K , = equilibrium ratio of COz/CO k = reaction rate constant, min-1 atm-’ k , = mass transfer coefficient, cm/sec m = proportionality constant, k , = m u nc = gram-moles of carbon deposited ncOn = gram-moles of COS P = total pressure, atm A P = pressure drop across the bed Apw = weight of the bed per unit cross-sectional area R = gasconstant ro = constant associated with reduction rate (gram-oxygen removed)/(gram-oxygen) (min) T = temperature t = time, min u = superficial velocity U P = superficial velocity corrected for carbon deposition u: = oxygen concentration in iron oxide, gram-atom oxygen/gram-Fe tu0 = initial oxygen concentration in iron oxide, gramatom oxygen/gram-Fe Y d = mole fraction of Con generated because of carbon deposition ys, yb = mole fraction of CO2 a t the particle surface and in the bulk phase, respectively Greek Letters
PI,
p2 = constants for carbon deposition rate, gram-atom carbon deposited per unit fractional reduction
Phys. Chem., 53,240-55 (1967); CA, 66,107136~(1967). In Ger. Bogdandy, L. V., Engell, H. J., “The Reduction of Iron Ores.” pp 256-63, Springer-Verlag, New York, N.Y., 1971. Bogdandy, L. V., Riecke, H. G., Arch. Eisenhuttentc., 29, 603-9 (1958); CA, 53,2025b (1959). In Ger. Ess, S. Y. M., Wild, R.. J . Iron Steel Inst.. London, 194, 211-21 ( 1960). Feinman, J . , Ind. Eng. Chem., Process Des. Decelop., 3, (3), 241-7 (1964). Hansen, J. P . Bitsianes, G., Joseph, T. L.. Blast Furn. Coke Oven, Raw Mater., Proc., 19, 185-99 (1960). Kettenring, K. N.,Manderfield, E. L., Smith, J. M., Chem. Eng. Propr. 46. 139-45 (1950). Kunii:D. Levenspiel, O., Ind. Eng. Chem., Process Des. Decelop., 7 (4), 481-92 (1968). Kunii. D.. Suzuki. M.. Int. J. Heat Mass Transfer. 10. 845-52 (1967). McKewan, W. M.. Trans. AIME, 221, 140-5 (1961). Meissner, H . P., Schora, F. C., ibid., 221,1221-5 (1961). Okura. A , , Matsushita, Y., Tetsu To Hagane, 50, 159-65 (1964); CA, 63,6646f (1965). In Jap. Osman, M. A,, Manning, F. S., Philbrook, W. O., Amer. Inst. Chem. Eng. J., 12,685-92 (1966). Resnick, W. E., White R. R., Chem. Eng. Progr. 45, 377-90 (1949). Ricetti, R. E., Thodos, G., Amer. Inst. Chem. Eng. J . , 7, 442-4 (1961). Richardson, J. F.. Szekely, J., Trans. Inst. Chem. Eng., 39, 21222 (1961). Smith, N. D., McKewan, W. M., Blast Furn. Coke Oven Ratc Mater. Proc., 21,3-13 (1962). Stelling, D.. J . Metals, 10, 290-5 (1958). Themelis, N. J., Gauvin, W . H.. Trans. AIME, 227, 290-300 (1963). Udy. M., Lorig, C., ibid., 154, 162-81 (1943).
Literature Cited Agarwal, J. C., Davis, W. L., Jr., Chem Eng. Progr Symp. S e r , 62 (67), 101-10 (1966). Boetticher, H., Bogdandy, L. V., Foerster, E., Schierloh, U., Z
Receiced f o r revietc October 26, 1972. Accepted April 11. 1973. Work supported bx a grant from the Cnited States Department of the Interior.
