,H, flow rate, l./min
Solvent
Air-C,H,
flamea
Solvent rate, ml/min
Solvent consumed, %
flow
N,O-C,H,
nb
MIBK 1.85 2.6 51 n-Butyl acetate 2.30 2.4 43 2-Octanol 2.70 1.8 13 40 % ( v / v ) aqueous 2.70 3.0 13 ammonia aAir flow rate = 7.5 l./min; N,O flow rate = 4.5 l./min. b n is the solvent consumed).
C,H, flow rate, I./min
Solvent flow rate, ml/min
4.50 4.50 5.20 6.50
5.2 5.0 2.5 5.0
5.7 4.4 1.0
1.7
flame0 Solvent consumed, %
nb
38 33
7.9 6.6
10 10
1.0
5.0
normalized product of the (solvent flow rate) x (%
Table IV. Normalized Solvent-Compound or Complex Effects hr-C,H, Ammonium phosphomolybdate Solvent
E.F.a
nb
NC
Mo(CO), E.F.a
nb
flame
N,O-C,H,
[MoCP(CO),Iz NC
E.F.0
nb
NC
Mo-cupferrate E.F.0
MIBK 1.6 5.7 3.1 1.0 5.7 1.7 1.3 5.7 1.0 1.4 n-Butyl 1.7 4.4 2.6 1.1 4.4 1.4 1.8 4.4 1.1 1.1 acetate 2-Octanol 2.9 1.0 1.0 3.4 1.0 1.0 1.0 aE.F. is the enrichment factor, see Table 111. b n is the normalized product sumed) for each solvent. CN is the normalized product of E.F. x n.
Mo(CO),
flame [MoCP(CO),lz
nb
N C
E.F.0
nb
NC
5.7 4.4
8.0 4.8
1.2 1.0
7.9 6.6
3.8 2.6
E.F.@
nb
Nc
1.2
7.9 6.6
1.3
1.1
1.0
1.0 1.0 2.5 1.0 1.0 of the (solution flow rate) X ( % solution con-
MIBK in a carbon-rich N ~ O - C ~ Hflame. Z e) The behavior of [MoCp(CO)3]2 is anomalous judging from its sensitivity relative to that of Mo(CO)e and other compounds which contain metal-oxygen bonds. In lean flames (both air-acetylene and NzO-acetylene), molybdenum absorbance signals are weak even with quite high Mo concentrations. In an air-acetylene flame, a 50 pg/ml aqueous ammonia solution of Moo3 produced, by the “small-area” absorbance technique, an absorbance ratio of 27:l in a carbon-rich vs. a lean flame (air flow rate = 7.5 1./ min; C2Hz flow rate = 1.0 l./min). The absorbance ratio given by the organomolybdenum compounds-Mo(C0)6 and [MoCp(CO)3]2-was 12:l and 11:l respectively, i.e., less than the ratio for MOO,; this suggests that carbon or carbon-containing species play a role in the enhancement of the absorbance. Deviations from Dissociation Equilibrium. It has been well documented (31, 32) that in and above the reaction zones of hydrocarbon flames, equilibrium conditions do not prevail. Certain flame radicals may exist far in excess over their equilibrium concentrations. Reactions with these radicals may produce free metal atoms far in excess over their equilibrium concentration. This excess of free metal atoms is produced by the reduction reaction: MO + C M CO (where C = carbon or carbon-containing species), it being thermodynamically very favorable. I t is highly probable that the above reaction is responsible for production of free Mo atoms in the highly-reducing flames employed. The above argument suggests that the concentration of MOO should decrease with an increase in the reducing nature of the flame. Coker et al. ( 1 1 ) reported a decrease in the emission intensity of MOO and a corresponding increase in the absorbance of the Mo atoms as the flame was varied from lean to rich. These observations agreed with the results of the present studies-a carbon-rich nitrous oxide-acetylene flame gave increased sensitivity for all the Mo compounds or complexes. Increased dissociation in a nitrous oxide-acetylene flame also results from the higher temperature and the longer
-
680
+
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
BO
>
$ mz I-
z
z eo-
BE 3 50y W
2
5
40-
L z 30-
20 -
HEIGHT OVER BURNER TOP, mm
Figure 1. % Relative emission intensity of MOO as a function of
burner height in a N20-C2H2 flame CPH2= 4.5 I./min: N 2 0 = 4.5 I./min. 0 800 @g/mlMo(CO)e in MIBK. A 800 pglml [MoCp(CO)slnin MIBK
residence time provided by it. The longer residence time is due to a 40% reduction in the vertical rise velocity of this flame as compared to the air-acetylene flame (33). Figure 1 is a plot of the emission intensity of the MOO band head a t 610.8 nm ( 3 4 ) in the nitrous oxide-acetylene flame, obtained by nebulizing solutions of M o ( C O ) ~or [ M O C ~ ( C O in ) ~MIBK ]~ containing the same mass of Mo. Under otherwise identical experimental conditions, the emission intensity of MOO formed by [MoCp(CO)3]2 is a t most about 80% of that formed by M o ( C O ) ~As . expressed by Equation 8, the free atom population and, hence, the sensitivity of the atomic absorption measurement depends on the partial pressure of gaseous metal monoxide in the
