Determination of Trace Metals in Petroleum Wet Ash-Spectrographic Method L. W. GAMBLE
and
W. H. JONES Oil Co.,
Esso Laboratories, Esso Standard
Louisiana Division, Baton Rouge, La.
Previous investigators have established that the quantitative determination of trace metals in petroleum and its products is of considerable significance. Data are presented in this paper which confirm that trace metals in some petroleum distillates are more completely recovered by a wet ashing or partial sulfuric acid decomposition method than by the conventional dry ashing procedure. A wet ashing-spectrographic procedure utilizing magnesium nitrate as an ashing and spectrographic aid has been developed. Nickel, vanadium, and manganese can be accurately determined at the 0.1- to 2.0-p.p.m. level using 10- to 50-gram samples. This procedure should also be applicable to trace quantities of iron, sodium, and copper, when magnesium nitrate free of these elements becomes available.
V
ARIOUS investigators have reported that small amounts
of metaIlic elements in crude petroleum and its products are of considerable significance in the petroleum industry. The presence of vanadium in fuel oils is undesirable because of its corrosive effect on turbine blades and its destructive effect on fire clay brick (4, 7 , 10). Trace quantities of nickel, vanadium, copper, and iron in catalytic cracking charge stocks contaminate the catalyst and cause greatly increased gas and coke yields and reduced gaeoline production ( 2 , 9).
has been reported by Milner and others (IO) and Karchmer and Gunn (8). Milner and coworkers presented a wet oxidation method for determining metals in crudes, residua, and overhead stocks. I n this procedure the sample is partially decomposed (coked) with sulfuric acid and the carbonaceous residue burned in a muffle furnace with added oxygen a t 886' to 976" F. Themetals are determined colorimetrically and ,polarographically. These authors concluded that this novel sulfation or wet ashing method should be used on distillate fractions to prevent loss of metah by volatilization. Milner has further pointed out that few data are available to show that metals are lost during the conventional ignition procedure. However, most of the investigators reporting no loss of metals by dry ashing have dealt with crudes, residual stocks, or synthetics prepared from nonvolatile metal organic compounds. I t is reasonable to postulate that the loss of metals, percentagewise, during dry ashing would be greater on distillate stocks which contain volatile metal compounds. Koodle and Chandler (16) concluded that crude oils contain iron, nickel, and vanadium compounds which volatilize during vacuum distillation. Wrightson ( 1 6 ) has shown that in the preparation of catalytic cracking charge stocks by vacuum distillation of reduced crudes, the metals content increases with depth of cut. Data (Figure I ) obtained in the authors' laboratory on the distribution of nickel and vanadium in the redistillation of heavy gas oils are supporting evidence for metal volatility. STUDY OF ASHING PROCEDURES
Considerable work has been done in this laboratory on the determination of metals in catalytic cracking feed stocks. Both the dry ashing (8) and the wet ashing (10)techniques have been studied, using colorimetric and spectrographic methods for determining the various metals. The wet ashing technique of Milner has been modified so that the operation is relatively simple and a considerable saving of time is effected. Using a ratio of 1 ml. of concentrated sulfuric acid to 1 gram of oil, samples up to about 100 grams can be quickly coked in Vycor beakers by rapidly heating the mixture on a high temperature hot plate (800' to 1000° F.). The coking reaction for most samples takes place in a fairly vigorous manner, and foaming or frothing over is prevented by skillful stirring and handling during the critical period which usually lasts only a few minutes. Heating of the carbonaceous residue on the hot plate is continued with occasional stirring until oxides of sulfur are no longer 0
IO
20
30
40
50
60
70
80
90
100
WT. 4: OIL DISTILLED
Figure 1.
Redistillation of heavy gas oils
Table I. Effect of Burning Kate on Metal Recovery by Dry Ashing 700-Gram Samples Dish size, ml. Burning rate, grams/hour
0. Laboratory vacuum distillation
.
Molecular distillation on bottoms from vacuum still
100 30
750 175
IO0 30
950 175
Total Ash, Nickel (as N i ) , P.P .M. P.P.hI. (Average Values) Heavy Gas Oil
For yeam the usual method for determining the mineral content of petroleum oils has consisted of burning the oil to a char and igniting the residue in a muffle furnace to remove the carbonaceous material. A brief resume of the analytical methods used in determining ash and its metallic components
hl K D C
A
1456
26.4 9.4 8.8 8.1
8.6
17.4 4.0 4.0
3.3 3.6
1.3 1.1 0.9 1.0
..
