Rico Canadinhas | Assassin’s Creed | 98 Secrets D'Une Victoire FRENCH HDTV 720p 2018

Selective reduction of nitrobenzene to nitrosobenzene over different kinds of trimanganese tetroxide catalysts

Selective reduction of nitrobenzene to nitrosobenzene over different kinds of trimanganese tetroxide catalysts

/ ELSEVIER APPLIED CATALYSS I A:GENERAL Applied Catalysis A: General 133 (1995) 81-93 Selective reduction of nitrobenzene to nitrosobenzene over di...

626KB Sizes 0 Downloads 12 Views

/ ELSEVIER

APPLIED CATALYSS I A:GENERAL

Applied Catalysis A: General 133 (1995) 81-93

Selective reduction of nitrobenzene to nitrosobenzene over different kinds of trimanganese tetroxide catalysts W a n g W e i m i n , Y a n g Y o n g n i a n , Z h a n g Jiayu * Institute of Chemistry and Molecular Engineering, Peking University, Beijing, People's Republic of China Received 4 April 1995; revised 5 July 1995; accepted 19 July 1995

Abstract

Selective reduction of nitrobenzene has been measured on several kinds of Mn304 catalysts obtained by different methods of preparation. Attention is mainly given to the relationship between catalytic performance of different Mn304 materials and their redox behaviour. It was found that the catalyst having the most active redox behaviour favoured the formation of highly reduced product (e.g. aniline etc) but, the catalyst having the least active redox behaviour showed poor catalytic activity. It s e e m s that the catalytic as well as the redox behaviour of Mn304 could be related to the Jahn-Teller effect. Keywords: Jahn-Teller effect; Manganese oxide; Nitrobenzene reduction; Nitrosobenzene

1. Introduction The high reactivity of nitrosobenzene in many reactions makes it an important intermediate for the production of antioxidants and other organic chemicals [ 1 ], thereby stimulating the study and development of a one-step reduction of nitrobenzene in place of the well known two-step process, which produces troublesome waste products, e.g. Zn and Cr compounds. It was claimed in patents [2,3] that modified manganese oxides show high activity and selectivity in the one-step reduction of nitrobenzene to nitrosobenzene. Recently, it has been reported that unpromoted pure manganese oxide is also a good catalyst for this reaction and the active phase is shown to be Mn304 (hausmannite) [4]. * Corresponding author. Fax. ( + 86-1 ) 2564095. 0926-860X/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSD! 0 9 2 6 - 8 6 0 X ( 95 )OO186-7

82

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

It was first proposed [5] that the reaction scheme of this selective reduction could be described by the redox mechanism postulated by Mars and Van Krevelen [6], namely, the nitrobenzene molecule inserts one of its oxygen atoms into the lattice oxygen vacancies of Mn304, itself being selectively reduced to nitrosobenzene. According to this scheme, the catalytic reduction activity of Mn304 should be closely associated with its redox behaviour. In the previous papers, we have reported the preparation of two kinds of Mn304 [7,8] (i.e. the non- and nearstoichiometric) and a supported Mn304 [9], and have also studied their redox behaviour in detail by temperature-programmed techniques (TPR, TPO and DTA etc). In this paper, measurement of their catalytic performance was carried out in order to ascertain whether there is a close relationship between the catalytic performance of the different Mn304 prepared and their redox behaviour.

2. Experimental 2.1. Catalyst preparation

Non-stoichiometric Mn304 (i.e. MnO > 1.33) was prepared by decomposition of Mn(OH)2 in air at 130°C for 24 h. The hydroxide was obtained by precipitation from an aqueous solution of manganous acetate (A.R.) with concentrated NH3" H20 at pH = 9. After 12 h, the precipitate was separated from the solution by centrifugation, followed by washing free of extraneous ions with demineralized water [ 7]. Near-stoichiometric Mn304 (i.e. MnO ~ 1.33) was prepared by a one-step procedure [8], namely, Mn(OAc)2.4H20 was decomposed in air at 350°C for 0.5 h, followed by crystallization in a nitrogen flow at 500°C for 12 h, in order to obtain sharp X-ray diffraction lines. Alumina supported Mn304 was prepared by two methods [9]: ( 1) Incipient wetness impregnation of alumina powder with an aqueous solution of manganese(II) acetate, followed by rapid drying at about 120°C and subsequently calcining in air at 350°C for 0.5 h. With this method, either the monolayer dispersed phase or the so called paracrystalline phase [ 10] of Mn304 could be obtained depending on the Mn loading. (2) Thorough co-milling of Mn (OAc)2" 4H20 with alumina in an agate mortar, followed by thermal decomposition in air at 350°C for 0.5 h. This method is not suitable for preparing monolayer dispersed Mn304 as has been shown in Ref. [9]. 2.2. Temperature-programmed reduction (TPR) measurements TPR measurements were carried out using the same apparatus as described elsewhere [ 11 ].

