creases as pH decreases, which, in general, is what would be expected from the pH dependence of copper-organic chelates. To explain these data in that context, however, the interfering species would have to be particularly basic so that a small decrease in p H a t pH 8 would lead to a large shift in copper concentration. The same experiment carried out on the San Diego water resulted in a small shift in potential on changing pH from 8 to 5 (20 mV), but a large decrease in potential on changing pH from 5 to 3 (50 mV). The addition of copper spikes to the San Diego seawaters developed a curve similar to that shown in Figure 5 for the Freeport water but with a more reasonable “Nernstian slope” a t high copper concentrations ( > 2 ppb)-ie., 60 mV/decade rather than 125 mV/decade. The actual potentials observed for the San Diego samples were also displaced to more negative values for a given amount of copper relative to those observed for the Port Isabel samples but not as negative as observed from the Freeport samples. For example, the San Diego samples containing 3.8 ppb copper generated a potential of -75.8 mV; at the same concentration, the Port Isabel sample generated a potential of -55 mV. Atomic absorption spectrometric analysis indicated the presence of 0.3 i~ 0.12 ppb copper in the San Diego water, which was sufficient to have generated potentials of the order of -110 mV for the unspiked solution rather than the -140 mV actually observed. Rather than simply having an inherently low copper content, it would appear that the San Diego sample also contained interfering species, presumably organic chelating agents, but to a lesser extent than the Freeport sample.
As discussed, three different groups of seawater samples were studied. The most representative of the open ocean was the Port Isabel sample taken 5 miles offshore from a nonindustrialized area; the Freeport samples taken 1 mile offshore were most likely representative of nearshore water, the composition of which may have been influenced by industrial activity; and the San Diego sample was also of a nearshore water, the composition of which may have been influenced by industrial activity and especially heavy shipping. The analytical method described for the measurement of ionic copper was developed primarily with one large sample of the seawater taken off Port Isabel. Application of the method to the other two seawater samples indicated a much more complex chemistry. Shown in Figure 5 are the measured potentials after adding copper spikes to the Freeport sample, as received and a t p H 3. These potential readings were taken a t least 15 minutes after addition; in all cases, there was still a slow drift to more positive potentials. The “Nernstian slope” between the last two points was 125 mV/decade (copper additions >5 ppb were not made to avoid possible complications from pH changes induced by the p H 2 CuClz spikes); the initial potential was abnormally low (-171.0 mV) compared to what was anticipated from the AA analysis (0.4 ppb) and compared to the potentials observed for 0.2 ppb copper in 0.5M NaCl (-116.5 mV); and the residual copper concentration calculated from the potential change on adding the first copper spike of 0.63 ppb (and the 50 mV/decade slope already established for this electrode in the Port Isabel water) was 1.1 ppb. Further additions gave calculated residual copper values decreasing to 0.4 ppb with the final spike. All effects are essentially what would be expected from the presence of small quantities (10-7M) of chelating agents, as discussed; indeed, such materials are produced, used, and discharged into the ocean from industrial sites a t Freeport, Texas. The addition of copper spikes to this solution a t pH 3 resulted in a monotonic increase in potential rather than the titration effect observed at pH 8 (Figure 5). Figure 6 shows the steady-state response of the sensor in the Freeport sample as a function of pH. The implication of these data is that the free copper concentration in-
ACKNOWLEDGMENT The authors wish to acknowledge the assistance of Roy Deviney in performing the experimental measurements and the assistance of Gary K. Rice in evaluating the data. Received for review June 7 , 1973. Accepted October 9, 1973. This work was sponsored by Advanced Research Proiects Aeencv on Order No. 2199 and administrated bv the“ Office-of Naval Research under Contract NO. OOIC 72-C-0368.
