Selective catalytic activity toward hydrofluorocarbon refrigerants in mixed oxides of manganese and copper

Selective catalytic activity toward hydrofluorocarbon refrigerants in mixed oxides of manganese and copper

Applied Catalysis B: Environmental 24 (2000) 107–120 Selective catalytic activity toward hydrofluorocarbon refrigerants in mixed oxides of manganese ...

548KB Sizes 0 Downloads 3 Views

Applied Catalysis B: Environmental 24 (2000) 107–120

Selective catalytic activity toward hydrofluorocarbon refrigerants in mixed oxides of manganese and copper Jane K. Rice ∗ , Jeffrey R. Wyatt, Louise Pasternack Chemistry Division, Code 6110, Naval Research Laboratory, Washington, DC 20375-5342 USA Received 16 March 1999; received in revised form 15 April 1999; accepted 29 July 1999

Abstract We report selective activity of an oxidative catalyst made of manganese and copper oxides toward two ozone-safe refrigerants, HFC-236fa (CF3 CH2 CF3 ), and HFC-134a (CF3 CFH2 ). These refrigerants are among those in a new class which contain hydrogen and fluorine substituents rather than fluorine and chlorine. This catalyst is generally non-specific and is used to oxidize carbon monoxide, hydrogen, and a variety of hydrocarbons and nitrogen compounds in heated burners, however, its activity toward refrigerants has been unpredictable. For our application, the catalyst is used in a heated burner in the closed-air environment on board submarines. In the process of optimizing the burner conditions for maximal activity toward airborne hydrocarbon contaminants and minimal activity toward refrigerants, we discovered enhanced and selective oxidative decomposition of these refrigerants in a group of recently-formulated catalyst lots which were manufactured with a higher percentage of ‘fines’. We attribute the increase in oxidative decomposition to an increase in adsorption of the refrigerants on the recently-formulated catalyst. We also observed some enhancement of activity toward CO on the recently-formulated catalyst. In addition to absorption isotherm measurements, we present several characterizations of the catalysts including scanning electron microscopy, elemental analysis, X-ray photoelectron spectroscopy, and temperature-programmed reduction. ©2000 Published by Elsevier Science B.V. All rights reserved. Keywords: Hopcalite; Manganese oxide; Copper oxide; Selective catalytic oxidation; Hydrofluorocarbons; Refrigerants

1. Introduction A mixture of the oxides of manganese and copper produce an oxidative catalyst which was first used to convert carbon monoxide to carbon dioxide [1–3]. The catalyst also has oxidative activity towards hydrogen, numerous alkanes, alkynes, aromatics, and nitrogen-containing organic compounds generally resulting in complete oxidative decomposition to car∗ Corresponding author. Tel.: +1-202-767-0721; fax: +1-202-404-8119 E-mail address: [email protected] (J.K. Rice)

bon dioxide and water [4–8]. It is a black or dark brown granule or pellet, amorphous in appearance, and has a high surface area. Its composition varies among manufacturers, ranging from amorphous CuMn2 O4 to amorphous mixtures of manganese and copper oxides with ratios of Mn : Cu between 2 : 1 and 5 : 1. Other compounds such as alumina are also found in some formulations. The operating temperature spans from room temperature to ∼500◦ C. The lifetime of the catalyst is short at room temperature due to moisture and is moderately long at temperatures above 150◦ C. It has been used as an air-purifying catalyst in military, mining, and space applications.

0926-3373/00/$ – see front matter ©2000 Published by Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 3 3 7 3 ( 9 9 ) 0 0 0 9 8 - 3

108

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

The objective of this research was to optimize the operating conditions of submarine-based catalytic burners to accommodate the use of new ozone-safe refrigerants. The burners help maintain a breathable air environment while the submarines are submerged for extended periods of time. The burner conditions have been adjusted to have maximal activity toward CO, hydrogen, and other airborne contaminants and minimal activity toward small quantities of refrigerants resulting from refrigeration leaks in the closed-air environment. The catalyst used in these applications is a formulation of Hopcalite 1 , a mixture of manganese oxide and copper oxide made by Mine Safety Appliances Company. The chlorofluorocarbons (CFC) currently used in numerous refrigeration and air conditioning applications deplete the ozone layer and their production is banned in developed countries by the Montreal Protocol. These include refrigerants CFC-12 (CF2 Cl2 ) and CFC-114 (CF2 ClCF2 Cl) which are currently used in refrigeration and air conditioning units on submarines. Refrigerants containing no chlorine are ozone-safe and are already being used to replace the CFC refrigerants. However, two of the most promising replacement refrigerants, HFC-134a (CF3 CFH2 ) and HFC-236fa (CF3 CH2 CF3 ), decompose in the presence of Hopcalite catalyst [9,10]. One oxidative decomposition product is HF, which is toxic and highly corrosive. To accommodate the increased oxidative decomposition of the new refrigerants, the catalyst temperature of the air purifiers on Navy submarines has been lowered to reduce the unwanted refrigerant decomposition products. At lower temperature, the catalytic activity towards carbon monoxide, hydrogen, alkanes, and aromatic compounds, which is the primary function of the catalytic air purifiers, is maintained at a high level while the activity toward the refrigerants is minimized [9,10]. For the case of carbon monoxide oxidation by Hopcalite, the following reaction scheme was updated by Vepˇrek et al. [11] and based on earlier work of Schwab and Kanungo [12] 3+ CO + Mn4+ → CO+ ads + Mn

(1)

1 Hopcalite is a registered trademark of Mine Safety Appliances Company, P.O Box 426, Pittsburgh, PA 15230. See Reg. No. 1,530,415 for reference.