W. Weimin et al./Applied Catalysis A: General 133 (1995) 81-93

83

2.3. Catalytic measurements A glass continuous flow apparatus with a fixed bed reactor containing 200-600 mg of catalyst powder was used for the catalytic measurements at 1 atm and 330°C, using nitrogen as a carder gas and hydrogen as a reducing agent (N2/H2 = 100/ 5). The catalyst was first activated under a N2/H2 flow (5 ml/min) at the reaction temperature of 330°C for 1 h. Liquid nitrobenzene was then fed into the reactor concurrently with N2/H2 flow (5 ml/min) at a rate of 2.2096 g/h (within an error of 1% during 12 h of measurement). Under such conditions, the conversion of nitrobenzene can be kept below 10% (i.e. under differential conditions) for all experiments performed to avoid complications caused by extensive consecutive reactions. The activity of a catalyst is defined as the percentage of nitrobenzene converted. Selectivity for nitrosobenzene is expressed as the amount of nitrosobenzene related to the amount of nitrobenzene reacted. The areal activity (activity per unit of surface area) is not used, as the problem of the surface area measurement of the supported Mn304 has not been solved.

2.4. Product analysis The off gases from the reactor were passed into a cold trap, which was renewed every hour, the liquid product mixture in the trap was analyzed by spectrophotometry, using a Shimadzu UV-120-20 spectrophotometer. It is known [ 2,3,5 ] that the main by-product of the reduction reaction is aniline; the others are very small quantities of dimerized product as azo- and azoxybenzene. Accordingly, it is only necessary to perform spectrophotometric analysis of a three-component system containing nitrobenzene, nitrosobenzene and aniline. The sample analyzed was prepared by diluting about 20 mg of liquid product to a known volume. The UV absorption spectra of the individual compounds in methanol, i.e. nitrobenzene (A.R.), nitrosobenzene (Sigma Chem. Comp.), aniline (A.R. redistilled O O

I

'

'

'

'

I

'

'

'

'

1

'

'

'

'

(-)

I

P <2)

c

O

,/

o

g

b

r,,:,

g o

o

i-

v-.

o o o o

r

90

250

300 ~nm)

I

350 O

Fig. 1. UV absorption spectra of three substances in methanol, measured separately with Shidmadzu UV-2100, (a) nitrobenzene, (b) nitrosobenzene, (c) aniline.

84

W. Weimin et al. /Applied Catalysis A." General 133 (1995) 81-93 ro o o o

o o o

o

o

o

o o o

g

o o

o 0

L

190 0

0

0 ,

i

i

i

300

250

i

350

r

o

4O0 ~

trim)

o

Fig. 2. UV absorption spectrum of aniline in methanol, after adding HC1.

before used) have been measured separately with a Shimadzu UV-2100 spectrophotometer as shown in Fig. 1. Because of the ill-defined extinction of the solution due to the slow oxidation of aniline in air, a few drops of concentrated HC1 must be added to the product mixture solution, thus causing the UV absorption peak of aniline to shift to shorter wavelength (as a result of salt formation) as shown in Fig. 2. However, no effect of HC1 was observed on the UV spectrum of either nitrobenzene or nitrosobenzene. The analysis could thus be carried out without the interference of aniline in the wavelength range of 250-300 nm,

3. Results

3.1. Correctness and feasibility of the method of analysis used It was mentioned above that by adding HCI the spectrophotometric analysis of the liquid product could be simplified as an analysis of a two component system containing both nitro- and nitrosobenzene. According to the superposition principle (i.e. extinctions are additive [12] ), the spectrophotometric analysis of a solution containing nitro- and nitrosobenzene should be solved with a pair of simultaneous equations as follows: E(259)

= ° ° n b ( 2 5 9 ) ' C n b "~ ~ ' n s b ( 2 5 9 ) " C n s b

E(3oo)

= ~ ' n b ( 3 0 0 ) " C n b -~- ' ~ n s b ( 3 0 0 ) " C n s b

where 'nb' and 'nsb' denote nitro-and nitrosobenzene, respectively; E(259 ) and E(3oo ) are the extinctions of mixed solution at wavelengths 259 nm and 300 nm; eob(259~, e,b(300), ensb(259) and gnsb(300) are the extinction coefficients (e) of nitroand nitrosobenzene at wavelengths 259 nm and 300 nm, which are the wavelengths of absorption maxima as seen in Fig. la and lb; Cab and Cnsbdenote the concentrations of nitro- and nitrosobenzene, respectively, in the sample solution.