flame. The lower concentration of M a 0 formed by [MoCp(CO)3]2 may partially account for the lesser sensitivity of this compound compared to Mo(C0)e. An additional explanation of this decreased sensitivity may he the possibility of formation of gaseous (undissociated) molybdenum carbide in the extremely carbon-rich environment. A point is always reached in the flame stoichiometry where further increases in the carbon-to-oxygen ratio result in a decreased sensitivity. The larger amount of free atomic and solid carbon (soot) in these flames leads to the increasing possihility of formation of gaseous MozC. More favorable conditions for carbide formation exist in the case of [MoCp(C0)3]2 than in the case of M O ( C O )because ~ of the presence of large, sandwich-structured cyclopentadienyl rings. These provide intimate contact between the Mo and the carbon in the condensed phase during the process of vaporization. The fact that [MoCp(CO)& gives the lower sensitivities in all the solvents (see Table I) than those of M O ( C O )is ~ consistent with the formation of gaseous (undissociated) Mo2C. Rasmusson e t al. (14) also indicated the possibility of certain metals forming carbon-containing compounds in a carbon-rich N20-CzHz flame and a consequent depression in the absorbance. Extension of the above principles to the inorganic molybdenum compounds or complexes may well account for their lesser sensitivities as compared to the organomolybdenum compounds. These latter species are unstable a t the flame temperature (average D for Mo(C0)e = 39 kcal mol-', averaged from (33, 36)) and may readily form MOO. Further evidence on this will he presented later. Solvent Effects. It is well known that many organic solvents enhance an atomic absorption signal (relative to that given by an aqueous solvent), hut it is difficult to isolate the effect of the solvent on the atomic absorption signal. Enhancement of sensitivity may he the resultant effect of several factors. In the case of suitable organic solvents, these include the production of smaller and more volatile droplets of aerosol with a corresponding decrease in the size of the desolvated salt particles, and an increase in the rate of solution uptake, and, hence, in t h e quantity of analyte reaching the flame per unit of time. Also, unlike water, combustible organic solvents do not lower the temperature of the flame; on the contrary, their heats of combustion offset their heats of vaporization. Organic solvents may also produce a favorable change in the stoichiometry of the flame, increasing its reducing characteristics if its C/O ratio > 1: this is unlike water which provides a more oxidizing environment. Table I1 nresents the enrichment factors (defined below) for the systems. The absorbance given by the solution returned by the nebulizer chamber, Le., the condensate of the larger diameter droplets arrested by the nebulizer chamber, has been compared with the ahsorhance given by that part of the solution which has entered the flame. The ratio of these two absorbances is called the "enrichment factor". I t should be emphasized that identical conditions were maintained for each system throughout the experiment and the concentration of Mo in all solutions was kept as nearly equal as the experimental conditions permitted. In each case optimum fuel-to-oxidant ratios were maintained. The gas flow rates and the solvent flow rates employed are as shown in Table 111. These enrichment factors show no definite correlation with the physical properties of the solvent or the analyte, such as their hoiling points, heats of vaporization, vapor pressures, etc. A factor which may contrihute to this phenomenon, however, is the well-known effect of a solute on the surface tension of the solvent (37). In the case where the surface tension is decreased upon dissolution of the solute, an excess of solute in the interfacial region (i.e.,
2
I
O
I
2
HORIZONTAL TRAVERSE h m l
Figure 2. Absorbance profile for MO(CO)~in 2-octanol in an airC2H2 flame CeHg = 2.7
i./min; air = 7.5 I./min Contour interval. 0 005 absorbance unit
2
0
I HORIZONTAL
i
2
TRAVERSE Imml
Figure 3. Absorbance profile for Mo-cupferrate in MlBK in an airC,H2 flame C2H2 = 1.85 i./min; air = 7.5 I./min Contour interval, 0 005 absorbance unit
2
,
0
1
2
HORIZONTAL TRAVERSE Imm!
