Capture of Particles from an Iron and Steel Smelter with a Pulse-Energized Electrostatic Precipitator

A bench-scale pulse-energized electrostatic precipitator (ESP) was developed to study the capture of particles from exhaust gas streams from iron and steel factories. The morpho-structural characteristics, composition and size distribution of the particles were analyzed. The results showed that the particles consisted primarily of iron oxides and other compounds containing Ca, S, C, Si, Al, and Na elements. The particles had a broad size distribution ranging from 5 nm to several micrometers. The V-I characteristics of the ESP were studied with biased DC voltages of ± 10 kV and ± 10 kV with superimposed pulses (± 5.5 kV, rise time 60 ns, repetition rate 50 pps). The capture efficiency was investigated by means of a low pressure impactor and scanning mobility particle sizer. In the range of 155 to 6,650 nm, the ESP number-based capture efficiencies were > 99.6% and > 98.9% for negative and positive voltages, respectively. For sizes less than 50 nm, the negative pulsed mode was remarkably efficient (93.7 to 99.4%).


INTRODUCTION
The toxicity of the particles has received significant attention in recent years.Research has shown that the size of inhaled particles greatly determines the toxicological and immunological effects.In general, the effects are much greater for submicrometer (d < 1 µm) and ultrafine (d < 0.1 µm) particles (Dusseldorp et al., 1995;Biswas and Wu, 2005;Oberdörster et al., 2005;Maynard, 2006;Ny and Lee, 2011).Consequently, there is a need to limit the emission of such particles.On the other hand, it is well known that industries, such as iron and steel smelters, coal-fired power plants and solid waste incinerators, release a significant quantity of particles to the atmosphere (Dusseldorp et al., 1995;Lin et al., 2010;Ning and Sioutas, 2010).Of these releases, over half the particles go to on-site land disposal, and one quarter are fugitive or point source air emissions.Therefore, knowledge of the composition and size distribution of the airborne particles and development of efficient capture means are of great importance.
Electrostatic precipitators (ESPs) have emerged as a useful tool, and have been extensively used as particle control devices in industry.Their overall mass based capture efficiency typically exceeds 99% (Zukeran et al., 1999a;Huang and Chen, 2002).However, the capture efficiency of the ESPs used in industry in terms of number concentration [number of particles/m 3 (gas)] can be as low as 50-80% (Senior et al., 2000), since most of the submicrometer and ultrafine particles escape from the ESP due to a minimum in the combined effect of diffusion and field charging (Hinds, 1999;Friedlander, 2000).Therefore, the capture of such particles with a high efficiency is an important challenge for the scientific community.
These pulse-energized ESPs were used mainly for fly ash (Dinelli et al., 1991;Jayaram et al., 1996) and mercury (Liang et al., 2002a) removal as well as SO 2 and NO x emission control from industrial flue gases (Mok et al., 1999).The fly ash particles studied, nevertheless, were very large, i.e. in the micrometer size range (Dinelli et al., 1991;Jayaram et al., 1996).Capture efficiency of ultrafine particles by a pulse-energized ESP was investigated by Zukeran et al. (1999a).These particles, however, were generated from an incense combustion, which showed limited practical implications to industry.To the best of our knowledge, the capture of fine and ultrafine particles released from industry, in particular, an iron and steel smelter by using a pulseenergized ESP has not yet been investigated.
This paper describes the development of a bench-scale pulse-energized ESP for effective capture of particles having sizes smaller than 2.5 µm.The specific aim is to study the ESP capture efficiency of particles emitted from iron and steel factories.The capture performance of the ESP was analyzed under various conditions, including (a) positive/ negative DC voltages (± 10 kV), and (b) positive/negative biased DC voltages (± 10 kV) with superimposed pulses (± 5 kV).The number concentration based capture efficiency (number capture efficiency, hereafter) was calculated based on the online measurements of the particles by means of a scanning mobility particle sizer (SMPS) as well as a low pressure impactor (LPI).

