Influence of Oxygen on Hg0 Adsorption on Non-Impregnated Activated Carbons.

Both physisorptive and chemisorptive mechanisms play a role in the adsorption of mercury. The present publication investigates the influence of oxygen on the adsorption of Hg0 by breakthrough curve measurements and temperature-programmed desorption (TPD) experiments. The presence of O2 in the gas phase promotes chemisorption. Because of slow adsorption mechanisms, no equilibrium capacities of mercury chemisorption can be determined. For further investigations, coupled adsorption and desorption experiments with concentration swing adsorption and TPD experiments are performed. The results of TPD experiments are simulated and quantitatively evaluated by means of an extended transport model. From the number of desorption peaks, we obtain the number of different adsorption and desorption mechanisms. A detailed simulation of the peaks yields the reaction order, the frequency factor, and the activation energy of the desorption steps. The kinetic reaction parameters allow a mechanistic interpretation of the adsorption and desorption processes. Here, we suppose the formation of a complex between the carbon surface, mercury, and oxygen.

Introduction

Separation of mercury from the waste gas of large emitters such as coal-fired power plants, cement plants, and waste incineration plants and monitoring of emissions are state of the art. However, an efficient treatment of discontinuous waste gases from small- and medium-sized enterprises with strongly changing mercury concentrations is not yet guaranteed. In addition to the established large-scale processes (absorptive gas scrubbing[2] and flue gas adsorption[1]), fixed-bed adsorption is suitable for the separation of mercury from waste gases of discontinuous processes. Suitable adsorbents for Hg0 adsorption are for economic and technical reasons non-impregnated[3−14] and impregnated[3−5,7,15−27] activated carbons, whose development and investigation make up the largest part of the literature. In previous publications, we have already reported on the current status of literature on adsorption of Hg0.[28,29] To sum up, thermodynamic and kinetic investigations of the adsorption of Hg0 are very fragmentary in the literature because of difficult experimental conditions. Many times, only single aspects of physisorption and chemisorption of Hg0 were considered, which often led to antithetical interpretations of adsorption mechanisms. Therefore, the Chair of Thermal Process Engineering at the University of Duisburg-Essen systematically investigates the adsorption of Hg0 on activated carbons. In this study, we will focus on the influence of oxygen on chemisorption, and the results on physisorption have already been published.[28,29]

Experimental Section

Materials

To realize a systematic investigation, two commercial activated carbons (delivered by Carbon Service & Consulting GmbH & Co. KG) with substantially different properties were used, AC 01 and AC 02. Activated carbon AC 01 is based on hard coal, and AC 02 is based on coconut shells; both were activated with steam and delivered in granular form with a particle size of 1.6–2.0 mm. The main element of the adsorbents is carbon with small amounts of hydrogen, nitrogen, and sulfur (Table ). Activated carbon AC 02 has a higher oxygen content than AC 01. In an earlier publication, we have shown that these activated carbons have predominantly physisorptive sites for the adsorption of Hg0.[28]

Table 1

Origin and Chemical Composition of the Adsorbents AC 01 and AC 02

			ash content	C	S	N	H	O
activated carbon	raw material	activation method	[wt % of dry mass]
AC 01	anthracite	Steam	10.7	87.4	0.24	0.32	0.53	0.8
AC 02	coconut	Steam	2.9	90.4	0.44	0.23	0.51	5.5

Mercury is fed to the adsorber via a nitrogen stream as carrier gas. Some thermodynamic data of mercury are shown in Table . The nitrogen has a purity of 99.9999% and a dew point of <−80 °C. In order to investigate the interaction of mercury and oxygen at the surface of the activated carbons, the amount of oxygen in the gas phase was varied between 0 and 40 vol % O2. The oxygen content in the gas phase from 0 to 21 (air) vol % O2 is relevant in industrial applications. To highlight the influence of oxygen on the chemisorption of mercury, additional experiments with a higher oxygen content in the gas phase during loading of the adsorbents were investigated.