Recovery of Metallic Iron from Flotation Tailings by Pneumatolytic Transport II. Formation of Iron Pentacarbonyl from Partially Reduced Iron Oxide in Fixed Bed Chin S. Rhee,’ Chung S. Kim,* Kun Li,3 and Robert R. Rothfus Department of Chemical Engineering, Carnegie-Mellon University, Pittsburgh, Pa. 15213
A conceived process for recovery of valuable metallic iron from waste iron ore flotation tailings by pneumatolytic transport involves reduction followed by carbonyl formation. In this work, the formation of iron pentacarbonyl from partially reduced flotation tailings in a fixed bed was studied experimentally at temperatures from 93-177”C, pressures from ’7.8-21 atm, and gas flows from 0.14-0.52 cm/sec. The surface reaction was found to control the overall reaction rate which varies approximately as the second power of carbon monoxide pressure. A deactivation term was introduced to account for the increasing difficulty of interaction of carbon monoxide molecules and iron atoms as reaction proceeded. A model based on surface reaction and deactivation adequately describes the kinetics of carbonyl formation. The existence of an optimum temperature a t 120°C is believed to be associated with the mobility of adsorbed carbon monoxide molecules. Reduction of iron oxide by carbon monoxide a t low temperature and low pressure produces more reactive iron than reduction a t higher temperatures and pressures. Present address, Electronic Associates, Inc., West Long Branch, N . J . 07764. Present address, Graham Research Laboratory, Jones & Laughlin Steel Corp.. Pittsburgh, Pa. 15230. 3 To whom correspondence should be addressed. 730
Environmental Science & Technology
Iron pentacarbonyl is known to be unstable a t high temperatures, decomposing readily to iron and carbon monoxide. The kinetics of decomposition have been studied recently (Carlton and Oxley, 1965), but kinetic information on formation is limited. The only kinetic equation reported in the literature is that of Dufour-Berte and Pasero (1967) for carbonyl formation from hydrogen-reduced iron, apparently developed by curve fitting of conversion data using a topochemical model. No work has been found in the literature on the kinetics of formation of carbonyl from carbon monoxide-reduced iron, even though from a practical standpoint the use of carbon monoxide for reducing iron oxide might be preferable to the use of hydrogen. The rate of formation of iron pentacarbonyl from hydrogen-reduced iron is approximately proportional to the second power of carbon monoxide pressure according to Stoffel (1914) and Dufour-Berte and Pasero (1967). For a given pressure there exists an optimum temperature (Mond and Wallis, 1922; Okamura et al., 1949) or a range of optimum temperatures (Lewis et al., 1958). Too high a temperature is not favorable for carbonyl formation mainly because the equilibrium carbonyl concentration decreases rapidly with increase in temperature. Lewis et al. (1958) indicated that there is a maximum conversion attainable a t a given temperature and pressure regardless of the duration of reac-
tion, and that the higher the pressure the faster this maximum is reached. Stoffel (1914) showed that the rate of formation decreased rapidly with increase in elapsed time. He attributed this to the adsorption of carbonyl on iron sites. Pichler and Walenda (1940) reported that the flow rate of carbon monoxide has minor influence on the rate of formation. Okamura et al. (1949) stated that the yield of carbonyl was better for hydrogen-reduced than for carbon monoxide-reduced iron, although no quantitative information was given. The condition of reduction was said to have some influence on the subsequent carbonyl formation; lower reducing temperature and longer duration tend to give a better yield of carbonyl. However, Carlton and Goldberger (1965) obtained a better yield with a lower reducing temperature and shorter duration. Lewis et al. (1958) reported that reduction at 800°C gives the best yield. Small quantities of impurities such as ammonia, hydrogen sulfide, methanol, acetaldehyde, alumina, bismuth, nickel, copper, and sulfur are believed to enhance the reation rate, while impurities such as carbon dioxide and oxygen hinder the rate. Experimental Methods The same reactor system as described by Kim et al. (1973) was used for carbonylation experiments. Following a reduction run, the reactor was cooled to the desired carbonyl formation temperature by passing purified nitrogen through it. Cooling usually took about 3 hr. The reduced material was then subjected to carbonylation by switching nitrogen flow to carbon monoxide. With few exceptions, the runs lasted for 2 hr. The exit gas from the reactor was analyzed for carbonyl by means of a Burrell Kromotog K1 chromatograph. The adsorption column was 1/4-in. copper tubing filled with Burrell Kromat FB impregnated with Burrell 550 silicone oil. The column and the detector were maintained a t 82" and 100°C, respectively.