Experimental Setup Description
The experimental setup is schematically shown in Fig. 1, which consisted of an aerosol preparation system, a bench-scale ESP, and particle size/concentration online measurement systems.The aerosol preparation system was designed to entrain particles released from a representative iron and steel smelter.The experiments were performed under laboratory conditions (air temperature and relative humidity were 20-22°C and 17-20%, respectively), but the dust had the same composition and size distribution as in the iron and steel manufacturing industries.
It comprised a powder-air mixing system and a particulate matter concentration monitor.The powder-air mixing system had a powder reservoir and a mixing drum where the aerosol was re-circulated.The particles stored in the powder reservoir were continuously introduced into the air flow by means of a worm feeder.The aerosol concentration could be adjusted by changing the rotational speeds, both of the ventilator and the feeder.From the mixing drum, the aerosolized particles flowed to the particulate matter concentration monitor (Fig. 1).The instrument helped monitor the particle mass concentration of the aerosol, in order to maintain a steady mass concentration for the duration of the experiment.The mass concentration of the aerosol upstream of the bench-scale ESP was in the range of 245 to 255 mg/m 3 (for all experiments).The aerosol from the aerosol preparation system was introduced either to the ESP, or to the measurement systems.Although the size distribution of the particles at the inlet of ESP is not the same as the exhausted aerosols at the stack, the ESP capture efficiency measurement is not affected because the ratio out/in is calculated.

Particle Characterization
The particles from a full scale iron and steel smelter were first characterized using various methods to determine offline size distribution, composition, crystallinity, and morphology.The offline particle size analysis was performed using a laser diffraction size analyzer (Malvern Mastersizer 2000) with a measurement range between 20 nm to 2000 µm.X-ray diffractometer (XRD, Bruker AXS/D8 ADVANCE) was used for composition and crystallinity analyses.The chemical investigation of the particle surface was performed by X-ray photoelectron spectroscopy (XPS, VG ESCA 3 MkII-EUROSCAN Instr.spectrometer).A scanning electron microscope (SEM, HITACHI S2600N) was utilized for morphology analysis.The high resolution images of the nanoparticles were performed with a transmission electron microscope (TEM, Phillips CM 120).

Bench-scale Pulse-Energized ESP
The bench-scale ESP is schematically shown in Fig. 1.It had a wire-cylinder geometry and could be energized with DC or DC coupled with pulsed voltages.The geometrical configuration was similar to those reported in other works (Kulkarni et al., 2002;Li et al., 2009).The collecting electrode of the ESPs was made of stainless steel (L = 150 mm, I.D. = 50 mm).It was enclosed on the outside by a polypropylene tube (L = 250 mm, I.D. = 53 mm, O.D. = 69 mm).The discharge (corona emission) electrode with a diameter of 0.4 mm was also made of stainless steel.The excess length of the discharge electrode was sheathed using an insulating material (Teflon ® ) to prevent uncontrolled discharges.
The high voltages were applied across the collecting and discharge electrodes with special designed generators.High DC voltages could be adjusted in the range of 0 to ± 10 kV.In the pulsed operating regime a sequence of high voltage pulses was superimposed on DC voltage.The adjustable high voltage pulses of 0 to ± 10 kV had a repetition rate of 50 pulses per second (pps).The high voltage was measured with a high voltage probe (Tektronix P6015 A) and an oscilloscope (Tektronix TDS 3034 B).The current waveforms were recorded by means of a current shunt.
A typical experimental protocol for capture efficiency measurements consisted of: (1) starting the aerosol preparation system, ensuring stable aerosol generation (controlled by the particulate matter concentration monitor), (2) applying high voltage supply, DC (± 10 kV) or high voltage pulses train (± 5.5 kV) superimposed on a DC (± 10 kV) base, (3) introducing the flowing gas to the ESP, (4) flowing the aerosol through the ESP, and (5) measuring the size distribution of the particles exiting the ESP.

Online Particle Size and Concentration Measurements
The particle size and concentration measurements were alternatively performed at the outlets of the aerosol preparation system and the bench-scale ESP, by electrical mobility and gravimetric analyses.An SMPS consisting of a differential mobility analyzer (DMA, Grimm 4700) and a Faraday cup electrometer (FCE, Grimm 5705) was used to conduct electrical mobility analysis.The DMA has a measurable size range of 4 to 250 nm and the FCE is capable of detecting the particle number concentration in the range of 6 × 10 2 to 10 8 #/cm 3 .Gravimetric analysis was performed with a low pressure impactor (LPI, Dekati Ltd.), which has 13-stages and can classify airborne particles from 30 nm up to 7 µm.Since the concentration of the particles with small sizes (< 200 nm) was very low, an accurate evaluation by the LPI required a long sampling time (8 hrs).The collecting and discharge electrodes of the benchscale ESP were thoroughly cleaned every 2 hours to keep the surfaces clear of deposits.The blower blades from the aerosol preparation system were also cleaned.The same methodology was used to measure particles at the outlet of the ESP.The measurement of size distribution by the SMPS was performed by averaging 8 successive measurements taken every 2 hours: four on clean ESP, and the other four after 2 hours of full operation.For all SMPS and LPI measurements, three runs were performed for each condition and the average value was used as the final data.The measurement error was well below 5%.Throughout the experiments the aerosol flow rate was 10 l/min.

