Heterogeneous reactions of HNO3 with flame soot

PCCP

Heterogeneous reactions of HNO3 with flame soot generated under different combustion conditions. Reaction mechanism and kinetics M. S. Salgado Munõzy and M. J. Rossi* Laboratory of Air and Soil Polution (LPAS), Swiss Federal Institute of Technology (EPFL), CH-1015, Lausanne, Switzerland. E-mail: michel.rossi@epfl.ch Received 22nd April 2002, Accepted 7th June 2002 First published as an Advance Article on the web 18th September 2002

The reaction of HNO3 with decane (C10H22) soot generated in the laboratory has been studied in a Knudsen flow reactor. Two different types of soot were produced: Soot originating from a rich flame of decane and air (‘‘ grey ’’ soot) and soot generated from a leaner flame of decane and air (‘‘ black ’’ soot). Both HNO3 uptake and product release have been observed. Whereas soot from a rich decane flame leads to HONO as the main product, NO and small amounts of NO2 are observed from soot generated in a lean flame. A reaction mechanism is proposed in which HNO3 is reduced to HONO, which is released into the gas phase on grey soot, but undergoes decomposition reactions on black soot to NO and NO2, the latter of which also reacts on black soot to produce additional NO. Experiments on black soot show that at [HNO3] > 1011 molecule cm3 a higher order process is induced which is associated with an additional production of NO and NO2 . Further studies of the reaction of gaseous NO with adsorbed HNO3 reveal a slow reaction leading to HONO and NO2 . Both initial uptake (g0) of HNO3 as well as steady state uptake probabilities (gss) have been obtained: g0 ¼ (4.6 2.3) 103 and gss ¼ (5.2 1.3) 104 for reaction on grey soot and g0 ¼ (2.0 0.1) 102 and gss ¼ (4.6 1.6) 103 for black soot, using the geometric surface area over which the samples are spread.

1. Introduction Soot particles are formed by incomplete combustion of carbonaceous fuels. They are interesting principally because they are a chemically reducing substrate in an otherwise oxidizing atmosphere. The quantity of soot suspended in the atmosphere is too small to make a difference to global ozone; however, it may have effects on a local or regional scale.1 In the lower troposphere, the principal sources of soot are fossil fuel and biomass burning,2,3 whereas in the upper troposphere and lower stratosphere, the primary source of soot is jet aircraft.4 In addition to its toxic properties as far as public health is concerned, the ability of soot to absorb and scatter solar radiation has led to its inclusion in calculations of the earth’s radiation budget.5,6 Soot may also serve as cloud condensation nuclei and therefore indirectly affect climatic forcing.6,7 The oxidation of soot by HNO3 , NO2 and/or O3 has been proposed as a mechanism which may deplete stratospheric ozone in the absence of sunlight at all latitudes.8–11 Recently, it has been reported that the ratio [HNO3]/[NOx] in photochemical models, excluding soot reactions was overestimated by a factor of 5–10 in comparison to measurements11,12 performed in the free troposphere. Recent modeling calculations, including surface areas for heterogeneous reactions representative of soot, resulted in ratios closer to those measured in the free troposphere. The reduction of HNO3 to NO on soot has been suggested as a mechanism to bring calculated [NOx]/[NOy] values closer to those measured in the troposphere.12–14 Since the distribution of nitrogen oxides (NOx) is closely related to the photochemical O3 balance, a possible renoxification of HNO3 due to heterogeneous reactions on soot aerosols will have a direct impact on the pathways of y Present address: Departmento de Quı´mica Fı´sica, Facultad de Ciencias, Universidad de Castilla La Mancha, Avda. Camilo Jose´ Cela, 10. 13071 Ciudad Real, Espanã.

5110

ozone production and destruction. In this sense, recent laboratory studies of the heterogeneous uptake of nitric acid HNO3 on oxide, carbonate and mineral dust particles,9 and the heterogeneous formation of nitrous acid HONO on soot particles15–17 are highlighted as two examples of potentially important heterogeneous processes in the troposphere. The reaction of gaseous NO with HNO3 on borosilicate glass in the presence of water vapor has been studied previously18 and resulted in the formation of NO2 . Saliba et al.18 proposed the reaction HNO3 + NO ! NO2 + HONO as a first step. They concluded that this chemistry may be potentially important in renoxification of the boundary layer of polluted urban atmospheres. There is considerable uncertainty regarding the role of soot in the atmosphere. This uncertainty is mostly due to the lack of laboratory data on the reactivity of atmospherically relevant species on soot and the lack of knowledge on the reactive surface area and chemical composition of the particles in the atmosphere as well as in the laboratory. The reactivity of soot and its potential catalytic activity are thought to depend on the nature of the functional chemical groups present on the surface of the particles.19,20 HNO3 is the prototype of a sticky molecule having a significant residence time on all surfaces; this makes the diffusion of HNO3 from an external to an internal surface much slower than predicted by Knudsen diffusion within a pore. The heterogeneous reaction of HNO3 on soot has been investigated by Thlibi and Petit21 in a static reactor at 303 K; they reported the decomposition of HNO3 into NO and NO2 . Laboratory investigations of the reactions of HNO3 on soot have also been performed using Knudsen flow reactors loaded with commercially available amorphous carbon substrates (Degussa FW 2) which yield NO, NO2 and water vapor as products. A reactive uptake coefficient g of approximately 4 102 has been measured22 using the projection of the surface area onto the sample support, the so-called geometric surface area. A recent study23

Phys. Chem. Chem. Phys., 2002, 4, 5110–5118 This journal is # The Owner Societies 2002

