- Email: [email protected]
Catalytic oxidation of model soot by metal chlorides
B:ENVIRONMENTAL Applied Catalysis B: Environmental 12 (1997) 33-47
ELSEVIER
Catalytic oxidation of model soot by metal chlorides Guido Mu1
*,
Freek Kapteijn, Jacob A. Moulijn
Delji University of Technology, Industrial Catalysis, Julianalaan 136, 2628 BL Delft, The Netherlands
Received 11 March 1996; revised 28 June 1996; accepted 15 July 1996
Abstract Several metal chlorides were screened for their catalytic activity in the oxidation of model soot (Printex-U) in ‘loose contact’ by means of TGA/DSC. HgCl,, CaCl,, BaCl,, CoCI,, and NiCl, show little activity. Hydrated BiCl, and FeCl, are converted in air into BiOCl and FeOCl, which have a moderate soot oxidation activity. MoCl, is converted into the corresponding metal oxide and also shows a moderate ‘loose contact’ activity. PbCl,, CuCl, and CuCl are very active catalysts; the soot oxidation temperature is lowered by 200-275 K. The activity of metal chlorides is thought to be induced by in situ formation of intimate contact between the soot and the metal chloride via ‘wetting’ and/or gas phase transport. A correlation between the melting point and the catalytic activity was found. Furthermore, a catalytic cycle is proposed involving activation of oxygen on the surface of the (oxykhloride, followed by transfer of activated oxygen to the soot surface. DRIFT analyses showed that this results in the formation of carbon surface oxygen complexes. Decomposition of those complexes yields CO and CO,. Practical application of metal chlorides for the removal of soot from diesel exhaust is not recommended, because they suffer from instability or high vapour pressures. Keywords: Soot; Oxidation; Metal chloride catalysts; Oxychlorides; Mechanism
1. Introduction Removal of soot and NO, from diesel exhaust gas is necessary to protect the environment. Although diesel exhaust is rather clean compared to Otto engine exhaust, in practice the amount of soot and NO, emitted by diesel engines is much larger than that emitted by otto engines equipped with catalytic converters
[Il. * Corresponding author. Fax: 015-2784452, E-mail: [email protected]. 0926-3373/97/$17.00 Copyright 0 1997 Elsevier Science B.V. All rights reserved PII SO926-3373(96)00065-3
34
G. Mu1 et al./Applied
Catalysis B: Environmental
12 (1997) 33-47
Soot emission can be reduced in various ways [l-3]. Collection of soot in a monolithic filter and simultaneous oxidation is considered to be a good option to minimize the contamination of the environment. However, soot oxidation takes place at temperatures of 825-875 K, while the temperature of diesel exhaust is typically 500-675 K. Hence, a catalyst is needed to prevent accumulation of soot in the monolithic filter. Several investigators have focused their attention on the application of oxidic materials to lower the oxidation temperature of soot. However, the activity of metal oxides depends on the intensity of contact between the metal oxide and soot. It has been shown that metal oxides, like Co,O, or Fe,O, can be reasonably active if ‘tight contact’ between the soot and catalyst has been established mechanically (by ball-milling), and that they lose their catalytic activity completely under so-called ‘loose contact’ conditions (after e.g. spatula mixing) [4,5]. In a previous paper we discussed the catalytic activity of a Cu/K/Mo/(Cl) catalyst [6]. Copper chlorides are present and formed within this catalyst and it was shown that they are essential for the high soot oxidation activity of a Cu/K/Mo/(Cl) catalyst in ‘loose contact’. Also other authors have shown that CuCl, is an active catalyst for the carbon oxidation reaction [7]. It was proposed that the mobility and volatility of CuCl or CuCl, results in an in situ establishment of ‘tight contact’. Furthermore, the higher activity of copper chlorides compared to copper oxide (even in ‘tight contact’ conditions) was tentatively explained by favourable redox properties of the copper (oxy)chlorides [6]. Little is known about the catalytic activity of other metal chlorides in soot oxidation and their stability with respect to this reaction. In order to find design rules for an active catalyst in soot oxidation we evaluate the catalytic activity of several metal chlorides with respect to their melting point. We also pay attention to the stability of the most active chlorides in air at the temperatures needed to combust the soot, and evaluate the necessity of a chloride ion for the activation of oxygen.
2. Experimental The metal chlorides investigated were purchased from Baker (BaCl,, CaCl,, FeCl, . 6H,O, (analytical grade)), Aldrich (CuCl, BiCl,, HgCl,, CoCl, * 6H,O, MoCl,, NiCl, . 6H,O (analytical grade)) and Merck (CuCl, * 4H,O, PbCl, (analytical grade)). The corresponding metal oxides were purchased or prepared as described elsewhere [4]. The oxychlorides of copper and lead were prepared by solid state reactions of CuCl, . 4H,O and CuO (Merck, p.a.> at 525 K for 2 h [8,25] and PbCl, and PbO (Aldrich, > 99%) at 575 K for 2 h in static air [9]. Since it is difficult to obtain diesel soot with constant properties (the composition depends on the engine load) a model soot was applied (Printex-U, a flame soot purchased from Degussa). This soot has a N,-BET surface area of 96
G. Mu1 et al/Applied
Catalysis B: Environmental
12 (1997133-47
35
m2gg’ and contains approximately 5 wt.-% of adsorbed hydrocarbons and 0.2-0.4 wt.-% sulfur [4]. Oxidation temperatures were determined in a thermobalance (STA 1500H) after mixing the soot and the catalyst in a ratio of 1: 2 by weight with a spatula (i.e. ‘loose contact’) and diluting the sample with Sic in order to prevent thermal runaways [4]. About 4 mg catalyst, 2 mg soot and 54 mg SIC were applied as a sample. This way, smooth DSC curves could be obtained. A heating rate of 10 K/min and a flow rate of 50 ml/min 21 ~01% 0, in N, were used. The maximum of the DSC curve was defined as the oxidation temperature. This temperature coincided with the temperature of the maximum weight loss rate within 10 K. TG/DSC analysis of the metal oxides was performed in ‘loose’ and ‘tight contact’ mode. Samples referred to as ‘tight contact’ were intensively milled in an agate ball mill for one hour, before dilution with SIC and thermal analysis. The temperatures determined for the metal oxides were reproducable within 5 K. Stability tests of CuCl were performed by TG/MS in N, (50 ml/min) and air (21 ~01% 0, in N,, 50 ml/min). About 15 mg of the chloride was analyzed without SIC dilution. A quadrupole mass spectrometer (Fisons instruments) was used, which was coupled to the balance with a heated capillary. In this way, products of decomposition could be detected. Carbothermic reduction of Cu,OCl, was recorded in the thermobalance in N, (Cu,OCl,:soot = 4:l by weight). A TG/DSC profile of Cu,OCl, without soot was recorded under similar conditions. Partially converted CuCl, PbC12, Pb,OCl,, FeOCl and BiOCl catalyzed soot samples were prepared (without Sic dilution, and a catalyst:soot ratio of 2:l by weight) isothermally in the thermobalance at 550 K (CuCl), 610 K (PbCl,, Pb,OCl,) and 645 K (FeOCl, BiOCl). The final temperature was reached with 10 K/min. Approximately 20 min were required to obtain soot conversion levels of 20%-60%. Diffuse reflectance infrared fourier transformed (DRIFT) spectra were recorded on a Nicolet Magna 550 spectrometer, equipped with a Spectratech DRIFT accessory. Samples were analyzed ex situ, after partial conversion of the soot in the thermobalance and dilution with KBr (1:lOO wt.-%). They were recorded against a soot in KBr background (soot:KBr = 1:lOO wt.-%). A resolution of 8 cm-’ and 256 scans were applied to obtain the spectra. Spectra are displayed without any further data processing.
3. Results 3.1. Screening
ofmetal chlorides
A typical TG/DSC profile of soot oxidation (PbCl, shown in Fig. 1. The oxidation activity is characterized
is used as a catalyst) is by the maximum of the
G. Mu1 et al. /Applied Catalysis B: Enuironmental12 (1997) 33-47
36
60.6
60.2
p
59.6
z 6 s
59.4
59.0
56.6
200
300
400
500
600
Temperature
Fig. 1, TG/DSC
analysis of PbCl,
catalyzed
700
800
900.
(K)
soot oxidation.
TG: dashed line, DSC: solid line.
exothermic heat effect, located at 683 K (Tl). Non-catalytic oxidation of Printex-U (the model soot applied) occurs around 875 K. A slight endothermic heat effect (T2) is located at 773 K, due to melting of PbCl,. This is accompanied by a gradually increasing weight loss above this temperature. Oxidation temperatures for the metal chlorides and metal oxychlorides are given in Fig. 2. This figure also contains the soot oxidation temperatures for the corresponding metal oxides in ‘loose’ (without ball milling) and ‘tight contact’ (with ball milling). Obviously the activity of the metal oxides is strongly contact dependent.
