Enhanced removal of ultrafine particles from kerosene combustion using a dielectric barrier discharge reactor packed with porous alumina balls

The experiments for the UFP removal were conducted by using the set-up shown in figure 1. The air was supplied by using an air pump with a polytetrafluoroethylene-sintered filter (15 × 10 × 51 mm³, Suzhou Mingrui, China), which can remove nanoparticles from 10 to 350 nm, with a filtration efficiency greater than 99.4% [41]. Three air flow meters (FM1-3, LZB-3WB, Changzhou Xinwang, China) were used. FM1 was used to control the air with a fixed flow rate of 2 l min⁻¹ to a quartz tube (50 × 45 × 800 mm³) in which a kerosene lamp was installed. The kerosene lamp was used to generate the UFPs from kerosene (Zippo, USA) combustion. The gas from the kerosene lamp was dehydrated using an ice bath. FM2 was used to dilute the gases from the quartz tube with the air into the gas mixer. The gas mixtures were supplied to the DBD reactor. As shown in figure 2, five types of DBD reactors were used. All of the DBD reactors generally consisted of a stainless steel frame, two alumina plate blocks, two alumina plates (purity 96%, 115 × 115 × 1 mm³), two stainless steel electrodes (95 × 95 × 0.3 mm³ or 48 × 95 × 0.3 mm³), two organic glass spacers (115 × 10 × 5 mm³), and two organic glass holders (115 × 10 × 2 mm³). The organic glass spacers were used to keep a space distance of 5 mm between the two alumina plates and to allow all of the gas mixtures to pass the space between two alumina plates. PA balls (purity 95% Al₂O₃, Zibo Jianlong, China) were placed in the space. The diameters of the PA balls were around 5 mm, with a number of about 19 × 19 balls in total in Reactor B. In Reactors C and D, PA balls were placed downstream and upstream of the discharge regions, respectively. The PA balls had a Brunauer–Emmett–Teller (BET) surface area of 384 m² g⁻¹, an average pore size of 4.31 nm and a pore volume of 0.413 cm³ g⁻¹. These were measured by using a N₂ adsorption apparatus (Micromeritics JW-BK 132F, Beijing JWGB, China). The packing density was around 0.722 g ml⁻¹.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A pulse power supply (DP-12K5-SCR, PECC, Japan) was used to supply pulsed voltage to the stainless steel electrode terminals of the DBD reactor. The pulsed discharges were generated within the discharge space between the two alumina plates covered with the two stainless steel electrodes. The waveforms of discharge voltage and current were measured by using a voltage probe (VP, P6015A, Tektronix, USA), a current transformer (CT, TCP 0030, Tektronix, USA) and a digital phosphor oscilloscope (DPO 3034, Tektronix, USA).

The particle number concentrations in the gases at the inlet and outlet of the DBD reactor were detected by using a scanning mobility particle sizer (SMPS + E, Grimm, Germany) equipped with a differential mobility analyzer in an aerodynamic diameter ranging from 6.25 to 1093.95 nm. In addition, a condensation particle counter that can measure particles in a number concentration range of 0–10⁷ cm⁻³ (the number of particles per cubic centimeter) was simultaneously utilized.

The energy injection, P, in J Hz⁻¹ from the pulse power supply to the DBD reactor was calculated using equation (1):

$$\begin{eqnarray}&&P=\displaystyle \displaystyle \sum _{i}\left(\displaystyle \frac{{V}_{i}+{V}_{i+1}}{2}\right)\left(\displaystyle \frac{{I}_{i}+{I}_{i+1}}{2}\right)\left({t}_{i+1}-{t}_{i}\right)\end{eqnarray} \tag{ 1 }$$

where V_i and V_i+1 in V were the discharge voltage at the discharge times t_i and t_i+1 in s; and I_i and I_i+1 in A were the discharge current at t_i and t_i+1, respectively. V and I were obtained from the data sequences of the discharge voltage and current waveforms, i ranged from 0 to 9999, and f was the pulse frequency in Hz.

The total UFP removal X_t was defined as equation (2). The sized UFP removal for a fixed diameter x_j was calculated by using equation (3), where n_j and N_j were the particle number concentrations in cm⁻³ in the outlet and inlet gases of the DBD reactor, respectively, and j was an index for the particle of an aerodynamic diameter d_i ranging from 20 to 100 nm.

$$\begin{eqnarray}&&{X}_{{\rm{t}}}=\displaystyle \frac{\sum {n}_{j}}{\sum {N}_{j}}\times 100 \% \,\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{x}_{j}=\displaystyle \frac{{n}_{j}}{{N}_{j}}\times 100 \% .\end{eqnarray} \tag{ 3 }$$

The mechanism of UFP removal by PA balls generally includes inertial collision, direct interception, diffusion deposition, gravity settlement and electrostatic attraction. As the diameters of the UFPs considered in this work were between 20 and 100 nm, the inertial collision, direct interception and gravity settlement were neglected, and only diffusion deposition and electrostatic attraction were thought to be the mechanism of UFP removal by PA balls.