DOI: 10.1039/b203912p

has examined the uptake of HNO3 on both Degussa FW 2 and soot from hydrocarbon combustion obtaining g values of 0.023, 0.060 and 0.067 using the geometric surface area of the different types of soot studied. On hydrocarbon soot no product formation has been observed and it was23 concluded that no chemical reaction was taking place. In other experiments the initial uptake coefficient for physical adsorption of HNO3 averaged g0 ¼ 1.5103 using the geometric surface area. However, no products resulting from that interaction were found.24 The heterogeneous loss of HNO3 was also investigated in a large aerosol chamber for interaction times of up to several days25 obtaining reaction probabilities of g < 3 107 using the accessible surface area of the fractal particles. In this case no significant HNO3 depletion other than wall loss or significant NO2 formation was detected in the presence of soot aerosol. FTIR studies of HNO3 interacting with carbonaceous samples from a spark generator have also been performed. IR absorption bands attributable to the products of soot oxidation and nitrogen containing species have been obtained and estimated uptake coefficients ranging from 2 103 to 2 106 have resulted,19 using the accessible surface area of 200 m2 g1; this is up to a factor of two smaller than the BET surface area. Long-pathlength infrared absorption spectroscopy was used by Disselkamp et al.,26 who observed HNO3 uptake for three different kinds of soot, and who obtained a lower limit for g of 1.4 106 using BET specific surface areas for the calculation of g. These and other studies thus reveal the uncertainty in the results of HNO3–soot reactions related to the detailed nature of the soot sample and the experimental conditions. Part of the uncertainty in the results stems from the ambiguity of whether to use the geometric or the internal surface area as measured for instance by the BET technique. However, the present case is clear-cut in that HNO3 is a very sticky molecule whose Knudsen diffusion into the bulk of the sample is exceedingly slow. We have therefore used the geometric surface area in interpreting the uptake rate data. It has been shown that commercially available carbonaceous substrates are not necessarily good surrogates for soot found in the atmosphere, and also that different types of soot react differently.27 The control of combustion conditions in soot generation taking place in the laboratory is necessary to obtain reproducible results and also because the substrate reactivity is affected by the fuel/oxygen ratio during soot generation.27 Therefore, in this study, we have investigated soot generated under different and controlled combustion conditions for decane burning in air and have used gas phase HNO3 as a probe for the reactivity of soot.

2. Experimental Soot samples from the combustion of decane have been produced using a simple co-flow system described previously,27 so only a brief description is given here. It basically consists of a diffusion flame maintained in a stream of air, which is

controlled by a mass flowmeter. Ceramics of different porosity allowed us to regulate coarsely the fuel flow feeding the flame. Soot samples were collected from the burnt gases approximately 1 cm above the flame on glass plates of 19.6 cm2 surface area. Two different types of soot from burning decane in air at both extremes of combustion conditions were produced: one type generated in a lean flame (low fuel/oxygen) and the other in a rich flame (high fuel/oxygen). The flame temperatures in the leaner flame are considerably higher, as manifested by a yellow/white as opposed to an orange/red appearance from a rich flame. The leaner flame was obtained using ceramic cylinders with small pores (ø ¼ 4–10 mm) and air flows between 1.4 and 1.6 l min1, whereas a rich flame was maintained using ceramic cylinders with large pores (ø ¼ 17–40 mm) and air flows between 1.1 and 1.3 l min1. Soot produced in a rich flame has a greyish colour when observed with the naked eye, whereas soot from the leaner flame is pitch black. Therefore, these two different soot samples will be referred to as ‘‘ grey ’’ and ‘‘ black ’’ decane soot. The fuel/air ratio, which is a key parameter in combustion may only be estimated and qualified as ‘‘ high ’’ or ‘‘ low ’’ because of the uncertainties in the fuel and entrained air flow rates. Previous experiments using this diffusion flame demonstrate the reproducibility of the obtained results. In practice, it is necessary to make continuous fine adjustments of the air flow during soot collection in order to maintain a flame of constant visual appearance. The reactivity of these two different soot samples towards HNO3 was examined in a low pressure flow reactor (Knudsen cell) using modulated molecular beam mass spectrometry for the quantification of reactant uptake and product release into the gas phase. The design and operation of the Knudsen cell flow reactor have been described in detail28 and only a brief description is given here. The apparatus consists of a vacuum line to produce, store and mix gaseous reactants and from which molecules are introduced into the reactor at a measured flow rate. The gaseous species interact with the soot in the twochamber Knudsen reactor. The gas is continuously injected into the reactor via a glass capillary inlet. The residence time of the gas in the reactor is controlled by means of a variable size orifice through which molecules escape into the vacuum chamber. By changing the area of the orifice the molecular rate of effusion out of the reactor changes, along with the residence time (t), by two orders of magnitude. Molecules escape from the reactor forming an effusive (thermal) molecular beam and are detected by an electron impact quadrupole mass spectrometer QMS lodged in the lower part of a differentially pumped vacuum chamber. Details of the Knudsen cell are summarized in Table 1. Each sample was pumped for 5 min prior to an uptake experiment and subsequently isolated from the flow using an O-ring sealed movable plunger. Subsequently, HNO3 was introduced into the reactor as a continuous flow for steadystate experiments, which was measured using the pressure decrease with time in a calibrated volume. The mass of soot used in these experiments was varied between 3 and 20 mg spread out over 19.6 cm2 and [HNO3] was in the range

Table 1 Measured parameters of the Knudsen Cell used in this work Surface areab ¼ 1300 cm2

Volumea ¼ 2000 cm3 Orifice diameter/mm

kesc/s1

Residence time/s

kesc/s1(HNO3), 300 Kc

1 4 8 14

0.013 (T/M)0.5 0.245 (T/M)0.5 0.796 (T/M)0.5 1.880 (T/M)0.5

76.5 (M/T)0.5 4.04 (M/T )0.5 1.23 (M/T )0.5 0.528 (M/T )0.5

0.034 0.299 1.231 3.254

a

The total volume is increased by 1% upon opening the sample chamber. Therefore, no distinction concerning the volume is made between a reference and an uptake experiment. b Estimated total surface area of the Knudsen Cell. The sample surface area is 19.6 cm2.