Figure 4. Absorbance profile for ammonium phosphomoiybdate in *butyl acetate in an air-CnHp flame C Z H ~= 2.3I./min: air = 7.5 i./min. Contour interval, 0.005 absorbance unit
2
I
0
1
2
HORlZONTaL TRaVERSE h r n l
Figure 5. Absorbance profile for MO(CO)~in MlBK in a N20-C2H, flame N20 = 4.5 i./min: C2He = 4.5 I./min. Contour interval, 0.005 absorbance unit
ANALYTICAL CHEMISTRY, VOL. 48. NO. 4, APRIL 1976
681
on the surface of the solution) will be present. Since the surface area per unit mass is inversely proportional to the particle diameter, the total surface area for a given volume of liquid increases greatly as the particle diameter diminishes, and, hence, the excess solute present on the surface of the extremely small droplets entering the flame will be considerable. Consequently, the part of the solution which enters the flame would be enriched in solute and the part of the solution which is arrested and returned from the nebulizer chamber would be depleted in solute. In the above case, the enrichment factor would be greater than 2 ' 0 I 2 unity. The enrichment factor would be less than unity for P O P l Z C Y T A L T R G V E P S E l m r r l those solute-solvent systems in which the solute increases the surface tension of the solution relative to that of the Figure 6. Absorbance profile for MOOSin aqueous ammonia in a pure solvent. N20-C2H2 flame Changes in the surface tension (measurements not reN20 = 4.5 i./min: CzH2 = 6.5 I./min. Contour interval, 0.005 absorbance ported) of the solvents upon addition of the analytes were unit not large enough to provide conclusive evidence even for saturated solutions. There is, however, the further possibility of other kinds of solute-solvent interactions which may make the combined effect unpredictable on the basis of the surface tension alone. This is evident from Table 11, which shows that the enrichment factors may vary from the airacetylene to the N z O - C ~ Hflame ~ even with the same solute-solvent combination. The rate of flow and the yield of solvent (the ratio of the amount of solvent entering the nebulizer chamber to the amount entering the flame, expressed as a percentage) for both flames are listed in Table 111. T o consider more fully the effect of the nature of the original molybdenum compound or complex on the sensitivities, it is first necessary to take account of the factors contributing to the net amount of the analyte reaching the flame. These include the solution aspiration rate, the yield, and the specific enrichment factor (E.F.) for each Mo compound or complexsolvent system. The value n , which is the normalized product of the first two of these factors, is presented in Table 111. Table IV takes account of the specific enrichment fac0' d ; ; b Ib I: ,b HEIGHT OVER BURNER TOP, m m tor for each of the Mo compound or complex-solvent systems, and presents the normalized final values, N , which is Figure 7. Relative absorption (YO) by M o atoms or relative emission the normalized product of n X E.F. Thus, N is a relative intensity from MOO band head as a function o f height over the burnmeasure of the amount of analyte actually entering the er top in a N20-C2H2 flame flame in unit time. C2Hz = 4.5 I./min; N 2 0 = 4.5 I./min. 0 % absorption by Mo atoms-the On the basis of N , the order of sensitivity of the solvents test solution contained 50 pg/ml Mo as Moo3 in aqueous ammonia. A % for a given Mo compound or complex in both flames is preabsorption by Mo atoms-the test solution contained 5 fig/rnl Mo as dicted to be: MIBK > n-butyl acetate > 2-octanol. The exMo(C0)e in MIBK. 0 % emission by Moo-the test solution contained 800 pg/ml Mo as Moos in aqueous ammonia. A % emission by Moo-the test perimentally observed sensitivities have the predicted solution contained 800 pg/ml as Mo(C0)s in MIBK order for [MoC,(C0)3]2 and Mo-cupferrate in the air-CaH2 flame, and Mo(CO)F;and [MoC,(C0)3]2 in the N ~ O - C ~ H Z flame. The reverse order was experimentally observed for free-atom population density, but the overall spatial distriammonium phosphomolybdate and Mo(CO),j in the airbution pattern remains unaltered. CzH2 flame. We can offer no explanation for this reversal. A second feature common to the absorbance profiles is It is evident from the above study that not only the solthe symmetrical distribution of free atoms on both sides of vent hut also the nature of the Mo compound or complex the slot of the burner (optical axis). Willis (38) has suggested that this is the result of the initial size-distribution of employed in the construction of calibration curves for molybdenum must match as closely as possible those of the liquid droplets. The burner slot directs the droplets into a unknown samples. thin vertical stream up the center of the flame. The moAbsorbance Profiles. All