Research Article 1

Forced triboelectrification of fine powders in particle wall colli- 2

sions 3

Leonard Hansen 1*‡, Martin Weers 1‡, Annett Wollmann1 and Alfred P. Weber 1 4

1
Institute of Particle Technology, TU-Clausthal; tower@mvt.tu- 5

clausthal.de 6

* Correspondence: leonard.hansen@tu-clausthal.de; 7
‡ Both authors contributed equally to this work 8

Abstract: 9

Triboelectric separation as an inexpensive and environmentally friendly technique could contrib- 10

ute to material-specific sorting. However, the application as a widespread method is limited due 11

to the complexity of the process. In particle-wall collisions, various parameters like collision en- 12

ergy and angle, work function of the contact partners, humidity, surface roughness, etc. influence 13

the particle charging in a hardly predictable way. This study investigates the possibilities of 14

forced triboelectric particle charging by applying an electrical potential to the metal contact part- 15

ner (copper/steel pipe). Variations include different pipe lengths (0.5 m to 3 m), particle material, 16

and particle sizes for limestone. A distinction is made between the net charge of the particles and 17

the positive, negative and neutral mass fractions. The work functions of the investigated materi- 18

als vary from about 3.2 eV to >8.5 eV for glass, limestone, artificial slag, and lithium aluminate 19
Citation: Lastname, F.; Lastname, F.; particles. With the applied high voltage potential, the particle net charge 20
Lastname, F. Title. Minerals 2021, 11,
can be shifted linearly. For limestone, it is shown that the neutral fraction 21
x. https://doi.org/10.3390/xxxxx
is highest at the Point of Zero Net Charge (PZNC). This observation may 22
Academic Editor: Firstname Last- identify an approach for the material selective separation of one target 23
name
component from a multi-material mixture. 24

Received: date
Accepted: date Keywords: triboelectric, bipolar charge, sorting, minerals, fine powders, 25
Published: date recycling 26

27
Publisher’s Note: MDPI stays neu-
tral with regard to jurisdictional
claims in published maps and institu-
tional affiliations. 1. Introduction 28

Triboelectric separation is industrially employed in mineral processing 29

Copyright: © 2021 by the authors.
Submitted for possible open access
publication under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/license
s/by/4.0/).
mostly for separating two components from each other such as feldspar 30
from quartz [1], carbon from ash-coal [2], or calcite from quartz [3]. The 31
benefits of easy operation, low energy consumption, and no uptake of wa- 32
ter and chemicals are offset by the challenges regarding a limited under- 33
standing of the particle charging behavior where standard triboelectric se- 34
ries can only be used as rough orientation [4]. For Engineered Artificial 35

Minerals 2021, 11, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/minerals

Minerals 2021, 11, x FOR PEER REVIEW 2 of 22

Minerals (EnAM) particles, i.e. slags encountered in the recycling processes within circular econ- 36
omy, this challenge becomes even more demanding due to the broad mix of components (e.g. in 37
the recycling of lithium-ion batteries). Therefore, an improved understanding of the triboelectric 38
charging and specific manipulation of the particle charge is needed for an improved separation 39
process. 40
Triboelectric charging (or frictional electrification) is used in a broad sense to account for the 41
transfer of charge when two materials come in contact (as charge separation or charge transfer) 42
[5]. While in metal-metal contacts, electrons are transferred according to the work functions of the 43
contacting metals from the partner with the lower to the partner with the higher work function, 44
the situation for the contact of insulators with other insulators or metals is less clear. Here the 45
surface structure of a particle is of paramount importance, since the surface electronic states may 46
only be generated by defects or impurities, which determine the so-called effective work function. 47
Among the different concepts for electron transfer [6], only the condenser model will be discussed 48
here since it relates directly to the approach taken in this study [7]. It states that the charge trans- 49
ferred between two surfaces is given by: 50

𝜀0 ∙𝜀𝑟 ∙𝑆 Φ1 −Φ2
∆𝑞 = 𝑘𝐶 ∙ 𝐶 ∙ ∆𝑉 = 𝑘𝐶 ∙ ∙ ∆𝑉 = 𝑘𝐶 ∙ 𝐶 ∙ ( + 𝑉𝑒𝑥 ) 51
𝑧0 𝑒
(1) 52

where kC is the charging efficiency (depending on the collision energy, collision angle, defor- 53
mation, etc.), C the capacity, ε0 the electrical permittivity of the vacuum, εr the relative permittivity 54
of the gas, S the contact area, and z0 the critical distance over which electrons can be exchanged. 55
Neglecting pre-charges and space charge effects, the potential difference ΔV contains the differ- 56
ence of the effective work functions (Φ1-Φ2) and an externally applied electric potential Vex. In this 57
study this external potential is varied to control the particle charging. A similar model was used 58
by Ireland, calling it Ohmic charging model [8]. 59
The evaluation of Eq. 1 requires the work functions of both contact partners. Several attempts 60
have been made to order the work functions of different materials, resulting in the so-called tribo- 61
electric series. However, various, partially contradictory series were reported. Part of these dis- 62
crepancies could be due to different surface roughnesses of the contact partners, as recently high- 63
lighted by Zou et al. [4]. They eliminated the roughness influence by realizing conforming contacts 64
using a liquid metal so that the intrinsic character of polymers for gaining or losing electrons was 65
revealed [4]. For the used contact partners, the work function values differ within the literature, 66
depending on the measurement technique and the origin of the sample (cf. Figure 1Figure 1). 67
Nevertheless, they should provide a rough estimate of the charging behavior. The lower part of 68
Figure 1Figure 1 shows the work function of the metal-contact partner used in the experimental 69
setup, i.e. coiled pipes of copper or steel, whereas the upper part shows the work functions for the 70
particle materials. Since the surface of copper is oxidized in air, the values for CuO and Cu2O are 71
also given (all copper containing species are indicated in green). 72
Minerals 2021, 11, x FOR PEER REVIEW 3 of 22