Phys. Chem. Chem. Phys., 2002, 4, 5110–5118

5111

2.1.

Fig. 1 Typical uptake of HNO3 (monitored at m/z 46) and products released from soot generated under rich fuel conditions (grey soot). The product was HONO monitored at m/z 47. Neither NO (m/z 30, corrected) nor NO2 (m/z 46, corrected) are observed. [HNO3] ¼ 9.1 1012 molecule cm3, soot mass of 16 mg, 4 mm diameter escape orifice, interaction time of 5 min.

1.0 1012–9.1 1012 molecule cm3 for grey soot and 1.9 1011 – 5.9 1012 molecule cm3 for black soot. When the flow had stabilized, the isolation plunger was lifted and the reaction monitored. As can be seen in Fig. 1, a large initial rate of disappearance of HNO3 was observed at m/z 46 (corrected when necessary) that quickly saturated within a few minutes. At the same time, formation of HONO as a reaction product was observed. It proved possible to identify different reactivities for the two different types of soot. The fuel/oxygen ratio is therefore a key parameter influencing the reactivity of soot towards HNO3 . The products obtained depended on the conditions of soot generation, i.e. lean vs. rich combustion. Dry gaseous HNO3 was obtained from the vapor above a mixture (1 : 3) of HNO3/H2SO4 . NO2 (Carbagas SA), NO (Matheson Inc.) and HONO which was generated in situ by flowing gaseous HCl through a vessel coated with solid KNO2 (FLUKA puriss.p.a q98.8%) were used as calibration gases. The detection of HNO3 and its decomposition products was performed by monitoring the MS signals at m/z ¼ 30, 46, 47 and 63. Since both HNO3 and its decomposition products (NO and NO2) have common mass fragments, calibration of the MS signals and subtraction of the contributions of other gases to the fragment peak intensity of the gas analyzed is necessary to obtain the pure component desired. This correction is unnecessary when a gas is monitored at its molecular ion peak, for example at m/z 63 for HNO3 or m/z 47 for HONO. The soot samples were characterized using scanning electron microscopy (SEM), measurement of the BET internal surface area and elemental analysis. BET measurements were performed with 400 mg of soot which had been recycled from uptake experiments using a Sorptomatic 1990 (Fisons Instruments) with nitrogen as adsorbate. For elemental analysis a PE 2400 CHN analyser (Perkin Elmer) was used. Each sample of approximately 2 mg was prepared just prior to use.

Table 2

Characterization of soot

The presence of any products of the reaction of HNO3 with soot remaining adsorbed on the soot has not been detected. However, in other studies using FTIR absorption R–O–NO2 and R–NO2 surface functionalities on the soot surface have been identified.19 In addition to these bands, a decrease in the broad IR absorption at 3500 cm1 was observed, indicating the consumption of surface OH and C–H groups, respectively, in the range 3000–2800 cm1. The reaction mechanism proposed here is in agreement with these results. The decrease in the intensity in the OH stretching vibrations may be explained by the formation of hydrogen bonds with HNO3 , irreversibly adsorbed on the soot. Possible stable products of this interaction may be organic nitrates and nitro compounds which have been identified by IR absorption. On the other hand, the partial disappearance of the C–H groups may be related to the extent of the chemical reduction of HNO3 which effectively consumes hydrogen. Grey and black decane soot have previously been examined27 using elemental analysis and the measurement of the BET internal surface area in order to characterize the soot substrates. Table 2 presents a survey of these results. It may be seen that the BET surface area is higher by a factor of 3 for black than for grey decane soot. Elemental analysis is a bulk analysis method, but owing to the large surface/volume ratio of soot, this analysis also addresses the surface properties to a certain extent. The oxidative conditions in the leaner flame determine the results of the elemental analysis. The oxygen content for black soot is a factor of 1.9 higher than for grey soot, with the opposite trend observed for the hydrogen content.

3. Results and discussion 3.1. Results of HNO3 reacting with decane soot from a rich flame (grey soot) The reaction of HNO3 with grey soot from a rich flame of decane was examined in a series of steady-state experiments by exposing the soot sample to a continuous flow of HNO3 . Fig. 1 shows the results of a typical uptake experiment. When the soot was exposed to HNO3 a large and instantaneous rate of uptake was observed. The MS signals at m/z 30, 46 and 63 correspond to HNO3 . Partial saturation of the sample seems to occur after some time at the chosen flow rate. At the same time, an important signal at m/z 47 is observed. This signal remains almost constant during the whole reaction and indicates the formation of HONO. Some experiments have been performed at long reaction times of up to 90 min, where both the loss of HNO3 and formation of products were observed. No measurable amounts of NO2 monitored at m/z 46 or NO monitored at m/z 30 have been observed after correction for the presence of HNO3 at both fragment masses. For uptake experiments on grey soot, m/z 46 was used to monitor the flow

Elemental composition and BET surface areas of two types of soot generated in this work including results from previous studies Elemental analysis

Reference 27 33,34 23

5112

BET surface area/m2 g1

Soot

C (wt.%)

H (wt.%)

N (wt.%)

O (wt.%)