Table 2

Physical Properties of Hg0

Property
molar mass [g•mol–1][30]	200.59
melting point [°C][30]	–38.83
vapor pressure (20 °C) [Pa][30]	0.1622
saturation concentration in nitrogen (20 °C, 1013 mbar) [μg m–3][31]	13.344
Enthalpy of evaporation (25 °C) [kJ•mol–1][32]	61.40

Conditioning and Volumetric Characterization of Adsorbents

Nitrogen isotherms at 77 K were measured with a volumetric measuring device (Belsorp-Max of Bel-Japan, Inc.) to characterize the adsorbents. The activated carbons were prepared under vacuum (<10–3 Pa) at 175 °C for about 2 h before starting the measurement. The pore size distribution (see Figure ) was determined by the NLDFT method using a slit pore model.[33] The specific surface area was measured using the BET method according to DIN ISO 9277, and the micropore volume was calculated using the Dubinin–Radushkevich method according to DIN 66135.[34]Table shows the structural properties of the adsorbents.

Figure 1

Logarithmic representation of pore volume vs pore diameter of AC 01 and AC 02.

Table 3

Structural Properties of AC 01 and AC 02 Derived from Nitrogen Isotherms

property	AC 01	AC 02
BET-surface [m2•g–1]	1076	956
total pore volume [cm3•g–1]	0.4960	0.3942
micro pore volume [cm3•g–1]	0.3821	0.3761

Experimental Approach

The experimental plant for measuring the concentration during the adsorption and desorption of Hg0 is shown in Figure and has already been described in detail in earlier studies.[28,29] In the gas mixing chamber, a defined mixture of Hg0, nitrogen, and oxygen (0–40 vol %) is provided using a mass flow controller (MFC). The adsorption process takes place in two glass vessels (reactor 1, reactor 2) with different temperature controls. In both systems, the feed gas is first heated to adsorption temperature in a tube coil and then fed into the vertical fixed bed. In reactor 1, temperatures in the range of 20–120 °C are achieved using a mineral oil-cooled double jacket, and in reactor 2, temperatures in the range of 20–550 °C are realized using a heating jacket. Reactor 1 is used for the measurement of breakthrough curves, and reactor 2 is used for coupled adsorption and desorption experiments. The concentration of Hg0 is measured continuously at the adsorber outlet with a VM 3000 atomic absorption spectrometer from Mercury Instruments GmbH. The measuring device of the experimental plant detects only elemental mercury and no other mercury species or oxygen. The exhaust gas is then decontaminated by two chemisorptive adsorbers filled with sulfur-impregnated adsorbents.

Figure 2

Flow diagram of fixed bed test unit; a = water bath; b = evaporator; c = cooler; d = tempered reactors 1 and 2; MFC = mass flow controller; AAS = atomic absorption spectrometer according to refs (28) and (29).

The activated carbons are prepared as described above before starting the experiments. Then, they are filled (still hot) into one of the two reactors and purged with nitrogen. After reaching adsorption temperature, the mercury is dosed into the gas phase through the bypass with the reactor closed and the concentration is measured until it is constant over a period of 20 min. Then, the mercury vapor is fed to the fixed bed and the experiment is started. To measure single breakthrough curves, the mercury-containing gas is passed over the bed until a defined maximum experimental duration is reached.

Breakthrough curves are compared graphically for the qualitative evaluation of the experiment. In addition, the loading of adsorbent X is determined by integrating the area above the measured breakthrough curve. Assuming that the density of the gas at the inlet of the adsorber is equal to the density at the outlet and that only mercury is adsorbed, the global mass balance around the adsorber can be used to derive the relationship shown in eq to determine the loading

The parameter mHg,ads is the mass of the adsorptive, mS is the mass of the adsorbent, and V̇G is the volume flow of the gas. In previous publications,[28] a maximum error of 4.5% was observed in the calculation of loadings.