Results Approximately 15 runs were made with Sample I and 20 runs with Sample 11. Compositions and particle sizes of these samples have been given by Kim et al. (1973). The primary data obtained in each run were the concentrations of iron pentacarbonyl measured by the chromatograph as functions of time. These data were then used to calculate, by mass balance, the percent conversion based on the "free iron" initially present in the partially reduced oxide. The amount of free iron was determined by subtracting from the total mass of iron produced in the reduction step, the mass of iron combined with carbon (assumed to be cementite). A typical graph of the mole percent of carbonyl against time is shown in Figure 1 and the corresponding graph of conversion against time in Figure 2. In all runs, the mole fraction of carbonyl increases to a maximum and then decreases with time, and the maximum occurs sooner with increasing gas flow rate. The initial increase of carbonyl concentration appears to indicate dispersion of gas in the empty space above the bed (coupled with time lag through the connecting line from the reactor to the chromatograph). Consequently, residence time distribution of carbon monoxide in the reactor under similar operating conditions, but with no reaction occurring, was experimentally determined so that the concentration of carbonyl a t the bed exit could be calculated from that measured by the chromatograph. The effects of system variables on the rate of carbonyl formation were examined in terms of "overall conversion," i.e., percent of free iron converted to carbonyl in 2 hr. An
important factor is the operating condition for reduction, which appears to affect the reactivity of iron for subsequent carbonyl formation. For a given starting material (Sample I or 11), overall conversion decreases with increasing reduction temperature. The decrease from 593649°C is greater, however, than that from 649-704°C. Similarly, overall conversion tends to decrease with increasing reduction pressure, and the effect of pressure becomes very small above about 30 atm. The flow rate of CO for reduction had a negligible effect on the reactivity of iron. T o isolate the effects of reduction on carbonyl conversion from those of operating variables for the carbonyl reaction itself, a series of runs was made in which the condition for reduction was kept constant a t 30 atm, 649"C, and a CO flow rate of 4.36 cm/sec. Sample I1 was used for
0.35
0.30
Reduction: 3.0 ATM, 6 4 9 "C 4.36 cm/sec CO E X P CALC 9 3 ° C I ........... 121 "c b 148 " C A
c
-
I I
Y
o,051 0.10tti
0.00 1
I 40
1
20
0
I 80 TIME IN MIN. 1 60
I
I
100
120
Concentration of iron pentacarbonyl measured by chromatograph Figure 1.
Sample II
35
2 1 A T M . 0 2 7 2 c m sec C O
Reduction 3 0 ATM, 649 C 4 3 6 c m sec C O
30
EXP
93 c
121
148
25
c
C
b
,
CALC
I
-
b A
_____
20
I I
15
10 5 C 0
20
40
60
80
100
120
TIME IN MIN. Figure 2.
Percent conversion of free iron to carbonyl , Volume 7 , Number 8, August 1973
731
this part of the investigation. In Figure 3, the percent overall conversion is plotted against temperature a t two levels of pressure. The points for 21 a t m indicate that the overall conversion is highest near 121°C. Similar behavior has been observed by Mond and Wallis (1922) who reported maximum conversion a t 200°C for all pressure levels between 100 and 300 atm and by Okamura et al. (1949) who indicated this temperature to be 180°C a t 200 atm and 130°C a t 100 atm. The effect of pressure on conversion is shown in Figure 4. Conversion increases with increasing pressure, approximately as the second power. During a given run, the pressure drop across the bed was observed to remain relatively constant. For the majority of runs, the ratio of pressure drop to weight of bed per unit area is less than 0.35 even though the gas velocity is several times greater than the calculated incipient fluidization velocity based on the average particle size of
Sample 'IT 155 ATM 21 ATM
A
Kinetic Model
5t\
O 90
120
150
TEMPERATURE I N "C
Figure 3. Effect of temperature on percent conversion of free iron to carbonyl in 2 hr
35 30 25
-
Sample It
/
93 "C 121 "c 149 "C
A
CO
-
+
Fe
co +
4.36 cm/sec CO
Surface reaction: Deactivation: Fe
20-
W
>
z 0 u