0.9 0.4
0.3 0.5
..
1457
V O L U M E 2 7 , NO. 9, S E P T E M B E R 1 9 5 5 Table 11. Comparison of Wet and Dry Ashing by Nickel Analyses on Heary Gas Oils Xickel in Oil, Parts per Million _ _ ~ _ _ _ ~ Dry Ashing ~ _ Individual .4ver-4verage determinations age
Heavy -~- .. . "etL4&g___Gaa Indi~-idual dt.trrininations Oil A
0.7,0.7, 1 0.0.7,0.9.0.7
0.8 0.5,0.5,O.B
B
1.1, 1 . 1 , 1 . 4
1.2
c
1.7,1.9,2 0. I 7
D
*
0 7, 0 6.0.!1
0.5 0.7
1.8 0.8, 1.2, 1.0
1.0
1.9, 2.0
2.0
0 9
E
0.6,0.7,0.8,0.0
0 . 7 0 . 2 . 0 2, 0 . 3
0.2
F G
1.7, 1.5
1 . 6 0.8, 1 0
0 9
2.3, 2.3, 2 . 3
2.3
1 . 0 , 1 . 3 , 0 4. 0 G
1 . 3 , 1.3, 1 . 3
I J
0 . 8 ; 0 . 7 , n . ~ . 0 . n , a . s , o . ~ , o 0. o. 7 0 . 3 , 0 . 3
K
2.5,2.5,27.2.7
2 4. 1 . 7 . 1 . 5
~~~
1.4 0.3
1 . 8 1 . 0 , 1 . 0 , 1 . 0 , 0 . 9 1.0 1.0. 1 . 1 1.1
2.6
~ ~ ~ ~ ~ _ _ _ _ _ _
evolved. The dry coke is finally burned to an inorganic residue in a muffle furnace at 1000" F. The ash is dissolved in concentrated hydrochloric acid for subsequent metals analyses. The colorimetric methods for iron, nickel, and vanadiud as described by Milner (IO), as well as the colorimetric method for vanadium as recommended by Wrightson ( 1 6 ) , have been used. T h e spectrographic method of Gamble and Kling (6) has been applied to both wet and dry ash residue from samples of 700 grams or more. I n applying the dry ashing procedure to heavy gas oils, the recovery of metals varies with the burning rate. The data in Table I show lower recoveries of total ash and of nickel when i00gram samples were allowed to burn freely in 4 hours using 750-ml. dishes aa compared to 24 hours employing 100-ml. dishes. Thus, Borne loss of ash material might be expected from burning a sample of any size. A comparison of dry ashing and m t ashing has been made on several gas oils. The results in Table I1 show that the nickel content is, on the average, 75% higher by the ITet ashing procedure. While the wet ashing method gives consistentlj- higher metal recovery than dry ashing, there is no absolute proof of complete recovery. Obviously, the preparation of known synthetics which sininlate natural samples in volatility characteristics of metah present is a difficult problem a t the present time. However, in the investigation of the wet ashing procedure an increase in the weight ratio of sulfnric acid to oil from 1 to 16 and also an increase in time of contacting at l o n w temperatures gave no greater nickel recovery. Thus, the wet ashing technique probably gives essentially complete metal recovery and only this type of ashing is reliable for diqtillate fractions.