In coupled adsorption and desorption experiments, the fresh activated carbon is first loaded with mercury for 1 h. Then, a change in concentration (CSA) is performed by purging the activated carbon with pure nitrogen at 25 °C, and physically bound mercury atoms are desorbed. Because the chemically bonded mercury atoms do not desorb during the CSA, temperature-programmed desorption (TPD) experiments need to be performed. Consequently, as soon as physisorptive desorption can no longer be observed, the temperature of the reactor is increased at a rate of 5.1 K min–1 while still purging with nitrogen. By detecting the mercury concentration in the outgoing gas stream, a temperature-dependent profile with characteristic desorption peaks can be measured. It should be noted that the heating rate has an influence on the pattern of the TPD peaks. The height and width of the peaks depend largely on the energy input of the heating and therefore on the selected temperature ramp. A change in the temperature ramp may provide a possibility to better quantify and localize the individual peaks. In the chosen example, experimental results have shown that a temperature ramp of 5.1 °C•min–1 seems to be the most appropriate (see Supporting Information, Figure S1). This implies that the results of different working groups can only be compared considering the heating rate. GHSV in coupled adsorption and desorption experiments with CSA and TPD was approximately 2230 m3 madsorbens–3 h–1.

For the evaluation of the experiment, the values for the Hg0 loading of the activated carbons during adsorption XAds and the masses of Hg0 desorbed during CSA (XCSA) and TPD (XTPD) are calculated using eq . The input concentration cHg,in of the desorption experiments corresponds to zero. The ratio of desorbed mass (during CSA + TPD) to adsorbed mass is calculated using eq .

A large number of replicate measurements show a ratio of desorbed mass to adsorbed mass ≥85%.

Computational Methods

For the evaluation of the TPD experiments, dynamic simulations were performed with the solver Aspen Custom Modeler from Aspen Tech. The fixed bed of the adsorber is discretized into a large number of volume increments, and for each increment, mass and energy balances are solved with a finite difference method according to Euler. For the solution of the differential equation system, the following assumptions are made

Idealized plug flow in the fixed bed

Ideal gas behavior of the fluid phase

Single-component adsorption (the carrier gas does not adsorb)

Uniform diameter of spherical adsorbent particles

Radial gradients of concentration and temperature are neglected

Pressure drop in the fixed bed is not taken into account

Axial dispersion is calculated according to the approach of Wakao[35]

Bonding of adsorptive on adsorbent is slow compared to intraparticular diffusion

Negligence of readsorption

Adsorption enthalpy has no influence on energy balances due to small loadings

Mercury complexes formed by the chemical reaction decompose to Hg0

With these assumptions, the mass balance of the fluid phase is calculated[36,37]

The parameter c is the concentration of Hg0 in the fluid phase, Dax is the axial dispersion coefficient, V̇G is the volume flow, εL is the bed porosity, Asp is the specific particle surface area of the bed, ρs is the apparent particle density, and A is the column cross-section.

For experiments on single particles or with small amounts of sample, the Polanyi–Wigner equation is often used, which describes thermal desorption kinetics by means of a simple potential function (eq ). For a temperature ramp, the temperature is related to time by the slope .[38]Here, n is the reaction order of the desorption reaction. It is usually assumed that the activation energy EA and the pre-exponential factor k0 are independent of the degree of coverage.

Because Hg0 adsorption often cannot be measured to equilibrium because of very slow adsorption kinetics and very high capacity of impregnated activated carbons for Hg0,[28,29] the monomolecular loading cannot be determined. For this reason, the loading X is used instead of the degree of coverage to describe the desorption rate in analogy to the rate law of chemical reactions.

The pre-exponential factor k0, the activation energy EA, and the reaction order n are fitted to experimentally measured curves. The regression coefficient R2 with a value of 0.9999 is used as a criterion for the least squares method. The following bounds of fit parameters have been set

The lower limit of the activation energy of desorption EA is equal to the adsorption enthalpy of physisorption of Hg0,[28] assuming that chemisorption must be stronger than physisorption. The upper limit corresponds to the averaged reaction enthalpy of several chemisorptive adsorption processes, which has been extended by a tolerance range.[39] The value range of the frequency factor k0 is difficult to estimate because of the unclear definition. In literature, values for first- and second-order reactions are given in the range from 1•× 104 to 3•× 1017 s–1.[40] For the lower limit, it was assumed that the frequency factor cannot be zero and the upper limit is limited by the mathematical limits of the simulation program. The reaction order is usually in the range between 1 and 4.