Gas/Solid Reaction. For the carbonyl formation reaction in the present system, three successive steps are involved: transfer of carbon monoxide from the gas phase into the solid phase, reaction between gas and solid, and transfer of carbonyl from the solid phase into the gas phase. The first and third steps actually include both the transfer of gas across the boundary layer surrounding the solid particle and diffusion through the particle. Based on a correlation by Kunii and Levenspiel (1968), the rate of mass transfer was found to be a t least 18 times greater than the actual rate of reaction. Diffusion through the particle is not likely to contribute a significant resistance in view of the small size of particles, hence very short diffusion paths. Even if the mean pore size is extremely small, on the order of 5 A, the mass transfer rate calculated from the estimated Knudsen diffusivity is about twice the reaction rate. It can, therefore, be assumed that the concentration of gas is uniform throughout the particle and equal to that in the bulk stream. Based on some experimental evidence and the results of previous workers (Stoffel, 1914 and Hayward and Trapnell, 1964), it is postulated that the reaction between gas and solid involves the following steps: Adsorption:
Reduction:
z
-0
the sample. Examination of the bed after the completion of a run showed that there were no distinguishable voids in the bed and that the particles near the bottom of the bed were usually stuck together. These observations suggest that fluidization was probably not achieved and that the reactor behaved more like a fixed bed. An indication of the overall efficiency of the pneumatolytic transport process is provided by the conversion based on the total iron initially present in the sample. For a 300-gram charge, the initial oxidic iron is 118 grams in Sample I and 80 grams in Sample 11. With 2 hr of reduction and 2 hr of carbonylation, the conversion of total iron to carbonyl varies from 2-6% for Sample I and from 3-15% for Sample 11. Reduction a t 593°C produces more reactive but smaller amounts of free iron (typically 5-6 grams from a 300-gram charge). Conversion of these samples lies between 40-6070 on the free iron basis, and 3-570 on the total iron basis. On the other hand, samples reduced at 649" and 704°C contain more free iron (15-20 grams) but have lower conversion of free iron to carbonyl (5-3670). However, conversion based on total iron is generally higher than for samples reduced at 593°C.
e
u
-
CO.Fe
(la)
cow
Ob)
deactivated Fe
Complex formation: overall 15-
CO.Fe
3
+
n COW
10 -
+
(4
- n ) CO Fe(CO)5.u
+
(n
-
1)u (2b)
Desorption: 5-
ob
732
I
I
I
I
5
10
15
20
Environmental Science & Technology
25
Fe(CO)5w Fe(CO), + u (3) where u denotes a vacant site, Fe a reacting iron site capable of forming carbonyl. Reaction 2b involves a series of successive steps by which a reacting iron site gathers the necessary five molecules of carbon monoxide to form an adsorbed complex which subsequently converts to a carbonyl molecule attached to a vacant site.
The deactivation step was introduced based on the experimental fact that the rate of carbonyl formation decreases with time more rapidly than the rate of consumption of the solid reactant. One possible cause for the deactivation is that as iron atoms are removed from discrete points of the surface, the probability of adsorbed CO molecules being adjacent to a reacting iron site becomes increasingly small. Another might be the gradual "poisoning" of the reacting iron sites by deposited carbon (a slight carbon deposition reaction was noted in some runs), carbonyl, or other trace impurities in the gas. The rate of adsorption of CO estimated conservatively on the basis of Wagener's (1957) measurements was exceedingly high compared to the observed rate of carbonyl formation. The desorption of carbonyl is not likely to be rate controlling since the over-all reaction rate varies as the second power of pressure. If the desorption step were controlling, the rate would be almost independent of pressure. It follows that the rate of overall reaction is most likely controlled by one or more of the elementary steps involved in the complex formation. Assuming that the adsorption and desorption steps are a t equilibrium, and that the fractional coverage of CO is far greater than that of Fe(C0)5, the rate of carbonyl formation per unit area of free iron surface may be written.