of the processes that occur in mentum of the larger droplets causes them to be less readithe flame affect the spatial distribution of free atoms of the ly deflected than the smaller ones. The highest concentraanalyte in the electronic ground state ( 2 4 ) . Absorbance tion of solute particles and, hence, free atom vapor, is profiles can be used to help elucidate the mechanisms intherefore located in the plane passing vertically through volved in the formation of free atoms in flames. Figures 2-6 the burner slot with the free-atom population decreasing are typical absorbance profiles obtained for molybdenum symmetrically on both sides of this plane to the edges of in the carbon-rich air-acetylene and nitrous oxide-acetythe flame. In the outer part of the flame, the entry of ambient air causes secondary combustion reactions to occur, lene flames. Changes in any of the following variables: solvent, Mo creating turbulence in the flame gases; also, the laminar compound or complex, or fuel flow rates (provided a highly flow of the flame gases breaks down in the outer partreducing environment is maintained), result in vertical both these factors create uncertainty in the outlying conshifts only in the coordinates for the site of the highest tour positions. 682
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
I
2
o
1
2
HORIZONTAL TRAVERSE Imml
Figure 8. Absorbance profile for ammonium phosphomolybdate in MIBK in an air-C2Hn flame C2H2 = 2.6 I./min; air = 7.5 i./min. Contour interval. 0.005 absorbance unit
~. I
*
,
'
I
2
HORIZONTAL TRAVERSE lmml
Figure 9. Absorbance profile for Mo(C0)e in MIBK in an air-CzHs flame C2H2 = 2.3 I./min: air = 7.5 I./min. Contour interval, 0.005 absorbance unit
Similarity in the absorbance profiles obtained from approximately equal masses of molybdenum contained in various Mo compounds or complexes does not imply that all the Mo is present as free atoms in the flame. I t may merely indicate that the concentration of Mo free atoms is controlled by an equilibrium that is independent of the salt in the original solution hut dependent on the composition of the flame gases a t various points in the flame, On the
basis of the above observations, it is reasonable to postulate the formation of a species common to all of these analytesthe species being predominantly MOO and the reaction being reduction of the MOO. By changing the solvent hut otherwise maintaining constant experimental conditions, the following order of increasing height for the position of the absorbance maximum was obtained 2-octanol < butyl acetate < MIBK. This order parallels the expected droplet size under the experimental conditions used. The mean droplet size can be calculated approximately by the empirical equation of Nukiyama and Tanasawa (39).Upon nebulization, the smallest droplets are formed from 2-octanol, the largest from MIBK. Since the time for vaporization of the analyte particles is proportional to the square of their radius, the above order is expected. Within a particular Mo compound or complex-solvent system, there is a shift in the position of the absorbance maximum to greater heights with increasing CzHz flow rates, indicating that one or more reactions of the atomization process (evaporation, vaporization, dissociation) become slower with increasing CzHz flow rates. After the C/O ratio = 1 has been reached in a flame, any further increase in the C2H2 flow rate does not increase its reducing cbaracteristics-the increasing CzH2 flow rate only decreases steadily the flame temperature (15, p 151). Also, molybdenum in such a flame which contains solid carbon (soot) is likely to form gaseous (undissociated) MozC. The lower flame temperatures (with higher CzHz flow rates), also make reduction of MOO (17, 18) with carbon-containing species and the dissociation thermodynamically less favorahle. The effect of variation of the molybdenum compound or complex on the height of the absorbance maximum is not as pronounced as that of the solvent, probably because the thermal stabilities of the species a t the flame temperature are not very different. However, the following order for the Mo compounds or complexes with respect to increasing height for the position of the absorbance maximum was observed ammonium phosphomolybdate 5 molybdenum
2.0 0.1
w o
0 "I "7
30
m
4 0.1
0
c
(13
02
01 0
187m9511
25ea,.s,.o~.Joo.s,ol.aeaz.e HORIZONTAL
TRAVERSE, mrn
0.1 0
2 5 Z D I . O , . 0 0 . 5 0 0 ~101.52.02.3
HORIZONTAL
TRAVERSE, mm
Figure 10 and Figure 11. Spatial distribution of the absorbance across the width of an air-acetylene flame of a IO-cm burner at various heights The ten Solution contained 30 gg/ml ammonium phosphomolybdate in MIBK. C2H2 = 1.85 I./min; air = 7.5 I./min ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRiL 1976
683
10
..
