The Kinetics of Hydrodenitrogenation over a Zeolite Catalyst I.E. Maxwell and J.A. van de Griend Koninklijke/Shell-Laboratorium, Amsterdam (Shell Research B.V.) P.O. Box 3003, 1003 AA Amsterdam, The Netherlands As a part of a study aimed at providing a more basic understanding of hydrodenitrogenation (HDN) catalysts, a kinetic study has been carried out using quinoline as a model feedstock and a bifunctional catalyst containing zeolite Y. The results show that the overall kinetics of quinoline HDN can be well described by means of a Langmuir-Hinshelwood type of relationship. This implies that there is relatively strong adsorption of N-containing intermediate and/or product molecules on the catalyst surface, which further plays a dominant role in determining the overall rate of HDN. The relatively simple model developed can be used, for example, to calculate the overall steady-state coverage of the catalyst surface with N-components under various reaction conditions. This type of approach could be used to develop a better insight into the relationship between catalyst physical/chemical properties and HDN kinetic parameters. INTRODUCTION The current world-wide trend towards processing heavier feedstocks in refineries has imposed severe demands on existing hydrotreating catalysts. In particular, the relatively slow heteroatom removal reactions such as hydrodenitrogenation are becoming of critical importance in oil conversion processes such as hydrocracking. A number of fundamental studies on hydrodenitrogenation (HDN) have been reported in the open literature [1-5), but these have, in general, been focussed on the understanding of reaction mechanism rather than on reaction kinetics. The present study is intended to provide more insight into the kinetics of hydrodenitrogenation, in particular as they relate to a zeolitebased catalyst system. EXPERIMENTAL Quinoline (pro-analyse) dissolved in a paraffinic oil (ONDINA 68, boiling range 287-525 °C) was used as a model feedstock, simulating organic nitrogen in the concentration range 50 to 1500 ppmw N. Carbon disulphide was also added (2-5 %w) to maintain 1 bar H2S partial pressure and thereby ensure that the metal function on the catalyst was maintained in the sulphided form. The catalyst used in these studies was a proprietory commercial second-stage hydrocracking catalyst based on zeolite Y and containing a Ni/W hydrogenation function. The hydrodenitrogenation kinetics experiments were carried out in a trickleflow microreactor, using 8 ml of catalyst (0.3-0.8 mm particle size). A total reactor pressure of 125 bar was applied and the H2/feed ratio was 1500 Nl.(kg feed)-l at a WHSV of 1.0 kg.l-l.h- l• The total nitrogen content of both the feedstock and the liquid product were analysed using a Dohrmann, DNlO chemiluminescence nitrogen analyser.
795
796 (CA-6-2) RESULTS AND DISCUSSION The results obtained for the overall rate of hydrodenitrogenation are given in Table 1. Since numerous previous measurements [1-5] had demonstrated pseudofirst-order kinetics for quinoline RON, the rate constants for given temperatures and quinoline concentrations (k app' see Table 1) were derived as follows: kapp = WHSV In where [N) feed [N)product
[N) feed
_.....;:;.=c.;,--_
[N)product
total nitrogen concentration in feed (ppmw) total nitrogen concentration in liquid product (ppmw)
Table 1. Measured and calculated kinetic data for quinoline RON over a Ni/W/zeolite Y catalyst
°C
ppmw
Measured. ka pp• kg.l- l.h- l
300 300 300 300 300
60 150 383 552 1479
6.1 3.18 0.80 0.31 0.10
5.1 2.53 0.81 0.47 0.09
315 315 315 315 315
60 150 383 552 1479
13.7 8.5 3.95 1.80 0.51
13.4 8.6 3.77 2.43 0.56
325 325 325 325 325
60 150 383 552 1479
23.8 16.1 11.3 6.58 1.77
23.2 17.0 8.94 6.22 1.71
Temp ••
[N) feed.
Ca Lcu I at ed f
kapp• kg.l- l.h- l
a) Parameters used for calculated values were: 30 kcal/mol (1 kcal = 4.2 kJ) 30 kcal/mol 9.8 kg.l-l.h- l (at 300 °C) 0.00645 ppmw- l (at 300 0C) Plots of (1/k a pp)1/2 against [N)feed. at constant temperature. were found to show a linear relationship (Figure 1).
I.E. Maxwell and J.A. van de Griend
1
1/2
(k-)
:.:~
opp
3,Or 2.8
2.6
2.4 2.2
2.0 1.8 1.6
1.4
1.2 1.0
0.8
O.()I--_ _---JL...-_ _..J.-_ _--L_ _- - '
o
500
1500 2000 IN FEED, ppmw
Fig. 1: Measured and calculated rates of quinoline HDN over a Ni/W/Zeolite Y catalyst This marked retardation of HDN with increasing inlet quinoline concentration is indicative of self-inhibition behaviour. Such systems, where strong adsorption of reactant or product molecules occurs, can often be adequately described by means of a Langmuir-Hinshelwood kinetics [6). Empirically, for the present reaction, a kinetic expression may then be written as follows: k
app
where ka pp ks b
[N) feed
=
k
s ---=-----". (1+b[N)feed)2
apparent first-order rate constant for HDN, intrinsic HDN rate constant, overall nitrogen-adsorption constant, the quinoline concentration at the reactor inlet.