Fig. 2. Soot oxidation temperatures as determined for several metal chlorides and metal oxychlorides (‘loose contact’) and their corresponding metal oxides (‘tight contact’ (T), and ‘loose contact’ CL)). The solid horizontal line indicates the non-catalytic soot oxidation temperature (875 K).
G. Mu1 et al./Applied
Catalysis B: Environmental
I2 (1997) 33-47
31
Several metal chlorides have hardly any effect on the soot oxidation temperature. HgCl, does not lower the soot oxidation temperature. Also BaCl, and CaCl, only show a small catalytic effect. Upon heating, CoCl, . 6H,O and NiCl, . 6H,O lose crystal water; in air they are (partially) converted into the corresponding oxides ([ 10,111, and references therein). The soot oxidation activity of partially oxidized nickel and cobalt chlorides lies in between that of the ‘tight contact’ and ‘loose contact’ activity of the metal oxides. MoCl, is completely oxidized in air at relatively low temperatures. The activity of oxidized MoCl, is equal to MOO, in ‘tight contact’. FeCl, . 6H,O and hydrated BiCl, are first converted into rather stable oxychlorides [12,13]: FeOCl and BiOCl, respectively. FeOCl decomposes into Fe,O, (Hematite) and Cl, at approximately 725 K [12]. BiOCl is more stable and does not liberate Cl, below 850 K [ 131. Both (oxyjchlorides have a somewhat higher activity in the soot oxidation reaction than their corresponding metal oxides in ‘tight contact’ (i.e. after ball milling of the metal oxide and soot). Also CuCl, and PbCl, have higher soot oxidation activities than ball-milled samples of the oxides. The activity of the oxychlorides, Cu,OCl, and Pb,OCl,, is even better: maxima in the DSC curves were located at 625 and 660 K, respectively. CuCl is extraordinarily active: the soot oxidation temperature is lowered by 28.5 K. As CuCl is the most active catalyst found, a thorough TG/DSC and DRIFT analysis of CuCl was carried out, in order to reveal the active component and the mechanism by which CuCl is operative. As PbCl,, Pb,OCI,, FeOCl and BiOCl have higher activities than their corresponding oxides and a reasonable stability in the temperature range of interest, these compounds were also further investigated. 3.2. Determination
of active phases and stability thereof
The TG and DSC profiles of CuCl in a nitrogen atmosphere are shown in Fig. 3. Two heat effects can be detected at 685 K (Tl) and 698 K (T2), characteristic for CuCl, due to a phase transition [14] and melting [ 151, respectively. A considerable weight loss is observed above 700 K, due to evaporation of CuCl. This compound has an appreciable vapour pressure above this temperature [16]. The TG and DSC profiles of soot oxidation catalyzed by Cu,OCI, (a) and CuCl (c) and the oxidation of CuCl in air (4 mg, without soot, but diluted with Sic (1:15)) (b), are shown in Fig. 4. The CuCl and Cu,OCl, catalyzed soot oxidation temperatures are located at approximately 595 K (Tl) and 625 K (T5) respectively. After 100% soot conversion, an increase in weight can be observed in the range of 660-690 K, accompanied by two heat effects (T2 and T3): one of which is positive, due to oxidation, and one a superimposed negative heat effect, due to melting. Similar heat effects are observed for pure CuCl. The
38
G. Mu1 et al. /Applied
Catalysis B: Environmental
Temperature
Fig. 3. TG/DSC
12 (1997) 33-47
(K)
analysis of CuCl in N,. TG: dashed line, DSC: solid line
interpretation is the (re)oxidation of CuCl into Cu,OCl,, which is thermally stable in air up to 740 K (T4). At this temperature a weight decrease is observed, due to decomposition of the oxychloride, yielding CuO and gaseous chlorine. The formation of chlorine upon Cu,OCl, decomposition was verified by TG/MS analysis of pure CuCl in air (20 mg, without Sic dilution). The TG/DSC analysis of carbothermic reduction of Cu,OCl, in nitrogen is shown in Fig. 5. A weight loss, accompanied by an exothermic heat effect is located at approximately 600 K (Tl). This weight loss agrees with the formation of CuCl, CO and CO,. The formation of CuCl is corroborated by the heat
Tl 6O
_.-.-.-.-
._.-._._. ._._.
-__... -._.
120 ‘\ 60
a &. z .P
59-
I
56-
G I
40
z 6 E
0
Y =
57-40
56
-60 600 Temperature
700
600
900
(K)
Fig. 4. TG/DSC analyses of Cu,OCl, and CuCl catalyzed soot oxidation CuCl (without soot, curves b). TG: dashed lines, DSC: solid lines.
(curves a, c) and the oxidation
of
G. Mu1 et al./Applied
Catalysis B: Enuirotunental
I2 (1997) 33-47
39
-9 -6 -3 -0 -
-3
-
-6 -9 -12
600 Temperature
Fig. 5. TG/DSC
analysis of Cu,OCl,,
650 (K)
recorded in N,, with (dashed lines) and without soot (solid lines).
effects at 685 K (T2) and 698 K (T3), which are characteristic for CuCl (Fig. 3). These phenomena are not observed in the TG and DSC patterns of Cu,OCl, without soot. The weight loss above approximately 700 K is ascribed to the volatilisation of CuCl and Cu,OCl,. Carbothermic reduction of Pb,OCl, was not observed at the temperatures where this compound is active in soot oxidation. Oxidation of PbCl, after 100% soot conversion (Fig. l), did not take place either. Hydrated BiCl, is converted at low temperatures (450-550 K) into BiOCl. Decomposition and carbothermic reduction of the latter compound was not observed. Finally, FeOCl (formed by decomposition of FeCl, .6H,O) was not carbothermally reduced, but decomposed into Fe,O, and Cl, at approximately 720 K. 3.3. DRIFT analysis The DRIFT spectra of a partially converted CuCl/soot mixture (50% conversion, 550 K), and a BiOCl/soot mixture (20% conversion, 645 K) are depicted in Fig. 6. The spectrum of soot after a heat treatment for 30 min at 645 K, is also shown for comparison. The DRIFT spectra contain three main absorptions located at 1738 cm-‘, 1607 cm-’ and centred around 1257 cm-‘. The 1607 cm-’ absorption is caused by aromatic stretching vibrations of the soot, which are enhanced by polar functional groups like quinone [ 171. The other two absorptions have been assigned to oxygen complexes formed on the soot surface: lactones (1738 cm- ‘1, and ether-like complexes (1257 cm- ‘> respectively [ 18,191. Clearly CuCl causes an enhancement of the amount of surface oxygen complexes (SOC). The increase of the amount of SOC formed in the presence of a Cu/K/V catalyst, whose activity we believe to be induced by
G. Mu1 et al./Applied
40
8 5 e 8 $
Catalysis B: Environmental 12 (1997) 33-47
1605
2000
1265
1000 Wavenumbers (cm’)
Fig. 6. DRIFT analysis of partially converted soot samples. Non-catalytic catalyzed (50% conversion), and BiOCl catalyzed (20% conversion).
(3% soot conversion),
CuCl
CuCl [6], has already been reported by Ciambelli et al. [20,21]. According to Ciambelli the complexes mainly desorb as CO, in the presence of the Cu/K/V catalyst, and as CO and CO, in the uncatalyzed oxidation. Pure CuCl has no infrared absorptions in the spectral region recorded (400-4000 cm- ‘>. However, the absorptions located at 905, 855 and 810 cm- ’ can be ascribed to water adsorbed on CuCl [22]. A strong band below 600 cm- ’ is indicative for Cu,OCl,. As this band is not present in the spectrum displayed in Fig. 6, the DRIFT analysis confirms that oxidation of CuCl into Cu,OCl, does not occur during catalytic soot oxidation at 550 K. The DRIFT spectrum of soot, partially converted in the presence of BiOCl (20%), is included in Fig. 6. BiOCl also catalyzes the formation of surface oxygen complexes. The observed absorptions are similar to the CuCl sample. The differences in intensities of the SOC vibrations are caused by the different soot conversion levels, which were induced by different pretreatment temperatures and catalytic activities. The absorption band at 530 cm- ’ can be ascribed to BiOCl [23]. DRIFT spectra of partially converted PbCl, (15%), Pb,OCl, (35%) and FeOCl(60%) are shown in Fig. 7. Similar absorptions as in Fig. 6 can be found. Apparently, these (oxy)chlorides also catalyze the formation of surface oxygen complexes. Again, the differences in intensities of the IR bands, ascribed to SOC, are caused by different soot conversion levels of the samples. For Pb,OCl,, the DRIFT spectrum collected after partial soot conversion contains a rather large absorption band below 500 cm- ‘, indicative for the presence of the oxychloride. This band is not observed in the PbCl, sample. The absorptions indicated with an asterisk (at 1375 cm-’ and 1113 cm-‘) might be due to
G. Mu1 et al./Applied
Catalysis B: Enuironmental12
(1997) 33-47
41
0.15-
1605
2000
1000 Wavenumbers
(cm-l)
Fig. 7. DRIFT analysis of partial converted soot samples. PbCl, catalyzed (35% conversion) and FeOCl catalyzed (60% conversion).
catalyzed
(15% conversion),
Pb,OCl,
carbonate formation. Furthermore, the 1250 cm- ’ band is absent in the Pb,OCl,/soot spectrum. This is still under investigation. Nonetheless, the DRIFT analyses confirm the observations of the TG/DSC analysis, that neither carbothermic reduction of Pb,OCl,, nor (bulk) oxidation of PbCl, takes place of FeOCl into hematite is during catalytic soot oxidation. Decomposition confirmed by changes in the large band below 800 cm- ’ (not shown).