Spectrochemical Method for the Determination of 36 Elements in Industrial Effluent V. M. LeRoy and A. J. Lincoln lnstrumental Analysis Laboratory, Engelhard Minerals & Chemicals Carp., P.O. Box 2307, Newark, N.J. 071 1 4
A quantitative spectrochemical method is described for the determination of 36 elements in industrial effluent with limits of element sensitivities for the analytical curves in the 0.0005-0.1 mg/100 ml range. One Set Of analytical curves is used for all samples. The samples are weighed salt residues obtained through evaporation to dryness at 120 “C. Any sample volume can be used. Matrix effects are significantly reduced by combining total energy burns in an argon-oxygen atmosphere with 1:6 salt to high purity graphite powder dilutions. Gerrnaniurn dioxide, added to the graphite powder, is the internal
standard. Conditions of analysis, salt weight ratios, preparation of standards, precision, and accuracy are discussed.
Industrial waste and inadequate domestic sewage treatment plants have become the primary targets for initiating programs to achieve responsible control of the two major sources of water pollution. In the precious metals industry, a continuous study of plant waste streams has always been a natural sequence in the refining and manufacturing processes. In addition to the analyses for pre-
ANALYTICAL CHEMISTRY, VOL. 46, NO. 3, MARCH 1974
369
cious metal content, the need to determine the presence and concentrations of possible toxic and other impurity elements in effluent streams accelerated the development of an emission spectrochemical method. A survey of the literature revealed a large volume of different analytical techniques including polarography, atomic absorption spectrophotometry, neutron activation, spark source mass spectroscopy, colorimetry, and X-ray fluorescence, in addition to the standard wet chemical methods published by ASTM ( I ) and the Environmental Protection Agency ( 2 ) . A number of spectrochemical methods have been published including a comprehensive review of spectrographic techniques made by Hitchcock and Starr (3) for the analysis of sea water. A copper spark method for the determination of Sr in water was described by Skougstad ( 4 ) . An ion-exchange spectrographic technique for the determination of As and P in river water, in addition to an evaporation-dilution technique for 12 trace elements were reported by KO (5, 6) and Daniel and KO (7). Several excellent papers have been published by Kopp and Kroner (8-11) and Kopp (12, 13), describing multiple techniques suitable for photographic and direct reading spectrographic instrumentation for some 19 trace elements. Barnett (14) reported on a common matrix residue method for the determination of 23 trace elements in natural waters. Eremenko (15) described a chemical extraction spectrographic procedure for 8 elements in natural waters. Concentration spectrochemical methods were reported by Silvey (16) and Silvey and Brennan (17). Since the published methods did not satisfy the requirements of our project, a residual salt-weight ratio method was developed to permit the use of one set of analytical curves.for the analysis of all our waste discharge streams. The described method proved more than adequate for assuming a major portion of the effluent analytical work.
