FIa me S pect rometric
Determinati o n of Te IIu riu m
JOHN A. DEAN and JOHN C. SIMMS' Department o f Chemistry, University o f Tennessee, Knoxville, Tenn.
b In the reaction zone of an acetylene flame the line spectlwm of tellurium appears in a strength that is analytically useful. The doublet situated at 238.3 and 238.6 mp is the most intense feature of the spectrum although usable lines are located at 214.3 and 225.9 mp The region of maximum intensity i s sharply localized and lies immediately above the inner cone of an acetylene-oxygen flame. The useful emission sensitivity of the line at 238.6 m,u is 2 pg. per milliliter of tellurium. The working curve is linear up to a t least 4000 pg. per milliliter of tellurium. When necessary, tellurium is isolated Irom most other elements b y a single extraction with sodium diethyldithiocarbamate in CC14 from an aqueous solution a t p H 4 in the presence of EDTA. To improve the emission intensity, the CC14 is removed and replaced b y methyl isobutyl ketone. Possible excitation mechanisms are discussed.
C
on the excitation characteristits of the reaction zone of flames have rwealed t h a t the line spectrum of tellurium is analytically useful and highly discriminatory. Use of a light-guide (3) confines the area of observation to that offwing the highest line emission while minimizing the effect of the intense conlinuous radiation which emanates from this portion of the flame. The useful emission sensitivity of the 238.6-mp line of tellurium is 2 pg. per milliliter. This transfers tellurium from the class of metals with weak and unsatisfactory flame spectra to the class of those determinable with good sensitivity by flane photometry. The sensitivity compares favorably with the iodotellurite colorkmetric method and surpasses the methods employing a stabiiized tpllurium sol. I n gravimetric methods, 0.1 mg. in 5 ml. represents the lower liniit for tellurium in recovery and weighing difficulties. With any of the methods, preliminary separations are essential in all but the most simple solutions. Separations are less often required when employing the flame-spectrometric method, but when necessary, telluriiim can be isolated from most other elements by a ONTINUIXG STUDIES
1 Present address, Department of Chemistry, North Georgia College, Dahlonega, Ga.
single extraction with sodium diethyldithiocarbamate in CCL from a weakly acid solution in the presence of EDTA. To improve the emission intensity, the CCl, was removed by evaporation and the residue was dissolved in methyl isobutyl ketone and the ketone solution was aspirated directly into the flame. A typical analysis on a simulated steel sample required about 45 minutes elapsed time, including the preliminary extraction step. EXPERIMENTAL
Special Reagents. Standard tellurium solution, 1.00 mg. per milliliter, was prepared by dissolving 1.000 gram of tellurium metal (99.9% purity) in t h e minimum amount of and HC1 plus a few drops of "03, diluting t o 1 liter with methyl isobutyl ketone. Alternately, one could dissolve 1.250 grams of tellurium dioxide, prepared from C.P. sodium tellurite, in the minimum amount of HC1, and dilute to volume with the ketone. More dilute solutions were prepared by appropriate dilution. Sodium diethyldithiocarbamate, approximately 0.06% (w./v.) solution, was prepared by saturating a t room temperature 100 ml. of CCL with the reagent. Apparatus. Extractions were performed with a Burrell wrist-action shaker. The Beckman Model DU flame spectrometer was employed. Data were recorded on a 10-mv. Bristol recorder having a 2/3-second pen response by the Beckman Energy Recording Attachment (ERA). The burner height v a s adjusted by means of a rack-andpinion arrangement. Brooks rotameters, Sho-Rate Model 1356, were used to regulate the flow of gases. The concave backing mirror in the burner housing was covered and a light-guide was employed ( 3 ) . The following instrument settings were employed: selector switch, 0.1 position; sensitivity control, 70 ERA % adjust; phototube resistor, 22 megohms; phototube (RCA 1P28), 65 volts per dynode; slit width, 0.10 mm.; half-band width (at 238.6 mp), 0.17 mp; height above tip of burner (axis of light-guide), 6 mm.; acetylene flow, 1.18 liters per minute; hydrogen flow, 2.7 liters per minute; oxygen flow, 3.3 liters per minute (both burners); oxygen pressure, 15 p.s.i.; aspiration rate (ketone solution), 2.2 ml. per minute. Procedure. Select a sample size to contain not less than 100 pg. of tellurium when the final dilution is