73

Figure 1: Overview over the different work functions of the used materials: Cu [9]; Stainless Steel 74

[10],[11]; Al2O3 [9][12]; CaCO3 [13][14]; SiO2 [9][14]; LiAlO2 (cf. Appendix A); Li2O [15][16] (Copper 75

containing species are indicated in green, self measured LiALO2 is indicated in grey with white 76

stripes) 77

As mentioned above, triboelectric charging depends on a large variety of influencing param- 78

eters related to the materials used (effective work functions, humidity, adsorbates, chemical sur- 79
face modifications, etc.) and on the way of contacting the two surfaces (number, intensity, and 80
duration of contacts) which themselves may depend on the environmental conditions such as rel- 81
ative humidity [17],[18],[19] and temperature [14]. 82
Essential for implementing an effective triboelectric sorting process with high separation 83
sharpness is to minimize the mobility overlap, i.e. the overlap of electrophoretic migration velocity 84
of the different components. As the electrical mobility of the particles depends on their size and 85
their charge, broad distributions in both parameters have to be avoided. In particular, bipolar 86
charge distributions for particles of the same material make it impossible to achieve high-quality 87
sorting. Many authors found such bipolar charging (e.g. [19],[20],[21]) and it was also observed in 88
our investigations for the triboelectric charging of calcite and talcum particles [22]. Several ap- 89
proaches were made to explain the bipolar contact charging of particles of the same material rang- 90
ing from surface curvature effects on the work function [23] to the heterogeneity of the particle 91
surface [24] and the material transfer during collisions [25]. 92
But even for powders exhibiting only one charge polarity, the phenomenon of equal mobility 93
can occur, hampering the sorting of different materials. That means, e.g. a bigger particle with a 94
higher electric charge can have the same mobility as a small particle with a lower electric charge, 95
i.e. both particles migrate with the same velocity in an electric field, suppressing any (material- 96
specific) sorting effect. While this phenomenon already represents a challenge for a two-compo- 97
nent powder system, it becomes nearly unsolvable for systems composed of a large variety of 98
different materials. The only fraction for which the equal-mobility problem and, in addition, the 99
Minerals 2021, 11, x FOR PEER REVIEW 4 of 22

bipolarity problem does not play a role is the neutral powder fraction of the target material. There- 100
fore, the objective of this work is to find triboelectric charging conditions, which move the highest 101
share of the target powder into the neutral state. 102
While the understanding of the origin of the bipolar charge distribution is still under debate, 103
a promising approach to shift the complete charge distribution was presented by Linsenbühler 104
[26] as well as Oberrauner et al. [27]. They investigated the influence of high voltage applied to 105
the contacting surfaces [26],[27]. The applied high voltage influences the average particle polarity. 106
Linsenbühlers [26] results resemble an (inverse) zeta-potential curve, showing an isoelectric point. 107
In contrast, Oberrauner et al. used an electric field between a PE-covered rotor (grounded) and 108
also covered stator (potential) as a tribocharging unit, varying the relative humidity from 20 % up 109
to 70 % and the applied stator potential in a range of -6 kV to 6 kV. As humidity increases, reflect- 110
ing higher particle surface conductivity and gas conductivity, the accumulated particle charge 111
decreases. Therefore, a lower potential is required to achieve a zero net charge of the used calcite 112
particles. Modeling results of Shen [28] show that the kinetics of electron reflux is slow, and fast 113
separation of the contact partners results in a non-equilibrium state, where charges remain on the 114
particles' surface. Increasing moisture content may result in a faster electron backflow kinetic, 115
which was also underlaid by experimental results by Shen [28]. Decreasing charge density is cor- 116
related with increased relative humidity. Wang [29] investigated the contact charge transfer of 117
glass material with different surface roughness. It was found that in each case, the smoothest con- 118
tact partner is more positively charged, whereas a higher roughness results in a negative net 119
charge (glass–glass pairing). That indicates an electron transfer direction from the smoothest to 120
the roughest material and might be the matter for the observed bipolar charging of identical ma- 121
terials. 122
Both investigations by Linsenbühler [26] and Oberrauner [27] provide persuading arguments 123
to implement an external high voltage to adjust the particle charge and improve the triboelectric 124
sorting process. In this work, non-conductive particles of different work functions are conveyed 125
and specifically charged by helices subjected to high voltage. Special attention is also paid to the 126
resulting influence on bipolar charging. 127

2. Materials and Methods 128

To get an overview of the experiments, this chapter introduces the used materials first. The 129
experimental setup for the triboelectric charging by a conducting coil and the electrostatic separa- 130
tion for a more detailed investigation are described afterwards. 131

2.1. Particulate materials 132

The particle materials used in the experiments with their main components are listed in Table 133
1Table 1. All particles except the glass spheres were produced by grinding. 134

Table 1: Table with properties of the used materials 135

Material Supplier Components

Saxolith 8 Sh minerals GmbH CaCO3
Saxolith 40 Sh minerals GmbH CaCO3
Saxolith 70 Sh minerals GmbH CaCO3
Lithiumaluminate Sigma Aldrich LiAlO2
Minerals 2021, 11, x FOR PEER REVIEW 5 of 22

1
EnAM slag Insitute of metallurgy - 2X2SiO4 (larnit), (for-
RWTH Aachen sterit), CaAlSiO4 (melilit
solid solution) and
LiAlO2
Glass spheres Samore GmbH SiO2
Spodumene LiAl(SiO3)2 Börkley Keratech GmbH 65% SiO2, 26% Al2O3,
and 7% Li2O (supplier
information)
1
Engineered Artificial Minerals, X may be calcium or magnesium
2
136
In addition to pure minerals such as Saxolith or glass, complex material mixtures (spodumene 137
or EnAM slag) are also used in the experiments concerning future slag materials from recycling 138
processes. 139
The cumulative particle size distributions of all materials were measured with laser diffraction 140
(HELOS®, Sympatec GmbH) using dry dispersion (RODOS®, Sympatec GmbH) and are given in 141
Figure 2Figure 2. The limestone is named Saxolith (or Sax) 2, 8, 40, and 70 and the corresponding 142
measured medians are 3.5 µm, 7.5 µm, 50 µm, and 60 µm respectively. These medians are used 143
for all further calculations. 144