0.16 1 Reduction: 3 . 0 ATM, 6 4 9 "C 4 . 3 6 crn/sec CO
D DIMENSIONLESS DISTANCE Figure 5. Calculated axial concentration profile of iron pentacar bony1 in bed at end of 2 h r
From the initial rate data, n is approximately equal to 3. In the absence of a known mechanism for the deactivation of reacting iron sites, it is assumed that the activity of reacting iron sites, uFe, decreases exponentially with time aFe =
aFe0 exp (
- b ~ )
(5)
This assumption is purely empirical and its only justification is agreement with experimental data. Since the details of free iron surfaces are not known, and examination under a scanning electron microscope of the reacted solid material showed no well-defined reaction interfaces, it seems reasonable to consider the reaction as being homogeneous. Thus, Equation 1 can be converted to a volumetric basis through the relationship r , = AFemFeo(1
- S)r,
(6)
For convenience, ru is defined as the rate of carbonyl formation per unit volume of the bed so that it can be used directly in the conservation equation for the bed. Combining Equations 4, 5, and 6 and assuming the ideal gas law and the constancy of A F give: ~
r , = h,,rnF,O(l - S ) exp(-bT)
(&)
where X , is the mole fraction of carbonyl and h,, the overall reaction rate constant. Reactor Bed. Since fluidization was not attained, the reactor used in the present work approximated a fixed bed with dispersed plug flow. The conservation equation for carbonyl and iron may be written in the dimensionless form
(9)
with the initial and boundary conditions
S(0,Y ) = 0; S ( r ,0 ) = 0 (11) The effective diffusivity of iron pentacarbonyl in the bed, appearing in the Peclet number, Npe, was estimated by means of the equation (Hoogschagen, 1955) Deff =
t DCO-Fe(COh'5)
-
(12)
7t
where Dco-Feccoi(5)is the binary diffusivity of the COFe(CO)5 system calculated from the Chapman-Enskog formula, and i f the tortuosity factor (taken to be 1.5). The set of Equations 8 through 11 were solved numerically using the Crank-Nicholson implicit method on a Univac 1108 digital computer. The values of k , , and b were obtained for every individual run by minimizing the standard deviation of the computed carbonyl concentration from the measured.
Discussion To test the validity of the proposed two-parameter model, the calculated mole percent Fe( C0)5 and percent conversion, as functions of time, were compared with experimental data. For all runs, the model closely follows the shape of the experimental curve and satisfactorily predicts the carbonyl concentration. In terms of percent conversion, the agreement is even closer. These results were taken to indicate that the proposed model adequately describes the process of carbonyl formation in the present system. As a point of interest, the calculated variation of carbonyl concentration with axial distance along the bed is illustrated in Figure 5. Some calculations with the diffusion term dropped from Equation 8 showed a difference of less than 1% in the predicted carbonyl concentration. While an estimated bed porosity of 0.59 was used in the calculation for all runs, practically identical results were obtained from a test calculation with c = 0.45. Of the two Volume 7, Number 8, August 1973 733
Table I. Overall Activation Energy ( E r ) and Frequency Factors ( k r m o )for the Rate Constant Sample
I
Temp range, "C 90-1 20 E r , kcal/g-mol 15.9 k r m o , cm6/(g-mol sec g Fe) Reduction: 3.0-4.4atm 593°C 649 704
II
120-50 6.40
90-1 20 6.03
120-80 1.42
, , , 2.76 x 104 , . . 35.7 2.29 x 109 1.21 x 104 6.13x 103 16.9 , , , 4.52x io3 5.00 x 103 13.7
-
TEMPERATURE IN "C 180
50.0
160
140
1
I
120 I
100
Sam --..iple
uction:
"M
593 O C 6 4 9 "C 7 0 4 "C
III A
A
0 0
-.