10 ml. Dissolve the sample in an appropriate solvent, while avoiding any excess of "03 or other strong oxidant. Dilute to approximately 100 ml. and adjust the pH to 4 with aqueous ammonia. For each 1 gram of metal present, add 5 grams of disodium salt of EDTA and, if necessary, readjust the pH to 4. Transfer the solution to a 250-ml. separatory funnel, add 50 ml. of 0.06% sodium diethyldithiocarbamate solution in CC4, and shake for 2 minutes. After the phases have separated, drain the CClr-layer into a 400-ml. beaker. Wash the aqueous phase with a 10-ml. portion of CC1, and add this to the extract. Evaporate, without boiling, the CC1, on a low-temperature hot plate until the volume is reduced to 2 or 3 ml. Transfer the solution to a 30-ml. beaker and rinse the original beaker with several 2-ml. portions of CC4. Evaporate the solution until only a dark oily paste remains. Pipet exactly 10 ml. of methyl isobutyl ketone on to the paste and swirl until everything dissolves. Aspirate the ketone solution directly into the flame and scan an appropriate portion of the spectrum. Record the line signal and background signal a t one of the pairs of wavelengths listed in Table I. Bracket the sample with suitable aliquots of the standard tellurium solution. I n the absence of interferents, particularly iron(lII), an alternate procedure can be followed. To the dissolved sample, 3.5N in HC1 (nitrate absent) and contained in a 50-ml. centrifuge tube, add an equal volume of hydrazine hydrochloride, 15% solution saturated with Son, t o reduce the tellurium to the free metal, centrifuge, and wash the precipitate with water. Dissolve the tellurium in 5 ml. of warm aqua regia and evaporate to incipient dryness. Add exactly 10 ml. of methyl isobutyl ketone and proceed as before. RESULTS
Tellurium(IV), as the diethyldithiocarbamate, can be extracted almost selectively a t pH 4 in the presence of EDTA. At p H 4 one equilibration sufficed whereas a t pH 8 two equilibrations were necessary. Also, a t pH 4 the aqueous phase is much lighter in color which enables an operator to observe whether any precipitate of ferric oxide forms (from insufficient EDTA) since hydrous ferric oxide is a notorious carrier for traces of tellurium. The diethyldithiocarbamate was dissolved in CCL to avoid the rapid deVOL. 35, NO. 6, MAY 1963
* 699
Table 1.
Wavelength, mp
20s. 1 214.2
Emission Lines of Tellurium
(Oxygen-acetylene flame; methyl isobutyl ketone aerosol; and optimum instrument settings) Emission LogaConcentration sensitivity, Background, rithmic range, pg. per m1.a mp slope pg. per ml. b
0.93 0.78
10 to 125 125 to 1000
6.7 228.4 or 226.5 238.3 3.7 237.8 or 239.0 238.6 2.0 237.8 or 239.0 0.98 At a noise level of lY0 for a time constant of 1 second. Barely detectible with 1000 pg. per milliliter.
10 t o 4000
213.5 or 214.7
6.7
22.5 9-
a
b
Table II. Effect of Slit Opening on Emission Signals (Optimum instrument settings with an oxygen-acetylene flame. Measurements taken at 238.6 mp with 100 pg. per ml. of Te) Slit Scale divisions, 0.1 mv units Ratio: line t o width, Tellurium, Background, mm. 238 6 m p 240 mp Koisea background
n- . n_._ .5 0.08 0.10 0.15 0.20 a
b
0 37 0.25 0.24 0.20 0.17
9 5
3. . .i
6 10 18 29
5 0 . 12 rto 20 50.35 *o. 84
2A
4 5 times the background noise. Zero reading reset by means of zero suppressor control
composition of the reagent which occurs in aqueous solutions whose p H .
I
m Z
W
I-
N
z Z
0 m
cn
n!
ZI
N
5
h
W
I Ia 1 W
DISCUSSION
Spectrum. It is evident t h a t the lines observed for tellurium (Table I) exhibit a sensitivity adequate t o satisfy many analytical requirements. Here t h e useful emission sensitivity (1) is that concentration of tellurium for which the mean net signal exceeds the 700
ANALYTICAL CHEMISTRY
r I
WAVELENGTH, Figure 1 . lurium
mG
Emission spectrum of tel-
tion) is 0.14 scale division, the useful emission sensitivity for a 1% noise level is calculated to be
2.0 pg./ml./O.l mv.