EXPERIMENTAL Apparatus. The spectrographic instrumentation used in this study includes a 2M Applied Research -Laboratory Spectrograph with both 24,000 and 36,000 linesiinch gratings and the ARL Multisource Model No. 5700. A Jarrell-Ash Recording Microphotometer Model No. 23-100 was used to obtain photometric measurements for computer calculations. The 70% argon-30% oxygen atmosphere is attained with a Stallwood Jet. For mechanical grinding and blending operations Spex Mixer/Mills No. 5000 and No. 8000 were used. ASTM, Part 23, Water; Atmospheric Analysis (1970). "Standard Methods for the Examination of Water and Wastewater," 1 3 t h ed.. U. S. Government Printing Office, Washington, D.C. 1971. R. D. Hitchcock and W. L . Starr, Appl. Spectrosc.,8,5-16 (1954). N . w. Skougstad, U . S . Geol. Surv.. Water-Supply Pap.. 1496-6, 19-31 (1961). R . K O , U.S. At. Energy Comm., HW-59008,14 pp (1959). R. K O , U. S. At. Energy Comm., HW-48770,1 1 pp (1957). J. L. Daniel and R. KO, U.S. At. Energy Comm., HW-43096 (1956) J. F. Kopp and R. C. Kroner, Appl. Spectrosc., 19, ( 5 ) , 155-9 (1 965) J. F. Kopp and R. C. Kroner, J. Water Pollut. Contr. Fed., 39, (Pt l), 1659-68 (1967). J . F. Kopp and R. Kroner, Develop. Appl. Spectrosc., 6, 339-52 (1967)( P u b . 1968). R. C . Kroner and J. F. Kopp, Anal, Instrum., 4, 91-9 (1966) (Pub. 1967). J. F . Kopp, "Trace Subst. Environ. Health-3, Proc. Univ., Mo. Annu. Conf., 3rd 1969" ( P u b . 1970),Delbert D. Hemphill, E d . , Univ. of Missouri, Columbia, Mo., pp 59-73. J. F. Kopp, "Methods for Emission Spectrochemical Analysis," American Society for Testing and Materials, Philadelphia, Pa., E-2 SM-11-16(1971). P. R. Barnett, "Methods for Emission Spectrochemical Analysis," American Society for Testing and Materials, Philadelphia, Pa.. E-3 SM-11-17(1 971 ) . V. Ya. Erernenko, Gidrokhim, Mater., 29,248-53 (1959). W. D. Silvey, U. S. Geol. Survey Water-Supply Pap., 1540-6,11-22 (1961). W . D. Silvey and R. Brennan, Anal. Chem., 34,784-6 (1962).
Procedure. Primary plant waste, streams and composite samples taken around the Newark facilities were studied for dissolved and total solid content, acidity-alkalinity, and anion-cation composition. The preliminary data showed a predominant chloride system containing minor amounts of nitrates and sulfates in addition to their free acid counterparts. Although most of the basic streams were ammoniacal, some solutions containing free sodium hydroxide were also encountered. Sodium and calcium were found to be the only persistent matrix elements and these were in varying concentrations. Preliminary synthetic samples prepared with several important impurity elements added to sodium chloride matrices ranging from 20 to 300 mg N a p 0 0 ml showed no signifi-. cant matrix effects on the impurity results. The effluent samples studied over a period of several months indicated a 20 to 25% concentration of alkali metals and/or alkali-alkaline earth combinations in the residual salts. Therefore, no common or synthetic matrix was added to the sample solution or the residual salt except in certain special cases discussed under "Sample Preparation." Since the total and/or dissolved solid content in all effluent samples is routinely determined, the dry salt residue obtained provided a suitable spectrographic sample that included pre-concentration of the impurity elements and the weight per volume data. The weights obtained from the original experimental samples ranged from 0.1 to 0.5 gram/100 ml with an occasional total solid content as high as 1.5 grams/100 ml. By using an average weight of approximately 0.75 gram/100 ml and adding an adjustable synthetic matrix such as sodium chloride to the standard solutions, reasonably constant weight residual salt standards could be obtained. The ratio derived from the sample weight per volume and standard weight per volume adjusts the results read from the analytical curves to the original sample and provides additional flexibility for evaporating any volume of sample to meet the minimum salt requirements of the method. To reduce the matrix and interlement effects to a minimum, total energy burns in a 70% argon-30% oxygen atmosphere combined with 1%dilution of salt with high purity graphite powder