1 2 O2

2+ + Cu1+ → O− ads + Cu

(2)

− CO+ ads + Oads → CO2

(3)

Cu2+ + Mn3+  Cu1+ + Mn4+

(4)

Steps (1) and (2) correspond to the adsorption of the reactants on the surface and Step (3) corresponds to the surface-enhanced chemical reaction. Step (4) is the redox reaction that returns the catalyst to an active state. The activity of mixed manganese and copper oxide catalysts towards carbon monoxide oxidation was reported to vary depending on the synthetic method of production [13,14]. The most amorphous materials had greatest activity; lower activity was correlated to more crystalline material and more segregation of copper and manganese oxides. Puckhaber et al. [13] also reported that the most active catalysts had more chemisorbed oxygen on the surface as compared to oxide material. However, there have been no studies on the mechanism of hydrocarbon or halogenated hydrocarbon oxidation on these catalysts. The oxidative decomposition of a variety of atmospheric contaminants found in the submarine environment have been tested in the presence of Hopcalite catalyst [4–8] at this laboratory. These include alkanes, alkenes, aromatic hydrocarbons, ammonia, and halogenated hydrocarbons (refrigerants). Although, no mechanistic experiments were performed, various correlations were noted. Christian and Johnson [4] found that lower molecular weight hydrocarbons were more resistant to oxidation. They also determined that most compounds produced complete combustion products; the exceptions were methyl chloroform and HCFC-11 (Freon-11, CFCl2 H) which produced other organic halides. Musick and Williams [5,6] examined the oxidative decomposition of 19 refrigerants and related compounds in the presence of Hopcalite catalyst. About half of the compounds decomposed in the presence of Hopcalite catalyst. They concluded that the stability of the halogenated hydrocarbons was enhanced by the absence of hydrogen, the presence of fluorine, and the absence of a double bond. They found no correlation between catalytic stability and thermal stability, bond dissociation energies, or heats of formation. In this paper we report the selective activity of the Hopcalite catalyst for oxidative decomposition of

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

HFC-134a and HFC-236fa replacement refrigerants. The activity of Hopcalite catalyst towards these hydrofluorocarbon refrigerants has been significantly affected by a change in the manufacturing process. The enhanced activity from a particular lot of catalyst was first observed in a full-scale burner by McCarrick, Jastrzebski, and Daley [15,16]. Our measurements indicate that variations in catalyst activity exist and that the most active Hopcalite catalyst lots have remarkably enhanced activity toward HFC-236fa, some enhancement toward CO, and no enhancement toward ethane and several other compounds [17]. In addition, we have performed numerous physical and chemical characterization measurements on ‘typically’ active and ‘superactive’ catalyst samples to try to determine the cause behind the increased adsorption and enhanced activity.

2. Methods Activity measurements were conducted in a plug flow reactor using the following procedure. An air stream at 50% relative humidity (RH) containing a test gas at concentrations of 10–100 ppm was flowed through a packed catalyst bed maintained at a preset temperature. The composition of the gas stream was monitored both before and after passing through the catalyst using Fourier transform infrared (FTIR) spectroscopy. A schematic view of the apparatus is shown in Fig. 1. Laboratory compressed air was purified by passing it through a dual tower molecular sieve based purifier (Puregas Heatless Dryer, Model PCDA1121-B032) removing water, carbon dioxide, and other contaminants. Removal of the atmospheric carbon dioxide was essential in order to measure accurately the amount of carbon dioxide produced by the oxidative decomposition of the test compounds. After purification, the air was humidified to a relative humidity of 50% by bubbling it through water. The purpose of the humidity was to mimic conditions in submarine atmospheres. Gaseous compounds were added to the humidified air using a dual dilution system. The system was constructed with Tylan mass flow controllers and calibrated using a Bios International DSC-ISC flow calibrator. The catalyst containers were constructed from 316 stainless steel pipe nipples with conflat ends. Each

109

catalyst bed contained catalyst to a depth of approximately 12 cm (4.7 in) and a diameter of 2.2 cm (0.87 in). The space velocity was designed to match that on submarines at 21 000 h−1 or 5.8 s−1 . For consistency, 42 g of 4 × 8 mesh catalyst were loaded in each catalyst bed. The catalyst beds were placed in an oven. The temperature was adjustable from room temperature to 500◦ C and measured to an accuracy of 1◦ C using a Type K thermocouple. The gas was preheated by passing through a 3/8 in i.d. × 18 in long stainless steel bellows coiled inside the oven prior to entering the catalyst beds. The effectiveness of the heat transfer to the gas was demonstrated by measuring a temperature change of less than 2◦ C in the catalyst bed when the gas flow was interrupted. The gas analysis was performed using FTIR spectroscopy with a Mattson Instruments Cygnus 100 spectrometer. Infrared analysis was selected because of its ability to detect a wide range of products. The infrared analyzer was equipped with a multi-pass White cell having a volume of 9.6 l and a path length of 21.4 m (Infrared Analysis Model 5-22PA). Spectra were collected by adding 128–256 scans acquired at a resolution of 4 cm−1 . Concentrations of the compounds of interest were calculated by integrating the area under the infrared absorption band. The amount of oxidative decomposition was calculated from the test gas absorption intensity by ratioing the area under an absorption band for the catalyst bed outlet and inlet samples. When possible, more than one band was used to check internal consistency. The dependence of integrated area on concentration for the C–H stretch of ethane near 3000 cm−1 was linear in the 1–100 ppm concentration range. For small molecules such as CO2 , the area under the absorption band was not linear with concentration. A calibration curve for integrated absorption intensity for CO2 using the area under the absorption band at 2349 cm−1 was well represented by a second-order equation of the form C = aA2 + bA + c

(5)

where A is the absorbance and C the concentration of CO2 . The coefficients a, b and c are specific to the White cell, resolution setting, and other spectrometer parameters and were 0.0128, 0.863, and −1.45, respectively. The equation is valid for concentrations greater than 3 ppm of CO2 . Whenever possible, the

110

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

Fig. 1. Schematic view of the laboratory scale plug flow burner.