69 218 89 46 91

Grey decane soot Black decane soot n-Hexane n-Hexane Kerosene

97.27 0.05 96.39 0.22 87–92.5

0.83 0.04 0.19 0.01 1.2–1.6

0.20 0.18 0.27 0.09 —

1.65 0.19 3.22 0.25 6–11


Fig. 2 HNO3 uptake monitored at m/z 46 and HONO production monitored at m/z 47 on grey soot as a function of [HNO3]. Yields are measured for 4 min of interaction. Mean sample mass is 16 0.8 mg, 4 mm diameter escape orifice reactor (kesc ¼ 0.299 s1), [HNO3] ranges from 1.0 1012 to 9.1 1012 molecule cm3.

of HNO3 because no other products contributing to this peak were observed and because the intensity at m/z 63 is very low. Both HNO3uptake and HONO release increase with [HNO3]. This is seen in Fig. 2 for the 4 mm diameter escape orifice integrated over 4 min. A good 1 : 1 correlation of the data as well as a linear increase with [HNO3] may be seen without any signs of saturation. The concentration of HNO3 was in the range 1.9 1011–9.1 1012 molecule cm3. Fig. 3 summarizes the observed HONO yields, defined as the ratio of the amount of HONO released to the amount of HNO3 taken up during the same time, plotted against the mean [HNO3]. Yields of HONO are in the range 34–68%, and it is concluded that HNO3 reacts with reducing surface sites on the soot to yield HONO. The missing fraction of adsorbed HNO3 not forming HONO corresponds to HNO3 , which is irreversibly adsorbed on the soot. A mechanism similar to that proposed16,27 for NO2 reacting with black and grey soot is now considered: HNO3 þ fC Hgred ! HONO þ fCgox

ð1Þ

HNO3 ! fHNO3 g

ð2Þ

where the species in curved brackets refer to adsorbed species and where the subscripts ‘‘ red ’’ and ‘‘ ox ’’ designate oxidizable and oxidized surface sites on soot for the reaction with HNO3 . A second series of experiments was performed where NO reacted with grey soot, after reaction with a known quantity of HNO3 . NO is unreactive towards fresh unexposed soot

Fig. 3 HONO yield monitored at m/z 47 relative to HNO3 (monitored at m/z 46) taken up on soot generated under rich fuel conditions (grey soot) as a function of [HNO3]. Yields are measured for 4 min of interaction. Mean sample mass of 16 mg, 4 mm diameter escape orifice (kesc ¼ 0.299 s1), [HNO3] ranges from 1.0 to 9.1 1012 molecule cm3.

Fig. 4 Interaction of NO with grey soot samples that had previously reacted for 15 min with HNO3 . Both NO2 (monitored at m/z 46, corrected) and HONO (monitored at m/z 47) signals are detected, but only HONO production is displayed. [HNO3] ¼ 8.22 1012 molecule cm3, the quantity of adsorbed HNO3 is 2.39 1016 molecule cm2 from the previous HNO3 uptake experiment, [NO] ¼ 4.89 1012 molecule cm3.

samples,29,30 a result checked often in reference experiments for the present soot samples, indicating that the observed reaction corresponds to NO reacting either with HNO3 adsorbed on the soot or with a reaction product following the reaction of HNO3 on soot. A small rate of HONO desorption is observed as may be seen in Fig. 4. Most of the adsorbed HNO3 still remains on the soot after reacting with NO, and only a small quantity of absorbed HNO3 reacts to produce HONO in the slow reaction according to reaction (3):18 fHNO3 g þ NO ! HONO þ fNO2 g

ð3Þ

HONO is released into the gas phase by grey soot and NO2 remains adsorbed on the soot or reacts27 to produce additional HONO. 3.2. Results of HNO3 reacting with decane soot from a lean flame (black soot) A series of experiments with black decane soot was performed akin to the measurements with grey soot discussed above. Fig. 5 shows a raw uptake experiment for a total reaction time of 30 min. Again a large and instantaneous uptake of HNO3 was observed, with NO being the main product detected at m/z 30 after correction for HNO3 contributing to m/z 30. The uptake of HNO3 and the formation of NO integrated over 4 min increased with [HNO3], as seen in Fig. 6. The irreversible uptake of HNO3 is the main process in this case. Yields of NO are in the range 7–23%, and increase with [HNO3]. The dependence of HNO3 taken up as a function of the mass of soot is scattered but does not show any dependence on mass. As

Fig. 5 Raw data of a typical uptake experiment of HNO3 monitored at m/z 63 on soot generated under lean fuel conditions (black soot). [HNO3] ¼ 3.4 1012 molecule cm3, soot mass of 16 mg, 4 mm diameter escape orifice (kesc ¼ 0.299 s1), interaction time of 30 min.


5113

Fig. 6 HNO3 uptake (monitored at m/z 63) and NO formation (monitored at m/z 30 corrected) on black soot as a function of [HNO3]. Yields are calculated for 4 min of interaction. Mean sample mass is 16 0.9 mg, 4 mm diameter escape orifice reactor (kesc ¼ 0.299 0.299 s1), [HNO3] ranges from 1.97 1011 to 5.9 1012 molecule cm3.

has been proposed previously for NO2 reacting on soot,27 it may be assumed that HONO is initially produced in the same process that occurs when HNO3 reacts with grey soot. However, the initially formed HONO decomposes on the black soot to NO, which is released into the gas phase and detected, and also to NO2 , which remains on the substrate or undergoes additional reactions with soot according to reaction (4): 2 fHONOg ! NO þ fNO2 g þ fH2 Og

ð4Þ

When Fig. 1 and Fig. 5 are compared, a steady state uptake of HNO3 on black soot is found, which is larger by a factor of 10 compared to experiments on grey soot, as discussed below. Experiments performed at long interaction times show that the samples are generally not exhausted even after 2 h of reaction. Other important observations may be made on Fig. 5 for experiments with black soot. After a certain reaction time, which depends on [HNO3], a second loss process for HNO3 is observed. This loss is accompanied by an important increase in the rate of formation of NO and the appearance of a small but significant signal of NO2 , which becomes visible when m/ 46 is corrected for the contribution of HNO3 . Fig. 7 shows the MS signal at m/z 30, which is proportional to the flow rate of NO, and gives the corresponding production of NO vs. reaction time for the experiment presented in Fig. 5. An increase in the production of NO after approximately 1250 s may be clearly seen. This surprising result, valid exclusively for black soot experiments, apparently depends on the [HNO3], thus suggesting that the decomposition involves a bimolecular or higher order process which becomes observable at high surface