4. Discussion 4.1. Catalytic activity of metal (oxykhlorides Metal (oxylchlorides with high melting points (CaCl,, BaCl,, NiCl, and CoCl,) are less active in ‘loose contact’ than metal chlorides with relatively low melting points and high vapour pressures. Milling a metal chloride and soot hardly effects the catalytic soot oxidation temperature. These facts indicate that the high catalytic activity of several metal chlorides in ‘loose contact’, can be explained by in situ distribution of the chlorides over the soot surface (resulting in ‘tight contact’) and is related to the melting point of the compounds. Neeft [4] found a correlation between the ‘loose contact’ catalytic activity and the melting point or volatility of metal oxides. The activity of metal chlorides, expressed by the soot oxidation temperature, correlated to their melting point, as is shown in Fig. 8. The non-catalytic soot oxidation temperature is indicated by the horizontal solid line. Unfortunately, several metal (oxyjchlorides do not have a well defined melting point in air. They are (partially) transformed before they melt into the corresponding oxide and Cl, (FeOCl (Hematite), MoCl,, BiOCl, CoCl,
42
G. Mu1 et al./Applied
Catalysis B: Environmental
12 (1997) 33-47
1000
Melting or Decomposition Point (K)
Fig. 8. Correlation between the melting point of metal chlorides and the determined soot oxidation temperature. Solid dot: well defined melting point. Open dot: decomposition or oxidation takes place before melting. The horizontal solid line indicates the non-catalytic soot oxidation temperature.
and NiCl, [ 161) or decompose otherwise (CuCl,, yielding CuCl [ 161). Therefore the decomposition temperatures given by Knacke [16] or determined by ourselves (FeOCl) are plotted in the figure, except for MO, whose melting point of the oxide (MOO, [16]) was used. These metal chlorides are indicated with an open circle in Fig. 8, while a solid circle indicates that the metal chlorides (and MOO,) have a well defined melting point. Fig. 8 clearly shows that the higher the melting point of the metal chloride, the less ‘loose contact’ activity it displays. Although FeOCl, CoCl, and NiCl, are partially oxidized around the temperatures where they display catalytic activity, they are more active than the corresponding oxides in ‘loose contact’. Apparently, a rather intimate contact between the (partially decomposed) chloride and the soot can still be established. Although HgCl, does have a well defined melting temperature, it is not on the curve, obviously because HgCl, has evaporated before it can exert its catalytic influence. Whether the in situ ‘tight contact’ formation occurs by ‘wetting’ or gas phase transport has yet to be established. Xie et al. [24] have investigated the spreading (or ‘wetting’) behaviour of many inorganic salts on several carrier materials (like Al,O,, TiO, and activated carbon), and found that CuCl, was able to wet the surface of alumina [24]. Previously we have shown that gas phase transport of copper chlorides also occurs [6]. The (oxyjchlorides of Cu, Pb, Fe and, to a lesser extent, Bi are even more active than their corresponding oxides in ‘tight contact’. This might be the result of an even better contact obtained by ‘wetting’ or condensation than obtained after ball-milling. However, several experimental observations indicate that chlorine ions also effect the activation of oxygen, as will be discussed in paragraph 4.3.
G. Mu1 et al. /Applied
Catalysis B: Environmental 12 (1997) 33-47
43
4.2. The active phase in metal chloride catalyzed soot oxidation In the TG profile of soot oxidation catalyzed by Cu,OCl,, an increase in weight can be observed after complete soot conversion at 650-680 K. A similar weight increase can be observed after catalytic oxidation by CuCl and in the TG analysis of heating CuCl in air without soot. Also the heat effects are similar. Apparently, even in the presence of 20% 0, in N,, an in situ conversion of Cu,OCl, into CuCl during soot oxidation has taken place. This is corroborated by the TG profile of heating Cu,OCl, and soot in nitrogen, which showed a weight decrease around 600 K, indicating carbothermic reduction of Cu,OCl, by soot (Fig. 5). Therefore, we conclude that during soot oxidation Cu,OCl, is reduced to CuCl. The TG/DSC analysis in nitrogen of Pb,OCl, mixed with soot showed that carbothermic reduction does not occur at temperatures below the melting point. On the other hand, Fig. 1 shows that oxidation of PbCl, (upon soot oxidation) does not take place either, which is corroborated by the absence of an IR band below 500 cm-’ in the DRIFT spectrum of the partially converted soot/PbCl, mixture. Apparently, the active phase of Pb is determined by the starting compound. This conclusion is confirmed by the activity data: Pb,OCl, is more active than PbCl,. The DRIFT spectrum of a BiOCl/soot mixture after 20% conversion (Fig. 6) showed the presence of BiOCl. Furthermore, BiOCl reduction was not observed in a TG/DSC experiment. Hence, during soot oxidation the active phase of Bi is the oxychloride. FeOCl is rather unstable and decomposes into hematite during soot oxidation. This is confirmed by the DRIFT spectrum of a partially converted soot/FeOCl mixture, which contains the features of Fe,O, [23]. 4.3. Mechanistic
aspects
Soot oxidation, catalyzed by metal oxides, is often thought to proceed through a reduction/oxidation mechanism (Mars and van Krevelen). In a first step the metal oxide is reduced by the soot, and in a second step the catalyst is reoxidized by air. A lot of metal oxides can be reduced by soot at the temperatures at which they catalyze its oxidation (after ball milling) [26]. Another mechanism is based on a spill-over effect: oxygen is activated on the surface of a metal oxide, and subsequently transferred to the soot surface, where it reacts yielding surface oxygen complexes (SOC) and CO and CO, [27,28]. The catalytic activity of CuCl is based on the latter mechanism, which is illustrated in Scheme 1. Starting soot oxidation with CuCl, the first step is oxygen activation on the surface of CuCl (Rl). Transfer of activated oxygen (indicated by 0 *> occurs according to reaction R2. DRIFT analysis has shown that this results in the
G. Mu1 et al. /Applied Catalysis B: Enuironmental I2 (1997) 33-47
44
cue
+ % Cl> 1 R4
R5
%
Clz
CuClz Scheme 1.