Table 1 Percent oxidative decompositiona,b,c of test compounds (space velocity of 5.8/s) at 260 and 316◦ C Compound

CFC-12, CF2 Cl2 CF3 H CFC-114, CF2 ClCF2 Cl HFC-134a, CF3 CFH2 HFC-236fa, CF3 CH2 CF3 CO Ethane

260◦ C

316◦ C

Typical (%)

Superactive (%)

Ratio of k of superactive to typical lota

Typical (%)

Superactive (%)

Ratio of k of superactive to typical lota

<1 <1 <1 2.2 ± 1 1 ± 0.4 100 24 ± 5

<1 <1 <1 5±2 7±2 100 28 ± 5

– – – 2.3 7.3 – 1.2

<1 <1 <1 7.4 ± 1 8±2 100 52 ± 6

<1 2 2 14 ± 2 24 ± 3 100 57 ± 7

– >2 >2 2.0 3.3 – 1.1

a The percent oxidative decomposition is converted to a rate constant (s−1 ) by using the equation: k = ln (1 − A/100)/τ , where A is the b percent oxidative decomposition given above and τ b is the resident time in the flow reactor. The ratio of the rate constants is shown. b Values above 2% are determined by the averages of the disappearance of the reactant and the appearance of the CO product. Values 2 below 2% are determined from the appearance of the CO2 product alone. c The errors in the table represent the averaging of oxidative decomposition percentages over several years and include any variations which may have existed in Hopcalite samples. Errors in separate trials of the same material taken the same day are ∼ ± 1 for values in the 5–98% range and are smaller for values <5%, due to the use of CO2 production to measure those oxidative decompositions. We can detect <1 ppm of CO2 .

Table 2 Percent producta formation from oxidative decomposition of ethane and HFC-236fa at 316◦ C over typical and superactive catalyst Compounds/products

HFC-236fa, CF3 CH2 CF3 Ethane a

Typical

Superactive

CO2

CF3 H

CO2

CF3 H

∼7% 52%

∼0.5 –

21% 57%

∼3 –

The product HF is not accurately measured but does appear from the decomposition of HFC-236fa.

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

amount of oxidative decomposition of the test gas was determined by measuring the disappearance of the test compound (parent) and appearance of CO2 produced. The values given in Tables 1 and 2 represent averages of the percent parent decomposition and the percent appearance of CO2 . The two values are typically within a few percent of one another. The oxidation of carbon monoxide was measured at several temperatures and analyzed to determine the activation energy of the reaction for each catalyst. The analysis was carried out assuming first order heterogeneous decomposition kinetics. Low concentrations of the test gases, i.e., in the 50–100 ppm range, allowed us to assume a constant volume during the reactions. In this case, the plug flow integrated forms of the rate equations are the same as those of the batch reactor. Activation energies were calculated using the Arrhenius equation  k = Aexp

−Ea RT

 (6)

where the rate constant k, is defined as k=

−ln (1 − x) τb

(7)

x is the mole fraction of reacted parent calculated directly or calculated from the amount of CO2 product formed, and τ b is the residence time. Surface area measurements were made using a Quantachrome Monosorb instrument. Prior to measurement, some of the samples were pretreated by flowing oxygen at 260◦ C for 1/2 h to fully activate the catalyst. All samples were then dried and degassed with a flowing mixture of 30% nitrogen in helium at 170◦ C for 1/2 h. Although pretreatment with oxygen does affect the activity of the catalyst, it was not expected to affect the surface area and in fact, no differences were observed between pretreated and untreated samples. Scanning electron microscopy (SEM) was carried out using a Hitachi Instrument, Model S4000, at 27 kV. Intact, cleaved and crushed surfaces were examined to determine if long-term exposure to air caused observable differences. Samples were prepared in air, placed on conducting tape, and placed in the vacuum chamber for analysis.

111

HFC-236fa and ethane adsorption tests 2 were performed using a Micromeritics ChemiSorb 2800 instrument. Total adsorption and chemisorption pressure isotherms were measured and compared. The samples were heated to reduce the equilibration time for adsorption and it was necessary to heat the HFC-236fa sample to a higher temperature (100◦ C) than the ethane sample (50◦ C). Sample weights were between 0.5 and 2.0 g for all samples. The free space, low pressure dose, and equilibration interval were 16–17 cm3 , 0.2 or 0.5 cm3 /g STP, and 10 or 20 s, respectively. A blank tube was run with ethane gas exposure at 50 C. This resulted in ∼0.16 cm3 /g STP adsorption at ∼600 Torr. X-ray photoelectron spectroscopy (XPS) was carried out with a Fisons 220iSL instrument using monochromatic Al K␣ X-rays. All samples were oxidized and dried prior to placement in the X-ray photoelectron chamber by heating to 260◦ C while flowing oxygen over them for 1 h. X-ray fluorescence spectroscopy (XRF) was carried out on a Spectrace Model 9000 instrument. 5 g of each catalyst sample were sent to E and R Microanalytical Laboratory, Inc. in Corona, NY for elemental analysis by atomic absorption. The several pellets were ‘blended’ for each of two separate trials. Temperature-programmed reductions (TPR) were run on an Autochem 2910 Micromeritics instrument with a 50 ml/min flow rate of 10% helium in argon. The heating rate was 10◦ C/min. Sample preparation included three steps: a cycle of heating 10◦ C/min to 250◦ C for 1 h with flowing helium, a similar cycle of heating with flowing oxygen and an hour of purging with argon. The cycles were separated by 60 min. Masses were 0.0730 g and 0.0792 g for the typical (Lot 3) and the superactive (Lot 24) samples, respectively. The catalyst samples were obtained from the US Navy stock system and were manufactured by the Callery Chemical Division of Mine Safety Appliances Company. The manganese and copper content is presented in the results section of this paper. The ‘typical’ sample referred to in this report represents lots manufactured prior to November 1991 and the ‘superactive’ sample represents lots manufactured after April 1994. No catalyst was manufactured for the Navy between the above dates. Prior to November 1991, the catalyst 2