145

Figure 2: Cumulative particle size distributions (Q3) of the used materials obtained with laser 146

diffraction 147

2.2 Methods 148

Figure 3Figure 3 shows an overview of the experimental setups used to investigate the ma- 149
nipulation of the triboelectric charge. A screw conveyor feeds the particles in an electrically 150
grounded dispersing unit (RODOS® – Sympatec), based on the injector principle and operated 151
Minerals 2021, 11, x FOR PEER REVIEW 6 of 22

with 6 m3/h dried air as dispersing gas. The relative humidity in the setup was kept constant at 152
5% for all experiments. After the dispersion, the particles are passed through an insulating tube 153
into a metal helix, connected to a high voltage (HV) source, adjustable from -30 kV up to +30 kV. 154
Due to breakdowns at -30 kV, the potential was varied between -25 kV and +30 kV. The helix 155
length can be varied from 0.5 m to 3 m. As helix material, copper and stainless steel were used. 156
The inner tube diameter is 10 mm, and the bending radius is 15 cm. 157
A partial flow is sampled for the FCE and the electrostatic precipitator. All experimental re- 158
sults are mean values of double determinations. 159
160

161

162

Figure 3: Experimental setup – The particles are suspended in a carrier gas (dried air) by a 163

dispersing unit and charged by a copper or stainless steel helix to analyze the 1.) Net charge and 164

2.) Charged and neutral fractions of the particles via an electrostatic precipitator 165

2.2.1 Net charge measurements 166

After the charging process, the net charge is measured with a Faraday cup electrometer (FCE). 167
An electrometer (Keithley, model 6514) is connected to an isolated inner Faraday cup containing 168
a particle filter holder enclosed by an outer Faraday cup to reduce external interference with elec- 169
tromagnetic radiation. The charge per particle mass (µC/g) was determined from the measured 170
current and the known particle mass flow rate. 171

2.2.2 Electrostatic separation 172

An electrostatic precipitator with a turbulent flow (Re = 11700) as described in Hansen et al. 173
[20], was used to quantify the charged and neutral fractions. The applied electric field was 174
2 kV/cm. The charged particles were deposited on the electrodes due to their high electrical mo- 175
bility, and the neutral particles were led into a cyclone at the end of the separator. In order to 176
Minerals 2021, 11, x FOR PEER REVIEW 7 of 22

assure complete separation of the particles, the separation efficiency curve was determined in pre- 177
liminary experiments. The charged fractions are analyzed offline by pulling out the coated elec- 178
trodes and sucking the particles onto filters for weighing. 179

3. Results 180

To get a rapid overview of the sensitivity of the triboelectric charging on different parameters 181
(helix-length, particle size, particle material), the net charge of the powders was measured by an 182
FCE as a function of the applied voltage. Finally, the electrostatic precipitator was used for more 183
detailed investigations of the positive, negative, and neutral mass fractions for one material 184
(Sax 8). 185

3.1 Net charge - length dependency 186

Eq. 1 suggests that applying an external electric potential to one contact partner can influence 187
the triboelectric charging process. Figure 4Figure 4 shows this behavior for limestone particles 188
(Saxolith 8) by changing the applied potential on the copper helix. The points where 0 kV is ap- 189
plied (grounded helix) represent pure triboelectric charging. For negative potentials and positive 190
potentials up to 15 kV, the particles are charged negatively. At 15 kV, the particle charge switches 191
to positive (cf. Figure 4Figure 4a). This point is defined as Point of Zero Net Charge (PZNC) in 192
the following. The PZNC varies slightly for the different helix lengths, whereas the slope rises 193
significantly with increasing helix length. Due to more impacts in a longer helix, the charging 194
efficiency rises, explaining the increase of the slope. As shown in the model of Matsusaka [7], the 195
slope depends on the capacity (C) and charging efficiency (kc). The PZNC, however, remains vir- 196
tually constant for a specific material as expected from Eq. 1. As Matsusaka pointed out, the par- 197
ticle impact frequency in a pipe remains constant, hence the number of impacts is proportional to 198
the pipe length. [7]. With Eq. 2, the obtained charge growth in a pipe with the length L and the 199
characteristic length of particle charging L0 can be calculated out of the fit (see table in Figure 200
4Figure 4b) [7]. In the present case, the equation was modified taking the applied external poten- 201
tial (Vex) into account. The initial particle charge without helix was negligible (e.g. 0.3 µC/g for 202
Sax 8). 203
204
𝐿 𝑆𝑒𝑓𝑓 𝐿
∆𝑞̅𝑚 = 𝑞̅𝑚∞ ∙ {1 − 𝑒𝑥𝑝 (− 𝐿 )} , 𝑤𝑖𝑡ℎ 𝑞̅𝑚∞ ~ ∙ ∆𝑉 𝑎𝑛𝑑 𝑆𝑒𝑓𝑓 ~ 𝐿 205
0 𝑧0 0
(2) 206
207
where the maximum charge is 𝑞̅𝑚∞ , the critical gap between particle and wall is z 0 and Seff is 208
the effective contact surface of the particle. 209
After a certain number of collisions, the particles cannot acquire additional charges due to 210
their limited capacity (Seff = SParticle). Thus, Eq. 2 generates a suitable fit for the measured data points 211
in Figure 4Figure 4b. 212
Minerals 2021, 11, x FOR PEER REVIEW 8 of 22