W. Weimin et al. / Applied Catalysis A: General 133 (1995) 81-93

85

Table 1 Extinction coefficients of nitro- and nitrosobenzene

A (nm)

substance

linear fitting equation

e (A)

259 259 300 300

nb nsb nb nsb

Y= 0.707089X+0.0364272 Y=O.366577X+O.0211982 Y=O.O756503X+O.0327412 Y=O.644328X+O.0325947

0.707089 0.366577 0.0756503 0.644328

The slope (e) values of the straight line plots (i.e. extinction vs. concentration) of nitro- and nitrosobenzene at wavelengths 259 nm and 300 nm have been determined and are shown in Table 1. According to the actual conversion of the reaction performed (below 10% conversion of nitrobenzene), we prepared three known solutions containing aniline, nitrosobenzene and nitrobenzene as calibration solutions to check the correctness of the method used. The results of the check calculation using the aforementioned equations are shown in Table 2. It seems that the relative error of the method used for nitrosobenzene is within 10% when the concentration ratio of nitrosobenzene to nitrobenzene is above 4%, the error becomes larger when the concentration ratio is within 1%. However, it might be adopted for comparison only because of the good reproducibility as confirmed by tests.

3.2. Results of catalytic measurements Seven different supported and unsupported Mn304 catalysts with different states and redox behaviour used in this study are summarized in Table 3. The symbol HFB (high frequency band) and LFB (low frequency band) denote the IR characteristics of two kinds of Mn304 and the numbers in parenthesis denote the weight ratio of Mn304 to support. Generally, Mn304 prepared by thermal decomposition of manganous acetate shows LFB IR characteristic as mentioned in detail elsewhere [ 8 ]. The apparent state of Mn304 prepared was characterized by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and temperature-programmed reduction (TPR) [9]. Table 2 Analysis results of calibration solutions Code No.

I 2 3

Nominal conc. ( 10-4 m)

Calculated conc.

nb

nsb

an a

nb

nsb

3.19 3.19 3.19

0.255 0.127 0.0255

0.0259 0.129 0.259

3.18 3.19 3.22

0.277 0.119 0.0166

" 'an' denotes aniline.

Relative error an ~

nb

nsb

-0.2% +0.1% +0.9%

+9% -7% -35%

Cnsb Cnb

8% 4% 0.8%

86

W. Weimin et al. / Applied Catalysis A." General 133 (1995) 81-93

Table 3 Summary of seven different Mn304 catalysts Type

Notation Preparation

HFB Mn304 ( MnO > i 33)

Ai

LFB Mn304(MnO_ ,~3)

A2

LFB/CK300(30/100) paracrystalline

A~

LFB/CK300(30/100) mixed state (dispersed + crystalline )

A4

LFB/D-AI~O3 (30/100) mixed state

A5

LFB/CK300(2.5/100) dispersed state

A6

LFB /quartz ( 25 / 100 )

A7

thermal decomposition of Mn (OH), thermal decomposition of Mn (OAc)2.4H20 incipient wetness impregnation, N2 treated at 500°C for 12 h co-milling CK300 with Mn (OAc) 2 4H~O, calcined in air at 350°C, 0.5 h co-milling D-A1203 with Mn (OAc) 2" 4H20, calcined in air at 350°C, 0.5 h incipient wetness impregnation co-milling quartz with LFB Mn304 (i.e. A2)

Loading weight (mg)

Ref.

200

[7 ]

560

[8 ]

500

19 ]

500

191

500

[9]

400

[ 9]

680

[9]

Note: CK300 and D-A1203 stand for Ketjen/Cynamid A1203 and Degussa AI203, respectively.