Particle Characteristics
The characteristics of the particles released from the iron and steel smelter were determined.The particles were collected upstream of an industrial ESP in the smelter.The ESP had a flow rate of 610,200 m 3 /h, and gas temperature of 110°C, with a capture efficiency of 98% for particles less than 2.5 µm.The typical size distribution of the particles collected upstream of the ESP is shown in Fig. 2. The particles ranged from 0.2 to 100 µm, most of them being in the range of 0.4 to 20 µm, with a maximum at 4 µm.The corresponding SEM and HRTEM images are shown as an inset in Fig. 2, in which mainly agglomerated micrometer and nano-sized particles were observed.
The particle composition and crystallinity of the particles collected upstream of the industrial ESP was identified by their XRD pattern (Fig. 3).The majority of the species consists of a mixture of iron oxides (hematite, maghemite,  From Table 1, iron oxides and silicates were clearly observed.In addition, other chemical compounds, such as carbides, alumina, calcium oxides, sulphates, and sodium salts, were also detected, indicating that the particles were a complex mixture.

ESP Voltage-current Characteristics
The voltage-current (V-I) characteristics of the bench-scale ESP were investigated under various conditions.The measurements were carried out with the ESP energized with DC high voltages (± 10 kV) and DC high voltages with superimposed pulses (± 5.5 kV).The applied voltage pulse had a rise time of 60 ns (measured at 10 to 90% of maximum) and a duration of about 1.5 µs (measured at half magnitude).The main corona current had a duration of 75 ns (measured at half magnitude), and a rise time of ~33 ns (measured at 10 to 90%).The same values were measured for both polarities (±).The V-I measurements were performed on the ESPs, with either normal clean air or flowing aerosol (dust loaded) (Fig. 4).The voltage of 10 kV was chosen since it provides an optimal regime for capture efficiency.Fig. 4(a) presents typical exponential graphs of the discharge currents with DC voltage for clean/dust loaded ESPs.The shape of the V-I curve indicates a low to medium dust resistivity.As seen in Fig. 4(a) the slope of the DC discharge currents is nearly the same for all voltages higher than 9 kV.The curves corresponding to the dust loaded ESP move to higher currents (left shift to the reference curves, with no dust loaded).The shift of the negative DC voltage curve is much stronger than those for the positive ones; this is associated with a higher capture efficiency observed for negative DC, than for positive DC energization.In the present case, there was no back-corona phenomenon.The left shift could be attributed to the increase of the discharge electrode emissivity, caused by dust deposition on the wire surface.This increased the local field, compensating the field reduction due to the particle space charge (Bacchiega et al., 2004;2006).Fig. 4(b) shows the dependence of the peak current on the pulsed voltage (clean ESP, no aerosol).The results indicate a linear increase of pulsed discharge current, (no DC voltages).This current-voltage dependence could be attributed to the ESP operation in the pre-corona inception region for very short pulses application (corona inception voltage is much higher compared to that of DC conventional energization).It is important to note that the pulse duration is considerably shorter than the transient time for the motion of the ions to the collecting plate.Fig. 4(c) shows the same dependence, with a voltage of 10 kV superimposed with a pulse.In this case, the curves fit very well a linear function.Also, contrasting the DC corona discharges, the positive pulsed discharge currents are much higher than the negative pulsed discharge currents for both, no dust and dust loading cases.This may be explained by the peculiarities of the positive short-pulse energization.The positive short-pulse energization of ESPs develops both streamers and corona discharges (Deuxan et al., 2003).The positive streamer discharges have high ionizability and propagate rapidly from discharging electrode to collector electrode with the aid of ionizing by light.Also, the positive pulsed streamer discharges produce a significant free electron charging that is an important process for charging particles in positive short-pulse corona discharges and leads to a significant increasing of discharge current shown by V-I characteristics from Fig. 4(b).The current values in DC and pulsed modes are summarized in Table 2.  99.20 89.00 -10 kV & -5.5 kV pulse 0.8-0.9A (pulse peak value) 99.75 99.20

ESP Capture Efficiency
The number capture efficiency is more appropriate for small particles and was used in this work (Zukeran et al., 1999a).

 
1 100 where η N is the number based capture efficiency (%), N o and N I are the number concentrations of particles (#/cm 3 ) at the outlet and inlet of the ESP, respectively.The number concentrations of the particles were analyzed by using the LPI and SMPS.