213

214
Minerals 2021, 11, x FOR PEER REVIEW 9 of 22

Figure 4: Specific net-charge for different lengths and voltages (0.5 m to 3 m) for limestone 215

(Saxolith 8) with a) constant lengths and b) constant voltages 216

It is interesting to note, that in the original publication of Matsusaka ΔV (in Eq. 2) was only 217
considering the difference of the work functions. However, Figure 4Figure 4b shows, that the 218
proportionality between 𝑞̅𝑚∞ and ΔV also holds true when including an externally applied po- 219
tential to one of the contacting partners. This observation confirms the applicability of the plate 220
condenser concept for an externally applied potential and extends the range of triboelectric parti- 221
cle charging (also mentioned by Matsusaka [30]) as further outlined below. The maximum value 222
of the specific charge is changing by the applied voltage (insert Figure 4Figure 4b), resulting in 223
the same dependence of the applied voltage like the net charge but with a steeper slope. The PZNC 224
(Figure 4Figure 4a) and the characteristic length of particle charging (table in Figure 4Figure 4b) 225
changes for +20 kV. 226

3.2 Net charge – size dependence 227

In addition to the number of contacts (e.g. length of the helix), the collision intensity for dif- 228
ferent particle sizes may affect the extent of charging due to their different collision kinetics. 229
Hence, the net charge of different limestone fractions (Saxolith 2, 8, 40, and 70) was measured as 230
a function of the applied potential at constant helix length and volume flow rate (cf. Figure 5Figure 231
5a). All Saxolith materials originate from the same open pit area, so that the materials hardly vary 232
neither in composition nor in morphology. 233
Minerals 2021, 11, x FOR PEER REVIEW 10 of 22

234

235
Minerals 2021, 11, x FOR PEER REVIEW 11 of 22

Figure 5: a) Specific net charge per mass for different sizes of Saxolith (limestone) and varied po- 236

tential at the copper helix; insert: zoom for Saxolith 2 particles; b) Dimensionless deposition veloc- 237

ity versus dimensionless relaxation time for turbulent deposition, calculated for Saxolith 2, 8, 40, 238

and 70 particles, and compared to literature data [31],[32] (calculation in Appendix B) 239

The slope of the mass specific net charge vs. applied potential is found to be nearly constant 240
throughout the particle sizes of Sax 8, Sax 40 and Sax 70, while Sax 2 shows a different behavior 241
as discussed below. To understand the constant slope, the particle wall collision is assumed as a 242
single Hertzian contact of a spherical particle. Then the charge transferred between an isolating 243
particle and a conducting wall (charge relaxation time τel >> contact duration ∆t) in a single colli- 244
sion can be calculated in the following way [33], [34], [35]. 245
246
𝜀0 𝜀𝑟 𝑆 ∆𝑡
∆𝑞𝑠𝑖𝑛𝑔𝑙𝑒 = ∆𝑉 247
𝑧0 𝜏𝑒𝑙
4
𝑑𝑝 2 1−𝜈𝑝 2 1−𝜈𝑤 2 5
𝑤𝑖𝑡ℎ: ∆𝑡~ 𝑢 𝛼 ; 𝑆~𝑑𝑝 𝛼 ; 𝜏𝑒𝑙 = 𝜀0 𝜀𝑟,𝑝 ∙ 𝛽𝑝 ; 𝛼~ [𝜌𝑝 ( + )] 248
𝑝 𝐸𝑝 𝐸𝑤
(3) 249
250
where νp, νw, Ep, and Ew are the Poisson ratio and the Young modulus of the particle or wall, 251
respectively. Furthermore, z0 is the critical gap between the particle and the wall, up is the particle 252
velocity and S the surface of the particle in contact with the wall. The electrical relaxation time is 253
τel and the electrical resistivity of the particle material is βp. 254
However, the number of collisions must be known to compare the totally transferred charge 255
along the helix (Eq. 1) with the transferred charge in a single collision (Eq. 3). As an approxima- 256
tion, the concept of turbulent deposition in a fully developed turbulent pipe flow can be employed. 257
Here the deposition rate (also called deposition velocity) is calculated from the particle flux to- 258
wards the wall and the particle number concentration outside of the viscous sublayer [36]. Ac- 259
cording to Liu and Agarwal [31], the deposition velocity as a function of the dimensionless aero- 260
dynamic relaxation time (τae+) increases first in a quadratic manner (calculated by Eq. A1 to A3 – 261
Appendix C) and reaches later a plateau (cf. Figure 5Figure 5b). The dimensionless relaxation time 262
is the ratio of the particle stop distance to the characteristic turbulence length. Regarding the tur- 263
bulent flow in the helix, the flow field can be assumed to be fully developed, but becomes more 264
complicated compared to a straight tube since additionally Dean vortices occur. Due to the com- 265
parable curvature of the pipe, Reynolds numbers, and Stokes numbers, the work of Noorani et al. 266
[37] can explain the occurrence of Dean vortices in the applied setup of this work. The simulations 267
reveal that the higher the Stokes number or the tube curvature, the more the particles follow the 268
Dean vortices within the pipe on a macroscopic level but experience inertial effects on a micro- 269
scopic level, leading to wall collisions. 270
However, in a first approximation the situation in the viscous sublayer remains unchanged 271
when comparing straight tubes with helical tubes. Transferred to the limestone particles consid- 272
ered here, Sax 8, Sax 40 and Sax 70 are located on the plateau of a constant dimensionless deposi- 273
tion velocity of approximately 0.1 (Figure 5Figure 5b). Therefore, the experimentally found con- 274
stant ratio of transferred charge and particle mass for particles > 8 µm is a consequence of the same 275
particle collision rate with the wall, which is independent of the particle size in this size range. 276
The smaller particles (Sax 2) have a lower deposition velocity of 0.01, meaning that the particle- 277
Minerals 2021, 11, x FOR PEER REVIEW 12 of 22