trace quantities of iron, sodium, and copper when magnesium nitrate free of these elements is made available. Apparatus and Materials. SPECTROGRAPH. The instrument ARL Model 2060, 1.5-meter grating spectrograph, is equipped with a 24,000-line per inch concave grating xyith 5 cm. ( 2 inches) _ of _ ruled surface. The linear dispersion is 7 A. per mm. in first order. A cylindrical quartz lens of 5-inch focal length was used to form an image of the arc a t the grating. An intensity control stand employing metallized quartz filters was used for adjusting light intensity. EXCITATION SOURCE.ARL Model 4700, high precision multisource unit. COXPAR TOR-DENSITOVETER. BRL Model 2050, nonrecording type u i t h voltage regulator. DEVELOPIVG Uxrr. ARL, temperature controlled rocking machine. FILM. Eastman Kodak 30. SA-1. ELECTRODES. Special high purity graphite obtained froin National Carbon Co., United Carbon Co. (see Figure 2). MAGXESIUM SITR~T hlerck's E . c.P.; 0.20 gram per ml. of ethyl alcohol. COBALT SAPIITHENATE.Nuodex Concentrate; 0.015 mg. of cobalt per ml. of ethyl alcohol. Analytical Procedure. PREPARATION OF STASDARDS.Synthetic samples for calibration mere prepared by diluting chemically analyzed metal concentrates with iso-octyl alcohol. Analyses of the metal concentrates are as follows: Wt. 72 Metal in Concentrates 5 . 4 2 i 0.09 2.84 i 0 . 0 5 1 . 3 6 =t0 . 0 1 1.93 i 0.05 0.61 i0.02 0.95 i 0.03
Compound Piickel oleate Iron oleate Vanadium oleate Copper naphthenate Chromium oleate hfanganese oleate
Using these compounds, individual synthetic stock solutions containing 100 p.p.m. of metal were prepared. -4composite stock solution containing 10 p.p.m. of each metal was then prepared, By further dilution of this blend, synthetics used in the spectrographic calibration were obtained. FILJI EJfcLsros ~ . ~ L I B R . % TION CL-RVE. Using an iron arc, selected homogeneous lines between 3200 and 3300 -1.suggested by Dieke ( I ) irere employed in conatructing the curve.
h S A L Y T I C .4 I.
Q'L-
WET ASH-SPECTROGR4PHIC METHOD
In analyzing oils for metal3 at the 0.1-p.p.m. level, it is usually necessary to wet ash separately 100-gram or larger samples for each element in order to obtain reliable colorimetric determinations. The requirement of large samples is objectionable from the standpoint of time, size of beakers, muffle space, and quantity of sulfuric acid used. Also, in many cases the amount of ra-nple available may be limited. A wet ash-spectrographic method has been developed for the simultaneous determination of several metals. Using samples as small as 10 to 20 grams, nickel, vanadium, chromium, and manganese ran be determined with reasonable accuracy a t the 0.1-p.p.m. level Lon-er concentrations can be determined, of course, by using larger samples. Briefly, the method consists of wet ashing the oil sample to which magnesium nitrate and cobalt naphthenate, both dissolved in alcohol, have been added. The magnesium nitrate act3 as a carrier (mechanically and spectrographically), buffer, and intensifier, while the cobalt serves as the internal standard. Studies show that this procedure should also be applicable to
K O R KIS0
CL-RVES. T h e s y n t h e t i c standard samples yere ashed and arced under the same conditions as outlined in the proc-edure. The curves !yere obtxined by plotting micrograms of metal versus intensity ratio. Typical calibration curves are shown in Fignres 3, 4, and 5. SULFCRIC ACID DECOJIPOSITIOS . ~ N DASHING. The oil nple, warmed if necessary, iq weighed into B Vycor h e a k e r . T h e s a m p l e size Iwies depending on the metal r o n cen t r a t ion level. Tengram samples give ideal photographic densities €or metal concentrations ranging from 0.3 t,o 2.0 p.p.m. Below the Fipirte 2. hlodified 0.3-p.p.m. level, larger samplatform electrode ples of the order of' 20 to 50 grams are recommended. For samples up t o about 25 grams, a 25O-mI. Vycor beaker is adequate. Two milliliters of the alcoholic magnesium nitrate solution and exactly 1 ml. of the cobalt naphthenate solution are added to the oil and thoroughly mixed. For each gram of oil, 1 t o 2 ml. of metal-free 95% sulfuric acid is added. This mixture is then heated rapidly on a hot plate. Considerable frothing occurs when the sulfuric acid begins to react, and cautious stirring and heating are necessary a t this point. When the foaming subsides, the reaction is essentially complete and the oil has been reduced in only a few minutes to a char. Some refractory samples, such
1458
ANALYTICAL CHEMISTRY
as refined paraEnic mineral oil, require longer times of heating. The charred mass is heated on the hot plate for 1 to 3 hours until sulfur trioxide(a) fumes are no longer evolved. The resulting char is then ashed in the muffle furnace at 1000" F. The time required for complete ashing varies from 4 to 16 hours, depending on the type and quantity of oil sample. I n routine work the sample is normally left in the muffle overnight. The ash is then cooled and transferred Kith the aid of a camel's hair brush t o a small agate mortar for mixing.
