Evaluation of Methods for Determination of Metallic Iron in Reduced Iron Ores S. C. Blum and T. D. Sear1 Esso Research and Engineering Company, Linden, N . J . 07036 IRONIS BEING PRODUCED in substantial quantities by the direct reduction of iron ores. The reduced ores contain gangue and iron oxides in addition t o metallic iron. An accurate method for the determination of metallic iron in the reduced product is required for research and commercial purposes. Some procedures have been available in the past, and several have appeared recently. When these were used with reduced ores, conflicting results were obtained. Absence of a National Bureau of Standards’ standard or similar standard of reduced iron ore necessitated the development of a technique for the evaluation of these methods. To d o this, a series of iron ores in various stages of reduction were analyzed for all elements. In particular, the analysis for oxygen was considered to be essential. The availability of neutron activation equipment afforded the opportunity t o measure oxygen. These analyses and the values for metallic iron and iron oxides obtained by the metallic iron method being evaluated were used to obtain a material and an oxygen balance. Three basic analytical techniques for metallic iron with modifications of each were evaluated. The first ( I ) employed mercury(I1) chloride. A variation of this is used as a standard by the Metal Powder Industry Federation (MPIF) ( 2 ) . The second (3), which uses silver thiocyanate, has been employed by the Institute D e Recherches D e La Siderurgie (IRSID) in France. A modification ( 4 ) tested by the association of German Foundrymen (VDEh) was also studied. Finally, a bromine-methanol reagent with (5) and without ( 6 ) reflux was studied. The latter modification was rejected, as it was not applicable t o reduced iron ores. A similar method ( 7 ) using a bromine-ethanol solution was not evaluated. EXPERIMENTAL
Reference Samples. Large quantities of five reduced iron ores ranging from 90 to 30% in metallic iron content were ground to pass 100 mesh and were homogenized. Each sample had been reduced independently in a pilot plant; none were prepared by blending other samples. Complete chemical analyses were performed by four laboratories, in duplicate, for total iron, phosphorus, sulfur, carbon, and the gangue elements. Total oxygen content was determined by neutron activation at two laboratories and the average value was used. Evaluation Technique. Reduced iron ores are composed of gangue, metallic iron, and iron(I1, 111) oxides. I n evaluating a metallic iron procedure, material balances and oxygen balances are obtained o n the series of reduced ores, by summing up the gangue values, calculated as the compounds, with (1) E. Merck, Z. Anal. Chem., 41,710 (1902). (2) Metal Powder Industries Federation, New York, MPIF Standard 7-61. (3) F. Marion and J. Aubrey, Chem. Anal., 41, # l o (1959). (4) Dr. G . Kraft, Metallgesellschaft, Frankfurt (Main)/Postfach 2609, private communication, January 1965. (5) G. Kraft and J. Fischer, Z. Anal. Chem., 197, 217 (1963). (6) S. Wakamatzu, Bunseki Kagaku, 14, 41 (1965). (7) W. Classen, “Analysenverfahren fur den Kraftswerksbetrieb,” Vulkan-Verlag 5.625, Essen (1962). 150
Table I. Composition of Reduced Iron Ores Sample Total iron, Total gangue,
z
2 4 5 6 7
93.7 91.6 87.5 88.6 78.7
z
5.4 4.7 7.4 4.6 5.1
the values obtained for metallic iron and the iron oxides by the procedure being studied. If a reagent did not dissolve all of the metallic iron, the remainder would be determined as iron oxides. The material balance in this case would total greater than loo%, and the oxygen calculated from the oxides would be greater than that determined by neutron activation. Conversely, if iron oxides were dissolved by a reagent and erroneously determined as metallic iron, the material balance would be less than loo%, and the calculated oxygen would be less than that determined by neutron activation. Analytical Method. For the modified bromine-methanol procedure, 0.5 gram of the reduced ore was refluxed for 15 and 60 minutes with 100 ml of a solution of 5% bromine in methanol. The resultant iron bromide was separated from the insoluble iron oxides by filtration. The oxides were dissolved by refluxing with hydrochloric acid in a nitrogen atmosphere. Iron(I1) was determined o n one aliquot by direct titration with dichromate. Iron(I1) and (111) was determined o n a separate aliquot by reduction of Iron(II1) with stannous chloride, followed by