-
-
10
1.0
09
0.9
W
4'm
-
09-
K
:: O B -
-
- 08
0.0
-
5 8
0.7
-
D
0.6
-
0.5
-
0.4
-
a
9 07 I0.6 c r 05 -
-07
-
06
W
-05
0
W
;
04-
a
B
a0.2
- 04
s'
-
03
0.3
-
- 02
d
a2
-
a1 I
Figure 12. Normalized integrated absorbance as a function of height over the burner top in an air-acetylene flame The test solution contained 30 pg/ml ammonium phosphomolybdate in MIBK. C2Hp = 1.85 I./min; air = 7.5 I./min. A Peak absorbance. 0 Integrated absorbance (area). 0 integrated absorbance (summation over discrete horizontal points)
cupferrate = MOO3 = Mo(C0)s = [MoCp(CO)3]z in the air-CzH2 flame, and Moos < [MoCp(CO)3]2< Mo(C0)s in the NzO-C2Hz flame. The position of the Moo3 system in the above order is not certain since it was the only compound for which an aqueous solution was employed. Another interesting feature of the absorbance profiles is that the absorbance maximum for all the organomolybdenum systems occurs higher in the flame than the Moo3 in aqueous ammonia system. Since the absorbance maximum marks the position of balance between the increase in the number of free atoms and their loss with increasing height, the height of the absorbance maximum is determined by the rates of these opposing processes. It is probable that because of their low decomposition temperatures, e.g., 429 K for Mo(CO)6, a significant portion of the organomolybdenum compounds may be decomposed inside the hot burner top. In such circumstances, it is likely that MOO is rapidly formed as soon as the analyte enters the flame. The result is that a relatively large quantity of MOO is present low in the flame; hence, the relatively high emission intensity of MOO low in the flame. As a result, the Mo atom population is low close to the burner top, increases with height and then decreases because of increasing dilution with increasing height. The initial increase in the absorption with the increasing height may be the result of the combined effects of the increased time (height) for further thermal dissociation, and also, of the reduction of the MOO by carbon (or carbon-containing species). However, if the analyte is oxygen-bonded, such as in the MOORin aqueous ammonia system, successive dissociation from the M403 to MOO may occur, resulting in an increasing concentration of the MOO with time (height) until the effect of dilution by flame gases predominates. Figure 7 is a plot of the percentage relative emission by MOO molecules and absorption by Mo atoms when the M o ( C O ) ~ in MIBK and the MOOSin aqueous ammonia systems were nebulized into a carbon-rich NzO-CzHz flame. Assuming local thermodynamic equilibrium, the concentration of MOO was plotted in arbitrary units (on a 0-100 scale) from the measured emission intensity of the MOO band head a t 610.8 nm. The position of the maximum concentration of MOO is lower (at 6 mm) in the flame in the case of the M o ( C O ) ~in MIBK system than in the case of the Moo3 in aqueous ammonia system (at 8 mm). Despite the fact that the flame stoichiometries are 684
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
I
I
I
I
I
1
I
I
2
3
4
5
6
7
HEIGHT
OVER
BURNER
,
I
I
8
9
10
TOP,
mm
Figure 13. Normalized integrated
absorbance as a function of height over the burner top in an air-acetylene flame A Moo3 in aqueous ammonia. C2H2 = 2.7 I./min; air = 7.5 I./min. 0 Ammonium phosphomolybdate in MIBK. C2H2 = 1.85 I./min; air = 7.5 I./min. 0 Mo-cupferrate in MIBK. C2H2= 1.85 I./min; air = 7.5 I./min
HEIGHT OVER
BURNER
TOF:
mm
Figure 14.