The carbon balance (CB) was calculated by using equation (4):

$$\begin{eqnarray}&&CB=\displaystyle \frac{\sum {S}_{{\rm{in}}}-\sum {S}_{{\rm{out}}}}{\sum {S}_{{\rm{in}}}}\times 100 \% \end{eqnarray} \tag{ 4 }$$

where $\sum {S}_{{\rm{in}}}$ and $\sum {S}_{{\rm{out}}}$ were the sums of mass of all the particles at a range of 20–100 nm from the exhaust gases of kerosene combustion at the inlet and outlet of the DBD reactor, respectively.

The efficiency of a single PA ball due to diffusion deposition was given by the Péclet number (P_e) as [42]:

$$\begin{eqnarray}&&{\eta }_{{\rm{d}}}=3.998\gamma {{P}_{{\rm{e}}}}^{-\displaystyle \frac{2}{3}}\end{eqnarray} \tag{ 5 }$$

where γ was the laminar hydrodynamic factor, and given by equation (5) [43], and calculated using the porosity of the PA balls (ε, 0.62). P_e was obtained using equation (6).

$$\begin{eqnarray}&&\gamma =\displaystyle \frac{1.31}{\varepsilon }\end{eqnarray} \tag{ 6 }$$

$$\begin{eqnarray}&&{P}_{{\rm{e}}}=\displaystyle \frac{{U}_{0}{d}_{{\rm{B}}}}{D}.\end{eqnarray} \tag{ 7 }$$

U₀ was the face velocity in m s⁻¹; d_B was the mean diameter of the PA balls in m; and D was the particle diffusion coefficient in m² s⁻¹ given by:

$$\begin{eqnarray}&&D=\displaystyle \frac{kT{C}_{{\rm{u}}}}{3\pi {\mu }_{{\rm{g}}}{d}_{{\rm{p}}}}\end{eqnarray} \tag{ 8 }$$

in which k was the Boltzmann constant (1.38 × 10⁻²³ J K⁻¹), T was the gas temperature (298 K), and ${\mu }_{{\rm{g}}}$ was the air dynamic viscosity (1.84 × 10⁻⁵ Pa s). C_u was the Cunningham coefficient given by:

$$\begin{eqnarray}&&{C}_{{\rm{u}}}=1+\displaystyle \frac{2{\lambda }_{{\rm{g}}}}{{d}_{{\rm{p}}}}\left[a+b\,\exp \left(-c\displaystyle \frac{{d}_{{\rm{p}}}}{2{\lambda }_{{\rm{g}}}}\right)\right].\end{eqnarray} \tag{ 9 }$$

λ_g was the mean path of an air molecule (66 nm at 25 °C and atmospheric pressure). a = 1.165, b = 0.483, and c = 0.997 were obtained from Kim et al [44].

The electrostatic attraction by a single PA ball was evaluated using equation (9):

$$\begin{eqnarray}&&{\eta }_{{\rm{e}}}=\displaystyle \frac{1+2{\gamma }_{{\rm{s}}}}{1+{K}_{E}}{K}_{{\rm{E}}}\end{eqnarray} \tag{ 10 }$$

$$\begin{eqnarray}&&{K}_{{\rm{E}}}=\displaystyle \frac{{\eta }_{{\rm{e}}}}{1+2{\gamma }_{{\rm{s}}}-{\eta }_{{\rm{e}}}}\,\end{eqnarray} \tag{ 11 }$$

γ_s was the collector polarization coefficient. K_E was the dimensionless parameter of the Coulombic force [45, 46].

The total collection efficiency ${\eta }_{i}$ by a single PA ball was then calculated using

$$\begin{eqnarray}&&{\eta }_{i}=1-(1-{\eta }_{{\rm{d}}})(1-{\eta }_{{\rm{e}}})\end{eqnarray} \tag{ 12 }$$

The total collection efficiency ${\eta }_{{\rm{t}}}$ by 19 PA balls in the gas stream was then calculated using

$$\begin{eqnarray}&&{\eta }_{{\rm{t}}}=1-{\left(1-{\eta }_{i}\right)}^{19}=1-{(1-{\eta }_{{\rm{d}}})}^{19}{(1-{\eta }_{{\rm{e}}})}^{19}.\end{eqnarray} \tag{ 13 }$$

The typical waveforms of discharge voltage and current for five types of DBD reactors filled with or without PA balls were shown in figure 3, where the peak voltages of each voltage pulse for five types of the DBD reactors were set at approximately 9.0 kV. Each voltage waveform had positive and negative pulses. As shown in table 1, the rise time (t_r) and full-width at half-maximum of the voltage pulse (t_w) were 3.3–4.4 and 19.0–20.8 μs, respectively. The peak discharge currents and energy injection were in the order of Reactor B > Reactor E > Reactor A > Reactors C and D.