Tests were performed at Micromeritics, Norcross, GA.

112

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

was made by precipitation from soluble forms of the metals. First manganese sulfate, water, and sulfuric acid were mixed. Potassium permanganate was added forming manganese dioxide as a precipitate. The resulting sludge was rinsed to lower the pH. Copper sulfate and sodium carbonate were added resulting in copper carbonate. Carbon dioxide gas was released leaving copper oxide as a precipitate. This mixture containing both metal oxides was rinsed, vacuum settled, and oven dried. This material was mixed in a 2 : 1 ratio with ‘fines’ from previous batches. The material was pressed, milled and screened to the proper size and then activated by a final heating step. Production of the catalyst by this method was stopped in November of 1991. The production was resumed in 1994 using ‘fines’ from previous manufacturing cycles. This material was re-pressed, sized, and activated and met all specifications for Navy Hopcalite catalyst. In this report, Navy Lots 3 and 6 (without LiOH) and Callery Lot 218 were used to represent the ‘typical’ samples and Navy Lots 24 and 25 (without LiOH) and Callery lot 237 were used to represent the ‘superactive’ samples.

3. Results 3.1. Performance of the catalyst 3.1.1. Activity tests The oxidative decompositions of several refrigerants, ethane, and CO in the presence of typical and superactive catalyst at 260 and 316◦ C are presented in Table 1. A wider variety of test gases which reflect categories of contaminants present in submarine atmospheres, including alkanes, alkenes, alcohols, aromatics, ammonia, personal care products, and chlorinated solvents is presented in a separate report [17]. The report also includes results from a variety of mixed manganese and copper oxide catalysts from several manufacturers. The ratios of the rate constants of oxidative decomposition in the superactive lot compared to the typical lot are also presented in Table 1. The HFC-134a and HFC-236fa refrigerants show significantly more oxidative decomposition in the presence of superactive catalyst. The other refrigerants may show enhance-

ment but the oxidative decomposition values are small enough that ‘selectivity’ has little meaning. Ethane shows no enhancement. CO exhibits an enhancement of ∼1.5. Preliminary data [17] from compounds which do not completely decompose at 260◦ C, such as hexane, C2 Cl4 and C2 HCl3 , have enhancements below 1.4. Other compounds were tested at temperatures too high to determine whether enhancements exist [17]. 3.1.2. Oxidative decomposition products of HFC-236fa For most compounds studied, the only products observed were H2 O and CO2 . The amount of CO2 produced corresponds to the amount of parent decomposed. However, the primary products from the oxidative decomposition of HFC-236fa (CF3 CH2 CF3 ) at 316◦ C are CO2 , H2 O, and HF, and a small amount of CF3 H. The percentages of products formed are listed in Table 2 and an IR spectra of the oxidative decomposition products are shown in Fig. 2. HF is not typically observed in the product spectrum since copper wool is placed before the White cell to prevent HF corrosion of the mirrors. The only other compound we have tested which produces a carbon product (other than CO2 ) is methyl chloroform. At 260◦ C, we observe vinylidene chloride as a minor product from both the typical and superactive catalysts. 3.1.3. Carbon monoxide oxidation under low temperature conditions As the oxidation of carbon monoxide on the catalyst is best understood, oxidation of carbon monoxide at several temperatures between 40 and 110◦ C were carried out to see if enhancement in the presence of superactive catalyst could be detected. The oxidation and rate data are shown in Table 3. The superactive catalyst exhibits an enhancement of about 1.5 when temperatures below 85◦ C are averaged. 3.1.4. Activation energies Temperature-dependent measurements to determine the activation energies for the typical and the superactive catalysts were carried out with carbon monoxide and HFC-236fa. Representative plots are given in Fig. 3. The activation energies for the oxidation of ethane and HFC-134a in the presence of the typical catalyst were also measured. The range of temperatures exam-

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

113

Fig. 2. FTIR spectra of (a) 100 ppm of HFC-236fa before oxidative decomposition; (b) 100 ppm of HFC-236fa after oxidative decomposition over the superactive catalyst at 316◦ C and (c) a spectrum of 2.9 ppm of CF3 H shown as a reference. The CF3 H is a partial oxidative decomposition product. The bands between 750 and 1500 cm−1 are HFC-236fa; the bands between 1500 and 2000 cm−1 are water vapor. Spectra b and c are offset for comparison only. Table 3 Percent oxidationa of CO (space velocity 5.8/s) over typical and superactive catalysts Temperature (◦ C)