Normalized integrated absorbance as a function of height over the burner top in an air-acetylene flame
0 Mo(C0)e in MIBK. C2H2 = 1.85 I./min; air = 7.5 I./min. A [MoCp(C0)3]2 in MIBK. C2H2 = 1.85 I./min; air = 7.5 I./min
not identical when solvent effects are taken into account (MIBK vs. aqueous ammonia), the MOO (from Mo(CO)~) emission intensity is higher a t all points in the flame and reaches a maximum earlier in the flame than the Moo3 in aqueous ammonia system; also, the absorption by Mo atoms is low a t the base of the flame and increases with height, in agreement with the above explanation. As is indicated in Figure 7 , the absorption by Mo atoms in the Moo3 in aqueous ammonia system reaches a maximum close to the burner top. The reason for this behavior is not known. The fact that the maximum Mo atom population occurs higher in the flame for M o ( C O ) ~than for oxygen-bonded systems cannot be explained in terms of formation of gaseous MozC during decomposition, thus requiring a higher temperature (and hence, greater height in the flame) for dissociation since this compound exhibits the greatest sensitivity of all those studied. It is also evident from Figure 7 that there is no correlation between the heights of maximum MOO concentration and atomic absorption as would
be expected from Equation 8. This equation was derived solely on the basis of thermal dissociation of the MOO.I t is evident, however, that this is not the only pathway for formation of Mo atoms; reduction of MOO by carbon (or carbon-containing species) is also important. Temperature Profiles. Temperature profiles constructed for both the air-CzH2 and NzO-CzH2 flames (figures not shown) offer no definite indication of the extent of the influence of temperature on the production of Mo free atoms. The position of maximum free-atom population does, however, lie in the regions of high temperature. Since reducing capacities of carbon and carbon-containing species such as CN increase with temperature (17, 18), it is reasonable to conclude that a high temperature in combination with a highly-reducing environment favors the existence of Mo free atoms. Figures 8 and 9 are absorbance profiles obtained in a very carbon-rich flame. These “population” profiles bear a close resemblance to the temperature distribution obtained for such a flame in which the regions of maximum temperature are off axis to the center of the flame (40). I t is possible that in these highly reducing flames, gaseous molybdenum carbide is formed which is dissociated to a greater extent in the hotter regions of the flame, thus accounting for the higher Mo free-atom population in the high-temperature regions. Rate of Release of Atoms in the Flame. The progress of metal atom formation may be followed by intergrating the total absorbance across the width of the flame a t various heights (41). Such integrated absorbance plots allow comparison of the various Mo compound or complex-solvent systems as regards their rates of release of atoms in the flame. One such plot is presented in Figures 10 and 11, which reveals an initial growth in the total absorbance with increasing height above the reaction zone. The numbers shown inside the curves in these figures are integrated absorbance values obtained by graphical integration of the areas under the curves with a planimeter. The numbers within parentheses are normalized absorbance values. An approximate integration of the absorbance values across the width of the flame at a given flame height can also be made by summing the absorbance values a t short intervals. The former method gives true integration whereas the latter method gives a summation (an approximate integration). The integrated absorbance is a measure of the total number of free atoms in a horizontal plane across the width of the flame a t various heights and, as such, its magnitude reflects any change in the total number of free atoms in that particular plane. Figure 12 ( A and B ) present curves for the normalized integrated absorbance given by both methods (the curves will be called the integrated absorbance profiles), and also for absorbance in the plane passing along the burner slot and perpendicular to the burner top (the curve will be called the peak absorbance profile). Both curves were obtained with the same quantity of ammonium phosphomolybdate in MIBK. Both methods for the integration of absorbance gave almost identical curves, but the method of summation has the advantage of requiring shorter time. I t is instructive to compare the information content and the dignificance of the integrated absorbance profile and the peak absorbance profile. The integrated absorbance increases and decreases faster than the peak absorbance. This could only be the result of lateral diffusion of Mo free atoms from the center of the flame to the edges. The maximum in the curves occurs a t slightly different heights over the burner top-for curve A a t 5.5-mm height, for curve B a t 6-mm height. Differences in the height for the maximum between the integrated and peak absorbances for the same
system under identical experimental conditions were observed for some other (but not all) Mo compounds or complex-solvent systems studied. From the peak absorbance profile (Figure 12B), it is not possible to say whether the decrease in the peak absorbance above the height of 6.0 mm is due to a decrease in the total number of Mo free atoms as a result of compound formation (presumably MOO) or dilution of flame gases by inrushing air. The integrated absorbance profile (Figure 12A) indicates that there is an actual decrease in the total number of Mo free atoms. The maximum in the two curves arises from different processes. The maximum in the integrated absorbance curves arises from two opposing processes: atomization tending to increase the total number of free atoms, and recombination tending to decrease the total number of free atoms. The maximum in the peak absorbance curves arises from three processes: atomization tending to increase the concentration, recombination, and flame dilution (competing processes) tending to decrease the comentration. The sharp linear increases and decreases in the integrated absorbance indicate that the rates of the processes involved in the atomization and depopulation are fairly rapid. Sharp peaks in the integrated absorbance profiles arise from those absorbance profiles (Figures 2-6) in which high absorbance values are localized in a relatively small area in the flame, and are characteristic of elements which form stable monoxides in flames. I t is clear from the above that the information content and the significance of integrated absorbance profiles are much greater than those of peak absorbance profiles. Figures 13 and 14 present normalized integrated absorbance data for the Mo compound or complex-solvent systems as a function of height over the burner top. Since these curves were obtained with different quantities of Mo (a consequence of the different sensitivities of the various Mo systems and the necessity of working with adequate concentrations to obtain a full range of absorbance values), the magnitude of the integrated absorbance of different Mo systems is not directly comparable. Under the experimental conditions used, the maximum in the curves occur a t the following heights over the burner top: 4 mm for the Moo3 in the aqueous ammonia system, 5.5 mm for the ammonium phosphomolybdate in MIBK system, 6.0 mm for the Mo(C0)s in MIBK system, 6.0 mm for [MoCp(CO)3]2 in MIBK system, and 6.5 mm for the Mo-cupferrate in MIBK system.