formation of surface oxygen complexes (SOC), indicated by C-O,*. Decomposition of the oxygen complexes results in the formation of CO and CO,. (Surface) oxidation of CuCl has already been proposed in the seventies as an important step in the catalytic conversion of HCl to Cl, (the Deacon reaction, Eq. (2)) and also in the oxychlorination of e.g. ethene (Eq. (3)) [29]: 4HCl+
0, t) 2C1, + 2H,O
2C,H,
+ 4HCl+
(2)
0, t) 2C,H,Cl,
+ 2H,O
(3)
The catalytic activity of CuCl, in these processes can be explained by reactions 4a (Deacon) or 4b (oxychlorination), followed by reactions 5 and 6: 2cuc1,
@ 2CuCl+
2CuC1, + C,H,
Cl,
f) 2CuCl+
(4a) C,H,Cl,
(4b)
4CuCl+
0, H 2Cu,OCl,
(5)
Cu,OCl,
+ 2HCl t) 2CuC1, + H,O
(6)
CuCl, is reduced to CuCl (reactions 4a and 4b) yielding Cl, or C 2H&l, respectively. Subsequently CuCl is oxidized to the oxychloride (Cu,OCl,) under reaction conditions. After the formation of Cu,OCl, (5), oxygen is transferred to hydrochloric acid (6), yielding water. However, bulk oxidation of CuCl does not occur during soot oxidation, as was discussed in Section 4.2. Instead, carbothermic (bulk) reduction of Cu,OCl, occurs around 600 K (if this compound is applied as the starting material), yielding CuCl and CO and CO,. Cu,OCl, carbothermally reduces at lower temperatures than CuO (600 K vs. 685 K), and even in the presence of oxygen (Fig. 4). This indicates that a chlorine ligand affects the reducibility of CuO. Once CuCl is formed, it exerts its activity according to Scheme 1. Starting soot oxidation with CuCl, results in a higher soot oxidation temperature, than starting with CuCl or Cu,OCl,, because decomposition of CuCl, (Eq. (4a), i.e. reaction R5 in Scheme 11, which
G. Mu1 et al/Applied
Catalysis B: Enuironmentall2
(1997) 33-47
45
is essential for an efficient activation of oxygen (on CuCl), takes place at higher temperatures ( > 700 K). Reaction R4 in Scheme 1 indicates the possible decomposition of the active (oxy)chloride into the oxide and chlorine by reaction with the activated oxygen. For CuCl this reaction does not occur in the temperature region where it is catalytically active (550-650 K). However, for other metal chlorides bulk oxidation occurs at relatively low temperatures, even in the presence of soot, as was observed for MoClJMoO,. At these low temperatures transfer of activated oxygen to the carbon (soot) surface is not fast enough to compensate for reaction R4. MoCl, does not show an increased activity compared to the corresponding metal oxide in ‘tight contact’. This observation indicates that redistribution of metal chlorides does not per se result in better contact than ball milling (in that case a higher activity for MOO, formed by decomposition of MoCl, was expected), as was argued in Section 4.2. In the case of lead, the active phase present during soot oxidation (the metal chloride or the metal oxychloride), is dependent on the starting compound. Neither bulk reduction of Pb,OCl,, nor oxidation of PbCl, was observed. Apparently oxygen is activated both on Pb,OCl, and on PbCl, which are stable at the temperatures where they perform their activity. Hence ‘CuCl’ in Scheme 1 can be replaced either by ‘PbCl,’ or by ‘Pb,OCl,‘. The higher activity of Pb,OCl, might be explained by better oxygen activation properties. Although we cannot exclude surface reduction, lattice oxygen of the latter compound is not consumed during soot oxidation, as was observed in the case of Cu,OCl,-. TG/DSC and DRIFT measurements have shown that BiOCl is not carbothermally reduced upon soot oxidation. Hence, the catalytic activity might be explained by an activation of oxygen on the surface of BiOCl, followed by reactions R2 and R3. Interestingly, the application of FeOCl leads to the formation of surface oxygen complexes. We have shown that catalytic soot oxidation by Fe,O, does not result in the formation of those complexes [19]. This is another indication that chlorine chemically affects the catalytic soot oxidation activity of metal chlorides and that Scheme 1 also holds for FeOCl. Evaluating the experimental results, the high activity of metal (oxy)chlorides in the oxidation of soot is induced by their high mobility or volatility. However, several arguments, previously given, lead to the conclusion that chlorine ions are not only necessary for the establishment of ‘tight contact’, but also induce oxygen activation and/or facilitate transfer of the activated oxygen to the soot surface. As can be deduced from this study, a major problem of the application of metal chlorides in soot oxidation is deactivation of the catalyst by evaporation and/or transformation of the active compounds. A detailed deactivation study of the Cu/K/Mo/Cl catalyst, which is based on the activity of CuCl, has confirmed that there is a progressive loss of catalytic material [4].
46
G. Mu1 et al./Applied
Catalysis B: Enuironmental12
(1997) 33-47
Another aspect of soot oxidation catalyzed by (copperlchlorides worthwhile mentioning is the possibility that carbon-chlorine bonds are formed by an oxychlorination like reaction. When this reaction is followed by reaction with oxygen, very toxic compounds can be formed. Luijk et al. [30] have demonstrated that during oxidation experiments of an activated carbon catalyzed by CuCl,, a relatively small burn-off of the chlorinated carbon surface gives rise to the production of chlorinated compounds such as chlorobenzenes and chlorophenols. Especially chlorophenols are very reactive precursors in the formation of polychlorinated dibenzo-p-dioxines at carbon surfaces. In this respect chlorine containing soot oxidation catalysts are less attractive for practical applications. This study has shown a relationship between the melting point of catalytic active materials and the soot oxidation activity. In our group eutectic mixtures of several oxidic compounds (which do not contain chlorine) are currently tested for their ‘loose contact’ soot oxidation activity.
5. Conclusions. * Several metal (oxy)chlorides appear to be more active in the soot oxidation than their corresponding oxides in ‘loose contact’. Especially these of Cu, Pb, Fe and Bi are very active. - The high activity of metal chlorides can be partially explained by the in situ formation of intimate contact between the soot and the active metal chloride by ‘wetting’ or through the gas phase. Metal chlorides which, due to their high melting points, cannot ‘wet’ the soot surface (like BaCl,, CaCl,, CoCl,, and NiCl,) exhibit little activity. Metal chlorides which have a too high volatility (HgCl,) neither show any catalytic activity. - The catalytic activity of metal chlorides and metal oxychlorides can be further explained by the activation of oxygen, followed by a transfer of the activated oxygen to the soot surface, resulting in SOC formation. Finally SOC decomposition results in CO and CO, evolution. - The application of metal chlorides as catalysts for diesel soot oxidation is questionable, because loss of activity by evaporation or decomposition of the active species is a severe problem.
References [l] [2] [3] [4]
J.P.A. Neeft, Fuel Proc. Techn., 47 (1996) 1. T. Inui and A. Miyamoto, Catal. Today, 10 (1991) 45. ES. Lox, B.H. Engler and E. Koberstein, Stud. Surf. Sci. Catal., 71 (1991) 291. J.P.A. Neeft, Catalytic oxidation of soot - Potential for the reduction of diesel particulate Ph.D. Thesis, TU Delft, 1995, Ch. 5; Appl. Catal. B, 8 (1996) 57. [5] J. van Doom, J. Varloud, P. Mtriaudeau and V. Perrichon, Appl. Catal. B, 1 (1992) 117.
emissions,
G. Mu1 et al./Applied
Catalysis B: Environmental I2 (1997) 33-47
47
161 G. Mul, J.P.A. Neeft, F. Kapteijn, M. Makkee and J.A. Moulijn, Appl. Catal. B, 6 (1995) 339. [71 L. Singoredjo, Low temperature selective catalytic reduction (SCR) of nitric oxide with ammonia, Ph.D. Thesis, University of Amsterdam, 1992. [81 L. Walter-Levy, A.M. Goreaud and M. Goreaud, Bull. Sot. Chim. France, (1970) 2789, 191 F.W. Lamb and L.M. Niebylski, Anal. Chem., 23 (1951) 1388. [lo] Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Ni, 57 B (1966) 689. 1111 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Co, 58 Erg. (1961) 544. [ 121 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Fe, 59 B (1932) 279. [I31 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Bi, 19 Erg. (1964) 689. [14] M.R. Lorenz and J.S. Prener, Acta Cryst., 9 (1956) 538. [15] W. Biltz and W. Fischer, Z. Anorg. Chem., 166 (1927) 290. [16] 0. Knacke, 0. Kubachewski and K. Hesselman, Thermochemical Properties of Inorganic Substances, Springer Verlag, Berlin, 199 1. [17] C. Morterra and M.J.D. Low, Spectrosc. Lett., 15(9) (1982) 689. [18] Q.-L. Zhuang, T. Kyotani and A. Tomita, Energy Fuels, 8 (1994) 714. [19] G. Mul. F. Kapteijn and J.A. Moulijn, in Proceedings of 22nd Biennial Conference on Carbon, San Diego, 1995, pp. 554-555. [20] P. Ciambelli, A. Di Pietro, V. Palma and S. Vaccaro, Thermochim. Acta, 227 (1993) 19. [21] P. Ciambelli, V. Palma and S. Vaccaro, Catal. Today, 17 (1993) 71. [22] C.F. Ng, K.S. Leung and C.K. Chan, J. Catal., 78 (1982) 51. [23] R.A. Nyquist and R.O. Kagel, Infrared Spectra of Inorganic Compounds, Academic Press, New York, 1971. [24] Y.-G. Xie and Y.-Q. Tang, Adv. Catal., 37 (1990) 1. [25] L. Walter-Levy and M. Goreaud, Bull. Sot. Chim. France, (1971) 444. [26] D.W. McKee, J. Catal., 108 (1987) 480. [27] R.T.K. Baker and J.J. Chludzinski Jr., Carbon, 19 (1981) 75. [28] T. Kyotani, S. Hayashi, A. Tomita, J.A. MacPee and R.R. Martin, Fuel, 71 (1992) 655. [29] J.A. Allen and A.J. Clark, Rev. Pure and Appl. Chem., 21 (1971) 145. [30] R. Luijk, F. Kapteijn and H.A.J. Govers, Environ. Sci. Technol., 28 (1994) 312.