Typical

Superactive

Ratio of ka

40.2 50.8 60.5 69.4 83.6 98.3 107.3

3.0 16.5 24 45.5 66.5 87 91

4.5 20.5 36 61.5 81.5 93.5 94

1.5 1.3 1.6 1.6 1.5 1.3 1.2

The percent oxidation is converted to a rate constant (s−1 ) by using the equation: k = ln (1 − A/100)/τ b , where A is the percent oxidation given above and τ b is the residence time in the reactor bed. a

ined for carbon monoxide is only about 70◦ C, due to its complete oxidation at a low temperature. The results are given in Table 4. The data collected at 40◦ C are not included in the calculation. Although we observed higher oxidation percentages from the superactive catalyst than the typical catalyst, the activation energies were similar. This suggests the reaction mechanism is the same on the two catalysts and the increased rate of reaction on the superactive catalyst is due to more surface area or more reactive sites. The activation energies

Fig. 3. Arrhenius plots of HFC-236fa over typical (䊏) and superactive (䊉) catalyst. Rate information is based on CO2 product formation. Activation energies determined from the slopes are Ea = 21 and 22 kcal/mol for typical and superactive catalysts, respectively.

do not correlate with the percent decomposition of the test gases indicating that the reaction barrier height is not a predictive factor. This can be seen by comparing the oxidative decomposition of HFC-134a to ethane. They have reasonably similar activation energies and quite different oxidative decomposition efficiencies.

114

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

Table 4 Activation energies of CO, ethane and HFC-236fa (space velocity 5.8/s) over typical and superactive catalysts (kcal/mol deg) Chemical compound

Typical

Superactive

COa

13 ± 1 12 ± 1b 14 ± 3d 21 ± 1

13 ± 1 [13]c [17]c 22 ± 2

Ethane HFC-134a HFC-236fae a

Limited temperature range; Data from five temperatures between 40 and 112◦ C. b Ethane temperature range between 260 and 350◦ C. c Estimate from two temperatures, 260 and 315◦ C. d Estimate from three temperatures, between 250 and 315◦ C. e HFC-236fa temperature range between 260 and 380◦ C. Table 5 Single-point BET surface areas (m2 /g) (±10%) of catalyst samples Typical

Superactive

Ratio of surface area

133

153

1.15

3.2. Diagnostic measurements on the catalysts Due to the large variation in performance between typical and superactive catalysts, we carried out several tests to understand what could account for enhanced activity. Our tests fall into several categories. Some were designed to physically characterize the samples, i.e., surface area measurements and scanning electron microscopy (SEM). Others were designed to look at the adsorption properties. This included total adsorption HFC-236fa to measure the chemical binding of the refrigerant to the surface of the catalyst. We also carried out tests to determine if chemical differences between the catalysts could be measured. These included elemental analysis, X-ray photoelectron spectroscopy, X-ray fluorescence spectroscopy, and temperature-programmed reduction studies. 3.2.1. Surface area measurements Single-point BET measurements were made to determine the surface areas of the two catalysts. Results are shown in Table 5. We observe a small difference in the total surface area of the two catalysts but it is not large enough to explain the difference in activity. 3.2.2. SEM micrographs Typical and superactive catalyst samples were investigated by SEM for microscopic differences and the

results are shown in Fig. 4. Intact, cleaved and crushed samples prepared in air and subsequently placed under high vacuum were examined. No correlations of the surface appearance versus the surface preparation (intact, cleaved or crushed) were observed. The particles in the typical sample range from 0.2–1 ␮m and ∼10% have high contrast, indicating poor conductivity. Other micrographs of the typical catalyst (not shown) indicate irregular, less tightly-packed regions with a larger number of high contrast particles, similar to the superactive catalyst shown in Fig. 4, B2. The superactive catalyst has a more open structure with round, amorphous particles of 0.2–5 ␮m diameter. The typical sample contains smooth, tightly packed regions that are not observed in the superactive catalyst. However, both catalyst samples have open, airy surfaces. The particle sizes appear larger in the superactive catalyst micrographs shown, but the measured surface areas are very similar indicating that this is not the case in the samples as a whole. The surfaces of both catalysts are complex and we find no interpretable link between the morphology of the surfaces and catalyst activity.

3.2.3. Adsorption studies of HFC-236fa and ethane The adsorption isotherms of HFC-236fa and ethane were measured on the two catalysts samples. The results which represent the total adsorption are shown in Fig. 5. There is a dramatic difference in the absorption isotherms of HFC-236fa and ethane. Both catalysts adsorb similar amounts of ethane at 50◦ C, however, a factor of three difference is seen in the adsorption of HFC-236fa by the two catalysts. The HFC-236fa measurement was carried out at 100◦ C to reduce the equilibration time of adsorption on the surface, which was considerably higher than ethane. The repeat isotherm was found to be nearly the same as the first isotherm, indicating that nearly all the molecules were removed by vacuum, leaving a near zero ‘first minus repeat’ component, typically referred to as the chemisorption component. The ‘first minus repeat’ component for HFC-236fa over the superactive catalyst was ∼0.2 cm3 (per 0.568 g of sample) out of ∼18 m3 at 600 Torr, which is at the threshhold of detection for the experiment. This was the case for both ethane and HFC-236fa. Total adsorption of ethane in an empty tube yielded about ∼0.16 cm3 at 600 Torr.

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

115

Fig. 4. SEM micrographs taken at 25 kV. The typical catalyst cleaved at a magnification of 300×, A1; the typical catalyst cleaved at 6000×, A2; the superactive catalyst crushed at 300×, B1; and the superactive catalyst crushed at 6000×, B2.