LPI Analysis
The particle size distribution and number concentration were first evaluated by the low pressure impactor.The measurements at the stages corresponding to 28, 56, 94 and 155 nm have high uncertainties, due to the difficulties in weight measurement of very small amounts of powder collected on these stages.Therefore, for reliable analysis particles larger than 155 nm were mainly used.A precise analysis of particles in the range of ≤ 155 nm was done using the SMPS as referred in the next section.
Fig. 5(a) shows the size distribution of the aerosol at the outlet of the aerosol preparation system, i.e. the upstream of the ESP.The LPI analysis clearly shows a polydisperse particle size distribution with the number concentration varying from 10 5 #/cm 3 at d p = 90 nm to 10 3 #/cm 3 for d p > 2.5 µm.The results indicated that most particles have sizes below 2.5 µm.The ESP capture efficiencies evaluated by the LPI are shown in Fig. 5(b).The capture efficiencies are higher for negative DC voltages.For example, at 400 nm and 4 µm the capture efficiencies with negative voltages were 99.83% and 99.98%, respectively.For the positive applied voltages, the efficiencies were 99.40% and 99.92%, respectively.Since in negative corona there are excess negative ions that have a higher electrical mobility than positive ions, the negative corona results in more effective diffusion charging in comparison to the positive case, leading to a better capture performance (Kulkarni et al., 2002).From Fig. 5(b), it is also noted that the DC energization is better than the pulsed one, almost in all ranges.However in this range the DC energization is only apparently better than the pulsed one because it is possible the re-entrainment of particles from collection electrode (CE) by dissociation of agglomerated particles due to the corona discharges or/and due to the back corona discharges developed on the inner surface of the CE (Zukeran et al., 1999a, b).These phenomena occurred due to the fact that for measurements with LPI the ESP operated continuously 2 hours, without cleaning, and a thick dust layer was built up at the collecting surface; consequently the back corona discharges were possible to be promoted.Also, it is reported that an abnormal dust re-entrainment occurs because the low resistivity dust acquires the opposite charge by induction charging and pulled back to space, jumping away (Mizuno, 2004).

SMPS Analysis
As mentioned above, the previous measurements could not offer reliable results in the nanometer size range for several reasons: (a) very few measurement points in this range; (b) inaccuracy of the measured mass for the collected powder in the stages of the impactor (28, 56, 94, and 155 nm); (c) the problem of obtaining sharp cut-off diameters in the last stages.For a precise evaluation of the ESP behavior in the submicrometer to nanometer size range, the differential mobility analysis is useful.
The measurements were performed in the range of 4 to 243 nm.The results of the measurements on the aerosol size distribution at the downstream of the ESP (aerosol preparing system outlet) are shown in Fig. 6(a).The particle concentration was approximately 10 2 to 10 3 #/cm 3 for particles of 4 to 20 nm, and 10 3 to 10 5 #/cm 3 for the rest, with the exception of two regions 55 to 70 nm and 140 to 210 nm, where the concentration was very low (at the noise level of the SMPS).It should be noted that along experiments the broadness and position of these two regions fluctuated by ± 20%.
An objective of the study was to determine the optimal value of the voltage and polarity, especially for the capture of fine particles.Fig. 6(b) shows the graphs of the size distributions in DC and pulsed regime, with both polarities.Fig. 6(b) reveals a striking phenomenon: in positive pulsed mode, for sizes < 45 nm, the particle concentration is by three orders of magnitude higher than at the ESP inlet!In order to avoid any accidental errors, the measurements were repeated several times for both upstream and downstream of the ESP.This issue could be related to highly charged aerosols existing in pulsed mode at the downstream of the ESP.It is well known that a deviation from an equilibrium charge distribution of particles which are going to be classified, could lead to errors in electrical mobility spectrometers.In the literature numerous papers studied the charge distribution dependence on the neutralizer activity, residence time of the particles in the neutralizer and particle concentration (Coven et al. 1997;Ji et al., 2004;Barbounis et al., 2008).Therefore, to get reliable results, a low aerosol flow rate was used (higher residence time).On the other hand, however, to get a high particle concentration at the ESP outlet a high aerosol flow rate is required.Consequently, as a trade-off, a reasonable sample flow rate of 5 L/min was selected with a neutralizer AM-241 alpha-source of 3.7 MBq.However the above mentioned anomalous behavior still appeared under these conditions, due to the mobility of the neutralizer to bring charges down to the Boltzmann level.A few possible reasons may explain the phenomenon: (a) abnormal nanoparticle re-entrainment from collection electrode by dissociation of agglomerated particles due to the very intense pulse corona discharges when the dust resistivity becomes low (Zukeran et al., 1999b;Mizuno, 2004), (b) and/or due to a large number of ions escaping the ESP and entering into the particle measurement device (SMPS), resulting in new particle formation due to ion-ion nucleation (Li et al., 2009), (c) nanoparticles initiated and generated by evaporation of the discharge electrode material in high intensity plasma produced by strong current pulses.The evaporation phenomenon was demonstrated by additional experiments carried out on clean ESP and cleans hosepipes, with a "zero test filter" at the inlet, and energized with high voltage pulses superimposed on DC voltage.The measurements revealed at the outlet of ESP a high amount of nanoparticles in the range of 4 to 20 nm.Finally, the capture efficiency dependence on particle sizes and energization modes is presented in Fig. 7. From Fig. 6(a), it was noted that the inlet aerosol had particle concentrations above the SMPS noise limit, in three regions.Therefore the capture efficiency graphs were drawn separately for 18 to 52, 80 to 120 and 220 to 243 nm.In the range of 18 to 52 nm (Fig. 7(a)), the negative pulsed mode is remarkably efficient (93.7 to 99.4%) in capture of small particles.In DC negative mode there was an abnormal overwhelming increase of the particle concentration at the outlet and therefore the capture efficiency could not be plotted.An obvious important conclusion is that the negative pulsed mode is appropriate for capture of particles of sizes between 18 to 50 nm.Below 18 nm, no conclusion can be made.In the range of 80 to 120 nm, both DC and pulse modes showed very high capture performance (Fig. 7(b)).Similar to the LPI results, the negative DC mode also showed better performance than the pulse mode.Such an unexpected result could be induced by particles re-entrainment from collection electrode by dissociation of agglomerated particles due to the very intense pulse corona discharges.Also, for the same reason, the positive pulse mode is comparable with the positive DC mode, and even for the particles larger than 90 nm.In general the negative DC bias is more efficient than the positive one.The difference is particularly pronounced in the range of 220 to 243 nm (95 to 99.56% in comparison with 66 to 98%) (Fig. 7(c)).A comparison between LPI and SMPS becomes possible for the particles in range of 220 to 243 nm.Table 2 summarizes the results of capture efficiencies for the particle size of 230 nm by the two methods.The LPI method indicated higher capture efficiencies in comparison to the SMPS, particularly in positive modes.