wall collisions decrease drastically resulting in a lower particle charge as shown in the insert of 278
Figure 5Figure 5a. For particles larger than 40 µm, it is expected that the deposition velocity drops 279
again as already implied in the behavior of the 70 µm particles. Nevertheless, the observed net 280
charge of 70 µm particles is nearly the same as for 8 µm and 40 µm particles. This result may be 281
due to an accumulation of charged particles in turbulent flows near the walls as reported by Li et 282
al. [35]. On the other side for particles much smaller than 8 µm a reduction of the deposition ve- 283
locity is expected. It needs to be emphasized that the width of the particle size distributions (espe- 284
cially broad for Sax 8) may smear out size depending charging effects to a certain extend. Never- 285
theless, for Sax 2, which exhibits also a broad size distribution, a low net charge is observed pre- 286
sumably due to low deposition velocity. 287
The observed phenomenon provides also a guidance for the range of applicability of the real- 288
ized forced triboelectric charging in a helix. Small particles, typically in the submicron range, will 289
not collide with the wall of the helical tube at all, except for diffusional transport, which is rather 290
slow and is not expected to contribute significantly to the charging. For large particles, i.e. above 291
a few 100 µm, their radial acceleration is rarely sufficient to penetrate the viscous sublayer. These 292
particles are preferentially entrained with the carrier gas in axial direction along the helix. How- 293
ever, with a further increase in size, particles may start to settle and can be transported through 294
rolling and gliding along the tube wall. For this mechanism of triboelectric charging Li et al. [35] 295
provided a conceptual framework. From their investigation, it can be concluded that for particle 296
sizes below 100 µm triboelectrification by rolling contributes only a few percent of the collision 297
charging. 298
299

3.3 Net charge - material dependence 300

The triboelectrification of particles is influenced by the mechanical properties of the contact 301
partners, as well as the (effective) electron work function. Particles of various materials should 302
charge differently under otherwise identical conditions to promote sorting. 303
When a steel helix is utilized, the magnitude of specific charge changes by a factor of about 4 304
for pure triboelectric charging (see Figure 6Figure 6a – y-axis intercept), while the PZNC value 305
remains almost constant. Different intersections with the x-axis (PZNC) only indicate differences 306
in the ∆Φ (cf. Eq. 1), whereas the intersection with the y-axis is a result of a different charging 307
efficiency, combined with a different ∆Φ. The PZNC for both helix materials varies only slightly 308
indicating that the work functions of the two wall materials are close to each other. 309
A change in the charging efficiency (kc) could be due to differences in mechanical properties 310
of the contacting wall like the Poisson ratio and Young modulus of steel (Poisson ratio = 0.3; 311
E = 210 GPa) or copper (Poisson ratio = 0.35; E=125 GPa)). In the case of a Hertzian contact between 312
two smooth components, a number of factors, including the Young’s Modulus, particle density, 313
particle diameter, and impact velocity, determines the contact area. Assuming constant impaction 314
velocities of the limestone particles on copper and steel walls and taking the different material 315
properties into account, the difference in the contact area amounts to only about 5 % (calculated 316
by the formula of Li et al. [35] and with the Poisson ratio and Young modulus of limestone of 317
0.27 GPa and 50 GPa respectively). Therefore, the minor differences in the mechanical behavior of 318
the walls alone cannot explain the observed huge discrepancy in charging behavior (factor of 3). 319
Another factor influencing the electrification in two ways is the surface roughness, since the effec- 320
tive contact area is influenced by it [38]. Roughness measurements of the inner copper and steel 321
Minerals 2021, 11, x FOR PEER REVIEW 13 of 22

pipe surfaces after the experiments have been performed. It was found that the roughnesses of the 322
two materials differ by a factor of about 2 (RZ,copper=3.165 µm; RZ,steel=1.489 µm), which may result 323
in different charging efficiencies and thus different slopes in the forced triboelectrification. In ad- 324
dition it has been shown, that with increasing roughness, the turbulent deposition of finest parti- 325
cles increases, resulting in an upwards shift to higher deposition velocities in Figure 5Figure 5b 326
[39]. 327
Minerals 2021, 11, x FOR PEER REVIEW 14 of 22

328

329

Figure 6: a) Net charge for combinations of limestone with different helix materials (copper and 330

steel); b) Net charges for combinations of different particle materials with copper helix 331
Minerals 2021, 11, x FOR PEER REVIEW 15 of 22

The triboelectric charging of other powders, such as artificial slag, spodumene, lithium alumi- 332
nate (LiAlO2), and glass beads, confirm the material dependence of the PZNC (cf. Figure 6Figure 333
6b). For example, the PZNC of Saxolith 70 is approximately 16 kV, while that of lithium aluminate 334
is roughly 55 kV. The PZNC of the spodumene is between the PZNC of the glass spheres and the 335
lithium aluminate, indicating its material dependence due to the significant SiO2 content 336
(64.7 wt%) in it. Due to a small content of calcium and several minerals, the EnAM-slag has a lower 337
PZNC than the lithium aluminate, but higher than the limestone. These findings point to a repeat- 338
able and predictable material dependence, which is a requirement for sorting procedures. 339
340