If multiple peaks occur during the temperature-dependent gas phase desorption of Hg0, the mass balance of the solid phase consists of several additional terms.

Each term represents the desorption rate of a single adsorption mechanism described by eq .

In the dynamic energy balance of the solid phase S, the transiently stored heat, the heat transfer in the adsorption stream, and the energy exchange between the gas phase G and the solid phase S is taken into account. By algebraic rearrangement, we obtain[36]

The energy balance of the fluid phase includes the energy exchange between solid and fluid phase, the transport terms of convection and dispersion and the energy input by the heat flow of the solid phase. Additionally, the storage term of the gas phase and the heat loss through the adsorber wall are considered. Algebraic rearrangement yields[36]

The heat input to the adsorber through the wall is described by another energy balance, which contains a term for heat storage in the wall, a heat transfer term from the wall to the ambience, and a heat transfer term from the inside of the adsorber to the wall. This yields[36]

The parameters αp, αw,I, and αw,i represent the heat transfer coefficients from the fluid to the solid phase, from the gas phase to the adsorber inner wall, and from the adsorber outer wall to the environment. The quantities cp,s, cp,A, cp,G ,and cp,W describe the specific heat capacities of the adsorbent, the adsorptive, the gas phase, and the adsorber wall. In addition, the density of the gas phase ρG and the wall ρW and the disperse thermal conductivity coefficient λD are used. The parameters da and di are the outside and inside diameters of the adsorber.

An analysis of the individual terms of the energy balances has shown that only the convection term and the heat transfer term between the bed and the gas phase have a significant influence on the temperature profile in the bed. Because of the low loading of the adsorbent with Hg0 in the trace concentration range, the adsorption term in particular has only a very small influence. Therefore, the value of Δhads in the energy balance of the solid phase does not necessarily have to be known exactly.

For the solution of the differential equation system from mass and energy balances, initial and boundary conditions are given and auxiliary equations and quantities are defined, which are described in detail in Supporting Information.

Results and Discussion

Dynamics of Hg0 Adsorption on Activated Carbons in the Presence of O2 in the Gas Phase

Figure shows experimental breakthrough curves of Hg0 adsorption with an amount of 0–40% O2 in the gas phase on AC 01 (left) and AC 02 (right) at 25 °C (top) and 100 °C (bottom). The concentration of Hg0 was constantly maintained at approx. 132 μg•m–3.

Figure 3

Breakthrough curves of Hg0 with 132 μg•Hg•m–3 and 0–40 vol % O2 on AC 01 (left) and AC 02 (right) at 25 °C (top) and 100 °C (bottom).

The breakthrough curves at 25 °C are almost identical for different O2 contents and reveal no significant influence of O2 on Hg0 adsorption. They show a typical S-shape and reach the equilibrium state after 5 or 6 h. Here, we have a single adsorption mechanism (physisorption) of Hg0. Increasing the adsorption temperature leads to a significant change in the shape of the breakthrough curves. At 100 °C, a fast adsorption with an early breakthrough is followed by a fast increase in concentration. The area above the breakthrough curve (capacity) has decreased significantly compared to the measurements at 25 °C. This is typical for physical interactions as physisorption as an exothermal process has lower adsorption capacities at higher temperatures. Later on, the rise in the outlet concentration flattens significantly and the input concentration (equilibrium) is not reached within the specified experimental time. The reason could be an additional (slow) adsorption mechanism, which is stronger with activated carbon AC 01 and with a higher oxygen content in the gas phase. As it can only be seen at a temperature of 100 °C, we assume chemisorptive interactions. At 25 °C, the activation energy cannot be supplied so that the O2 content in the gas phase has no influence on the chemisorption of Hg0.