4.0 3.0
0 2.0
F U
U
".L
2
Figure 3.
3
4 5 6
20
810
U U
O
DENSITOMETRY.
i l l i i
!
i
INI 3050.8 IC0 3044.0
I .0
Excitation Potential, E.V. 4.05 4 14
Line, A. 3044.0
Element Cobalt (internal standard) Nickel
Analytical working curve for manganese
EXCITATIOK .4ND
-4nalytical Lines
30
The ash is packed in the modified latform electrodes and arced against a 1/8-inch upper electrode. g e t a i l s for excitation and film development are outlined in Table 111. Normally, three electrodes are arced for one analysis. Thus, by using a split filter, six intensity ratios per sample are available. The transmittance of selected analytical lines (Table IF-) is determined on a nonrecordina densitometer. The densitometer is adjusted to 100% transmittance for any background adjacent to the line. The intensity ratios, analytical line to cohalt line, are determined by applying the transmittance values t o the film calibration curve. Byreferring theintensity ratio to the appropriate anal?;tird working curve the metal content is detrrminetl. SPECTROGRAPHIC
PI-
Spectrograph Grating aperture Open 40 Slit, microns Filters in intensity control stand 2-6-9 % transmittance 32 Electrodes, Special High Purity Gpper 1 1 8 inch in inch Lower Modified platform Electrode gap, mm. 3 Power Type unit ARL high precision Capacitance, mfd. 60 Inductance mh. 360 Resistance.'ohms 25 Volts 300 Amps 10 Rotary gap pointer 90 Exposure Time, Sec. Preburn None 30 Exposure Film Type SA-1 Development Developer D-19 Time, min. 3 Fixing time min. 2 Drying time. min. in infrared dryer 2
Table IV. I
MANGANESE, MICROGRAMS
2.
Table 111. Spectrographic Conditions for Excitation and Film Development
3002 5
3030.8 3183.4 2576.1
Vanadium Manganese
4.07
3.89
..
compounds were found t o be satisfactory as an ashing aid, carrier, and buffer, magnesium nitrate was chosen because of the increased sensitivity that it provided. Acid Blanks. The sulfuric acid is composited in 3-gallon batches and both chemical and spectrographic metal analyses are obtained on 300- to 400-gram samples prior to use in coking. Typical analyses are illustrated in Table V. Accuracy and Precision. The calibration data on seven synthetic samples ranging from 0.1 to 2.0 p.p.m. were obtained over a 4-week period. A total of about 45 to 50 samples were ashed and analyzed. Thus, the analytical working curves prepared from these data include errors involved in both the ashing and s ectrographic parts of the procedure. The precision of the nietEod for nickel was further studied by msking replicate analyses of several gas oils in which the nickel content varied from 0.1 to 2 p,p.m. (Tahle V I )
0 .6 0 .6
INi
3002.5
ICO3044.0 0 .4 0 .3
0 2
I
i
I
2
3
!
I / , I i l
4 5 6
810
I
I 20
! 30
N I C K E L , MICROGRAMS
Figure 4. Analytical working curves for nickel Choice of Buffer. Murray and Plagge (11)described a method which employs silica as an ash-aid in the metal analysis of petroleum oils. Others (3, 5, 6, 1 4 ) have used lithium carbonate as a matrix for determining the composition of ash obtained by the dry anhing procedure. O'Conner and Heinzelman (12, IS) recommended the use of magnesium nitrate for trace metals in vegetable oils and fats. I n their work an alcoholic solution of magnesium nitrate was added prior t o dry ashing. I n this laboratory the use of alcoholic solutions of lithium nitrate and magnesium nitrate was investigated. Although both
0.2
1I
I
I
1
I I I l I l l
2
3
4 5 6
I
810
I l l 20
VAN A DI UM, MICROGRAMS
Figure 5.
-tnalytical working curve f o r vanadium
30
1459
V O L U M E 2 7 , NO. 9, S E P T E M B E R 1 9 5 5 Table V. Batch ..i
Analyses. P.P.M. in HzSOa Iron Nickel Vanadium 0.04 0.004