The catalytic measurements have been performed under comparable experimental conditions and the catalytic results obtained with different Mn304 catalysts as a function of reaction time are listed in Tables 4 - 9 , in which A 7 is not included as it shows no activity at all for the reduction reaction of nitrobenzene. All catalysts listed in Tables 4-9 show higher activity and selectivity after one hour and then drop slightly. The conversion then remains approximately constant at about 6-7% with a selectivity of about 20% for nitrosobenzene (except for A6) in later stages of reaction. This means that, on the one hand, there is little difference between these catalysts based on Mn304, and on the other hand there seems no effect of alumina support on the catalytic performance of Mn304. The latter fact is due to the weak interaction between Mn and A1203 as demonstrated in our previous Table 4 Catalytic results of A, Time (h)

1 2 3 4 5 7 8 9

Sample weight (g)

0.0216 0.0217 0.0217 0.0217 0.0218 0.0218 0.0218 0.0218

Cn~ ( 10 4)

3.27 3.32 3.31 3.32 3.33 3.34 3.33 3.33

Cn~b ( 10 4)

0.0968 0.0701 0.0587 0.0569 0.0519 0.0518 0.0503 0.0486

Conversion (%)

7.7 7.0 7.2 6.8 7.1 6.7 7.1 7.1

Selectivity nsb

others

0.36 0.28 0.23 0.24 0.20 0.21 0.20 0.19

0.64 0.72 0.77 0.76 0.80 0.79 0.80 0.81

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

87

Table 5 Catalytic results of A2 Time (h)

1 2 3 4 5 7 8 9

Sample weight (g)

0.0218 0.0217 0.0218 0.0219 0.0219 0.0219 0.0220 0.0217

C.b ( 10 4)

C.~b ( 10 4)

Conversion (%)

Selectivity nsb

others 0.59 0.69 0.72 0.77 0.8(I 0.81 0.79 0.81

3.27 3.33 3.35 3.35 3.36 3.36 3.38 3.34

0.125 0.0717 0.0583 0.0550 0.0468 0.0434 0.0465 0.0419

8.5 6.5 6.3 6.7 6.6 6.5 6.2 6.1

0.41 0.31 0.28 0.23 0.20 0.19 0.21 0.19

C.b ( 10 -4)

Cnsb ( 10 - 4 )

Conversion (%)

Selectivity

Table 6 Catalytic results of A3 Time (h)

I 2 3 4 5 7 8 9

Sample weight (g)

0.216 0.0217 0.0216 0.0217 0.0216 0.0216 0.0216 0.0216

nsb

others 0.50 0.70 0.76 0.76 0.76 0.83 0.80 0.85

3.13 3.31 3.29 3.34 3.32 3.30 3.31 3.32

0.210 0.0751 0.0621 0.0584 0.0552 0.0423 0.0471 0.0339

11.8 7. I 7.2 6.4 6.3 6.9 6.6 6.4

0.50 0.30 0.24 0.24 0.24 0.17 0.20 0.15

C.h(10 4)

Chub(10 4)

Conversion (%)

Selectivity

Table 7 Catalytic results of A 4 Time(h)

1 2 3 4 5 7 8 9

Sample weight (g)

0.0216 0.216 0.0216 0.0216 0.0217 0.0217 0.0216 0.0216

3.22 3.35 3.32 3.32 3.29 3.31 3.31 3.31

0.101 0.0517 0.0405 0.0339 0.0325 0.0274 0.0274 0.0274

9.3 5.4 6.4 6.4 7.5 7.1 6.7 6.7

nsb

others

0.30 0.27 0.18 0.15 0.12 0.11 0.12 0.12

0.70 0.73 0.82 0.85 0.88 0.89 0.88 0.88

paper [9]. The low selectivity for nitrosobenzene is likely due to the relatively strong reducing power of hydrogen which results in the creation of a higher density of oxygen vacancies and larger amounts of adsorbed hydrogen, which favours the formation of highly reduced product e.g. aniline etc.