containing germanium dioxide were used. The argon-oxygen atmosphere stabilizes the arc and promotes uniform emission of the impurity elements. In addition, the cyanogen interference is significantly reduced. The 1:6 mixtures provide a substitute graphite matrix and effectively control the vapor density in the gas column. A number of oxides and salts including lithium carbonate, indium oxide, the fluoride and chloride salts of strontium and barium were investigated as possible internal standards. Germanium, as high purity germanium dioxide, was selected because of its special characteristics including rarity in sample, medium volatility, spectrum simplicity, and relative freedom from interelement effect from both heavy and light alkali metals. The combination of these features gave greatly improved precision and accuracy to the method. Persistent surveillance of the intensity of the germanium internal standard lines provides an excellent monitor for detecting burn upsets or possible sample matrix problems. Usually these effects are observed as a significant weakening in the intensity of the internal standard lines. The primary burn shown in Table I (Type A) was biased for the determination of all impurity elements a t very low concentrations except mercury, potassium, and cadmium. The special burn for these elements (Type B) requires a large sample volume of the 1:6 mixture (200 mg), a special electrode, and an air burn. The block type Engelhard No. 150 electrode (18) is shown in Figure 1. Cadmium was added to this burn because Cd 2288 was excluded by the instrumental restrictions imposed by grating dispersion and filter arrangements used in the primary burn. A secondary burn (Type C) was included to determine high concentration levels for certain important elements. Occasionally, dilutions with sodium chloride were necessary since it was impossible to predict the number of elements and the extent of the concentration levels for all samples. Preparation of Standards. The original synthetic matrix selected for the standards was composed of sodium chloride as the major and calcium chloride as the minor constituent. During the investigation, it was observed that calcium did not exert any particular influence on the impurity elements and was, subsequently, eliminated in the standard preparation. For example, the element intensity ratios obtained from standards prepared with a sodium matrix produced curves with identical slopes as the sodium-calcium matrix curves. Because the weight ratio adjusts the values from the analytical (18) V . LeRoy and A. J. Lincoln, Develop. Appl. SpeCtfOSC.,8, 199-215 (1970).
ANALYTICAL CHEMISTRY, VOL. 46, NO. 3, MARCH 1974
Table I. Spectrographic Operating Parameters T y p e of burn W a v e l e n g t h , rl Voltage, dc Current, A E x p o s u r e , sec Slit W i d t h , microns Electrodes: A n o d e ( +) C o u n t e r I -) S a m p l e charge, m g Atmosphere F i l t e r system
Emulsion Developer F i l m calibration
A (primary) 2300-4000 290 15 120 40
2180-4600 290 15 30 40
H i g h purity g r a p h i t e E n g e l h a r d No. 1 5 0 Flat face 1/8 r o d
S-12 (ASTM) 40 ) 70 % Argon-30% oxygen P r i m a r y filtering at slit-split field 50 TjlOOT
S-4 (ASTM) 10 70% Argon-30 % oxygen Primary filtering at slit-split field 50 TI100 w i t h selected lines i n d e p e n d e n t l y filtered
200
Air P r i m a r y filtering at slit-split field 50 T / l O O
K o d a k SA-1 K o d a k D-19 5 min at 65 O F C o m p u t e r reduced data derived f r o m modified 2-line iron a r c m e t h o d (20)
Thiers, "Separation, Concentration, and Contamination in Trace Analysis,'' J . H . Yoe and H. J. Koch, Jr., Ed., Wiley, New York. N . Y . . 1957, p p 637-66. (20) A . Carnevale and A . J. Lincoln, Spectrochirn. Acta, 248, 313 (1969).