TPD of Hg0

For detailed investigation of mercury chemisorption, the activated carbons AC 01 and AC 02 were loaded for 1 h with an Hg0 concentration of approx. 132 μg•m–3 at an O2 content of 10 vol % in the gas phase at 25, 50, and 100 °C. Afterward, physisorptively bound mercury atoms were desorbed by a CSA with pure nitrogen at 25 °C. These samples were then used to perform TPD experiments in which further Hg0 was desorbed by increasing temperature. Figure shows the concentration curves of desorbed Hg0 in μg•m–3 as a function of temperature in °C for AC 01 (left) and AC 02 (right).

Figure 4

Hg0 concentration in TPD experiments on activated carbons AC 01 (left) and AC 02 (right) after adsorption at 25, 50, and 100 °C for 1 h.

Table lists the calculated values for the Hg0 loading of the activated carbons during adsorption and the masses of Hg0 desorbed during CSA and TPD. The ratio of desorbed (CSA + TPD) to adsorbed mass was calculated using eq and was ≥85% in all experiments.

Table 4

Detected Mass and Mass Ratio from Coupled Adsorption and Desorption Tests with CSA and TPD on AC 01 and AC 02 at 25, 50, and 100 °C

		mass Hg0
material	adsorption-temperature [°C]	Ads [μg•g–1]	CSA [μg•g–1]	TPD [μg•g–1]	mass ratio
AC 01	100	0.38	–0.11	–0.29	1.06
AC 01	50	0.74	–0.54	–0.11	0.88
AC 01	25	1.97	–1.59	–0.12	0.87
AC 02	100	0.23	–0.15	–0.09	1.04
AC 02	50	1.11	–0.84	–0.11	0.86
AC 02	25	2.30	–2.00	–0.08	0.91

The concentration curves for the experiments with loading at 25 and 50 °C show a desorption peak at 130 °C with a maximum height of 21 μg•m–3 for AC 01 and AC 02. In the identical experiment, with an adsorption temperature of 100 °C, this peak is no longer detectable. The increase in adsorption temperature therefore leads to a decrease in capacity at this peak, which suggests physical interactions. The fact that a small physisorbed amount of Hg0 is still detectable after adsorption at 25 and 50 °C even after desorption with nitrogen indicates a very strong physisorptive interaction, for example, with surfaces in submicropores (<0.45 nm), which cannot be reliably identified by current methods for determining the micropore volume.

Further peaks in the concentration curves of the TPD experiments occur at higher desorption temperatures (>140 °C). In experiments at adsorption temperatures of 25 and 50 °C, these concentration peaks are only slightly noticeable. In experiments at an adsorption temperature of 100 °C, however, two peaks are clearly visible. This indicates chemisorptive adsorption mechanisms because these are activated processes that benefit from a higher adsorption temperature of 100 °C. For activated carbon AC 01, these two peaks occur at 193 and 313 °C, and for activated carbon, they occur at AC 02 at 173 and 280 °C. The maximum concentration is significantly higher for activated carbon AC 01 (18 μg•m–3) than for activated carbon AC 02 (9 μg•m–3). Activated carbon AC 01 additionally shows a concentration plateau in the higher temperature range (<270 °C).

For further investigation, the activated carbons AC 01 and AC 02 were loaded with Hg0 (132 μg•m–3) with an O2 content of 0–40 vol % in the gas phase at 100 °C and purged with nitrogen at 25 °C. Subsequently, TPD tests were carried out. The high temperature was chosen to enhance chemisorption and to limit physisorption (see above and previous publications[29]). The temperature-dependent concentration curves of desorbed Hg0 in μg•m–3 as a function of temperature in °C are shown for the activated carbons AC 01 and AC 02 in Figure .

Figure 5

Hg0 concentration in TPD experiments on activated carbons AC 01 (left) and AC 02 (right) after adsorption for 1 h at 100 °C and 0–40 vol % O2.