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

88

Table 8 Catalytic results of A5 Time (h)

I 2 3 4 5 6 7 8 9

Sample weight (g)

0.0217 0.0216 0.0216 0.0216 0.0216 0.0216 0.0216 0.0216 0.0216

C.b ( 10 4)

3.25 3.32 3.31 3.36 3.34 3.32 3.32 3.35 3.32

C.~h ( 10 -4)

0.0805 0.0289 0.0175 0.0155 0.0156 0.0109 0.0109 0.0139 0.0109

Conversion (%)

8.5 6.1 6.3 5.1 5.5 6.2 6.2 5.4 6.2

Selectivity nsb

others

0.27 0.13 0.08 0.09 0.08 0.05 0.05 0.07 0.05

0.73 0.87 0.92 0.91 0.92 0.95 0.95 0.93 0.95

Close inspection of the results listed in Table 4 and Table 5 reveals that 200 mg A, possesses nearly the same catalytic performance as does 560 mg A2. This is because AI has a larger surface area than A2, e.g. 49 m2/g and 15 m2/g (see [8] ) rather than other factors. Moreover, it was found that A1 after use for catalytic reaction showed a somewhat similar TPR profile to that of A2 as shown by the diminution of the LTRP (low temperature reduction peak) and the increase of the Tm~xof the HTRP (high temperature reduction peak) of A I (see Fig. 3). However, the 'spent' AI still maintained its HFB IR characteristic, namely, no shift to lower wavenumbers of the IR bands at 650-500 c m - 1 was observed after 9 h of catalytic reaction. This observation is at variance with the report in Ref. [ 13 ], probably due to the shorter period of reaction in this study. Worthy of particular mention is the catalytic performance of alumina supported Mn304 catalysts, where Mn304 can be deposited on alumina with different states, e.g. two dimensional monolayer dispersed state and/or three dimensional, the so called paracrystalline state. It is of interest to note that A6, the monolayer dispersed Mn304, having a single TPR peak ( see [ 9 ] ) at ca 380°C [ well below the Tmax of A2 (500°C), see Fig. 3b ], Table 9 Catalytic results of A6 Time (h)

1 2 3 4 5 7 8 9

Sample weight (g)

0.0217 0.0217 0.0217 0.0216 0.0217 0.0216 0.0217 0.0216

C.b ( 10-4)

3.31 3.35 3.35 3.36 3.35 3.34 3.34 3.33

Cns b ( 10 4)

0.0751 0.0139 0.00401 0.00555 0.00236 -0.00414 -0.00579 -0.00733

Conversion (%)

6.8 5.9 5.7 4.9 5.7 5.6 6.0 6.0

Selectivity nsb

others

0.31 0.07 0.02 0.03 0.01 n.d. n.d. n.d.

0.69 0.93 0.98 0.97 0.99 n.d. n.d. n.d.

W. Weimin et al./Applied Catalysis A: General 133 (1995) 81-93

89

500

483

'

I

'

I



I

I

I



I



|



t

i

|

100 300 500 "C 100 3o0 500 700 "(2

Fig. 3. TPR profiles of (a) AI (HFB MnO>l.33), (b) A 2 (LFB MNO~1.33), (c) the 'spent' Al, N2 pretreated (raising the temperature to 600°C to remove the residual substances formed during reaction).

shows a conversion of ca. 6% (Table 9), in which only a trace of nitrosobenzene (within 1% of liquid product) can be detected during the steady-state operation. However, what remains unclear is the higher selectivity (ca. 30%) for nitrosobenzene of A6 during the early stage of reaction (within one hour of measurement). The 'spent' m 6, i.e. after having been subjected to catalytic reaction for 9 h, was subsequently characterized by X-ray powder diffraction, which revealed the same XRD pattern as that of the freshly prepared A 6, indicating that the structure of m 6 was presumably unchanged during catalytic reaction. Thus, the deactivation of m 6 towards selective reduction of nitrobenzene, in our opinion, may be attributed to the intrinsic property of A6, having a two-dimensional structure as expected from the results of Ref. [4], namely, the active phase is the three-dimensional spinel hausmannite, or attributed to the self-poisoning of A6 by carbonaceous like species formed by the reaction. This awaits further exploration. Comparison of A 3 (the so called paracrystalline Mn304) with its analogue A 2 ( L F B Mn304) reveals that 500 mg A3, i.e. 23% LFB/CK300 (30/100 weight ratio) containing 115 mg A2, is slightly better than or equal to 560 mg A2 in activity or even selectivity (see Table 5 and Table 6). Obviously, the superiority of A 3 over A2 is certainly a result of the increase of surface area, since there is no interaction between Mn304 and alumina as mentioned above. We should mention in passing that the pretreatment of A 3 (originally having two TPR peaks as shown in Fig. 4a) with nitrogen at 500°C for 12 h (Table 3) aimed to obtain the three-

90

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

L,(b)

(a) !

1

I

~

i

i

i

!