3.4 Characterization of the charge distribution 341

The electrostatic precipitator (ESP) was used to investigate the fractions of neutral and 342
charged particles. While charged particles are deposited on the electrodes, the neutral particles 343
flow through the ESP and are collected by a cyclone. According to Hansen et al. [22], 80 % of low 344
charged Talcum particles of similar particle size distribution was deposited in the same ESP, op- 345
erated at identical conditions (mass loading, air flow, and electric field). Therefore, it is assumed 346
that nearly 100 % of the higher charged limestone particles are deposited. 347
The length of the copper helix and applied high voltage were varied to investigate the effect 348
on the mass fractions of neutral and charged particles. In Figure 7Figure 7, the mass fractions of 349
positively charged, negatively charged, and neutral Sax 8 particles for a helix length of 3.0 m are 350
shown for several applied potentials (Figure 7Figure 7b). In addition, the different charge frac- 351
tions without helix are also included in Figure 7Figure 7a and indicated in Figure 8Figure 8 as 352
helix length = 0 m. The highest fraction of neutral particles was found at a potential of +16 kV, i.e. 353
at the PZNC. Starting from the most negative voltage towards the PZNC, the neutral fraction in- 354
creases almost linearly, while the positive and negative fractions decrease in the same manner. 355
Above the PZNC, the positive and negative fractions increase with increasing applied potential, 356
greatly reducing the amount of neutral particles. The natural potential difference of the materials 357
(difference in work function) can be superimposed or negated by the electrical potential. As soon 358
as the PZNC is exceeded, the natural potential is not only balanced, but determined by the poten- 359
tial. Further investigations for other helix lengths and applied potentials are shown in Appendix 360
C. 361
Minerals 2021, 11, x FOR PEER REVIEW 16 of 22

362

Figure 7: Charged and neutral fractions of Saxolith 8 for a) no helix attached, b) 3 m Copper-Helix 363

attached 364

By changing the lengths of the helix, another interesting effect occurs. Without applied poten- 365
tial, the neutral fraction increases with the helix length from 20 % to 46 %, whereas at the PZNC 366
(+16 kV) nearly 70 % was reached. (cf. Figure 8Figure 8). While the amount of charged particles 367
decreases equally for both signs for the pure tribocharging, the fraction of negatively charged par- 368
ticles decreases more than the positively charged particles at the PZNC. 369

370

Figure 8: Charged and neutral mass fractions of the particles with no external potential (0kV) and 371

potential at the PZNC (16kV) of Saxolith 372

The main goal of this investigation was to neutralize the majority of the particles with an ex- 373
ternal potential. At the PZNC this high neutral fraction was generated, indicated by the charge 374
Minerals 2021, 11, x FOR PEER REVIEW 17 of 22

distributions. In future investigations, this mechanism will be applied in a sorting apparatus 375
where the charging process of a mixture of two materials can be influenced before the sorting 376
process, leading to a better enrichment ratio. 377

3.5 Energy consumption 378

A further aspect of interest for an industrial process may would be the energy consumption 379
of the introduced method. For this purpose, the current flowing between the high-voltage power 380
supply and the helix was measured. For Sax 70 particles in a 3 m copper helix with a mass flow of 381
0.213 g/s and an external potential of -30 kV, there is a maximum current flow of approx. -190 nA, 382
ergo a power consumption of 5.7 mW. Applied on the particle mass flow, this results in a specific 383
energy consumption of 26.76 mJ/g. 384
385

4. Conclusion 386

Experiments with the focus of actively influencing the triboelectric charge of powders were 387
conducted. The particles were dispersed by a dry dispersing unit before being led through a coiled 388
copper or steel pipe. The investigated particulate materials were isolators of different effective 389
work functions and various mean diameters, ranging from roughly 3.5 µm to 60 µm. The main 390
objective was to influence the triboelectrification (with an external, electrical potential) so that a 391
high fraction of neutralized particles was obtained. For a multicomponent material sorting pro- 392
cess, the neutral fraction is the only possibility to reduce the problem of equal mobility and to 393
increase the sorting efficiency. In general, triboelectric charging leads to bipolar charge distribu- 394
tions and thus, the target material should ideally be neutral while the tailings are charged and 395
hence easily removable in an electrostatic precipitator. Through the applied voltage, the particle 396
net charge could be adjusted, also resulting in the highest neutral fraction at the Point of Zero Net 397
Charge (PZNC), which is material dependent. This observation can be used to concentrate e.g. 398
Lithiumaluminate in a material sorting process. 399
The investigations of limestone particles showed that the bipolarity could not be fully sup- 400
pressed, but significantly decreased at the PZNC, independent from the pipe length. However the 401
pipe length affects the slope of the relation between net charge vs. applied potential. The restricted 402
suppression of the bipolarity is one limitation in the application of the investigated method but it 403
is also important to notify that the sorting attribute is the elctron work function. Thus if the work 404
function of two materials is close to one another, the separation by the materials triboelectric 405
charge is hardly possible as well as the application of the shown method here. In this case, another 406
sorting process, aiming on other attributes of the materials (density, size, etc.) should be used. 407
For limestone particles lager than about 7 µm (Sax 8, Sax 40, Sax70), the mass specific net 408
charge of the particles was more dependent on the contact partner than on the particle size. For 409
small limestone particles (Sax 2, mean: 3.5 µm) the specific net charge was greatly reduced, pre- 410
sumably due to the low inertia in the turbulent flow. 411
Finally, the comparison of different pipe materials with similar work functions underlined the 412
importance of surface roughness affecting the effective contact area and hence the intensity of 413
charge transfer. 414

Acknowledgment 415
Minerals 2021, 11, x FOR PEER REVIEW 18 of 22

The authors thank the German Research Foundation (DFG) for the financial support of this 416
work within the SPP 2045 priority program (WE 2331/22-2). We acknowledge financial support 417
by Open Access Publishing Fund of Clausthal University of Technology. Furthermore, we want 418
to thank Theresa Rimbach, Mirco Dreeßen, and Silvia Löffelholz (IEC) for their help during the 419
experimental investigations. 420
421

Author Contributions: For research articles with several authors, a short paragraph specifying 422
their individual contributions must be provided. The following statements should be used. 423