ELSEVIER
Catalytic oxidation of model soot by metal chlorides Guido Mu1
*,
Freek Kapteijn, Jacob A. Moulijn
Delji University of Technology, Industrial Catalysis, Julianalaan 136, 2628 BL Delft, The Netherlands
Received 11 March 1996; revised 28 June 1996; accepted 15 July 1996
Abstract Several metal chlorides were screened for their catalytic activity in the oxidation of model soot (Printex-U) in ‘loose contact’ by means of TGA/DSC. HgCl,, CaCl,, BaCl,, CoCI,, and NiCl, show little activity. Hydrated BiCl, and FeCl, are converted in air into BiOCl and FeOCl, which have a moderate soot oxidation activity. MoCl, is converted into the corresponding metal oxide and also shows a moderate ‘loose contact’ activity. PbCl,, CuCl, and CuCl are very active catalysts; the soot oxidation temperature is lowered by 200-275 K. The activity of metal chlorides is thought to be induced by in situ formation of intimate contact between the soot and the metal chloride via ‘wetting’ and/or gas phase transport. A correlation between the melting point and the catalytic activity was found. Furthermore, a catalytic cycle is proposed involving activation of oxygen on the surface of the (oxykhloride, followed by transfer of activated oxygen to the soot surface. DRIFT analyses showed that this results in the formation of carbon surface oxygen complexes. Decomposition of those complexes yields CO and CO,. Practical application of metal chlorides for the removal of soot from diesel exhaust is not recommended, because they suffer from instability or high vapour pressures. Keywords: Soot; Oxidation; Metal chloride catalysts; Oxychlorides; Mechanism
1. Introduction Removal of soot and NO, from diesel exhaust gas is necessary to protect the environment. Although diesel exhaust is rather clean compared to Otto engine exhaust, in practice the amount of soot and NO, emitted by diesel engines is much larger than that emitted by otto engines equipped with catalytic converters
[Il. * Corresponding author. Fax: 015-2784452, E-mail: [email protected]. 0926-3373/97/$17.00 Copyright 0 1997 Elsevier Science B.V. All rights reserved PII SO926-3373(96)00065-3
34
G. Mu1 et al./Applied
Catalysis B: Environmental
12 (1997) 33-47
Soot emission can be reduced in various ways [l-3]. Collection of soot in a monolithic filter and simultaneous oxidation is considered to be a good option to minimize the contamination of the environment. However, soot oxidation takes place at temperatures of 825-875 K, while the temperature of diesel exhaust is typically 500-675 K. Hence, a catalyst is needed to prevent accumulation of soot in the monolithic filter. Several investigators have focused their attention on the application of oxidic materials to lower the oxidation temperature of soot. However, the activity of metal oxides depends on the intensity of contact between the metal oxide and soot. It has been shown that metal oxides, like Co,O, or Fe,O, can be reasonably active if ‘tight contact’ between the soot and catalyst has been established mechanically (by ball-milling), and that they lose their catalytic activity completely under so-called ‘loose contact’ conditions (after e.g. spatula mixing) [4,5]. In a previous paper we discussed the catalytic activity of a Cu/K/Mo/(Cl) catalyst [6]. Copper chlorides are present and formed within this catalyst and it was shown that they are essential for the high soot oxidation activity of a Cu/K/Mo/(Cl) catalyst in ‘loose contact’. Also other authors have shown that CuCl, is an active catalyst for the carbon oxidation reaction [7]. It was proposed that the mobility and volatility of CuCl or CuCl, results in an in situ establishment of ‘tight contact’. Furthermore, the higher activity of copper chlorides compared to copper oxide (even in ‘tight contact’ conditions) was tentatively explained by favourable redox properties of the copper (oxy)chlorides [6]. Little is known about the catalytic activity of other metal chlorides in soot oxidation and their stability with respect to this reaction. In order to find design rules for an active catalyst in soot oxidation we evaluate the catalytic activity of several metal chlorides with respect to their melting point. We also pay attention to the stability of the most active chlorides in air at the temperatures needed to combust the soot, and evaluate the necessity of a chloride ion for the activation of oxygen.
2. Experimental The metal chlorides investigated were purchased from Baker (BaCl,, CaCl,, FeCl, . 6H,O, (analytical grade)), Aldrich (CuCl, BiCl,, HgCl,, CoCl, * 6H,O, MoCl,, NiCl, . 6H,O (analytical grade)) and Merck (CuCl, * 4H,O, PbCl, (analytical grade)). The corresponding metal oxides were purchased or prepared as described elsewhere [4]. The oxychlorides of copper and lead were prepared by solid state reactions of CuCl, . 4H,O and CuO (Merck, p.a.> at 525 K for 2 h [8,25] and PbCl, and PbO (Aldrich, > 99%) at 575 K for 2 h in static air [9]. Since it is difficult to obtain diesel soot with constant properties (the composition depends on the engine load) a model soot was applied (Printex-U, a flame soot purchased from Degussa). This soot has a N,-BET surface area of 96
G. Mu1 et al/Applied
Catalysis B: Environmental
12 (1997133-47
35
m2gg’ and contains approximately 5 wt.-% of adsorbed hydrocarbons and 0.2-0.4 wt.-% sulfur [4]. Oxidation temperatures were determined in a thermobalance (STA 1500H) after mixing the soot and the catalyst in a ratio of 1: 2 by weight with a spatula (i.e. ‘loose contact’) and diluting the sample with Sic in order to prevent thermal runaways [4]. About 4 mg catalyst, 2 mg soot and 54 mg SIC were applied as a sample. This way, smooth DSC curves could be obtained. A heating rate of 10 K/min and a flow rate of 50 ml/min 21 ~01% 0, in N, were used. The maximum of the DSC curve was defined as the oxidation temperature. This temperature coincided with the temperature of the maximum weight loss rate within 10 K. TG/DSC analysis of the metal oxides was performed in ‘loose’ and ‘tight contact’ mode. Samples referred to as ‘tight contact’ were intensively milled in an agate ball mill for one hour, before dilution with SIC and thermal analysis. The temperatures determined for the metal oxides were reproducable within 5 K. Stability tests of CuCl were performed by TG/MS in N, (50 ml/min) and air (21 ~01% 0, in N,, 50 ml/min). About 15 mg of the chloride was analyzed without SIC dilution. A quadrupole mass spectrometer (Fisons instruments) was used, which was coupled to the balance with a heated capillary. In this way, products of decomposition could be detected. Carbothermic reduction of Cu,OCl, was recorded in the thermobalance in N, (Cu,OCl,:soot = 4:l by weight). A TG/DSC profile of Cu,OCl, without soot was recorded under similar conditions. Partially converted CuCl, PbC12, Pb,OCl,, FeOCl and BiOCl catalyzed soot samples were prepared (without Sic dilution, and a catalyst:soot ratio of 2:l by weight) isothermally in the thermobalance at 550 K (CuCl), 610 K (PbCl,, Pb,OCl,) and 645 K (FeOCl, BiOCl). The final temperature was reached with 10 K/min. Approximately 20 min were required to obtain soot conversion levels of 20%-60%. Diffuse reflectance infrared fourier transformed (DRIFT) spectra were recorded on a Nicolet Magna 550 spectrometer, equipped with a Spectratech DRIFT accessory. Samples were analyzed ex situ, after partial conversion of the soot in the thermobalance and dilution with KBr (1:lOO wt.-%). They were recorded against a soot in KBr background (soot:KBr = 1:lOO wt.-%). A resolution of 8 cm-’ and 256 scans were applied to obtain the spectra. Spectra are displayed without any further data processing.
3. Results 3.1. Screening
ofmetal chlorides
A typical TG/DSC profile of soot oxidation (PbCl, shown in Fig. 1. The oxidation activity is characterized
is used as a catalyst) is by the maximum of the
G. Mu1 et al. /Applied Catalysis B: Enuironmental12 (1997) 33-47
36
60.6
60.2
p
59.6
z 6 s
59.4
59.0
56.6
200
300
400
500
600
Temperature
Fig. 1, TG/DSC
analysis of PbCl,
catalyzed
700
800
900.
(K)
soot oxidation.
TG: dashed line, DSC: solid line.
exothermic heat effect, located at 683 K (Tl). Non-catalytic oxidation of Printex-U (the model soot applied) occurs around 875 K. A slight endothermic heat effect (T2) is located at 773 K, due to melting of PbCl,. This is accompanied by a gradually increasing weight loss above this temperature. Oxidation temperatures for the metal chlorides and metal oxychlorides are given in Fig. 2. This figure also contains the soot oxidation temperatures for the corresponding metal oxides in ‘loose’ (without ball milling) and ‘tight contact’ (with ball milling). Obviously the activity of the metal oxides is strongly contact dependent.