C (high concentrations)
2180-4600 290
15 30 40
curve to the sample, the evaporated salt weights for each standard within a series and between the various series had to be made comparable to each other to avoid multiple correction factors. The standard matrix salts and trace element additions per 100 ml were calculated to deliver approximately 0.75 gram/100 ml of residual salt weight following evaporation. The standard series were prepared by dissolving sodium and/or calcium chloride salts containing 200-300 mg Na/100 ml and 10-50 mg of Ca/100 ml in a 100-ml volumetric flask. When sodium chloride is used as a single matrix salt, 250-300 mg of Na/100 ml is used. A total of 11 series of standards was prepared for this investigation. Nine series were prepared as a chloride system, one as a nitrate system for silver, and one as a magnesium-calcium matrix for the determination of sodium. The impurity elements were added from successive dilutions of Spex Industries Standard Solutions and from 0.1, 0.5, and 1.0 gram/l. stock solutions containing nitrates, chlorides, and ammonium salts. The solutions were adjusted to volume with triple-distilled water and mixed thoroughly. Considerable care was taken to ensure compatible anion-cation mixtures. Each standard was evaporated to dryness in a porcelain dish a t 120 "C and the residual salt weighed. The salt was mechanically blended to ensure homogeneity. An overall average weight of 0.7607 gram/100 ml was obtained for nine series of chloride standards with a standard deviation of ztO.0107 gram/100 ml. Similarly, the average weight of the nitrate series for silver. 1.204 gramsilo0 ml, and the magnesium-calcium series for sodium, 0.9860 gram/100 ml, are used in the final calculations. The hygroscopic nature of many samples and standards suggests the work be done in a humidity-controlled atmosphere and stored in a desiccator. Sample Preparation. All effluent samples should be processed immediately. Both the total and dissolved solid concentration can be determined. 1. Dissolced Solids. To determine impurity elements in dissolved solids, the sample solution is filtered through 0.3-micron filter paper and 100 ml of the filtrate is evaporated. 2. Total Solids. The entire sample is homogenized on a magnetic stir plate. The length of blend time is dependent upon the physical state of the solids present. Evaporate 100 ml of the thoroughly blended solution to dryness on low heat (150 O F ) under infrared lamps. Prolonged baking under lamps is to be avoided. Evaporation chambers such as the one designed by Thiers (19) can be used if special protection from atmospheric contamination is necessary. The dry salt is transferred to a 120 "C oven for 2 hours, cooled in a desiccator, and the salt weight obtained. 3. Samples Containing Free NaOH. A 25-ml portion of the filtered or homogenized sample is evaporated to dryness in a weighing bottle. An additional 25 ml is transferred to a porcelain dish, acidified with HCI, and evaporated to dry in the described man(19) R. E.
B (Hg-Cd-K)
t-
0493.Ioo2_L/
I
I
1 7,003 0 150"
I
7, 0 156'-
0°'
t
I-') 10 125"
TO
FIT
S-
PEDESTAL
I
1, 1
I,