Fig. 8 Induction time for the higher order process observed in the reaction of black decane soot + HNO3 as a function of [HNO3].

coverages of HNO3 . This observation has previously been reported in laminar coated-wall flow tube experiments.23 The results suggest that when two HNO3 molecules adsorbed on the soot substrate are in close proximity, bond breaking and rearrangement processes form products such as NO and NO2 . As more surface sites are occupied by HNO3 with time, the second stage of HNO3 uptake is induced; this is observable in Fig. 7 at t ¼ 1250 s because a surface process faster than reaction (1) is kicking in, thus freeing up surface sites at an accelerated pace for increased adsorption of HNO3 . This increased rate of uptake combined with decomposition of HNO3 was further investigated by varying [HNO3]. In Fig. 8 the induction time, i.e. the time elapsed until the beginning of this accelerated decomposition, such as for example observed at 1250 s in Fig. 7, is plotted vs. [HNO3]. Two features are seen. There seems to be a threshold for the induction time which may be explained by the competition between a higher order and a first order loss process as reported previously31 for ozone reacting on activated carbon. This competition results in a minimum and constant value for the induction time of approximately 300 s at high [HNO3]. The fact that the induction time does not tend to zero at high [HNO3] suggests that the reaction mechanism is complex. On the other hand, when [HNO3] is lower than 1011 molecule cm3, the accelerated stage of HNO3 uptake is not observed even after 70 min of exposure. Apparently, a low rate of HNO3 adsorption never can achieve sufficient coverage for the higher order process to kick in owing to the competitive first order loss process, such as diffusion of HNO3 into the bulk of the soot. The following reaction mechanism is accordingly suggested for black soot: HNO3 þ fC Hgred ! fHONOg þ fCgox

ð1Þ

2fHONOg ! fNO2 g þ NO þ fH2 Og

ð4Þ

HNO3 ! fHNO3 g

ð2Þ

27

NO2 reacts on black soot to produce mainly NO, akin to its interaction on amorphous carbon30 according to eqn. (5): NO2 þ fC Hgred ! NO þ fCgox

ð5Þ

At high coverage by adsorbed HONO and HNO3 , reactions (6) and (4) may become important, because they are both effective processes second order in HNO3 : fHNO3 g þ fHONOg ! 2 NO2 þ fH2 Og

Fig. 7 Corrected MS signal at m/z 30 corresponding to the formation of NO based on the raw data in Fig. 5. A higher order process leading to increased release of NO starts at t ¼ 1250.

5114


ð6Þ

Reaction (6) is the reverse of the reaction which is usually taken to represent the dimer of NO2 reacting in the aqueous phase.18 In summary then, NO is the only stable product from the reaction of HNO3 with black soot because NO is unreactive, unlike all the other intermediates, such as adsorbed HONO, HNO3 and NO2 , which continue to react on this substrate. In contrast, NO is quite reactive on soot at high

Fig. 9 Typical rate of product release (HONO, NO2) for the reaction of NO with black decane soot that previously had reacted with HNO3 at [HNO3] ¼ 1.58 1013 molecule cm3. [NO] ¼ 2.25 1013 molecule cm3, the quantity of adsorbed HNO3 is 1.01 1017 molecule cm2 from the previous HNO3 uptake experiment, escape orifice diameter of 1 mm (kesc ¼ 0.034 s1).

temperatures and yields mainly N2 and CO, as is well documented in the automotive and engineering literature.35 A series of steady state experiments was performed in order to study the reaction of NO with soot which had previously reacted with HNO3 . Of course it should be remembered that NO is an observed product of the reaction of HNO3 with soot. A typical experiment is presented in Fig. 9 where an important rise in the corrected MS signal at m/z 46 indicates significant production of NO2 and a very small rate of HONO formation. The yields of NO2 are just a small fraction of the total quantity of HNO3 retained on the surface. HONO yields have also been obtained in similar experiments, and were small and non-reproducible; this is not surprising, as they are expected to undergo secondary heterogeneous reactions on black soot.27 The dependence of the rate of NO2 formation on [NO] in reaction (3) with previously adsorbed HNO3 has been investigated. Two experiments were performed, in which an identical NO concentration probed two different soot samples, each of which had been reacted with HNO3 beforehand to a different extent, leading to samples that had acquired two significantly different quantities of adsorbed, but presumably unreacted HNO3 . Fig. 10a and b show that the rate of NO2 production clearly depends on the duration of prior exposure to HNO3 and thus the quantity of HNO3 previously adsorbed. Therefore, NO2 is produced in reaction (3) resulting from the reaction of HNO3 with NO: fHNO3 g þ NO ! NO2 þ fHONOg

ð3Þ

followed by subsequent secondary reactions, because significant yields of HONO are not observed unlike with grey soot, as displayed in Fig. 4. Thus, reactions (4), (6) and (7) might occur: 2 fHONOg ! NO þ NO2 þ fH2 Og

ð4Þ

fHNO3 g þ fHONOg ! 2 NO2 þ fH2 Og

ð6Þ

fHONOg ! HONO

ð7Þ

The production of HONO is not significant because the desorption from black soot is not rapid. Changes in the MS signal for NO are not detected, due to the large [NO] used, such that NO2 becomes the only observable product under the current experimental conditions. It is therefore not possible to distinguish between reactions (6) and (4) as generators of NO2 . Nitrogen dioxide also reacts on black soot such that the observed quantities of NO2 may be only a small fraction of the total primary product yield of reactions (3), (6) and (4).