The calculated values for the Hg0 loading of the activated carbons during adsorption and the masses of Hg0 desorbed during CSA and TPD are shown in Supporting Information Table S2.

The masses of Hg0 desorbed during the CSA and TPD experiments are shown in Figure as a function of the O2 content in the gas phase for activated carbon AC 01 (left) and activated carbon AC 02 (right).

Figure 6

Desorbed mass of Hg0 during the CSA and TPD experiments for AC 01 (left) and AC 02 (right).

The two chemisorptive desorption peaks of activated carbons AC 01 and AC 02 increase with the O2 content in the gas phase during Hg0 adsorption. Physisorption however does not profit from the higher O2 content.

The chemisorptive desorption peaks are hardly visible (Figure ) when the activated carbons AC 01 and AC 02 are gassed with O2 and subsequently loaded with Hg0 for 1 h. The chemisorption of Hg0 only takes place when activated carbon and O2 are simultaneously present. The surface therefore acts either as a heterogenous catalyst for a gas phase reaction or as an active site for chemisorption on the surface. In a heterogeneously catalyzed reaction, Hg0 could react with O2 to form mercury oxide, which then adsorbs on the surface of the activated carbon. Alternatively, surface–oxygen–mercury complexes could form during adsorption.

Figure 7

Hg0 concentration in TPD experiments with simultaneous (continuous lines) and sequential (dashed lines) adsorption of Hg0 and O2 on AC 01 (left) and AC 02 (right) with 20 (top) and 30 (bottom) vol % O2.

Kinetics of Hg0 Desorption

The results from chapters 4.2 and 4.3 have shown that one or more peaks occur during the TPD of chemisorptively bound Hg0. Several desorption peaks can overlap and result in a broad asymmetric signal. Concentration peaks at different temperatures with different heights and widths indicate different adsorption mechanisms. The quantitative evaluation of the TPD experiments is performed in three steps

1. Estimation of the number of concentration peaks.

2. Calculation of the areas below the peaks.

3. Dynamic simulation of the peaks.

The concentration curves of the TPD experiments with previous mercury loading at 100 °C and oxygen content of 0–40 vol % for 1 h were fitted. In case of AC 01, three Gaussian bell curves and for AC 02 two Gaussian bell curves were used with a regression coefficient of R2 = 0.99. Each desorption peak is attributed to an adsorption mechanism, and the initial loadings for the simulation (Table ) can be determined from the areas below the curves using the mass balance around the adsorber according to eq .

Table 5

Initial Loadings Calculated from the Peaks of the TPD Experiments on AC 01 and AC 02 after Adsorption for 1 h

		mass Hg0 of peaks
material	oxygen content [vol % O2]	X1 [μg•g–1]	X2 [μg•g–1]	X3 [μg•g–1]
AC 01	10	0.075	0.106	0.107
AC 01	20	0.113	0.150	0.126
AC 01	30	0.160	0.156	0.143
AC 01	40	0.171	0.192	0.160
AC 02	10	0.030	0.045
AC 02	20	0.028	0.053
AC 02	30	0.041	0.071

Figure shows experimental and simulated concentration curves of TPD experiments with previous mercury loading on activated carbon AC 01 at 100 °C for 1 h with 10 vol % O2 (top left), 20 vol % O2 (bottom left), 30 vol % O2 (top right) and 40 vol % O2 (bottom right) in the gas phase.

Figure 8

Experimental and simulated concentration curves of the TPD experiments on AC 01 with previous Hg0 loading at 100 °C for 1 h and an O2 content of 10 vol % (top left), 20 vol % (bottom left), 30 vol % (top right) and 40 vol % (bottom right).

Table shows the values determined for the activation energy of desorption EA, the frequency factor k0, and the reaction order n. Additionally, rounded global values for the activation energy of desorption and the reaction order are given.