A kinetic expression of this type is suggestive of relatively strong adsorption of both reactant and product molecules on the surface of the catalyst [6). In fact, equation (1) represents a major simplification, since one would normally, if possible, include separate adsorption terms for the individual reactant, intermediate and product molecules, which are
797
798 (CA-6-2) known to be quite complex [4]. Nevertheless, this rather simplistic approach, which obviates the need to identify all the individual reaction products, is found to be quite useful. Although equation (1) is shown to provide a good description of the overall HDN kinetics at a given temperature (see Figure I), the temperature de-pe ndence of the system was found to be somewhat more complex. In part i c ul ar , the apparent activation energy, Ea was observed to vary as a function of quinoline concentration (see Figure 2).
100
Ee• keeL/mol
20
OL.------'----...1-------'----'
o
500
1000
1500
2000
[N] IN FEED, ppmw
Fig. 2: Measured and calculated apparent activation energies for quinoline HDN over a Ni/W/zeolite Y catalyst This phenomenon is in fact indicative of strong product adsorption on the catalytic surface 6 and may be described by means of a temperature-dependent adsorption constant. i.e. k~pp
where k~pp
exp( -Ea/RT) =
exp( -Es/RT) o (l+b exp(X/RT) [N]feed)2
k~ ----::.-~.......;:........:__:_-
(2)
pre-exponent of the apparent first-order HDN rate constant, kapp apparent activation energy
pre-exponent of the rate constant for HDN, ks intrinsic activation energy for HDN reaction overall heat of adsorption of nitrogen compounds pre-exponent of the overall adsorption constant for nitrogen compounds, b = reaction temperature (K) T [N]feed total nitrogen concentration in feed. The above expression implies that in the limiting cases, i.e. when [N]feed ..... 0,:
I.E. Maxwell and J.A. van de Griend Ea ~ ~s and if [N]feed 1S large, then Ea ~ Es + 21\. Therefore, one would expect an increase in Ea with increasing [N]feed until a maximum value is attained, after which Ea is nearly constant. This is, in fact, observed, as is shown in Figure 2. The experimental data have been fitted using the following parameters:
Es
30 kcal/mol (1 kcal = 4.2 kJ) 30 kcal/mol ks 9.8 kg.l-l.h- l (at 300 °C) 0.00645 ppm-l (at 300 0C). b As shown in Figures 1 and 2, with these parameters there is good agreement between the calculated and observed data with respect to variations in both temperature and quinoline concentration. The relatively high value required for the overall heat of adsorption, 1\ (30 kcal/mol), is consistent with strong chemisorption of intermediate and/or product molecules (i.e. basic N-compounds). Likely intermediate compounds with relatively high basicity include molecules such as l,2,3,4-tetrahydroquinoline, which has been shown [5] to be formed via rapid hydrogenation of the N-containing aromatic ring of quinoline. In order to gain some insight into the steady-state surface coverage of the catalyst with these strongly adsorbed N-compounds, we have used the following relationship. From the Langmuir adsorption isotherm relationship [6] it follows that: (3)
where eN = fractional surface coverage with organic nitrogen. Then e v = 1 - eN' fraction of surface not covered with organic nitrogen where e v and
(4)
1 + b[NJ
If we now assume that [N] = [N] feed, i.e. the integral of all adsorbed N compounds is equal to the total organic N concentration, then combining equation (4) with equation (1) yields: kobs
= ks
8v
2
(5)
from which 8 v can be calculated for a given temperature and [N]feed value. The results obtained are shown in Figure 3. Clearly, the model indicates that there is a marked increase in surface coverage with increasing feed nitrogen concentration. This is indicative of reaction retardation due to self-inhibition behaviour.
799
800 (CA-6-2)
0.2
Ol..-_ _---I...
a
500
.L-_ _--'-_ _
~
1000
Fig. 3: Calculated surface coverage, ON' of a Ni/W/zeolite Y catalyst with organic nitrogen as a function of quinoline feed concentration [N) T
= 300
°C, P
= 125
bar
CONCLUSIONS This study has shown that the overall kinetics of quinoline HDN over a commercial bifunctional zeolite-containing hydrocracking catalyst can be well described by means of a Langmuir-Hinshelwood-type model. This implies that there is relatively strong adsorption of N-containing intermediate and/or product molecules on the catalyst surface, which further plays a dominant role in determining the overall role of HDN. Clearly, the catalyst properties will be important in determining the degree of steady-state surface coverage of such intermediates for a given set of reaction conditions. Further, studies will, however, be required to provide insight into the relationship between catalyst physical and chemical properties and HDN kinetic parameters. In addition, although studies using model N-containing molecules such as quinoline are valuable in simplifying otherwise almost intractable kinetic and analytical problems, the translation of these results to the very complex mixtures of N-compounds which are found in, for example, hydrocracker feedstocks remains a major challenge. This fundamental understanding, however, is invaluable in providing the basis for developing significantly improved HDN catalysts in the future.
IoE. Maxwell and J.Ao van de Griend REFERENCES 1. R.A. Flinn, a.A. Larson and H. Beuther, Hydrocarbon Process. Pet. Ref., 42 (963) 129. 2. if."G. McIlvried, Ind. Eng. Chern. Process Des. Dev, , 10 (971) 125. 3. A.K. Abou1-Gheit and I.K. Abdou, J. Int. Petr., 59 ([973) 188. 4. J.F. Cocchetto and C.N. Satterfield, Ind. Eng. Chern. Process Des. Dev., 20 (981) 49. 5. ~S. Shih, J.R. Katzer, H. Kwart and A.B. Stiles, Amer. Chern. Soc. Div. Petro Chern. Reprints, 22 (1977) 3, p. 919. 6. C.N. Satterfield, "Heterogeneous Catalysis in Practice", McGraw-Hill, 1980, p, 46.
801