CONCLUSIONS Extensive formation of Mo free atoms takes place only in the highly reducing environments provided by carbon-rich air-acetylene and nitrous oxide-acetylene flames. Formation of Mo free atoms probably proceeds predominately through MOO, both for the inorganic and organomolybdenum compounds or complexes studied. In carbon-rich flames, formation of gaseous (undissociated) molybdenum carbide is a distinct possibility.
LITERATURE CITED ( 1 ) V. A. Fassel, R. B. Myers, and R N. Kniseley, Spectrochim. Acta, 19, 1187 (1963). (2) D. J. David, Analyst (London),86, 730 (1961). (3) M. D. Amos and J. B. Willis, Spectrochim. Acta, 22, 1325 (1966). (4) V. A . Fassel, J. 0. Rasmuson, R. N. Kniseley, and T. G. Cowley. Specfrochim. Acfa, Part 8,25 559 (1970) (5) B. M. Gatehouse and J. B. Willis, Specfrochim. Acfa, 17, 710 (1961). (6) C. L. Chakrabarti, G. R. Lyles, and F. B. Dowling, Anal. Chim. Acfa, 29, 489 (1963). (7) C. L. Chakrabarti, Anal. Chim. Acta. 42. 379 (1968). (8) C. L. Chakrabarti and D. P. D. McNeil. Can. J. Spectrosc., 20, 90-98 (1975). (9) C. L. Chakrabarti and S. P. Singhal, Spectrochim. Acta, Part B, 24, 663 (1969). (10) K. Fujiwara, H. Haraguchi, and K. Fuwa, Anal. Chem., 44, 1895 (1972)
ANAL.YTICAL CHEMISTRY, VOL. 4.5, NO. 4, ADRIL 1976
685
(11) D. T. Coker, J. M. Ottaway, and N. K. Pradhan, Nature (London), Pbys. Sci., 233, 69 (1971). (12) J. Y. Marks and G. G. Welcher, Anal. Cbem., 42, 1033 (1970). (13) G. De Maria, R. P. Burns, J. Drowart, and M. G. Inghram, J. Cbem. Pbys., 32, 1373 (1960). (14) J. 0. Rasmuson, V. A. Fassel, and R . N. Kniseiey, Spectrocbim. Acta, Part E, 28, 365 (1973). (15) E. V. L'vov, "Atomic Absorption Spectrochemical Analysis", Adam Hilger, London, 1970. A. G. Gaydon and H. G. Wolfhard, "Flames, Their Structure, Radiation, and Temperature", 2nd ed., Chapman and Hall, London, 1960. V S. Sastri, C L. Chakrabarti, and D. E. Willis, Can. J. Cbem., 47, 587 11969). . , C. W. Dannett and H. S. T. Eliingham, Discuss. Faraday SOC., 4, 126 (1948). G. F. Kirkbright, M. K. Peters, and T. S. West, Talanta, 14, 789 (1967). T. G. Cowley. V. A. Fassel, and R. N. Kniseley, Acta, Part . Spectrocbim. . 8, 23, 771 (1968). D. T. Coker and J. M. Ottaway, Nature (London), Pbys. Sci., 230, 156 (1971). G.F. Kirkbright and T. S. West, Talanta, 15, 663 (1968). C. S. Rann and A. N. Hambly, Anal. Cbem., 37, 879 (1965); A. N. Hambly and C. S Rann in "Flame Emission and Atomic Absorption Spectrometry", Voi. I, J. A. Dean and T. C Rains, Ed., Marcel Dekker, New York. 1969, pp 249-252. C. L. Chakrabarti, M. Katyal, and D. E. Willis. SDectrocbim. Acta, Part E, 25, 629 (1970). D. M. Kemp, Anal. Cbim. Acta, 27, 480 (1962). J. J. Eisch and R . B. King, "Organometallic Synthesis", Vol. I (Transition Metal Compounds), Academic Press, New York, 1965, p 110. C. L. Chakrabarti, R. Pal, and M. Katyal, Anal. Cbem., 43, 1704 (1971). W. S. Zaugg and R. J. Knox, Anal. Cbem., 38, 1759 (1966). R. F. Browner and J. D.Winefordner, Anal. Cbem., 44, 247 (1972). A. C. West. V. A. Fassel. and R. N. Kniseiev. Anal. Cbem.. 45. 1586 (1973).