The root mean square of the variability of the signal plus background and the background alone is essentially times the fluctuation of either. The emission sensitivity implied can be achieved with additional signal amplification. The limit of detectability is considered to be twice the useful emission sensitivity. The most intense feature of the tellurium spectrum (Figure 1) is the doublet situated a t 238.3 and 238.6 mp. For a slit width of 0.09 mm. or less, the doublet is resolved and the individual net line intensities bear the ratio (stronger: weaker) of 1.83. Less intense, but usable lines, are those located at 214.3 and 225.9 mp. In the vicinity of these lines the background emission is less noisy and less pronounced than in the locality of the doublet. In this far ultraviolet region of the spectrum, the reaction zone exhibits only the atomic emission line of carbon a t 247.8 mp and a continuous background due to the incandescent carbon particles on which is superimposed the Fourth Positive molecular band system of CO. Slit Width. The slit width may be increased t o 0.20 nim., at least, equivalent t o a spectral band n-idth of 0.34 mp, while retaining sufficient signal stability. The influence of ,slit width on the net emission intensity, the background intensity, the ratio of the net line emission to background, and the signal noise is gathered in Table 11. I n these measurements the axis of the lightguide wts 6 mni. above the tip of the integral-aspirator burner. The inner cone extended about 4 mm. above the base of the oxygen-acetyleneketone flame. S o significant change in the noise or background reading was observed for settings a t 4 or 8 mm. above the tip of the burner. Excitation Conditions. The tellurium lines attained their maximum brilliance 6 mm. above the tip of the burner with a n acetylene-oxygen flame (Figure 2). The region of maximum intensity is sharply localized and lies immediately above the inner cone. S o atomic lines of tellurium were detected in the plume of the flame. With a hydrogen-oxygen flame, the region of maximum intensity was slightly higher, 5 to 8 mm. above the tip of the burner, and less sharply localized. Use of a hollow light-guide of aluminum, 6 mm. in diameter and 140 nim. in length, facilitated location of the region of niazimum intrnsitv and increased
di
0
2
4
6
8
l
O
HEIGHT ABOVE TIP OF BURNER,
l
2
1
4
mm.
Figure 2. Emission intensity of tellurium lines as a function of flame region viewed. Acetylene-oxygen flame with methyl isobutyl ketone aerosol
both the net signal and the signal-tobackground ratio ( 3 ) . Variation of the signal of the 238.6mp line as a function of the acetylene flow and the oxygen flow is shown in Figure 3. Adjustment of the acetylene flow is critical; the emission intensity passes through a sharp maximum at a ratio of flow of oxygen to acetylene of 2.4. I n richer flames the signal diminishes rapidly. The signal in an oxygen-hydrogen flam: is less and the line intensity is critically dependent upon the amount of solvent introduced. Actually only a minute lean hydrogenoxygen flame is needed. When the organic aerosol is introduced the flame assumes normal size with a clearly defined reaction zone. Among different organic aerosols the emission intensity varies considerably, as shown by the data in Table 111. Although pure acetone gave a larger signal than did methyl isobutyl ketone, the steadier flame background of the latter and the smaller mmple consumption rate favors the ketone. Introduction of a small amount of water into an acetone solution markedly diminishes the signal. I n fact, the tellurium signal is almost 200 times larger from a ketone solution than it is from an aqueous solution. The working curve jor the 238.6-mp line is linear up to 4000 ,ug. per milliliter of tellurium; for higher concentrations the curve bends toward the concentration axis. Self-absorption is evident for the 214.2-mp line w e n a t low tellurium concentrations dthough it does not become very pronounced up to 1000 pg. per milliliter. Interferences. Since t h e tellurium spectrum is a line spectrum i n the far ultraviolet, where tl-e Model D U