100 300 500 700 "C Fig. 4. TPR profiles of (a) A3, before N2 treatment, (b) 'A'3, Ne pretreated (500°C).

dimensional phase of Mn304 only, since the LTRP of Fig. 4a represents the reduction of the monolayer dispersed phase of Mn304 [9]. In addition, comparing Table 7 with Table 6, one can see that A4, a mixture of dispersed and crystalline phase of LFB Mn304 having two TPR peaks like Fig. 4a, shows somewhat smaller activity for nitrosobenzene formation than A 3 with a crystalline phase only, confirming the likelihood of the low activity of the monolayer dispersed phase in the reaction in question. Finally, let us compare the catalytic results of As and A5 (Table 7 and Table 8), both of them have the same Mn loading (30/100) and have been prepared by the same method but with a different alumina as support (Table 3). It seems that the activity for nitrosobenzene formation of A 4 is twice as high as that of As. Obviously, this is because the specific surface area of CK300 is also twice as high as that of D-A1203.

4. Discussion As suggested in Section 1 of this paper, the activity of Mn304 towards selective reduction of nitrobenzene should be related to its redox behaviour, if the reaction follows the Mars and Van Krevelen [6] mechanism. We have already reported the redox behaviour (see Refs. [ 7-9 ] ) and the catalytic performance (this paper) of a variety of Mn304 catalysts. It turns out that there seems to be a close relationship between the two properties as will be discussed in what follows.

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

91

It has been reported [ 8 ] that the redox behaviour of A 2 (MnO ~ 1.33) is less active than that of AI (MnO> 1.33)" For example, A2 exhibits higher Tma× of TPR and TPRO (temperature-programmed reoxidation) than that of A~, that should render A2 less active than A~ in catalyzing selective reduction of nitrobenzene. However, A1 and A2 show little or no difference in activity or selectivity essentially as reported above. This will be discussed later. It is known [ 14] that Mn304 has a normal spinel structure with Mn ~ ions in tetrahedral sites and Mn m ions in tetragonally distorted octahedral sites. Both A1 and A2 show the same XRD pattern as reported in [8]; however, they are distinguishable by their different M n / O stoichiometry as measured by thermogravimetry and their different IR characteristic (see also Ref. [ 8 ] ). The important difference in the redox behaviour between A~ and A2 can in our opinion be attributed to the supra-stoichiometric oxygen, bound to the tetrahedral Mn. For example, it is suggested [7,8] that the LTRP at 272°C of A~ (Fig. 3a) is due to the supra-stoichiometric oxygen bound to the tetrahedral Mn In (for maintaining the electrovalent balance) and the tetrahedral Mn in can weaken the octahedral Mn HIO bonds that may cause the Tmaxof HTRP of A1 to decrease from 500°C(A2) to 473°C (Fig. 3). As already mentioned in Section 2.3 of this paper, the catalysts were activated before reaction under TPR gas (i.e. Nz/H2) at a temperature of 330°C for one hour, which is much higher than the Tmax (272°C) of AI. This pretreatment can undoubtedly remove the supra-stoichiometric oxygen of A1, which explains the small difference in catalytic activity between A~ and A2. This is shown also by the close similarity of the TPR profile of Al after being used for 9 h reaction (see Fig. 3c) to that of Ae (Fig. 3b). We consider that both AI and A: have their own advantage as well as disadvantage when used as catalyst for the selective reduction of nitrobenzene. For example, A1 (MnO > 1.33), prepared by low-temperature decomposition of manganese (II) hydroxide [ 7 ] has a twice higher surface area than that of A2 (MnO ~ 1.33), prepared by higher temperature decomposition of manganese(II) acetate [8] (see also Table 3). However, the method used for the preparation of A: is simpler and easier than that of A~, since no troublesome washing procedure is necessary, e.g. the hydroxide precipitate needs thorough washing before decomposition when manganese(II) nitrate is used as a precursor, otherwise the final product would be brownish black, corresponding more to Mn203 than the Mn304 due to oxidation of the residual nitrate [7]. Moreover, it is most convenient to disperse manganese(II) acetate rather than the hydroxide via impregnation onto a support with high surface area. On the thermal decomposition of the supported acetate, a very small particle - - so called paracrystalline [ 10] Mn304 - - can be obtained. This can be regarded as the maximum benefit from using alumina as the support of Mn304 catalyst as shown by the results of this study. As mentioned above, the monolayer dispersed ('surface') Mn304, i.e. the A 6 seems to be the most redox active of the catalysts under discussion. For example,