Table 6

Activation Energy of Desorption EA, Frequency Factor k0, and Reaction Order n for TPD Experiments on AC 01 with Previous Mercury Loading at 100 °C for 1 h and an O2 Content of 0–40 vol % in the Gas Phase

oxygen content	peak	k0 [g1–n•s–1•μg1–n]	EA [kJ•mol–1]	n [-]
10 vol %	1	1.76 × 106	62.15	1.999
	2	5.62 × 103	64.23	1.361
	3	16	55.13	1
20 vol %	1	2.65 × 107	71.00	1.941
	2	2.50 × 104	66.63	1.531
	3	13	51.74	1
30 vol %	1	8.88 × 106	68.70	1.934
	2	1.10 × 104	64.23	1.421
	3	4	45.87	1
40 vol %	1	1.06 × 107	69.76	1.897
	2	2.19 × 104	66.68	1.544
	3	12	51.18	1
global	1	8 × 106 to 3 × 107	68	2
	2	5 × 103 to 3 × 104	65	1.5
	3	4–16	51	1

The fitted values for the activation energy of desorption EA and the reaction order n of the individual peaks are close to each other for all O2 concentrations so that the experimental measurements can also be simulated with high accuracy using rounded global values (see Supporting Information, Figure S2). Because of the large deviations in the values of the frequency factor, no global value could be used here. Different authors interpret this parameter differently. Besides the temperature dependence, also a concentration dependence is discussed.[49] From Table , we can see that the kinetic parameters have slightly different values at different oxygen concentrations. These inaccuracies are probably due to the experimental error in carrying out coupled adsorption and desorption experiments with CSA and TPD and due to the heterogeneity of the activated carbons as a new adsorbent sample was used for each experiment.

Figure shows experimental and simulated concentration curves of the TPD experiments with previous mercury loading of the activated carbon AC 02 at 100 °C for 1 h with 10 vol % O2 (top), 20 vol % O2 (middle), and 30 vol % O2 (bottom) in the gas phase. Simulations were carried out with individual values for the activation energy of desorption and the reaction order of each experiment as well as with rounded global values for these parameters (see Supporting Information, Figure S3). For the simulations, experimental data up to 4250 s (approx. 290 °C) were supplied to the solver. The tailing at the end of the experiment is probably a measurement artifact. Mercury oxide could be formed in the gas phase at high temperatures (>200 °C), adsorb on the walls and the bed, and then slowly decompose to Hg0 (>300 °C).[30] The kinetic parameters are shown in Table .

Figure 9

Experimental and simulated concentration curves of TPD experiments on AC 02 with previous Hg0 loading at 100 °C for 1 h and an O2 content of 10 vol % (top), 20 vol % (middle), and 30 vol % (bottom).

Table 7

Activation Energy of Desorption EA, Frequency Factor k0, and Reaction Order n for the TPD Experiments on AC 02 with Previous Hg0 Loading at 100 °C for 1 h and an O2 Content of 0–30 vol %

oxygen content	peak	k0 [g1–n•s–1•μg1–n]	EA [kJ•mol–1]	n [-]
10 vol %	1	9.99 × 1011	102.57	2
	2	2.58 × 107	89.90	1.69
20 vol %	1	1.30 × 1012	96.13	2
	2	1.88 × 106	72.50	1.82
30 vol %	1	5.00 × 1011	96.19	2
	2	5.34 × 106	76.18	1.76
global	1	1 × 1011 to 5 × 1011	96	2
	2	6 × 105 to 3 × 106	77	1.5

Figure shows the theoretical curve of the potential energy as a function of the reaction progress for chemisorptive adsorption (from left to right) and for chemisorptive desorption (from right to left).

Figure 10

Simplified energy scheme for the progress of a chemisorptive reaction according to ref (41).

In chemisorption, the oxygen molecule adsorbs at a catalytically active surface site and then dissociates for the subsequent reaction with Hg0. For this step, energy must be supplied as an energy peak (activation energy of the adsorption EA adsorption). The dissociation energy contributes significantly to the activation energy of adsorption.[42] As the dissociation step depends on the catalytic activity of the site, the activation energy of adsorption is substance-specific. The structural and chemical properties of activated carbons differ, especially if the materials were produced from different raw materials (here: hard coal in case of AC 01 and coconut shells in case of AC 02). After dissociation, the reaction to the products takes place on the surface. The chemisorptive adsorption is an exothermic reaction, which is why the potential energy falls to an energetic state below the educts.