Fig. 10 Corrected MS signal atm/z 46 proportional to [NO2] resulting from the reaction of NO + black decane soot that previously had reacted with HNO3 . (a) 15 min of previous interaction of HNO3 + black soot leading to 2.05 1017 molecule cm2 of adsorbed HNO3 ; (b) 6 min of previous interaction of HNO3 + black soot leading to 5.53 1016 molecule cm2 of adsorbed HNO3 . [NO] ¼ 2.25 1013 molecule cm3, soot mass of 16 0.8 mg, 4 mm diameter escape orifice.

The effective rate constant of reaction (3) has been determined for the conditions of Fig. 10a by applying eqn. (8) and (9): r¼

d½NO d½NO2 ¼ dt dt

r ¼ k3 ½NOðgÞbHNO3 cads

ð8Þ ð9Þ

We obtain an initial value of k3 ¼ 5.4 1019 cm2 molecule1 s1 with r ¼ 2.5 1012 molecule s1 cm3 based on the rate of formation r of NO2 and [NO] given in Fig. 10a. We note that r is expressed in units of molecule s1 cm3 in analogy to units for gas phase reactions; it may easily be expressed in units of molecule s1 cm2 appropriate for surface reactions given the volume-to-surface ratio of the experimental system. The amount of adsorbed HNO3 for the experiment of Fig. 10a corresponds to 1.2 1014 molecules cm2 when we use the BET surface area of 218 m2 g1 from Table 2. This coverage corresponds to approximately 30% of a molecular monolayer of HNO3 at the mass loading of 0.8 mg cm2 given in Fig. 10a. The reaction of gaseous nitric oxide with nitric acid adsorbed on silica surfaces in the presence of water at room temperature has been investigated previously.18 The major gaseous product observed is NO2 , in agreement with the results from this study with the overall reaction stoichiometry corresponding to 3NO2 produced per NO reacted. Saliba et al.18 measured the decay of NO and the formation of NO2 in this reaction at different water coverages on the surface. The rate of reaction is affected by the number of adsorbed water layers and [NO]0 , the initial concentration of NO. The reaction rate obtained for a low value of [NO]0 and a low amount of adsorbed water is d[NO2]/dt ¼ 3.5 1013 molecule cm3 min1. The [NO] used in our experiments is lower by approximately two orders of magnitude. However, the influence of adsorbed water on the rate law has not been explored in this work. By contrast, in the present work a typical rate of NO2 formation is r ¼ 1.5 1014 molecule Phys. Chem. Chem. Phys., 2002, 4, 5110–5118

5115

cm3 min1, which is significantly larger than the rate obtained on rough silica substrates. We conclude that the rate coefficient for reaction (3) on soot is larger by at least three orders of magnitude compared to reaction on silica. Therefore, HNO3 adsorbed on soot may react efficiently with gas phase NO to generate NO2 in certain cases. This could be an effective process of renoxification of HNO3 in the boundary layer under special circumstances of polluted urban atmospheres, where soot particles and nitric oxides are both present in high concentrations. This process would therefore generate NO2 at the expense of HNO3 and would thus affect the relevant ratio. However, this chemistry would probably not have any influence on the measured [HNO3]/[NOx] in the free troposphere,11,12 barring any unexpectedly high soot concentrations.

3.4. Uptake kinetics of NO2 on grey and black decane soot Raw data on the reaction of HNO3 with grey decane soot clearly show that the rate of uptake of HNO3 strongly decreases during the course of an experiment, as displayed in Fig. 1. After an initial fast uptake, saturation is observed, a while after the start of the experiment with [HNO3] in the range 1.0 1012–9.1 1012 molecule cm3. Both the initial uptake (g0), and the steady state uptake probabilities (gss) have been plotted as a function of [HNO3]. The values obtained are constant within experimental uncertainty and are therefore independent of [HNO3] in the range tested. They are: g0 ¼ (4.6 2.3) 103 and gss ¼ (5.2 1.3) 104. For HNO3 reacting on black soot, a fast rate of HNO3 uptake was again observed, followed by a slow saturation process. The uptake tends to steady state before the accelerated stage of HNO3 decomposition is observed. Fig. 11 presents the values of the measured g0 and gss as a function of [HNO3]. The following mean values were obtained: g0 ¼ (2.0 0.1) 102 and gss ¼ (4.6 1.6) 103. The independence of the first-order rate constant and thus of the uptake coefficient of [HNO3] indicates that the rate law for HNO3 on black soot is first-order in nitric acid. The values of the uptake coefficients are one order of magnitude lower for grey soot than for black soot, reflecting the higher reactivity of black compared to grey soot. The increased polar character of the soot surface and the expected larger number of the surface OH groups on black soot may help explain its faster uptake of HNO3 compared to grey soot. In addition, it may also explain the surface reactions of the primary product HONO, which subsequently interacts with the reactive surface of black soot to yield the observed products NO and NO2 .

Fig. 11 Initial (g0) and steady state (gss) uptake coefficients of HNO3 interacting with soot obtained under lean combustion conditions (black soot) as a function of [HNO3] in the 4 mm diameter escape orifice, mean sample weight is 16 0.9 mg spread out over 19.6 cm2. Calculations have been performed using the geometric surface area of the sample.

5116


Fig. 12 Initial (g0) and steady state (gss) uptake coefficients of HNO3 interacting with soot obtained under lean combustion conditions (black soot), as a function of soot mass in the 4 mm diameter escape orifice, mean [HNO3] of 2.7 1012 molecule cm3.