Fig. 2. Soot oxidation temperatures as determined for several metal chlorides and metal oxychlorides (‘loose contact’) and their corresponding metal oxides (‘tight contact’ (T), and ‘loose contact’ CL)). The solid horizontal line indicates the non-catalytic soot oxidation temperature (875 K).
G. Mu1 et al./Applied
Catalysis B: Environmental
I2 (1997) 33-47
31
Several metal chlorides have hardly any effect on the soot oxidation temperature. HgCl, does not lower the soot oxidation temperature. Also BaCl, and CaCl, only show a small catalytic effect. Upon heating, CoCl, . 6H,O and NiCl, . 6H,O lose crystal water; in air they are (partially) converted into the corresponding oxides ([ 10,111, and references therein). The soot oxidation activity of partially oxidized nickel and cobalt chlorides lies in between that of the ‘tight contact’ and ‘loose contact’ activity of the metal oxides. MoCl, is completely oxidized in air at relatively low temperatures. The activity of oxidized MoCl, is equal to MOO, in ‘tight contact’. FeCl, . 6H,O and hydrated BiCl, are first converted into rather stable oxychlorides [12,13]: FeOCl and BiOCl, respectively. FeOCl decomposes into Fe,O, (Hematite) and Cl, at approximately 725 K [12]. BiOCl is more stable and does not liberate Cl, below 850 K [ 131. Both (oxyjchlorides have a somewhat higher activity in the soot oxidation reaction than their corresponding metal oxides in ‘tight contact’ (i.e. after ball milling of the metal oxide and soot). Also CuCl, and PbCl, have higher soot oxidation activities than ball-milled samples of the oxides. The activity of the oxychlorides, Cu,OCl, and Pb,OCl,, is even better: maxima in the DSC curves were located at 625 and 660 K, respectively. CuCl is extraordinarily active: the soot oxidation temperature is lowered by 28.5 K. As CuCl is the most active catalyst found, a thorough TG/DSC and DRIFT analysis of CuCl was carried out, in order to reveal the active component and the mechanism by which CuCl is operative. As PbCl,, Pb,OCI,, FeOCl and BiOCl have higher activities than their corresponding oxides and a reasonable stability in the temperature range of interest, these compounds were also further investigated. 3.2. Determination
of active phases and stability thereof
The TG and DSC profiles of CuCl in a nitrogen atmosphere are shown in Fig. 3. Two heat effects can be detected at 685 K (Tl) and 698 K (T2), characteristic for CuCl, due to a phase transition [14] and melting [ 151, respectively. A considerable weight loss is observed above 700 K, due to evaporation of CuCl. This compound has an appreciable vapour pressure above this temperature [16]. The TG and DSC profiles of soot oxidation catalyzed by Cu,OCI, (a) and CuCl (c) and the oxidation of CuCl in air (4 mg, without soot, but diluted with Sic (1:15)) (b), are shown in Fig. 4. The CuCl and Cu,OCl, catalyzed soot oxidation temperatures are located at approximately 595 K (Tl) and 625 K (T5) respectively. After 100% soot conversion, an increase in weight can be observed in the range of 660-690 K, accompanied by two heat effects (T2 and T3): one of which is positive, due to oxidation, and one a superimposed negative heat effect, due to melting. Similar heat effects are observed for pure CuCl. The
38
G. Mu1 et al. /Applied
Catalysis B: Environmental
Temperature
Fig. 3. TG/DSC
12 (1997) 33-47
(K)
analysis of CuCl in N,. TG: dashed line, DSC: solid line
interpretation is the (re)oxidation of CuCl into Cu,OCl,, which is thermally stable in air up to 740 K (T4). At this temperature a weight decrease is observed, due to decomposition of the oxychloride, yielding CuO and gaseous chlorine. The formation of chlorine upon Cu,OCl, decomposition was verified by TG/MS analysis of pure CuCl in air (20 mg, without Sic dilution). The TG/DSC analysis of carbothermic reduction of Cu,OCl, in nitrogen is shown in Fig. 5. A weight loss, accompanied by an exothermic heat effect is located at approximately 600 K (Tl). This weight loss agrees with the formation of CuCl, CO and CO,. The formation of CuCl is corroborated by the heat
Tl 6O
_.-.-.-.-
._.-._._. ._._.
-__... -._.
120 ‘\ 60
a &. z .P
59-
I
56-
G I
40
z 6 E
0
Y =
57-40
56
-60 600 Temperature
700
600
900
(K)
Fig. 4. TG/DSC analyses of Cu,OCl, and CuCl catalyzed soot oxidation CuCl (without soot, curves b). TG: dashed lines, DSC: solid lines.
(curves a, c) and the oxidation
of
G. Mu1 et al./Applied
Catalysis B: Enuirotunental
I2 (1997) 33-47
39
-9 -6 -3 -0 -
-3
-
-6 -9 -12
600 Temperature
Fig. 5. TG/DSC
analysis of Cu,OCl,,
650 (K)
recorded in N,, with (dashed lines) and without soot (solid lines).
effects at 685 K (T2) and 698 K (T3), which are characteristic for CuCl (Fig. 3). These phenomena are not observed in the TG and DSC patterns of Cu,OCl, without soot. The weight loss above approximately 700 K is ascribed to the volatilisation of CuCl and Cu,OCl,. Carbothermic reduction of Pb,OCl, was not observed at the temperatures where this compound is active in soot oxidation. Oxidation of PbCl, after 100% soot conversion (Fig. l), did not take place either. Hydrated BiCl, is converted at low temperatures (450-550 K) into BiOCl. Decomposition and carbothermic reduction of the latter compound was not observed. Finally, FeOCl (formed by decomposition of FeCl, .6H,O) was not carbothermally reduced, but decomposed into Fe,O, and Cl, at approximately 720 K. 3.3. DRIFT analysis The DRIFT spectra of a partially converted CuCl/soot mixture (50% conversion, 550 K), and a BiOCl/soot mixture (20% conversion, 645 K) are depicted in Fig. 6. The spectrum of soot after a heat treatment for 30 min at 645 K, is also shown for comparison. The DRIFT spectra contain three main absorptions located at 1738 cm-‘, 1607 cm-’ and centred around 1257 cm-‘. The 1607 cm-’ absorption is caused by aromatic stretching vibrations of the soot, which are enhanced by polar functional groups like quinone [ 171. The other two absorptions have been assigned to oxygen complexes formed on the soot surface: lactones (1738 cm- ‘1, and ether-like complexes (1257 cm- ‘> respectively [ 18,191. Clearly CuCl causes an enhancement of the amount of surface oxygen complexes (SOC). The increase of the amount of SOC formed in the presence of a Cu/K/V catalyst, whose activity we believe to be induced by
G. Mu1 et al./Applied
40
8 5 e 8 $
Catalysis B: Environmental 12 (1997) 33-47
1605
2000
1265
1000 Wavenumbers (cm’)
Fig. 6. DRIFT analysis of partially converted soot samples. Non-catalytic catalyzed (50% conversion), and BiOCl catalyzed (20% conversion).
(3% soot conversion),
CuCl
CuCl [6], has already been reported by Ciambelli et al. [20,21]. According to Ciambelli the complexes mainly desorb as CO, in the presence of the Cu/K/V catalyst, and as CO and CO, in the uncatalyzed oxidation. Pure CuCl has no infrared absorptions in the spectral region recorded (400-4000 cm- ‘>. However, the absorptions located at 905, 855 and 810 cm- ’ can be ascribed to water adsorbed on CuCl [22]. A strong band below 600 cm- ’ is indicative for Cu,OCl,. As this band is not present in the spectrum displayed in Fig. 6, the DRIFT analysis confirms that oxidation of CuCl into Cu,OCl, does not occur during catalytic soot oxidation at 550 K. The DRIFT spectrum of soot, partially converted in the presence of BiOCl (20%), is included in Fig. 6. BiOCl also catalyzes the formation of surface oxygen complexes. The observed absorptions are similar to the CuCl sample. The differences in intensities of the SOC vibrations are caused by the different soot conversion levels, which were induced by different pretreatment temperatures and catalytic activities. The absorption band at 530 cm- ’ can be ascribed to BiOCl [23]. DRIFT spectra of partially converted PbCl, (15%), Pb,OCl, (35%) and FeOCl(60%) are shown in Fig. 7. Similar absorptions as in Fig. 6 can be found. Apparently, these (oxy)chlorides also catalyze the formation of surface oxygen complexes. Again, the differences in intensities of the IR bands, ascribed to SOC, are caused by different soot conversion levels of the samples. For Pb,OCl,, the DRIFT spectrum collected after partial soot conversion contains a rather large absorption band below 500 cm- ‘, indicative for the presence of the oxychloride. This band is not observed in the PbCl, sample. The absorptions indicated with an asterisk (at 1375 cm-’ and 1113 cm-‘) might be due to
G. Mu1 et al./Applied
Catalysis B: Enuironmental12
(1997) 33-47
41
0.15-
1605
2000
1000 Wavenumbers
(cm-l)
Fig. 7. DRIFT analysis of partial converted soot samples. PbCl, catalyzed (35% conversion) and FeOCl catalyzed (60% conversion).
catalyzed
(15% conversion),
Pb,OCl,
carbonate formation. Furthermore, the 1250 cm- ’ band is absent in the Pb,OCl,/soot spectrum. This is still under investigation. Nonetheless, the DRIFT analyses confirm the observations of the TG/DSC analysis, that neither carbothermic reduction of Pb,OCl,, nor (bulk) oxidation of PbCl, takes place of FeOCl into hematite is during catalytic soot oxidation. Decomposition confirmed by changes in the large band below 800 cm- ’ (not shown).