Zoom In Zoom Out Reset image size

To understand the differences between discharge performances in the five types of DBD reactors, the relationship of energy injection as a function of peak voltage was investigated (figure 4). The energy injection gently increased when the peak voltage increased to a peak voltage of less than 5.0 kV, and greatly rose when the peak voltage was higher than 5.0 kV. The value of 5.0 kV was the inception voltage (V_in) for the pulsed corona discharges that occurred within the gas space between two alumina plates. However, when packing the PA balls in the discharge space (Reactors B and E), the energy injection was greater than those without PA balls (Reactors A, C and D). This was because when the PA balls were in the discharge space, the pulse voltage across the dielectric balls was lower than that across the gases of the same space without dielectric balls (as the relative permittivity of the dielectric balls was higher than that of the gases) [38]. This resulted in an increase in the pulse voltage across the reduced gas space. Thus, the energy injection into the reduced space in the DBD reactor packed with PA balls was higher than that into the DBD reactor without dielectric balls.

Zoom In Zoom Out Reset image size

The UFP removal without PA balls was initially evaluated by using Reactor A. Figure 5(a) shows the UFP number concentrations in the inlet and outlet gases. The UFP concentrations in the inlet gases had a peak of 3.4 × 10⁶ cm⁻³ at 48.3 nm in the range of 20–100 nm, while those in the outlet gases also had a peak of 3.2 × 10⁶ cm⁻³ at 48.3 nm. The UFP removal efficiency was only at a level of 10% with a gas residence time in the discharge space of 1.38 s (figure 5(b)). The low UFP removal was due to the fact that the UFPs could be finitely adsorbed on the surfaces of alumina plates in the DBD reactor.

Zoom In Zoom Out Reset image size

The effect of PA balls on the UFPs removal was then investigated by using Reactor B (figure 6(a)). The UFPs number concentration in the gases at the inlet had a peak value of 4.73 × 10⁶ cm⁻³ at 40.2 nm in the range of 20–100 nm. When there was no discharge, the UFPs number concentration in the outlet gases was decreased to 2.72 × 10⁶ cm⁻³ at 40.2 nm. The sized UFPs removal was enhanced to 60% at 20 nm and decreased with increasing UFPs diameter (figure 6(b)). This was due to the UFPs adsorption on the PA balls. When pulse voltages were applied on to the DBD reactor, pulsed discharges occurred within the reactor, resulting in the obvious decreases of the UFPs number concentration. The peak value of the UFPs number concentration decreased to 6.65 × 10⁵ cm⁻³ at 58.3 nm. The sized UFPs removal was around 90% in the range of 20–100 nm (figure 6(b)). It was indicated that the adsorption capacity of PA balls was enhanced during the discharges.

Zoom In Zoom Out Reset image size

The relationships of the total UFP removal at various peak voltages were evaluated. When using Reactor A without PA balls (figure 7(a)), the total UFP removal increased with an increase in the peak voltage. The total UFP removal was 32% at a 12 kV peak voltage. When the PA balls were fixed in Reactor B (figure 7(b)), the total UFP removal was 35.9% at 0 kV, indicating that those parts of UFPs were removed due to the adsorption by PA balls. When the peak voltage was increased to 7–11 kV, at which corona discharges occurred within the discharge spaces, the total UFP removal was enhanced to approximately 90%. The total UFP removal using Reactor B at a peak voltage higher than 7 kV was obviously higher than that using Reactor A, which again indicated that PA balls play an important role in the UFP removal. The main reason was that the particles were charged and more facilely captured by the dielectric PA balls [47].

Zoom In Zoom Out Reset image size

Figure 8(a) shows the evaluation of different types of DBD reactors on the UFP removal. It was found that total UFP removal was only 7.1% using Reactor A under the experimental conditions of 9.0–9.6 kV, 50 Hz and 2 l min⁻¹ gas flow rate. When packing the PA balls before the discharge space of Reactor D, the total UFP removal increased to approximately 35.0% due to the adsorption ability of the PA balls. Encouragingly, the total UFP removals improved to 84.1% and 85.6%, when using Reactors of C and E, respectively. The results indicated that total UFP removal is strongly affected by the location of PA balls in the DBD reactor. PA balls placed downstream of the discharge regions (Reactor C) were more efficient than those placed upstream of the discharge regions (Reactor D). The PA balls packed inside the discharge regions of Reactor E showed a relative improvement in UFP removal compared to the others. In Reactor B, 91.4% of UFPs were removed when the DBD reactor was fully filled with PA balls and equipped with electrodes, which had double the amount of PA balls compared to Reactor E during discharges. This can be attributed to the enhanced effect on particle charging and the adsorption by the PA balls.