titration with dichromate. Bromine and methanol were removed by evaporating the filtrate to a low volume. Traces of organic matter were then destroyed by wet oxidation with nitric and sulfuric acid. CAUTION:Complete removal of alcohol and bromine are necessary before addition of nitric acid to avoid the possibility of a violent reaction. Metallic iron was determined on the filtrate by reduction with stannous chloride followed by dichromate titration. RESULTS AND DISCUSSION A summary of the elemental analysis of five reduced iron ores is presented in Tables I and 11. Sample 5 contained 4.1 % SiOl and 0.16 carbon, while Sample 2 contained 0.08% carbon. The gangue values for the other samples are close to the median. A summary of the metallic iron results obtained by all the methods is presented in Table 111. At low levels, the variation in metallic iron content is prohibitive. Although this variation diminishes with increasing metallic iron content, the range of values a t the 90% level is still statistically significant. The data obtained with the two silver thiocyanate procedures are given in Table IV. With the sample of highest metallization, the methods give comparable results although the value with the p H 4.5 reagent is a little higher. At the lower metallic iron level (higher iron oxide level) the results with the p H 4.5 reagent are over 2% higher. The material balance becomes poorer and the disparity between calculated
ANALYTICAL CHEMISTRY, VOL. 43, NO. 1, JANUARY 1971
Comp
z
Si02
AhOa
2.1
1.8
Table 11. Median Values of Gangue Components TiOz CaO MgO MnO 0.24
Table 111. Comparison of Results Obtained by Metallic Iron Methods Metallic iron, Sample N o . 4 2 4 5 6 7 Method HgCl,, Wilner 89.5 78.7 71.2 64.7 26.1 HgC12, MPIF ... 76.9 AgCNS, pH 5 . 1 91.1 82.1 77.4 72.8 40.0 AgCNS,pH4.5 91.5 84.4 Br-MeOH,60Min 91.3 80.6 74.5 69.6 29.9 Br-MeOH, 15Min 91.3 80.7 74.3 69.8 28.9 Range 2.0 7.5 6.2 8.1 13.9 a Each value average of at least duplicate determinations.
z
Table IV. Evaluation of Silver Thiocyanate Method PH of Solution --+
Metallic iron,
z
Calcd oxygen Material bal Z Measda 5 . 1 4.5 5.1
- ~.
4.5
5.1
4.5
91.5 84.4
91.1 82.1 77.4 72.8 40.0
3.2 4.9
Sample 2 4 5 6 7 a
3.2 5.9 8.4 8.5 18.2
3.6 5.7 7.5 7.6 15.6
99.5 99.0
99.9 99.4 99.6 98.1 98.2
By neutron activation.
Table V. Evaluation of Mercury(I1) Chloride Method Wilner method Metallic Oxygen, Z Material Sample iron, Measure& Calcd balance, % 2 4 6 7
89.5 78.7 71.2 64.7 26.1
3.2 5.9 8.4 8.5 18.2
4
76.9
5.9
5
4.1 6.7 9.0 10.1 19.7
100.6 100.6 loo. 2 100.9 100.9
flux
time, Metallic iron Min 60 15 Sample
100.8
2 4 5 6 7
Table VI. Effect of Bromine-Methanol Reflux Time Reflux, Metallic min iron, % 22.7 27.1 28.9 29.3 29.4 33.0 ~
~~~
~
that the material balances exceed loo%, combined with a consistently high calculated oxygen, suggests that the metallic iron values are erroneously low. This suggests the possibility that the mercury sludge occludes some metallic iron, rendering it unavailable to the reagent. Bromine-Methanol Data. Data from a study made with Sample 7 on the effect of reflux time on metallic iron are presented in Table VI. As the data indicated that reaction times of 1 5 or 60 min gave small differences, both times were studied with the five samples (Table VII). At high levels of metallic iron, both dissolution times gave equivalent results. Below 70%, some differences were observed. This was supported by increasing differences in the material balances. Since the material balance is poorer with decreasing metallic iron content when a 60-minute reflux time is used, it suggests that this reflux time may be too severe. However, n o difference in oxygen balance was observed. In view of the relatively constant and nearly 100% material balance, it would appear that the 15-minute reflux time has some advantage over a wider range of metallic iron. It has a definite advantage in the shorter analytical time. Iron carbide reacts like metallic iron in the brominemethanol procedure, but its behavior with the other reagents is not known. However, the carbon content of the reduced ores, Table 11, is low and would not affect the conclusions drawn from this study. To obtain a measure of precision, the 15-minute methanol method was cooperatively tested by having four laboratories analyze the 5 samples in duplicate. The results presented in Table VI11 show a standard deviation of 0.6.
Table VII. Evaluation of Bromine-Methanol Method
Neutron activation.
2.5 5 15 30 60 240
S 0.01
0.028
Re-
MPIF method 7.2
C
pzo5 0.32