4. Discussion 4.1. Catalytic activity of metal (oxykhlorides Metal (oxylchlorides with high melting points (CaCl,, BaCl,, NiCl, and CoCl,) are less active in ‘loose contact’ than metal chlorides with relatively low melting points and high vapour pressures. Milling a metal chloride and soot hardly effects the catalytic soot oxidation temperature. These facts indicate that the high catalytic activity of several metal chlorides in ‘loose contact’, can be explained by in situ distribution of the chlorides over the soot surface (resulting in ‘tight contact’) and is related to the melting point of the compounds. Neeft [4] found a correlation between the ‘loose contact’ catalytic activity and the melting point or volatility of metal oxides. The activity of metal chlorides, expressed by the soot oxidation temperature, correlated to their melting point, as is shown in Fig. 8. The non-catalytic soot oxidation temperature is indicated by the horizontal solid line. Unfortunately, several metal (oxyjchlorides do not have a well defined melting point in air. They are (partially) transformed before they melt into the corresponding oxide and Cl, (FeOCl (Hematite), MoCl,, BiOCl, CoCl,
42
G. Mu1 et al./Applied
Catalysis B: Environmental
12 (1997) 33-47
1000
Melting or Decomposition Point (K)
Fig. 8. Correlation between the melting point of metal chlorides and the determined soot oxidation temperature. Solid dot: well defined melting point. Open dot: decomposition or oxidation takes place before melting. The horizontal solid line indicates the non-catalytic soot oxidation temperature.
and NiCl, [ 161) or decompose otherwise (CuCl,, yielding CuCl [ 161). Therefore the decomposition temperatures given by Knacke [16] or determined by ourselves (FeOCl) are plotted in the figure, except for MO, whose melting point of the oxide (MOO, [16]) was used. These metal chlorides are indicated with an open circle in Fig. 8, while a solid circle indicates that the metal chlorides (and MOO,) have a well defined melting point. Fig. 8 clearly shows that the higher the melting point of the metal chloride, the less ‘loose contact’ activity it displays. Although FeOCl, CoCl, and NiCl, are partially oxidized around the temperatures where they display catalytic activity, they are more active than the corresponding oxides in ‘loose contact’. Apparently, a rather intimate contact between the (partially decomposed) chloride and the soot can still be established. Although HgCl, does have a well defined melting temperature, it is not on the curve, obviously because HgCl, has evaporated before it can exert its catalytic influence. Whether the in situ ‘tight contact’ formation occurs by ‘wetting’ or gas phase transport has yet to be established. Xie et al. [24] have investigated the spreading (or ‘wetting’) behaviour of many inorganic salts on several carrier materials (like Al,O,, TiO, and activated carbon), and found that CuCl, was able to wet the surface of alumina [24]. Previously we have shown that gas phase transport of copper chlorides also occurs [6]. The (oxyjchlorides of Cu, Pb, Fe and, to a lesser extent, Bi are even more active than their corresponding oxides in ‘tight contact’. This might be the result of an even better contact obtained by ‘wetting’ or condensation than obtained after ball-milling. However, several experimental observations indicate that chlorine ions also effect the activation of oxygen, as will be discussed in paragraph 4.3.
G. Mu1 et al. /Applied
Catalysis B: Environmental 12 (1997) 33-47
43
4.2. The active phase in metal chloride catalyzed soot oxidation In the TG profile of soot oxidation catalyzed by Cu,OCl,, an increase in weight can be observed after complete soot conversion at 650-680 K. A similar weight increase can be observed after catalytic oxidation by CuCl and in the TG analysis of heating CuCl in air without soot. Also the heat effects are similar. Apparently, even in the presence of 20% 0, in N,, an in situ conversion of Cu,OCl, into CuCl during soot oxidation has taken place. This is corroborated by the TG profile of heating Cu,OCl, and soot in nitrogen, which showed a weight decrease around 600 K, indicating carbothermic reduction of Cu,OCl, by soot (Fig. 5). Therefore, we conclude that during soot oxidation Cu,OCl, is reduced to CuCl. The TG/DSC analysis in nitrogen of Pb,OCl, mixed with soot showed that carbothermic reduction does not occur at temperatures below the melting point. On the other hand, Fig. 1 shows that oxidation of PbCl, (upon soot oxidation) does not take place either, which is corroborated by the absence of an IR band below 500 cm-’ in the DRIFT spectrum of the partially converted soot/PbCl, mixture. Apparently, the active phase of Pb is determined by the starting compound. This conclusion is confirmed by the activity data: Pb,OCl, is more active than PbCl,. The DRIFT spectrum of a BiOCl/soot mixture after 20% conversion (Fig. 6) showed the presence of BiOCl. Furthermore, BiOCl reduction was not observed in a TG/DSC experiment. Hence, during soot oxidation the active phase of Bi is the oxychloride. FeOCl is rather unstable and decomposes into hematite during soot oxidation. This is confirmed by the DRIFT spectrum of a partially converted soot/FeOCl mixture, which contains the features of Fe,O, [23]. 4.3. Mechanistic
aspects
Soot oxidation, catalyzed by metal oxides, is often thought to proceed through a reduction/oxidation mechanism (Mars and van Krevelen). In a first step the metal oxide is reduced by the soot, and in a second step the catalyst is reoxidized by air. A lot of metal oxides can be reduced by soot at the temperatures at which they catalyze its oxidation (after ball milling) [26]. Another mechanism is based on a spill-over effect: oxygen is activated on the surface of a metal oxide, and subsequently transferred to the soot surface, where it reacts yielding surface oxygen complexes (SOC) and CO and CO, [27,28]. The catalytic activity of CuCl is based on the latter mechanism, which is illustrated in Scheme 1. Starting soot oxidation with CuCl, the first step is oxygen activation on the surface of CuCl (Rl). Transfer of activated oxygen (indicated by 0 *> occurs according to reaction R2. DRIFT analysis has shown that this results in the
G. Mu1 et al. /Applied Catalysis B: Enuironmental I2 (1997) 33-47
44
cue
+ % Cl> 1 R4
R5
%
Clz
CuClz Scheme 1.