'
\\
0.7 0.5 -3.0
1
2.2
\\
'9
2.3
2.4 2.5 1031~
2.6
2.7
OK
Figure 6. Arrhenius plot of
reaction rate constant ( k r m )
main factors responsible for the decrease in the rate of reaction with time, namely the (1 - S) term and the e - b r term in Equation 7 , the latter has the greater effect. The overall reaction rate constant, k,,, determined for an individual run was correlated in terms of the Arrhenius equation h r m = h r m o ~ X (P- E r I R T ) (13) as plotted in Figure 6. The data appear to indicate two different activation energies above and below 120°C with the higher value below 120°C. It is probable that the temperature of 120°C represents the transition point a t which the adsorbed CO molecules become appreciably mobile. Under the operating conditions investigated, the equilibrium surface coverage of CO is estimated to be less than 40% (Beebe and Stevens, 1940). Thus, some of the nonadjacent adsorbed CO molecules must diffuse over to a reacting iron site for the formulation of a carbonyl complex and their mobility may have significant effect on the reaction rate. Since the movement of chemisorbed molecules from site to site requires activation energy, most of them are mobile only beyond a certain minimum temperature. Although the minimum temperature for adsorbed CO on iron is not known, it is not unreasonable for this temperature to be about 120°C in view of the fact that the onset of mobility is considered to occur a t 149°C for CO on barium and 199°C for CO on titanium (Wagener, 1957). 734
Environmental Science & Technology
Where there are not enough data to establish the slope of the line, it is assumed that for a given sample, I or 11, the lines above and below 120°C are parallel and intersect a t 120°C. The values of overall activation energy and frequency factor are listed in Table I. It is to be noted that the overall activation energies reported here reflect the combined temperature dependence of the rate constant for the rate-controlling step and the equilibrium constants for other steps. While these values appear to be too low for surface reaction control, the activation energy for the rate-controlling step itself could be much higher. The data included in Figure 6 are for runs with samples reduced a t 3.0-4.4 atm. In this range, reduction pressure has little effect on the rate of carbonyl formation or k r m . Limited data indicate that k r m increases sharply with decreasing reduction pressure between 1.0 and 3.0 atm. According to the postulated concept of the deactivation of reacting iron sites, the deactivation constant, b, is expected to be a function of the type of solid material, the condition of reduction, the temperature, possibly pressure (which affects the carbon deposition reaction), and flow rate (due to carbonyl adsorption) for carbonyl formation. Unfortunately, no discernible trend can be established with respect to each factor due to limited data. These various factors, however, are not serious except for reduction pressures less than 3.0 atm. Excluding those runs with samples reduced a t pressures below 3.0 atm, the points for Sample I1 were correlated, albeit with considerable scatter, by the Arrhenius relation b = bo exp ( - E b / R T ) (14) The values of bo and Eb were determined to be 0.110 and 3.26 kcal/g-mol, respectively. Because of insufficient data, the points for Sample I could not be correlated in a similar manner. The values of b for Sample I, however, fall within the range for Sample 11. Hence, it may be assumed that the same values of bo and Eb are valid for Sample I. The optimum temperature noted in Figure 3 is apparently related to the transition point in the temperature dependence of the rate constant (Figure 6). This can be seen by integrating Equation 7 with respect to time, assuming the concentration term to be constant a t some average value and examining the combined temperature effect of k r m and b on conversion. Below 120"C, conversion increases with increasing temperature, while above 120°C it decreases with temperature. Additionally, as temperature increases, the concentration driving force decreases because of unfavorable equilibrium. Final justification of the model should, of course, be based on its ability to predict experimental data using correlated values of the parameters. The recalculated mole 0'9 carbonyl and 70conversion are indicated in Figures 1 and 2. The standard deviations of calculated concentrations from the measured are, in most cases, less than 22% and deviations in percent conversion are generally much smaller.