We have performed a chemisorptive adsorption step in an adsorption experiment which is then followed by a chemisorptive desorption step performed by the TPD experiment. In the chemisorptive desorption step, there is chemical degradation of the product back to the educts plus desorption of the educts (including mercury) from the surface. The energy peak that must be passed is the activation energy of desorption (EA desorption), which is the sum of the activation energy of adsorption and the reaction enthalpy. With different activation energies of adsorption (see above), the activated carbons AC 01 and AC 02 have also different activation energies of desorption. The activation energies of desorption are in the range of 51–68 kJ•mol–1 on AC 01 and at 96 and 77 kJ•mol–1 on AC 02 (see Table ).

One route of mercury chemisorption in the presence of oxygen may be the formation of mercury oxide. The enthalpy of formation of mercury oxide at an adsorption temperature of 100 °C is 94.22 kJ•mol–1.[43] This is above the activation energy of desorption for all desorption reactions at AC 01 and for the second desorption reaction at AC 02. Thus, the formation of mercury oxide is excluded here. In the case of the first desorption reaction at AC 02, the activation energy of desorption is of the same order of magnitude as the enthalpy of formation of mercury oxide, so that according to Figure , no more energy is available for the activation of adsorption. Therefore, the formation of mercury oxide is also implausible here.

Another mechanistic idea proposes the reaction of Hg0, O2, and the surface of the activated carbon to form a surface–oxygen–mercury complex. Hg0 and O2 approach the surface and physisorb on catalytically active sites. The binding of the oxygen atoms is weakened until dissociation of O2 occurs (Figure , top left).[42,44] At the same time, Hg0 is oxidized to Hg2+ (Figure , top right and bottom left). Finally, a stable complex is formed including the active site on the surface of the activated carbon, the mercury, and the oxygen (Figure , bottom right). During the reaction, a hydrogen atom is released from the surface and becomes part of the complex. Because three peaks are found during the TPD from the activated carbon AC 01, it can be assumed that during adsorption in the presence of O2 in the gas phase, Hg0 is chemisorptively bound to three energetically different binding sites and three different surface–oxygen–mercury complexes are formed. The experiments on AC 02 show 2 peaks, so that probably two complexes occur.

Figure 11

Proposal for the reaction sequence of chemisorption of Hg0 and O2 according to refs (42) and (44).

It has to be stressed that this mechanistic discussion of Hg0 chemisorption is speculative and sketchy. In particular, no valid explanation is given as to how mercury is oxidized and oxygen is reduced. It can be assumed that a mercury atom forms two covalent bonds, one with carbon on the surface and the other with oxygen. In a corresponding complex, mercury as the most electropositive element has the oxidation number +II as in many known mercury compounds. The complex in Figure fulfills these conditions. Nevertheless, the illustrations cannot show any concrete mechanistic proposal and therefore have a symbolic character.

Conclusions

The influence of O2 on the chemisorption of Hg0 has been shown by breakthrough curve measurements and TPD experiments. The observation of several desorption peaks at different desorption temperatures suggests several chemisorptive mechanisms. The experimental results could be simulated using an extended transport model. Reaction kinetic parameters (frequency factor, activation energy and reaction order) were determined for the desorption peaks.

The presence of O2 in the gas phase promotes chemisorption. Chemisorption benefits from higher adsorption temperatures and a higher oxygen content. Gassing the activated carbon with O2 prior to the adsorption process has no effect. The activation energy of desorption determined in the simulation is smaller than the standard formation enthalpy of mercury oxide HgO. It was shown that in this case, no HgO can have formed. Oxygen probably reacts in a cascade of elementary reactions with Hg0 and the surface of the adsorbents to form several surface–oxygen–mercury complexes.