formation of surface oxygen complexes (SOC), indicated by C-O,*. Decomposition of the oxygen complexes results in the formation of CO and CO,. (Surface) oxidation of CuCl has already been proposed in the seventies as an important step in the catalytic conversion of HCl to Cl, (the Deacon reaction, Eq. (2)) and also in the oxychlorination of e.g. ethene (Eq. (3)) [29]: 4HCl+
0, t) 2C1, + 2H,O
2C,H,
+ 4HCl+
(2)
0, t) 2C,H,Cl,
+ 2H,O
(3)
The catalytic activity of CuCl, in these processes can be explained by reactions 4a (Deacon) or 4b (oxychlorination), followed by reactions 5 and 6: 2cuc1,
@ 2CuCl+
2CuC1, + C,H,
Cl,
f) 2CuCl+
(4a) C,H,Cl,
(4b)
4CuCl+
0, H 2Cu,OCl,
(5)
Cu,OCl,
+ 2HCl t) 2CuC1, + H,O
(6)
CuCl, is reduced to CuCl (reactions 4a and 4b) yielding Cl, or C 2H&l, respectively. Subsequently CuCl is oxidized to the oxychloride (Cu,OCl,) under reaction conditions. After the formation of Cu,OCl, (5), oxygen is transferred to hydrochloric acid (6), yielding water. However, bulk oxidation of CuCl does not occur during soot oxidation, as was discussed in Section 4.2. Instead, carbothermic (bulk) reduction of Cu,OCl, occurs around 600 K (if this compound is applied as the starting material), yielding CuCl and CO and CO,. Cu,OCl, carbothermally reduces at lower temperatures than CuO (600 K vs. 685 K), and even in the presence of oxygen (Fig. 4). This indicates that a chlorine ligand affects the reducibility of CuO. Once CuCl is formed, it exerts its activity according to Scheme 1. Starting soot oxidation with CuCl, results in a higher soot oxidation temperature, than starting with CuCl or Cu,OCl,, because decomposition of CuCl, (Eq. (4a), i.e. reaction R5 in Scheme 11, which
G. Mu1 et al/Applied
Catalysis B: Enuironmentall2
(1997) 33-47
45
is essential for an efficient activation of oxygen (on CuCl), takes place at higher temperatures ( > 700 K). Reaction R4 in Scheme 1 indicates the possible decomposition of the active (oxy)chloride into the oxide and chlorine by reaction with the activated oxygen. For CuCl this reaction does not occur in the temperature region where it is catalytically active (550-650 K). However, for other metal chlorides bulk oxidation occurs at relatively low temperatures, even in the presence of soot, as was observed for MoClJMoO,. At these low temperatures transfer of activated oxygen to the carbon (soot) surface is not fast enough to compensate for reaction R4. MoCl, does not show an increased activity compared to the corresponding metal oxide in ‘tight contact’. This observation indicates that redistribution of metal chlorides does not per se result in better contact than ball milling (in that case a higher activity for MOO, formed by decomposition of MoCl, was expected), as was argued in Section 4.2. In the case of lead, the active phase present during soot oxidation (the metal chloride or the metal oxychloride), is dependent on the starting compound. Neither bulk reduction of Pb,OCl,, nor oxidation of PbCl, was observed. Apparently oxygen is activated both on Pb,OCl, and on PbCl, which are stable at the temperatures where they perform their activity. Hence ‘CuCl’ in Scheme 1 can be replaced either by ‘PbCl,’ or by ‘Pb,OCl,‘. The higher activity of Pb,OCl, might be explained by better oxygen activation properties. Although we cannot exclude surface reduction, lattice oxygen of the latter compound is not consumed during soot oxidation, as was observed in the case of Cu,OCl,-. TG/DSC and DRIFT measurements have shown that BiOCl is not carbothermally reduced upon soot oxidation. Hence, the catalytic activity might be explained by an activation of oxygen on the surface of BiOCl, followed by reactions R2 and R3. Interestingly, the application of FeOCl leads to the formation of surface oxygen complexes. We have shown that catalytic soot oxidation by Fe,O, does not result in the formation of those complexes [19]. This is another indication that chlorine chemically affects the catalytic soot oxidation activity of metal chlorides and that Scheme 1 also holds for FeOCl. Evaluating the experimental results, the high activity of metal (oxy)chlorides in the oxidation of soot is induced by their high mobility or volatility. However, several arguments, previously given, lead to the conclusion that chlorine ions are not only necessary for the establishment of ‘tight contact’, but also induce oxygen activation and/or facilitate transfer of the activated oxygen to the soot surface. As can be deduced from this study, a major problem of the application of metal chlorides in soot oxidation is deactivation of the catalyst by evaporation and/or transformation of the active compounds. A detailed deactivation study of the Cu/K/Mo/Cl catalyst, which is based on the activity of CuCl, has confirmed that there is a progressive loss of catalytic material [4].
46
G. Mu1 et al./Applied
Catalysis B: Enuironmental12
(1997) 33-47
Another aspect of soot oxidation catalyzed by (copperlchlorides worthwhile mentioning is the possibility that carbon-chlorine bonds are formed by an oxychlorination like reaction. When this reaction is followed by reaction with oxygen, very toxic compounds can be formed. Luijk et al. [30] have demonstrated that during oxidation experiments of an activated carbon catalyzed by CuCl,, a relatively small burn-off of the chlorinated carbon surface gives rise to the production of chlorinated compounds such as chlorobenzenes and chlorophenols. Especially chlorophenols are very reactive precursors in the formation of polychlorinated dibenzo-p-dioxines at carbon surfaces. In this respect chlorine containing soot oxidation catalysts are less attractive for practical applications. This study has shown a relationship between the melting point of catalytic active materials and the soot oxidation activity. In our group eutectic mixtures of several oxidic compounds (which do not contain chlorine) are currently tested for their ‘loose contact’ soot oxidation activity.
5. Conclusions. * Several metal (oxy)chlorides appear to be more active in the soot oxidation than their corresponding oxides in ‘loose contact’. Especially these of Cu, Pb, Fe and Bi are very active. - The high activity of metal chlorides can be partially explained by the in situ formation of intimate contact between the soot and the active metal chloride by ‘wetting’ or through the gas phase. Metal chlorides which, due to their high melting points, cannot ‘wet’ the soot surface (like BaCl,, CaCl,, CoCl,, and NiCl,) exhibit little activity. Metal chlorides which have a too high volatility (HgCl,) neither show any catalytic activity. - The catalytic activity of metal chlorides and metal oxychlorides can be further explained by the activation of oxygen, followed by a transfer of the activated oxygen to the soot surface, resulting in SOC formation. Finally SOC decomposition results in CO and CO, evolution. - The application of metal chlorides as catalysts for diesel soot oxidation is questionable, because loss of activity by evaporation or decomposition of the active species is a severe problem.
References [l] [2] [3] [4]
J.P.A. Neeft, Fuel Proc. Techn., 47 (1996) 1. T. Inui and A. Miyamoto, Catal. Today, 10 (1991) 45. ES. Lox, B.H. Engler and E. Koberstein, Stud. Surf. Sci. Catal., 71 (1991) 291. J.P.A. Neeft, Catalytic oxidation of soot - Potential for the reduction of diesel particulate Ph.D. Thesis, TU Delft, 1995, Ch. 5; Appl. Catal. B, 8 (1996) 57. [5] J. van Doom, J. Varloud, P. Mtriaudeau and V. Perrichon, Appl. Catal. B, 1 (1992) 117.
emissions,
G. Mu1 et al./Applied
Catalysis B: Environmental I2 (1997) 33-47
47
161 G. Mul, J.P.A. Neeft, F. Kapteijn, M. Makkee and J.A. Moulijn, Appl. Catal. B, 6 (1995) 339. [71 L. Singoredjo, Low temperature selective catalytic reduction (SCR) of nitric oxide with ammonia, Ph.D. Thesis, University of Amsterdam, 1992. [81 L. Walter-Levy, A.M. Goreaud and M. Goreaud, Bull. Sot. Chim. France, (1970) 2789, 191 F.W. Lamb and L.M. Niebylski, Anal. Chem., 23 (1951) 1388. [lo] Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Ni, 57 B (1966) 689. 1111 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Co, 58 Erg. (1961) 544. [ 121 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Fe, 59 B (1932) 279. [I31 Gmelins Handbuch der Anorganischen Chemie, 8. Auflage, Verlag Chemie, Weinheim, Bi, 19 Erg. (1964) 689. [14] M.R. Lorenz and J.S. Prener, Acta Cryst., 9 (1956) 538. [15] W. Biltz and W. Fischer, Z. Anorg. Chem., 166 (1927) 290. [16] 0. Knacke, 0. Kubachewski and K. Hesselman, Thermochemical Properties of Inorganic Substances, Springer Verlag, Berlin, 199 1. [17] C. Morterra and M.J.D. Low, Spectrosc. Lett., 15(9) (1982) 689. [18] Q.-L. Zhuang, T. Kyotani and A. Tomita, Energy Fuels, 8 (1994) 714. [19] G. Mul. F. Kapteijn and J.A. Moulijn, in Proceedings of 22nd Biennial Conference on Carbon, San Diego, 1995, pp. 554-555. [20] P. Ciambelli, A. Di Pietro, V. Palma and S. Vaccaro, Thermochim. Acta, 227 (1993) 19. [21] P. Ciambelli, V. Palma and S. Vaccaro, Catal. Today, 17 (1993) 71. [22] C.F. Ng, K.S. Leung and C.K. Chan, J. Catal., 78 (1982) 51. [23] R.A. Nyquist and R.O. Kagel, Infrared Spectra of Inorganic Compounds, Academic Press, New York, 1971. [24] Y.-G. Xie and Y.-Q. Tang, Adv. Catal., 37 (1990) 1. [25] L. Walter-Levy and M. Goreaud, Bull. Sot. Chim. France, (1971) 444. [26] D.W. McKee, J. Catal., 108 (1987) 480. [27] R.T.K. Baker and J.J. Chludzinski Jr., Carbon, 19 (1981) 75. [28] T. Kyotani, S. Hayashi, A. Tomita, J.A. MacPee and R.R. Martin, Fuel, 71 (1992) 655. [29] J.A. Allen and A.J. Clark, Rev. Pure and Appl. Chem., 21 (1971) 145. [30] R. Luijk, F. Kapteijn and H.A.J. Govers, Environ. Sci. Technol., 28 (1994) 312.