Acknowledgment The authors are grateful to the Jones and Laughlin Steel Corp. and U.S.Bureau of Mines, Pittsburgh, Pa., for their aid in determining the composition and size of solid materials. Nomenclature A F = ~ surface area per unit mass of free iron C L F ~ = activity of reacting iron site U F ~ ' = initial activity of reacting iron site b = deactivation constant bo = pre-exponential factor for deactivation constant
Ct = total molar concentration per unit volume of bed Dco-F~(coI( 5 ) = binary diffusivity in CO-Fe(C0)s mixture Deff = effective diffusivity of Fe(C0)5 in bed Eb = activation energy for deactivation constant, b E, = activation energy for overall reaction rate constant, krm
YH = reduced bed height, bed height/y, yo = reference distance, 7.28 cm y = axial distance along the bed from the distributor Greek Letters t = porosityofbed T = dimensionless time, t u o / y o T f = tortuosity factor
K , = equilibrium constant of carbonyl formation reaction, a F e ( C 0 )(51/ a C 0 5 k,, = reaction rate constant defined by Equation 7, cm6/g-mol sec g Literature Cited k r m o = frequency factor fork,, Beebe, R. A , , Stevens, N . P., J. Amer. Chem. Soc., 62, 2134-40 k, = reaction rate constant in Equation 4 (1940). MFe = molecular weight of iron, 55.847 g/g-mol Carlton, H. E., Goldberger, W. M., J . Metals, 17,611-15 (1965). mFe = mass of reduced free iron per unit volume of bed Carlton, H. E., Oxley, J. H., Amer. Inst. Chem. Eng. J., 11, 79-84 mFe0 = initial mass of reduced free iron per unit volume (1965). of bed Dufour-Berte, C., Pasero, E., Chim. Ind., 49, 347-54 (1967); CA, 67,24115a (1967). In Ital. Npe = Peclet number, youo/Deff n = number of CO molecules adsorbed on nonreacting Hayward, D. O., Trapnell, B. M . W., “Chemisorption,” pp 191-7, Butterworths, London, 1964. iron sites and involved in forming a molecule of Hoogschagen, J., Ind. Eng. Chem., 47,906-13 (1955). Fe(C0)5 Kim, C. S., Rhee, C. S., Li, Kun, Rothfus, R. R., Enuiron. Sci. P = total pressure l’ech., 7, 725 (1973). Kunii, D., Levenspiel, O., Ind. Eng. Chem. Process Des. Develop, Pi = partial pressure of species i 7, (4), 481-92 (1968). R = gasconstant Lewis, R. M., Cookston, J. W., Coffer, L. W., Stephens, F. M., r, = rate of carbonyl formation per unit surface area of Jr., J. Metals, 10,419-24 (1958). Mond, L., Wallis, A . E.,J. Chem. Soc., 121,29-32 (1922). free iron Okamura, T., Kozima, H., Masuda, Y . , Sei. Rep., Tohoku Uniu., r, = rate of carbonyl formation per unit volume of bed A l , 319-25 (1949). S = conversion of free iron to iron pentacarbonyl, (mFeo Pichler, H., Walenda, H., Brennst-Chem., 21, 133-41 (1940); CA, - mFe)/mFe’ 35,3207 (1941). In Ger. T = temperature Stoffel, A,, Z. Anorg. Chem., 84, 56-76 (1914); CA, 8 (l), 639 (1914). In Ger. t = time Wagener, S.,J . Phys. Chem., 61,267-71 (1957). U = dimensionless velocity, u / u o uo = reference velocity, 0.272 cm/sec Xu = mole fraction of Fe(C0)5 Received for review October 26, 1972. Accepted April 11, 1973. Work supported by the U.S. Department of the Interior. Y = dimensionless distance from the distributor, y / y o
Photolysis of NO, in Air as Measurement Method for Light Intensity Donald H. Stedman’ and Hiromi Niki Fuel Sciences Department, Ford Motor Co. Scientific Research Laboratories, P.O. Box 2053, Dearborn, Mich. 481 24
w We have investigated the photolysis of NO2 in air using a chemiluminescent O3/NO detector. The first 30 sec of photolysis provide a simple and rapid measurement of the light intensity k1. (NO2 + hu NO + 0 ) . The inhomogeneity of light intensity can also be measured. Both parameters are important in photochemical smog studies. This experiment also provides for a check on the 0 3 calibration by turning off the lights. Comparison of our k l and k d measurements confirms recent work which gives k l / k d = 0.64. According to the mechanism this gives the ratio of rate constants for 0 + NO2 + M NO3 + M to 0 + NO2 NO + 0 2 of 0.28 f 0.03.
-
-
+
The primary process in photochemical smog reactions is the photodissociation of NOz, and the rate of this reaction is a critical parameter for understanding the mechanism *To whom correspondence should be addressed. Present address. Chemistry Department, University of Michigan, Ann Arbor, Mich. 48104.
of smog formation. The relevant reactions have been reviewed recently (Schuck and Stephens, 1970).
+ + + - + + - + + - + + - + + + - + +
NO2 hv NO 0 0 0, M 0, M NO 0, NO* 0, SO? 0, KO3 o2 0 NO, NO O2 0 NO, M NO, M NO, NO:, NZOj KO NO, 2N0, 0 NO M NO, M 2NO 0, 2N0,
(1) (2)
(3) (4) (5) (6) (7) ( 8) (9) (10) The photolysis of NO2 in N2 is commonly used for measuring the light intensity in a photochemical smog chamber. The method consists of filling a photochemical reactor with a low concentration of NO2 in an inert gas (usually Nz) and monitoring the decay of NO2 as a function of
+ + + + +
---
+
Volume 7, Number 8, August 1973
735