Conceptualization, L. Hansen and M. Weers; 424

methodology, L. Hansen and M. Weers; 425

formal analysis, L. Hansen, M. Weers, A. Wollmann; 426

investigation, L. Hansen and M. Weers; 427

resources, A. P. Weber; 428

writing—original draft preparation, L. Hansen and M. Weers.; 429

writing—review and editing, A. Wollmann and A. P. Weber; 430

visualization, L. Hansen and M. Weers; 431

supervision, A. P. Weber; 432

project administration, A. Wollmann; 433

funding acquisition, A. P. Weber. 434

All authors have read and agreed to the published version of the manuscript.” Please turn to 435
the CRediT taxonomy for the term explanation. Authorship must be limited to those who have 436
contributed substantially to the work reported. 437
L. Hansen 438
M. Weers 439
A. Wollmann 440
A. P. Weber 441
Funding: This research was funded by the German Research Foundation (DFG), grant number 442
WE 2331/22-2. The publishing fee of the article was funded by the Open Access Publishing Fund 443
of Clausthal University of Technology. 444
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in 445
the design of the study; in the collection, analyses, or interpretation of data; in the writing of the 446
manuscript, or in the decision to publish the results. 447

Appendix A 448

In literature, values for the electron work function of lithium aluminate (LiAlO2) are not avail- 449
able. Therefore, measurements were conducted using self-produced LiAlO2 nanoparticles. The 450
particles were generated by nebulizing a solution of lithium nitrate and α-aluminum nitrate, fol- 451
lowed by a calcination step at 1000 °C within a tube furnace. The work function of these LiAlO 2 452
particles was above the upper limit of 8.5 eV of the employed home-built measurement system 453
Minerals 2021, 11, x FOR PEER REVIEW 19 of 22

(Aerosol Photoemission Spectroscopy - APES) developed by Röhrbein [40]: Therefore, a work 454
function >8.5 eV is assumed. 455

Appendix B 456

In a turbulent pipe flow, particles are deposited, if they pass the laminar and viscous boundary 457
layer at the wall of the pipe due to their inertia. To determine the deposition rate by Wood [41] 458
(Eq. A3) the (dimensionless) relaxation time is needed (Eq. A1). 459
460
𝜏𝑎𝑒 ∙𝑢 ∗2 𝜌𝑝 ∙𝑑𝑝 2
𝑓 0.316
𝜏𝑎𝑒 + = 𝑤𝑖𝑡ℎ: 𝜏𝑎𝑒 = 𝑢∗ = 𝑢̅ ∙ √8 𝑓= 4 461
𝜈𝑎𝑖𝑟 18𝜂 √𝑅𝑒
(A1) 462
463
The dimensionless relaxation time τae is calculated by the kinematic viscosity of air (νair), the
+
464
shear velocity u* and the relaxation time τae, where η is the dynamic viscosity. The shear velocity 465
itself is calculated by the friction factor (f) and the average air velocity in the pipe (𝑢̅), which cor- 466
responds to the quotient of the wall shear stresses and the fluid density under the square root. It 467
is also possible to simplify this calculation by using the friction factor by Schlichting [42],which is 468
calculated with the fluid Reynolds number (Re). The Schmidt number (Sc) can be calculated by 469
Eq. A2 [43], where the Stokes-Einstein equation gives the mass diffusivity (D) of a particle and CC 470
is the Cunningham correction (CC). 471
472
𝜈 3𝜋𝜂𝑑𝑝 𝜈
𝑆𝑐 = 𝐷 = 473
𝑘𝑇𝐶𝑐
(A2) 474
475
2
− +2
𝑢𝑑 + = 0.057 𝑆𝑐 3 + 4.5 ∙ 10−4 𝜏𝑎𝑒 476
(A3) 477
478

Appendix C 479

480

Figure 9: Charged and neutral fractions of Saxolith 8 for a) 0.5 m Copper-Helix attached, b) 1 m 481

Copper-Helix attached 482

Minerals 2021, 11, x FOR PEER REVIEW 20 of 22

483

References 484

1. Peretti, R.; Serci, A.; Zucca, A. Electrostatic K-feldspar/Na-feldspar and feldspar/quartz 485

separation: Influence of feldspar composition. Miner. Process. Extr. Metall. Rev. 2012, 33, 220– 486

231, doi:10.1080/08827508.2011.563156. 487

2. Dwari, R.K.; Hanumantha Rao, K. Non-coking coal preparation by novel tribo-electrostatic 488

method. Fuel 2008, 87, 3562–3571, doi:10.1016/j.fuel.2008.05.029. 489

3. Bittner, J.D.; Hrach, F.J.; Gasiorowski, S.A.; Canellopoulus, L.A.; Guicherd, H. Triboelectric 490

belt separator for beneficiation of fine minerals. Procedia Eng. 2014, 83, 122–129, 491

doi:10.1016/j.proeng.2014.09.021. 492

4. Zou, H.; Guo, L.; Xue, H.; Zhang, Y.; Shen, X.; Liu, X.; Wang, P.; He, X.; Dai, G.; Jiang, P.; et al. 493

Quantifying and understanding the triboelectric series of inorganic non-metallic materials. 494

Nat. Commun. 2020, 11, 1–7, doi:10.1038/s41467-020-15926-1. 495

5. Kelly, E.G.; Spottiswood, D.J. The theory of electrostatic separations: A review part III. The 496

separation of particles. Miner. Eng. 1989, 2, 337–349, doi:10.1016/0892-6875(89)90003-4. 497

6. Kronik, L.; Shapira, Y. Surface photovoltage phenomena: Theory, experiment, and 498

applications. Surf. Sci. Rep. 1999, 37, 1–206, doi:10.1016/S0167-5729(99)00002-3. 499

7. Matsusaka, S.; Maruyama, H.; Matsuyama, T.; Ghadiri, M. Triboelectric charging of powders: 500

A review. Chem. Eng. Sci. 2010, 65, 5781–5807, doi:10.1016/j.ces.2010.07.005. 501