Fig. 12 shows that g0and gss are independent of the mass of black soot. The well-documented stickiness32 of HNO3 keeps the HNO3 molecules within the top few layers, resulting in a small effect of the mass of the underlying soot layers on g. This independence of g on mass justifies the use of the geometric surface area for the interpretation of the uptake rate data in terms of uptake coefficients, at least within the time scale of the uptake rate measurements ranging from approximately 0.1 to 10 s. However, after a sufficiently long interaction time on the scale of tens of minutes the whole internal surface of the soot sample would eventually become available for the uptake of HNO3 . Owing to the properties of the Knudsen reactor, which operates in the molecular flow regime the present kinetic results address the intrinsic surface reactivity of HNO3 on soot, because gas phase diffusion of HNO3 and of its reaction products to and from the substrate are not rate limiting as there are no gas phase collisions at the pressures used. We therefore assert that the measured rate-limiting process in the present experiments is indeed the heterogeneous decomposition of HNO3 rather than the diffusion of molecules to and from the substrate surface. A final remark concerns the interpretation of the uptake coefficients g0 , gss and the use of the geometric surface area in the assessment of the rate coefficient in view of the highly structured soot substrate. Pore diffusion theory36 has been used to correct the measured uptake rate coefficient in cases where a moderately reactive gas interacts with a monodisperse powder. The theory has been applied successfully for powders of uniform size and for molecules of zero residence time ts on the substrate. This latter condition is necessary in order to attribute the decrease in the effective diffusion coefficient to the porosity y and the tortuosity t, two material parameters of the substrate.36 In the present case it is not possible to establish a characteristic length parameter of the soot which would lead to the Knudsen diffusion coefficient and the number of layers with which the gas is interacting because of the documented fractal nature of the substrate.25,37 This remains true even though we have obtained good quality SEM and TEM images showing the size of the primary soot particles to lie in the range 20 to 30 nm. In addition, the data displayed in Fig. 12 show that the observed values of gss are independent of the sample mass, which is proof of the interaction of only the uppermost layer on the time scale of the present experiment. That same conclusion has been reached from a study of HNO3 with alkali metal salts38 for which a very small Knudsen diffusion coefficient was deduced from the application of pore diffusion theory, owing to the large value of the surface residence time of HNO3 on that substrate. As expected, the value of the initial uptake coefficient g0 is

independent of mass, as may be seen in Fig. 12 because the interaction time of HNO3 with the substrate is insufficient for reaction in deeper substrate layers. We have given a pertinent example in the past30 where we have observed an identical value of g0 within the uncertainty of the measurement for the interaction of NO2 with well-characterized soot substrates (amorphous carbon FS 101, Printex 60 and FW 2) whose internal surfaces as measured by the BET surface differed by a factor of 26. In conclusion, we therefore abstain from applying the pore diffusion theory as it is not applicable in the present case because of (a) the fractal nature of soot and our inability to calculate a Knudsen diffusion coefficient and define a finite number of interacting substrate layers, (b) the observation of the mass-independence of gss , and (c) the confirmed ‘‘ stickiness ’’ of HNO3 as measured by ts on a variety of surfaces.

Conclusions The different combustion conditions for the production of grey and black soot from decane (C10H22) strongly influence its reactivity towards the uptake of HNO3 , both in terms of the reactive pathways and its kinetics. One reason may be that the surface of black soot is more polar than that of grey soot. The results of the elemental analysis show that the mass fraction of oxygen in black soot is approximately twice that in grey soot as a consequence of the leaner combustion conditions. Conversely, black soot contains less bulk hydrogen than grey soot. Whereas grey soot from a rich flame leads to HONO in yields of 34–68%, which is constant throughout the reaction, NO is the main product of HNO3 reacting with soot from a leaner flame. Traces of NO2 are also observed in this case. A reaction mechanism has been proposed for grey soot in which HNO3 is reduced to HONO, which is subsequently released into the gas phase. However, incipient HONO formed in the primary reaction stays adsorbed on black soot and decomposes to NO in a second step. NO is the marker used to monitor the reaction of HNO3 with black soot, whereas NO2 is expected to mostly stay adsorbed onto the substrate as it undergoes secondary heterogeneous reactions.27 Uptake experiments on black soot have shown that at [HNO3] > 1011 molecule cm3, a higher order process is induced, accompanied by additional production of NO and NO2 . The induction time decreases with increasing [HNO3] to a limit of approximately 300 s at 4.3 1012 molecule cm3. Beyond this concentration a constant induction time of approximately 300 s was obtained which may be explained by the competition between a higher order reactive and a first order loss process. The reaction of NO with soot previously exposed to HNO3 shows that a small fraction of adsorbed HNO3 may react with NO, enabled by the strong interaction with hydroxy groups on the soot. Small yields for this reaction lead to the conclusion that it is a slow process resulting in a rate coefficient of k 5.5 1019 cm2 molecule1 s1. The expected products are indeed observed and are mainly HONO for grey soot, whereas on black soot NO is the main product. Both initial and steady state uptake probabilities are presented in Fig. 11 as a function of [HNO3]. The values of g0 and gss have been calculated using the geometric surface area of the sample (19.6 cm2). No dependence of g0 and gss on [HNO3] was observed, thus confirming a rate law first-order in HNO3 . The stickiness of HNO3 explains that the gss values are independent of the mass of soot. Arguments are presented why the application of the pore diffusion theory is inappropriate in the present molecule–substrate combination.

Acknowledgements Generous support of this research was granted by OFES sponsoring the EU project NITROCAT within the Fifth EU Framework Program on ‘‘ Environment and Climate ’’. A postdoctoral fellowship obtained from the Spanish Ministry of Education, Culture and Sports on behalf of M. S. Salgado is gratefully acknowledged.