Zoom In Zoom Out Reset image size

Figure 8(b) displays the removal efficiency of sized UFPs versus the particle diameters. The sized UFP removals for a fixed diameter in Reactors B, C and E were higher than those in Reactors A and D at the range of 20–100 nm. The sized UFP removal of 20–30 nm in Reactor B showed better efficiency than that in Reactor E. The particle removal of diameters from 20 to 70 nm in Reactor C was less efficient than those in Reactors B and E. The carbon balances of total UFP removal for the five types of DBD reactors were 7.6%, 90.5%, 87.8%, 46.3% and 89.0%, respectively (figure 9). These values were proportional to the values of total UFP removal (figure 8(a)). The values of carbon balances for the five reactors were obviously different, mainly due to the differences in charge effect and the specific surface areas of the PA balls. The highest carbon balance value was found in Reactor B. Reactor E had half the amount of catalysts compared with Reactor B, and thus the carbon balance value was smaller than that in Reactor B. The charged UFPs in Reactor C were partly neutralized during the transfer processes. Therefore, the value of carbon balance was less than those in Reactors B and E. Moreover, the influences of pulse frequency and gas flow rate were analyzed. The total UFP removal was around 90% regardless of changes to the pulse frequency and gas flow rate (figure 10).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To clarify the mechanism of UFP removal using the DBD reactor packed with PA balls, the experimental results using Reactors B and C were compared to illustrate the effects of particle charging and PA ball adsorption. The discharge space was upstream of the PA balls in Reactor C (figure 2), and the particles in the gas stream were charged when passing through the discharge space. The total UFP removal was 84.1%, an approximately 8% decrease in comparison with that using Reactor B. This was due to the fact that the charged particles were immediately collected by the PA balls via the electrostatic force in Reactor B. However, parts of the charged particles were neutralized in Reactor C before crossing through the space packed with PA balls. Thus, fewer UFPs were adsorbed by the PA balls downstream of the DBD reactor.

In this work, a DBD reactor was utilized to remove UFPs from kerosene combustion. UFPs were positively or negatively charged by plasma discharges. These charged particles were facilely adsorbed by PA balls due to the electrostatic force. However, the total UFP removal did not approach 100%, due to the charging efficiency being unable to reach 100%. Fine particles, with diameter smaller than 100 nm, are mainly charged by the diffusion of ions to the surface of the particles. The final number of charges per particle depends on the particle size and on the charging conditions. With the short transit time in the DBD reactor, UFPs (diameter <60 nm) were not completely filtered by electro-collection due to the small fraction of charged particles [27]. For example, the charged fraction was less than 60% for particles between 5–35 nm in the corona charger [48].

Meanwhile, according to the calculation equations in section 2.2, the removal efficiency of UFPs was related to the particle size regardless of the modes of diffusion deposition or electrostatic attraction. The effect of the electrostatic forces on UFP removal was calculated using equation (11). Figure 11 shows the K_E values while gas flow rate, pulse voltage or pulse frequency were changed. It was found that the K_E values varied with particle size and generally had a peak at a diameter around 40 nm, and had an order of about 0.05 under different conditions, close to that given by Shapiro et al [46].

Zoom In Zoom Out Reset image size

Figure 12 shows the calculation results of UFP removal using equations (5)–(13). It was clear that the UFP removal by diffusion deposition was less than 20% and decreased while UFP diameter increased. The UFP removal by filtration (without discharge) was higher than that by diffusion deposition, but less than that with discharge. This finding suggested that the UFPs from the kerosene combustion were charged, and the removal efficiency was enhanced with the PA balls by electrostatic forces. The total UFP removal from calculation fits well with the experimental results, where the K_E value was calculated using equation (11) and fitted within 20–100 nm (with a good standard deviation (R²) of 0.979) to be:

$$\begin{eqnarray*}&&\begin{array}{l}{K}_{{\rm{E}}}=1.67\times {10}^{-7}{d}_{{\rm{p}}}^{3}-4.11\times {10}^{-5}{d}_{{\rm{p}}}^{2}+2.60\times {10}^{-3}{d}_{{\rm{p}}}\\ \,+1.15\times {10}^{-2}.\end{array}\end{eqnarray*}$$

Zoom In Zoom Out Reset image size

In brief, the removal efficiency with particle size distribution was consistent with the calculated K_E values versus particle sizes, verifying that the total UFP removal cannot approach 100%.