affords good resolution, even with a weak source, specificity is excellent. Examination of the compilation of spectral lines by Meggers, Corliss, and Scribner (7) provided a n indication of elements which might be expected to offer spectral interference either at the wavelength of the 238.6-mp line of tellurium or when attempting to measure the background signal on either side of the tellurium doublet. Only the Fe I1 spectrum gave serious spectral interference. When the iron mas present as the chloride, the Fe I1 spectrum was strong. By contrast, a sulfate medium delayed the dissociation of iron sufficiently and the Fe I1 spectrum did not appear. At slit widths 0.09 mm. or less, up to 100 pg. per milliliter of i-on could be tolerated in chloride medium; Le., the F e I1 lines at 238.2 and 238.9 mp could be resolved from the Te 238.6-mp line. Background would have to be measured a t 240 mp to avoid iron lines a t 237.2 and 239.6 mE*. Among the various types of samples containing tellurium, generally in amounts less than 1% by weight, tellurium will be accompanied by large quantities of matrix elements. The influence of many of these possible interferents is shown in Table IV. Although few offer serious interference the matrix elements will usually be present in high concentrations which is not conducive to ideal aspiration. Consequently, the extraction procedures were designed to separate tellurium from most matrix elements and from large amounts of the alkali and alkaline earth elements. Excitation Mechanism. Before concluding. it might be instructive t o speculate briefly on possible eucitation mechanisms. These facts are known. Tellurium evhibits its line spectrum only in the reaction zone of the flame. &laximum signal strength is achieved only when a n organic aerosol is injected into a n acetyleneoxygen flame. T h e relative emission intensity, as a function of the flame region viewed, parallels closely the pattern observed for the carbon atomic line, but occurs slightly higher in the flame than the band emission from C H and CZ molecules. Undoubtedly energetic conditions prevail in the flame. The excitation energy of the carbon line is 7.7 e.v.; for the F e I1 spectrum the ionization energy is 7.9 e.v. and the excitation energy of the ionic lines is 5.2 e.v. By contrast, the excitation energy of the tellurium lines is about 5.8 e.v.; a value too large for appreciable thermal excitation alone, as was obvious when aqueous solutions n-ere aspirated into an oxygen-acetylene flame. Many years ago Lundegsrdh (6) pointed out that the high reduction potential of the reaction zone induces
BACKGROUND
239.5
I
1
I
I
I
2
3
4
GAS FLOW, LITERS PER MINUTE Figure 3. Emission intensity of tellurium 238.6-mp line and background reading at 239.5 mp as a function of gas flow 0, ocetylene flow (oxygen flow = 3.3 liters per rnin.)
A, oxygen flow (acetylene flow = 1.18 liters per min.)
emissions from the negative elements of high ionization potential. The ionization potential of tellurium is 9.0 e.v. The chemiluminescent reaction postu-
Table 111. Influence of Solvent on Emission Intensity of Tellurium
(Uncorrected for sample feed rate) Emission sensitivity, Solvent p g . per ml. acetone, 1 0 0 ~ o 1.6 11.2 80% 6.0 Carbon tetrachloride Cyclohexane 4.0 Methj-l isobutyl ketone 3.0 Water 388
Table IV. Cation Interference Study (Present mas 125 pg. per ml. of tellurium and approximately 5000 pg. per ml. of each interferent examined) Tellurium Emission signal, lines, 0.1-mv. Element mr units Arsenic 237.1 60 238.1 rlntimony 238.3 59 58 Cadmium Chromium 23s:3 57 Cobalt 238.4 59 Copper 59 Iron 236:2 60 238.9 Lead 239.4 59
Manganese Nickel Tellurium (alone)
240.2
238:7
238.6
VOL. 35, NO. 6, MAY 1963
59
59.5
60
701
lated by Sternberg, and quoted by Gilbert (6), TeO
+C
+
Te*
+ CO
wherein the heat of formation of CO (11.1 e.v.) less the dissociation energy of TeO (4.0 e.v.) leaves 7.1 e.v. for the excitation of tellurium atoms, would seem feasible. The flame gases are undoubtedly saturated with carbon vapor and thus are capable of reducing the oxide (or preventing its formation). However, our failure to observe the atomic line of carbon in the reaction zone of hydrogen-oxygen-ketone flames indicates that the atomic carbon emission itself must be attributed to a different chemiexcitation process. The energy from the oxidation of CH to CO by atomic oxygen ( Z ) , which is exothermic to the extent of 7.6 e.v., might well be retained as electronic excitation of the CO. This highly energetic CO may be the precursor to the chemiluminescence of tellurium through the reactions (8):
+ CO* TeO + CO* Te
or
+
Te*
+ Te*
+ CO
+- Con
Some support for these postulations is found in the fact that the emission from the C H band system reaches its maximum intensity lower in the flame and apparently precedes the emission of the tellurium lines [see Figure 1 in Ref. ( 4 ) ] . Furthermore, it is known (2) that CH emission drops when hydrogen replaces a hydrocarbon; and the decrease is surprisingly rapid. Figure 2 shows a parallel trend among the emission intensity of tellurium and flame background which is composed largely of the Fourth Positive system of CO and whose intensity level is consistent with the thermal excitation for the triplet excited state of CO. Unfortunately atomic absorption equipment was unavailable to follow the concentration profile of atomic tellurium in different regions of the flame and under different excitation conditions.