(31) C. Th. J. Alkemade in "Flame Emission and Atomic Absorption Spectrometry", Vol. I, J. A. Dean and T. C. Rains, Ed., Marcel Dekker, New York, 1969, pp 121-123. (32) G. F. Kirkbright, M. K . Peters, and T. S. West, Atomic Absorption Symposium, Praha, 1967 (summary); G. F. Kirkbright, A. Semb, and T. S. West, Spectrosc. Left., 1, 7 (1968). (33) J. E. Willis. J. App. Opt., 7, 1295 (1968). (34) R. W. B. Pearse and A. G. Gaydon, "The Identification of Molecular Spectra", 3rd ed., Chapman and Hall Ltd., London, 1965, p 209. (35) F. A. Cotton, A. K. Fischer, and G. Wilkinson, J. Am. Chem. Soc.. 78, 5168 (1956). (36) D. R. Bidinosti and N. S. Mclntyre, Can. J. Cbem., 45, 641 (1967). (37) D. J. Shaw, "Introduction to Colloid and Surface Chemistry", 2nd ed., Butterworths, London, 1970, pp 68-70. (38) J. 8. Willis, Spectrocbim. Acta, Part E, 25, 487 (1970). (39) S. Nukiyama and Y. Tanasawa, Trans. SOC. Mecb. Eng. Jpn, 5, 62 (1939). (40) Unpublished results of the present authors. (41) J. H. Gibson, W. E. L. Grossman, and W. D. Cooke, Anal. Cbem., 35, 266 (1963).
RECEIVEDfor review August 19,1975. Accepted December 18, 1975. The authors are grateful to the National Research Council of Canada for financial support of this project. The authors also wish to thank D. R. Wiles for the use of his facilities for the neutron activation analysis. This paper was presented a t both the 56th Canadian Chemical Conference, June 4-6, 1973, Montreal, Canada, and the 4th International Conference on Atomic Spectroscopy, October 29November 2,1973, Toronto, Canada.
Precision of Flame Atomic Absorption Measurements of Copper N. W. Bower and J. D. Ingle, Jr." Department of Chemistry, Oregon State University, Corvallis, Ore. 9733 1
Repetitive atomic absorption measurements on copper solutions which yield absorbances in the range of 0-2 are made with a high resolution voltmeter and on-line computer and are used to evaluate the dependence of measurement precision on absorbance and instrumental variables. The experimental relative standard deviation in absorbance is compared to that predicted by a recently proposed theoretical equation. The study reveals that under the typical conditions used for copper atomic absorption measurements, noise sources associated with the lamp (Le., signal shot noise and source flicker noise), with the detection system (i.e., amplifier-readout noise, dark current noise), and with the background emission of the flame are not limltlng. Noise due to fluctuations in the transmission properties of the flame limits precision at small absorbances (i.e., A < 0.05), while fluctuations in the absorption properties of the analyte limit reproducibility for moderate and large absorbances.
varies with absorbance ( A )and instrumental variables and noise parameters. The purpose of this paper is to outline the experimental procedures for evaluation of the precision characteristics of AA measurements for a given element, on a particular AA instrument, under specific conditions. In particular, AA measurements on copper solutions ranging from 0.05 to 500 ppm were studied. The procedure involves measurement and calculation of the standard deviation of various signals and evaluation of instrumental variables and noise parameters so that theoretical calculations can be made. The net result of the procedure is that the relative contribution of the potential noise sources can be identified. For copper measurements with a 2-8, spectral bandpass, flame transmission fluctuations are limiting a t small absorbances while, at higher absorbances, fluctuations in the absorption properties of the analyte in the flame became limiting. Fluctuations in absorption properties are due to such factors as changes in the free atom population, the path length, the absorptivity of the analyte, or the sample and gas flow rates. For very small slit widths, signal shot noise and noise in the flame background emission became important.
The precision of flame atomic absorption (AA) measurements are significantly affected by the dynamic nature of flames. In a recent paper ( I ) , the previous experimental BACKGROUND and theoretical work dealing with precision and signal-toAppendix I contains the important equations and defininoise ratio (S/N) aspects of AA measurements was reviewed and new equations were presented which indicate tions to be used in this paper. The derivation of these equahow the relative standard deviation in absorbance ( ~ A / A ) tions and the assumptions made in the derivations have 686
ANALYTICAL CHEMISTRY, VOL. 48, NO. 4, APRIL 1976