Table 6 y-Intercept volumes adsorbed (top) in cm3 /g STP, and slopes (bottom) from total adsorption isotherms of HFC-236fa and ethane over typical and superactive catalysts Typical

Superactive

5.8 ± 0.1 0.0014 ± 0.0001 2.6 ± 0.1 0.0050 ± 0.0001

14.1 ± 0.1 0.0070 ± 0.0002 2.8 ± 0.1 0.0052 ± 0.0001

3.2.4. Chemical characterization We examined the bulk and surface composition to see if there were notable differences in the catalysts which would explain the enhanced adsorption and increased activity of HFC-236fa. We also investigated the oxidation states of the surface species and the relative number of binding sites present in the catalyst samples.

On the other hand, the absorption isotherms appear to have a sizable chemisorption component based on their shape. The y-intercept volumes adsorbed and slopes for the four isotherm curves are given in Table 6. The chemisorbed state may have a low binding energy thus accounting for a small ‘first minus repeat’ result. The enhanced adsorption of the HFC-236fa on the superactive catalyst does appear to explain its enhanced activity.

3.2.4.1. Elemental analysis. The two catalysts samples were analyzed using atomic absorption, and the results are reported in Table 7. They have nearly identical ratios of manganese to copper. If we assume the metals oxides are MnO2 and CuO, ∼90% of the total weight is identified: 74% MnO2 , 14% CuO, and 1.6% minor components. The oxygen content, based on the assumed stoichiometry and percent metal content of manganese and copper given in Table 7, accounts for

HFC-236fa at

100◦ C

Ethane at 50◦ C

116

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

Fig. 6. XPS large area scans of (a) typical catalyst and (b) superactive catalyst, offset for comparison only.

Fig. 5. Total adsorption isotherms for HFC-236fa at 100◦ C for typical catalyst 䊏 and superactive catalyst 䊉 ; and for ethane at 50◦ C for typical catalyst H, and for superactive catalyst N. Table 7 Elemental analysis (% by weight) of catalyst samples Elements

Typical

Superactive

Mna Cua Al K Na

46.9 11.0 2.3 2.3 0.5

45.7 10.5 0.5 2.6 0.5

a

The percent oxygen was not detected by atomic absorption, however, calculations yield ∼27% oxygen from MnO2 and ∼3% oxygen from CuO,for a total of ∼30% (assuming those oxide species). See text for more details.

∼30% of the total weight. There remains ∼10% undetermined, which we assume is water. There are small differences between the typical and superactive catalyst in the alumina content, however, we know from the study of other manufacturers’ catalyst that activity is not strongly correlated with the percentage of alumina [17]. Potassium [11,14] and sodium [14] were measured because they have been reported to poison

these catalysts by blocking reactant adsorption sites. Although the typical catalyst contains more potassium than the superactive catalyst, this correlation is not present when samples from other manufacturers are compared [17].

3.2.4.2. X-ray photoelectron spectroscopy and X-ray fluorescence spectroscopy. The samples were investigated using X-ray photoelectron spectroscopy (XPS) and energy dispersive X-ray fluorescence spectroscopy (XRF) to determine if any other contaminants were present on or near the surface of the catalysts and to determine if the manganese and copper were in the same oxidation state in the two samples. Samples from both typical and superactive lots were tested and found to have similar spectra, as shown in Fig. 6. The ratio of Mn : Cu was found to be slightly larger in the superactive catalyst (measuring a region of size 0.25 mm × 1 mm, 20–30 Å deep). This was confirmed by XRF measurements, which measured the Mn : Cu ratio to be 20% larger in the superactive catalyst over the typical catalyst (measuring a region of size 1 ␮m × 1 ␮m, 1.5 ␮m deep). However, the Mn : Cu ratio measured by elemental analysis, which measures the entire sample, was the same in both typical and superactive catalyst. It is not clear whether these dif-

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

ferences reflect radial gradients, uneven surface distributions within the pellets, or uneven distributions of copper and manganese among different pellets. In any case, it is difficult to believe that these slight differences in the Mn : Cu ratio significantly impact the oxidative decomposition rates since there are much larger differences in composition in catalysts from other manufacturers that have little effect on the oxidative decomposition rates, as reported by Pasternack et al. [17].

117

reduction of either CuO or Mn2 O3 . The most apparent difference is seen in the temperature of Peak 4. It is ∼20◦ C lower in the superactive sample. There are also more reducible sites observed in Peak 1. Further studies are needed to determine if these differences could explain the enhanced adsorption and increased activity of HFC-236fa on the superactive catalyst.