LITERATURE CITED
( 1 ) Alkemade, C. Th. J., Ph.D. thesis, University of Utrecht, 1954. (2) Broida, H. P., Shuler, K. E., J. Chem. Phys. 27, 933 (1957).
(3) Carnes, W. J., Dean, J. A., Analyst 87. 748 (1962). (4) Dean, J. A,,‘ Carnes, W. J., Ibid., p. 743. (5) Gilbert, P. T., Jr., Proceedings Xth Colloquium Spectroscopiurn Internationale, pp. 171-215, Spartan Books, Washington, D. C., 1963. (6) Lundeghrdh, H., Lantbruks-HdgslzolAnn. 3, 49 (1936). (7) Meggers, W. F., Corliss, C. H., Scribner, B. F., “Tables of Spectral-line Intensities,” Part I, National Bureau of Standards Monograph 32, Washington, D. C., 1961. (8) Sternberg, J. C., Beckman Instruments, Inc., private communication, December 1962.
RECEIVED for review November 7 , 1962. Accepted February 13, 1963. Presented at Southeastern Regional Meeting, ACS, Gatlinburg, Tenn., Xovember 1962. J. C. Simms is indebted to the Kational Science Foundation for a post-doctoral summer research participant’s fellowship in 1962, which made possible this work.
A N e w Instrument for the Continuous Measurement of Condensation Nuclei GEORGE F. SKALA Advanced Technology laboratories, General Electric Co., Schenectady,
b An instrument i s described which provides a continuous measurement of airborne particle concentrations in the size range down to 0.001 micron. The concentration range of the instrument i s from 10 to 10’ particles per cubic centimeter with a response time in the order of 1 to 2 seconds. The instrument is designed for continuous unattended operation and employs a unique dual beam, single phototube optical system in which the reference beam passes through the same optical elements as the measuring beam.
T
HE FIRST experiments which
demonstrated the existence of nuclei as centers of condensation can be traced back as early as 1841 to the work of Espy (6). Further investigations were conducted by Aitlien ( I ) , who concluded that, except for the presence of nuclei, there would be no fog, no clouds, no mist, and probably no rain. He is also credited with making the first instrument for the measurement of nuclei concentrations ( 2 ) . In recognition of his contributions to the investigations of condensation nuclei, these particles are often referred to as Aitken nuclei.
702
ANALYnCAL CHEMISTRY
N, Y .
Typical nuclei concentrations range from 500 per cc. a t mid-ocean to 500,000 per cc. in metropolitan traffic conditions. While conducting experiments based on the works of Aitken and others on the expansion of air saturated with water vapor, Wilson devised his famous cloud chamber for the detection of radioactivity ( I 6). It was not until 1938 that photoelectric cells were mentioned as being applied to the measurement of condensation nuclei, by Bradbury and Meuron ( 3 ) . About the same time, L. IV. Pollak applied the photocell to a photoelectric counter of his design, and together with P. J. Nolan described its calibration (9). With the improvements that have been made in this counter, and the accumulated knowledge of its performance, it has become a very useful laboratory instrument and serves in many instances as tt calibration standard. A smaller, battery operated, portable counter which employs an expansion into a partially evacuated chamber has been described by Rich (IO). This device is now being manufactured commercially by Gardner Associates, Schenectady, N. Y.
All the counters referred to above are manually operated, and are capable of a single measurement a t a time. A number of investigators have constructed automated versions of various counter configurations in which electrically operated valves substitute for the human operator. Cycling rates have varied from one reading every 15 minutes ( I S ) to as many as four a minute (6). These have generally served their purpose, but because of the cyclic nature of the output, they present problems of read out and recording. Also, zero stability and adequate water supply can be problems for long term unattended operation. A counter in which hydrogen chloride gas is used to achieve a continuous supersaturation, making it capable of reading continuously, is described by Holl and Muhleisen ( 7 ) . However, because of the low flows employed, the response time is of the order of 1 minute. A nuclei counter which achieves a continuous reading by rapidly cycling the expansion ratio mas described in 1949 by Vonnegut (IC). An improved version, employing a different expansion system, was described by Rich (11). The counter of this paper ia based on