References 1 J. H. Seinfeld and S. N. Pandis, Atmospheric Chemistry and Physics: from Air Pollution to Climate Change, Wiley, New York, 1998. 2 J. E. Penner and T. J. Novakov, Geophys. Res., 1996, 101, 19 373. 3 W. F. Cooke and J. J. Wilson, J. Geophys. Res., 1996, 101, 19 395. 4 U. Schumann, J. Strom, R. Busen, R. Baumann, K. Gierens, M. Krautstrunk, F. P. Schroder and J. Stingl, J. Geophys. Res., 1996, 101, 6853. 5 E. J. Jensen, O. B. Toon, A. Tabazadeh, G. W. Sachse, B. E. Anderson, K. R. Chan, C. W. Twohy, B. Gandrud, S. M. Aulenbach, A. Heymsfield, J. Hallett and B. Gary, Geophys. Res. Lett., 1998, 25, 9, 1363. 6 Special Report: Aviation and the Global Atmosphere, ed. J. E. Penner, D. H. Lister, D. J. Griggs, D. J. Dokken and M. McFarland, IPCC, 1999. 7 D. J. Lary, R. Toumi, A. M. Lee, M. Newchurch, M. Pirre and J. B. Renard, J. Geophys. Res., 1997, 102, 3671. 8 F. J. Dentener, G. R. Carmichael, Y. Zhang, J. Lelieveld and P. J. Crutzen, J. Geophys. Res., 1996, 101, 22. 9 G. M. Underwood, C. H. Song, M. Phadnis, G. R. Carmichael and V. H. Grassian, J. Geophys. Res., 2001, 106, 18 055. 10 J. E. Penner, D. Hegg and R. Leaitch, Environ. Sci. Technol., 2001, August, 332A. 11 H. B. Singh et al., J. Geophys. Res., 1996, 101, 1793. 12 D. A. Hauglustaine, B. A. Ridley, S. Solomon, P. G. Hess and S. Madronich, Geophys. Res. Lett., 1996, 23, 19, 2609. 13 B. Aumont, S. Madronich, M. Ammann, M. Kalberer, U. Baltensperger and F. Brocheton, J. Geophys. Res., 1999, 104, 1729. 14 V. H. Grassian, J. Phys. Chem. A, 2002, 106, 860. 15 M. Ammann, M. Kalberer, D. T. Jost, L. Tobler, E. Ro¨ssler, D. Piguet, W. Ga¨ggeler and U. Baltensperger, Nature, 1998, 395, 157. 16 A. Gerecke, A. Thielmann, L. Gutzwiller and M. J. Rossi, Geophys. Res. Lett., 1998, 25, 2453. 17 J. Kleffmann, K. H. Becker, M. Lackhoff and P. Wiesen, Phys. Chem. Chem. Phys., 1999, 1, 5443. 18 N. A. Saliba, H. Yang and B. J. Finlayson-Pitts, J. Phys. Chem. A, 2001, 105, 10 339. 19 U. Kirchner, V. Scheer and R. Vogt, J. Phys. Chem., 2000, 104, 8908. 20 D. M. Smith and A. R. Chughtai, J. Geophys. Res., 1996, 101, 19 607. 21 J. Thlibi and J. C. Petit, in Impact of Emissions from Aircraft and Spacecraft upon the Atmosphere, ed. U. Schumann and D. Wurzel, Ko¨ln, Germany, 1994. 22 C. A. Rogaski, D. M. Golden and L. R. Williams, Geophys. Res. Lett., 1997, 24, 381. 23 M. T. Leu and W. Choi, J. Phys. Chem., 1998, 102, 7618. 24 C. A. Longfellow, A. R. Ravishankara and D. R. Hanson, J. Geophys. Res., 2000, 105, 24 345. 25 H. Saathoff, K. H. Naumann, N. Riemer, S. Kamm, O. Mo¨hler, U. Schurath, H. Vogel and B. Vogel, Geophys. Res. Lett., 2001, 28, 1957. 26 R. S. Disselkamp, M. A. Carpenter and J. P. Cowin, J. Atmos. Chem., 2000, 37, 113. 27 D. Stadler and M. J. Rossi, Phys. Chem. Chem. Phys., 2000, 2, 5420. 28 F. Caloz, F. F. Fenter, K. D. Tabor and M. J. Rossi, Rev. Sci. Instrum., 1997, 68, 3172. 29 D. M. Smith, W. F. Welch, S. M. Graham, A. R. Chughtai, B. G. Wicke and K. A. Grady, Appl. Spectrosc., 1988, 42, 674. 30 K. Tabor, L. Gutzwiller and M. J. Rossi, J. Phys. Chem., 1994, 98, 6172. 31 S. Stephens, M. J. Rossi and D. M. Golden, Int. J. Chem. Kinet., 1986, 18, 1133. 32 T. G. Koch, H. van den Bergh and M. J. Rossi, Phys. Chem. Chem. Phys., 1999, 1, 2687.


5117

33 M. S. Akhter, A. R. Chughtai and D. M. Smith, Appl. Spectrosc., 1985, 39, 143. 34 M. S. Akhter, A. R. Chughtai and D. M. Smith, Appl. Spectrosc., 1985, 39, 154. 35 K. Matsuoka, H. Orikasa, Y. Itoh, P. Chambrion and A. Tomita, Appl. Catal. B, 2000, 26, 89.

5118


36 L. F. Keyser, S. B. Moore and M.-T. Leu, J. Phys. Chem., 1991, 95, 5496 and references therein. 37 D. F. Blake and K. Kato, J. Geophys. Res.D, 1995, 100, 7195. 38 F. F. Fenter, F. Caloz and M. J. Rossi, J. Phys. Chem., 1994, 98, 9801.