4. Discussion 4.1. Relative activities of refrigerants

3.2.4.3. Temperature-programmed reduction. The temperature-programmed reduction (TPR) technique was applied to determine if the catalyst samples contained different oxidation states. Temperatureprogrammed reduction of pure copper oxide (CuO) has been reported as a single peak at 380, 333, or 227◦ C depending on morphology and TPR experimental conditions [19–21]. Temperature-programmed reduction of manganese oxide is more complicated due to the number of oxidation states sampled from Mn4+ to Mn2+ [22]. Kapteijn et al. report a mechanism in which MnO2 may reduce through Mn2 O3 or directly to Mn3 O4 . This is followed by the reduction of Mn3 O4 to MnO [22]. Our TPR results are shown in Table 8 and Fig. 7. The samples have nearly identical total uptake of hydrogen at 200 ml/g. This agrees with our calculated value of 235 ml/g, assuming (1) the starting species are MnO2 and CuO, (2) reduction stops at MnO, (3) there is a 10% loss of water in the sample (based on experience) and 4) the atomic ratio of Mn : Cu is 80 : 20. We observe four clearly defined peaks in our TPR experiments which are presented in Table 6. Peaks 2, 3 and 4 have integrated areas of 5 : 1 : 3 for the typical sample and 3 : 1 : 2 for the superactive catalyst. A reduction mechanism from MnO2 to Mn2 O3 to Mn3 O4 to MnO would generate peaks in the ratio of 3 : 1 : 2 and a mechanism leaving out the Mn2 O3 intermediate would generate peaks in the ratio of 4 : 2. Reduction of CuO to Cu would have a hydrogen uptake of 20% that of Mn in our samples for a value of 1.2 relative to Mn. The peak at 200◦ C can be tentatively assigned to the reduction of MnO2 and the peak at 314◦ C (or 292◦ C in the superactive sample) can be tentatively assigned to Mn3 O4 . The sharp band is probably the

If we compare the activities of Hopcalite toward the refrigerants in Musick and Williams [5,6] with the two new refrigerants presented here, we confirm and extend their observed trends. None of the inert refrigerants contain hydrogen; they are all chlorofluorocarbons, e.g., CCl2 F2 , CClF2 CClF2 , CClF2 CF2 CClF2 , and CF3 CClFCClF2 or fluorocarbons, e.g., CF4 and cyclic C4 F8 . The refrigerants in Musick and Williams [5,6] which decompose extensively at 360◦ C, include but are not restricted to hydrogen-containing compounds, e.g., CHCl2 F (37%), CHClF2 (41%), and CF3 CH2 Cl (47%). The others are CCl3 F (37%), and CClF2 CCl2 F (5%), with the latter being the least active and a bridge between the two categories. We are adding CF3 CH2 CF3 and CFH2 CF3 to the list of decomposing refrigerants, which are both hydrogen-containing. If we scale our oxidative decomposition percentages to the temperature used by Musick and Williams [5,6], we have values of ∼5% for CF3 CH2 CF3 and ∼6% for CFH2 CF3 at 305◦ C from our typical catalyst. As the detection methods are different, we could not be confident of a direct comparison without having tested a compound on the Musick and Williams [5,6] list. We have measurements of tetrachloroethene and trichloroethene [9] at 260 and 316◦ C from our typical catalyst sample [17]. Our extrapolations of the oxidative decompositions for the two compounds at 305◦ C are 66 and 72%, respectively. Musick and Williams [5,6] report 73 and 94%, respectively. Their values are 1.1–1.3 times our values. With or without the scaling, the HFC-236fa and HFC-134a oxidative decomposition percentages appear to be lower than the hydrochlorofluorocarbons tested by Musick and Williams, but higher than the bridging molecule, CClF2 CCl2 F (1.5% at 305◦ C),

118

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

Table 8 Temperature-programmed reduction results over typical and superactive catalysts Peak no.

1 2 3 4

Typical catalyst

Superactive catalyst

Temperature at maximum (◦ C)

Volume of hydrogen (ml/g) STP

Temperature at maximum (◦ C)

Volume of hydrogen (ml/g) STP

130 205 244 314

8.0 103 23 67

108 200 238 292

12 88 35 65

Fig. 7. Temperature-programmed reduction plots of catalysts typical (- - -) and superactive (—).

which contains no hydrogen. It appears that molecules which can release HF, HCl, or HBr are oxidatively decomposed more readily. This may be a result of dehydrohalogenation from adjacent carbons. It also appears that the release of HCl leads to a higher oxidative decomposition than the release of HF. 4.2. Enhanced activity from the superactive catalyst Oxidative decomposition of gases in the presence of Hopcalite catalyst occurs by an undetermined mechanism. However, it can be described in four steps: (1) The reactant is adsorbed on the surface, (2) a reaction occurs under conditions of a reduced activation barrier, (3) the products are desorbed and (4) the oxygen balance within the catalyst is replenished from the catalyst or from O2 in the gas phase. The higher adsorp-

tion properties of HFC-236fa compared to ethane correlate well with the enhanced activity observed with HFC-236fa. The two and one-half fold increase in adsorption is similar to the threefold increase in observed activity. Therefore, it appears that the observed increase in adsorption is the key to understanding the different catalytic activities of the two Hopcalite samples. The absorption isotherms have a shape consistent with a physisorbed and chemisorbed component. Extrapolation of the adsorption isotherms yields a y-intercept values which correlate well with the increased activity seen for HFC-236fa in the presence of typical and superactive catalyst. We also conclude that this measurement should be able to discriminate between typical and superactive catalyst for screening purposes. The adsorption isotherms indicate that HFC-236fa is more readily adsorbed than ethane. The absorption

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

curve for HFC-236fa on the typical catalyst reaches saturation at a lower pressure than the HFC-236fa on the superactive catalyst. This suggests the molecules are bound more strongly to the typical catalyst. The adsorption curves of HFC-236fa also indicate that there are about 2.5 times more sites selective for HFC-236fa on the superactive catalyst. The activation energy data suggests that the reaction mechanism is similar on the two catalysts, but more sites are present on the superactive catalyst. Taken together, it appears that there are more sites on the superactive catalyst which are selective toward HFC-236fa but are not selective toward ethane, and sites on the superactive catalyst appear to bind molecules less strongly than sites on the typical catalyst. We have applied several characterization measurements to determine a physical property leading to the increased adsorption, and hence the increased activity, and have not been able to assign a specific characteristic, although we note several differences. There are three points of interest for future studies. A study of the adsorption properties of several chemically similar refrigerants would lead to a better understanding of the correlation between adsorption and increased activity. It also may be possible to model or calculate the attraction between the refrigerant and the catalyst surface using a simplified model and provide an explanation of this effect. Finally, further examination of the reducible sites using TPR may provide information on the selective nature of the HFC-236fa reaction on the superactive catalyst.