92

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

A 6 has the lowest Tmax (ca. 380°C) of the TPR. Thus, during steady-state operation, the A 6 should be in the more reduced form (e.g. MnO) with high density of oxygen vacancies. This favours the formation of highly reduced products like aniline and may be the reason of the poor selectivity of A 6. It has been reported elsewhere [9] that the A7, prepared by co-milling inert quartz with A2 (see Table 3) in an agate mortar, shows the same XRD pattern as that of pure A2, but with smaller d-values as shown in Table 4 of Ref. [ 9 ], indicating the lattice contraction of Mn304 induced by strong mechanical force. This makes A7 less redox active as confirmed by TPR and DTA [9], and also inactive in the catalytic reduction as reported above. In other words, only one reason can explain the inactivity of m7, namely, the lattice contraction of Mn304. One point is of particular importance. The effect of milling is only observed with Mn304, which has a distorted spinel structure (because of the Jahn-Teller effect), but it is absent with the non-distorted spinel like Co304, as shown in Ref. [9]. It appears that the process of lattice contraction occurs as a result of structure change from the metastable to the stable state. Accordingly, a mechanical treatment ('tribochemical') of the Mn304 removes the Jahn-Teller prolongation of the lattice constant and, simultaneously, suppresses the catalytic activity which is based on the Mars and Van Krevelen [6] mechanism. In some sense, the catalytic as well as the redox behaviour of Mn304 could be related to the Jahn-Teller effect as suggested earlier in Ref. [ 5 ].

5. Conclusions (1) In spite of their different redox behaviour, two kinds of Mn304 (i.e. MnO_ 1.33 and MnO> 1.33) exhibit the same catalytic performance. This is due to the removal of the supra-stoichiometric oxygen during activation and/or reaction. (2) Milling could induce lattice contraction of Mn304, resulting in catalytic as well as redox deactivation. (3) Monolayer dispersed Mn304 shows very low activity for nitrosobenzene formation. This may result from its over-active redox behaviour. (4) Because of the weak interaction between Mn304 and alumina, the maximum benefit from using alumina as the support of Mn304 is to obtain a very small particle, the so called paracrystalline Mn304 with high surface area.

Acknowledgements Financial support from the Beijing National Laboratory for Structural Chemistry of Unstable and Stable Species is acknowledged.

W. Weimin et al. /Applied Catalysis A: General 133 (1995) 81-93

93

References [ 1] [2] [3] [4] [5 ] [6] [7] [8] [9] [ 10] [ 11 ] [ 12] [ 13] [ 14]

H.G. Zengel, Chem. Ing. Tech., 55 (1983) 962. D. Dodman, K.W. Pearson and J.M. Woolley, Brit. Patent Appl. 1 322 531 (1973). H.G. Zengel and M. Bergfeld, Ger. Often. 2 939 692 ( 1981 ). T.L.F. Favre, P.J. Seijsener, P.J. Kooyman, A. Maltha, A.P. Zuur and V. Ponec, Catal. Lett., 1 (1988) 457. T.L.F. Favre, A. Maltha, P.J. Kooyman, A.P. Zuur and V. Ponec, Proc. XI Symp. Iberoam. Catal. Guanajuato, Mexico, Vol. II, 1988, p. 807. P. Mars and D.W. van Krevelen, Chem. Eng. Sci., 3 (1954) 41. Yang Yongnian, Huang Ruili, Chen Lin and Zhang Jiayu, Appl. Catal. A, 101 (1993) 233. Wang Weimin, Yang Yongnian and Zhang Jiayu, Recl. Trav. Chim. Pays Bas, 114 (1995) 22. Wang Weimin, Yang Yongnian and Zhang Jiayu, Appl. Catal. A, 131 (1995) 189. G.C. Bond, J.P. Zurita, S. Flamerz, P.J. Gellins, H. Bosch, J.G. van Ommen and B.J. Kip, Appl. Catal., 22 (1986) 361. Liu Xingquan, Yang Yongnian and Zhang Jiayu, Appl. Catal., 71 ( 1991 ) 167. J.A. Barnard and R. Chayen, Modem methods of chemical analysis, McGraw-Hill, London, 1965, p. 63. A. Maltha, T.L.F. Favre, H.F. Kist, A.P. Zuur and V. Ponec, J. Catal., 149 (1994) 364. D. Jarosch, Min. Petrol., 37 (1984) 254.