4 +E* Figure 1. High-purity spectrographic grade graphite electrode No. 150 ner. This is the spectrographic sample. Salt weights for both samples are recorded for use in the final calculations. 4. Excessively Hygroscopic Salt Samples. These samples may contain high concentrations of Zn, Ni, and Cu as nitrates which are too hygroscopic t o process. Twenty-five ml of the sample are evaporated in a tared weighing bottle and the residual salt weight is recorded. An additional 25 ml of solution is now added to an amount of sodium chloride equal to the salt weight of the first 25 ml of sample. The solution and salt are mixed and evaporated in the usual manner. The weights are recorded for use in the final calculations. 5. Single Matrix Samples Containing Traces of Na, Ca, Mg, and Other Impurities. This type sample can be easily processed but it is necessary to add sodium chloride. Only two samples of this type were encountered, one was an aqueous solution containing a large quantity of suspended A1203 particles and the other a totally organic matrix containing traces of sodium. Following evaporation of 100 ml of the blended solution, a 3:l mix of residual salt and NaCl was made to compensate for the sodium deficiency in the spectrographic sample and the weakening of the germanium internal standard lines. Other organic matrix samples containing minor amounts of sodium were analyzed according to the routine procedure. A master mixture of high purity conducting graphite powder containing 20% germanium dioxide (Spex Industries High Purity GeOz) is prepared by both manual and mechanical blending operations. Successive dilutions with graphite powder are made to provide 0.002 and 0.005% germanium dioxide working mixtures. T o prepare the electrode charges necessary to determine all low concentration elements including mercury, potassium, and cadmium, combine 80 mg of salt and 480 mg of 0.002% germanium
ANALYTICAL CHEMISTRY, VOL. 46, NO. 3, MARCH 1974
371
Table 111. Accuracy of Method: Comparative Spectrographic and AAS D a t a o n S y n t h e t i c Effluent HR-400
Table 11. Analytical Lines, Concentration Ranges, and Precision D a t a Element
Ge (IS) (IS) (IS) Na Fe Si Ni
Mn Cr
cu Be Zn A1
Pd
Rh Pt Ir Ru Ag
Mo Pb Ti
Zr AS
Sn Au Sb Bi
V Sr Te Cd
Ca Mg Hg
K P
co
Analytical lines, i
2691 .34 2709.63 3039.06 3302.32 2680.33 3047.60 2635.81 3490.57 2514.33 2436.16 3414.76 3105.47 2576.16 2933 .06 3021 .56 3005.06 2975.48 3273.96 2961.16 3131.07 3282.33 3075.90 3082.15 2652.49 3242.70 3434.89 2659 .45 3220.78 3436.74 3382.89 2816.15 2833.07 3372,80 3391.97 2780.20 3262.33 2675.95 2427.95 2598.06 3067.72 3217.02 3464.45 2385.76 2288.02 3261.06 3179.33 2398 .56 2779 .83 2790.79 2536.52 3447.70 2553.28 2554.93 3453.50 3449 .44
Concentration range, rng/100 rnl
Re1 std dev, 9%
Elements
0.1
-350
0.005 - 1 . 0 0.5 - 10 1.0 - 50 0.01 - 1 . 0 0.1 - 2.0 0.001 - 0 . 1 0.1 - 5.0 0,001 - 0 . 0 1 0.01 - 0.5 0.001 - 0 . 0 1 0.01 - 2.0 1.0 - 50. 0.001 - 0.25 0.1 - 50. 0.001 - 0 . 1 0.001 -200 1.0 - 50 0.001 - 0 . 1 - 20. 0.1 0.002 - 0 . 1 0.001 - 0 . 1 0.005 - 1.0 0.007 - 0 . 2 0.001 - 0.5 0.0005- 1 . 0 0.002 - 2.0 0.001 - 0 . 5 0.001 - 0 . 1 0.001 - 0 . 1 0.02 - 0 . 1 0.004 - 0 . 1 0.002 - 0.04 0.004 - 0 . 1 0.004 - 0 . 1 0.001 - 0 . 1 0.005 - 5 . 0 0.008 - 0 . 3 0.01 - 5.0 0.001 - 0.05 0.01 - 0 . 1 0.001 - 0 . 5 0.1 - 50. 0.001 - 0 . 2 0 . 1 - 2.0 0.05 - 10. 0.1 - 10. 0.1 - 5.0 0.5 - 10.0 0 . 0 0 1 - 0.01 0.01 - 1.0
13.3 10 .o
10 .o
Mg
cu
Ni Pb Ti*
9.3 11.1
Mn Pt Cd
9.8
Pd
Rh Au Cr
8.O 18.6 11.o
Mo Ber
7.3
...
984 26 22 5.4 4.0 5.2 5.2 1. 0 0.92 0.87 1.1 1. 0
1000 25 20 5 5 5 5 1 1 1 1 1 1
24.82 19.6 4.9 5.77 5.1 4.0 0.96 1.04 1.o 1.o
1 .o 1 .o
1.o 1.1
1 0.5 0.1 0.1 1 .o
0.42 0.11 0.14 0.12 0.98
1. 0
1.1
0.5
0.51
0.1
1.o 0.47 0.1 0.1 0.1
... ...
...
mg/100 ml (from analytical curves), B = sample weighti100 ml, C = sample weight/25 ml (as received), H = sample + HC1 weighti25 ml, iV = sample + NaC1 weight/25 ml, D = dilution factor, S = standard weighti100 ml. Calcuiation., f o r S a m p i e s inciuded in 1 and 2.
B 3 x
10 = mg/l.
X
40 X
-
S
X
40 x C = mg/].
B S
X
A x C a l ~ ~ l ~ t if o rnSsa m p l e s included in 3
A
X
C
-
S
H C
= mg/l.
Calculations ,for Samples inciuded in 4
A
10.4
X
C
-
N
Calculation,