5. Conclusions We observe selective oxidative decomposition of two hydrofluorocarbon refrigerants in the presence of Hopcalite catalyst. The oxidative decomposition is significantly enhanced in several recently-formulated lots. This is of concern for any application involving a heated burner, because these refrigerants are the likely replacements for the most widely used refrigerants on the market and their byproducts are hazardous and corrosive. Carbon monoxide oxidation is examined for comparison and a small enhancement is seen in the presence of ‘superactive’ catalyst. A clear correlation is observed between adsorption of HFC-236fa and increased catalytic activity. Although, we have carried

119

out extensive diagnostic tests and have revealed some differences, we are unable to determine what specific physical or chemical property of the superactive catalyst is responsible for the increased absorption and increased oxidative decompostion. However, the adsorption measurement can serve as a screening test for future catalyst lots.

Acknowledgements We thank Mr. Eric Anderson, Mr. John Milius, and other members of the staff at Mine Safety Appliances Company for discussions on the manufacture of Hopcalite catalyst. We thank Dr. Noel Turner of NRL for assistance with the XPS data and Dr. David Smith of Micromeritics of Norcross, GA for assistance with the adsorption and temperature-programmed reduction studies. We thank Dr. Jim Horwitz of NRL for assistance with the SEM measurements. We thank Dr. Michele Anderson of NRL for discussions on the chemisorption method and reading the manuscript. We thank Prof David Cocke of Lamar University and Dr. Debra Rolison of NRL for helpful discussions. Funding of this project was provided by the Naval Sea Systems Command.

References [1] A.B. Lamb, W.C. Bray, J.C.W. Frazer, J. Ind. Eng. Chem. 12 (1920) 213. [2] T.H. Rogers, C.S. Piggot, W.H. Buhlke, J.M. Jennings, J. Am. Chem. Soc. 43 (1921) 1973. [3] D.R. Merrill, C.C. Scalione, J. Am. Chem. Soc. 43 (1921) 1982. [4] J.G. Christian, J.E. Johnson, Int. J. Air Water Poll. 9 (1965) 1. [5] J. K Musick, F.W. Williams, Ind. Eng. Chem. Prod. Res. Dev. 13 (1974) 175. [6] J.K Musick, F.W. Williams, Am. Soc. Mech. Eng. ASME Tech. Report 75-ENAs-17. [7] J.K. Musick, F.S. Thomas, J.E. Johnson, Ind. Eng. Chem. Prod. Res. Dev. 11 (1972) 350. [8] R.W. McCabe, P.J. Mitchell, Ind. Eng. Chem. Prod. Res. Dev. 23 (1984) 196. [9] L. Pasternack, J.K. Rice, J.R. Wyatt, C.H. Douglass, Project report on the effects of alternative refrigerants on submarine atmosphere control: laboratory scale burner studies, NRL/MR/6110-94-7470, Naval Research Laboratory, Washington, DC, 31 March 1994.

120

J.K. Rice et al. / Applied Catalysis B: Environmental 24 (2000) 107–120

[10] J.K. Rice, L. Pasternack, J.R. Wyatt, C.H. Douglass, Project report on the effects of alternative refrigerants on submarine atmosphere control: laboratory scale burner studies – II, NRL/MR/6110-95-7761, Naval Research Laboratory, Washington, DC, 10 August, 1995. [11] S. Vepˇrek, D.L. Cocke, S. Kehl, H.R. Oswald, J. Catal. 100 (1986) 250 . [12] G.M. Schwab, S.B. Kanungo, Z. Phys. Chem. N.F. 107 (1977) 109. [13] L.S. Puckhaber, H. Cheung, D.L. Cocke, A. Clearfield, Solid State Ionics 32/33 (1989) 206–213. [14] L.S. Puckhaber, Bulk, surface and activity characteristics of pure and doped thermally-modified hopcalites, Dissertation, 1994. [15] A. McCarrick, Decomposition of HFC-236fa in the Mk V CO/H2 Burner, NAVSSES Final Report PM-9515 Ser 9212/142, 13 January 1997.

[16] A. McCarrick, Trifluoromethane Background from HFC236fa Decomposition, NAVSSES Final Report PM-9515 Ser 9212/052, 24 June 1997. [17] L. Pasternack, J.K. Rice, J.R. Wyatt, C.H. Douglass, Performance variation in hopcalite catalysts, NRL/MR/611097-8115, Naval Research Laboratory, Washington, DC, 31 December 1997. [18] B.L. Yang, S.F. Chan, W.S. Chang, Y.Z. Chen, J. Catal. 130 (1991) 52061. [19] M.F. Luo, Y.J. Zhong, X.X. Yuan, X.M. Zheng, Appl. Catal. A: General 162 (1997) 121–131. [20] S.J. Gentry, N.W. Hurst, A. Jones, J. Chem. Soc., Faraday Trans. 1 77 (1981) 603–619. [21] G. Fierro, M. LoJacono, M. Inversi, P. Porta, R. Lavecchia, F. Cioci, J. Catal. 148 (1994) 709–721. [22] F. Kapteijn, L. Singoredjo, A. Andreini, Appl. Catal. B: Environmental 3 (1994) 173–189.