CONCLUSIONS
A bench-scale pulse-energized ESP was developed for the capture of particles in exhaust gases of iron and steel factories.A detailed particle characterization indicated that they consisted mostly of iron oxides, and other Ca, S, C, Si, Al and Na compounds.The sizes of the particles ranged from 5 nm to several micrometers.The capture efficiency of the ESP was analyzed with a ± 10 kV DC voltage, and ± 10 kV DC voltage with superimposed pulses of ± 5.5 kV using LPI and SMPS.Both methods demonstrated that the negative DC energization resulted in a better performance, due to the generation of negative corona in which the negative ions have a higher electrical mobility than the positive ones, leading to a better capture efficiency.In general, the DC mode is better than the pulse mode.However, in the case of particles smaller than 100 nm, the pulse mode also resulted in capture efficiencies exceeding 99%.In the case of measurements by the SMPS, the study revealed an anomalous phenomenon in the range < 50 nm for pulsed positive mode and negative modes (higher particle concentrations downstream of the ESP than upstream).It was attributed to the surplus of charge introduced in the SMPS which could not be neutralized completely, or due to the particle re-entrainment, or discharge electrode evaporation due to strong pulsed currents.Additional future studies must be performed in the range < 50 nm, in order to further understand and elucidate the physical processes in investigations of ESPs by SMPS.

Fig. 1 .
Fig. 1.Schematic diagram of the experimental setup for laboratory scale studies of capture of iron-steel smelter dust in a pulse energized ESP.

Fig. 2 .
Fig. 2. Typical size distribution and SEM image of the powder collected upstream of the industrial ESP of an iron and steel factory.

Fig. 3 .
Fig. 3.The XRD pattern of the particles collected upstream of the industrial ESP.

Fig. 5 .
Fig. 5. (a) Size distributions at the inlet of the ESP measured by the LPI, (b) calculated capture efficiencies for different size particles.

Fig. 6 .
Fig. 6.The size distribution at the inlet (a) and outlet (b) of the ESP measured by the SMPS.

Table 1 .
XPS analysis results of dust collected from an iron-steel smelter.

Table 2 .
Summary of experimental conditions for laboratory scale studies.