8. Ireland, P.M. Triboelectrification of particulate flows on surfaces: Part II - Mechanisms and 502

models. Powder Technol. 2010, 198, 199–210, doi:10.1016/j.powtec.2009.11.008. 503

9. Fomenko, V.S.; Samsonov, G. V. HANDBOOK OF THERMIONIC PROPERTIES; 1st ed.; 504

Plenum press data division: New York, 1966; ISBN 9781468472950. 505

10. Wang, X.Y.; Li, D.Y. Mechanical, electrochemical and tribological properties of nano- 506

crystalline surface of 304 stainless steel. Wear 2003, 255, 836–845, doi:10.1016/S0043- 507

1648(03)00055-3. 508

11. Guo, L.; Hua, G.; Yang, B.; Lu, H.; Qiao, L.; Yan, X.; Li, D. Electron work functions of ferrite 509

and austenite phases in a duplex stainless steel and their adhesive forces with AFM silicon 510

probe. Sci. Rep. 2016, 6, 1–7, doi:10.1038/srep20660. 511

12. Mao, B.H.; Crumlin, E.; Tyo, E.C.; Pellin, M.J.; Vajda, S.; Li, Y.; Wang, S.D.; Liu, Z. In situ study 512

of the electronic structure of atomic layer deposited oxide ultrathin films upon oxygen 513

adsorption using ambient pressure XPS. Catal. Sci. Technol. 2016, 6, 6778–6783, 514

doi:10.1039/c6cy00575f. 515

13. Mirkowska, M.; Kratzer, M.; Teichert, C.; Flachberger, H. Principal Factors of Contact 516

Charging of Minerals for a Successful Triboelectrostatic Separation Process – a 517

ReviewHauptfaktoren der Triboaufladung von Mineralphasen für eine erfolgreiche 518

elektrostatische Trennung – ein Überblick. BHM Berg- und Hüttenmännische Monatshefte 2016, 519

161, 359–382, doi:10.1007/s00501-016-0515-1. 520

14. Ciccu, R.; Ghiani, M.; Ferrara, G. Selective Tribocharging of Particles for Separation. KONA 521

Powder Part. J. 1993, 11, 5–16, doi:10.14356/kona.1993006. 522

15. Etxebarria, A.; Koch, S.L.; Bondarchuk, O.; Passerini, S.; Teobaldi, G.; Muñoz-Márquez, M.Á. 523
Minerals 2021, 11, x FOR PEER REVIEW 21 of 22

Work Function Evolution in Li Anode Processing. Adv. Energy Mater. 2020, 10, 524

doi:10.1002/aenm.202000520. 525

16. Ullah, S.; Wan, G.; Kouzios, C.; Woodgate, C.; Cattelan, M.; Fox, N. Structure and electronic 526

properties of tin monoxide (SnO) and lithiated SnO terminated diamond (1 0 0) and its 527

comparison with lithium oxide terminated diamond. Appl. Surf. Sci. 2021, 559, 1–9, 528

doi:10.1016/j.apsusc.2021.149962. 529

17. Biegaj, K.W.; Rowland, M.G.; Lukas, T.M.; Heng, J.Y.Y. Surface Chemistry and Humidity in 530

Powder Electrostatics: A Comparative Study between Tribocharging and Corona Discharge. 531

ACS Omega 2017, 2, 1576–1582, doi:10.1021/acsomega.7b00125. 532

18. Pan, S.; Zhang, Z. Fundamental theories and basic principles of triboelectric effect: A review. 533

Friction 2019, 7, 2–17, doi:10.1007/s40544-018-0217-7. 534

19. Manouchehri, H.R.; Rao, K.H.; Forssberg, K.S.E. Triboelectric charge, electrophysical 535

properties and electrical beneficiation potential of chemically treated feldspar, quartz and 536

wollastonite. Magn. Electr. Sep. 2002, 11, 9–32, doi:10.1155/2002/46414. 537

20. Mirkowska, M.; Kratzer, M.; Teichert, C.; Flachberger, H. Principal Factors of Contact 538

Charging of Minerals for a Successful Triboelectrostatic Separation Process – a 539

ReviewHauptfaktoren der Triboaufladung von Mineralphasen für eine erfolgreiche 540

elektrostatische Trennung – ein Überblick. BHM Berg- und Hüttenmännische Monatshefte 2016, 541

161, 359–382, doi:10.1007/s00501-016-0515-1. 542

21. Lacks, D.J.; Duff, N.; Kumar, S.K. Nonequilibrium accumulation of surface species and 543

triboelectric charging in single component particulate systems. Phys. Rev. Lett. 2008, 100, 1–4, 544

doi:10.1103/PhysRevLett.100.188305. 545

22. Hansen, L.; Wollmann, A.; Weers, M.; Benker, B.; Weber, A.P. Triboelectric charging and 546

separation of fine powder mixtures. Chem. Eng. Technol. 2020, 1–10, doi:10.1002/ceat.201900558. 547

23. Wang, Z.L.; Wang, A.C. On the origin of contact-electrification. Mater. Today 2019, 30, 34–51, 548

doi:10.1016/j.mattod.2019.05.016. 549

24. Lee, V.; James, N.M.; Waitukaitis, S.; Jaeger, H.M. Collisional charging of individual sub- 550

millimeter particles: Using ultrasonic levitation to initiate and track charge transfer. arXiv 2018, 551

1–24. 552

25. Kwetkus, B.A.; Sattler, K. Electrification Curves. J. Phys. D. Appl. Phys. 1992, 25, 1400–1408. 553

26. Linsenbühler, M. Herstellung und Charakterisierung funktionalisierter Composite-Partikel. 554

Der Tech. Fak. der Univ. Erlangen-nürnb. 2005, 273. 555

27. Oberrauner, A.; Flachberger, H. Beitrag zur Untersuchung der triboelektrischen 556

Aufladecharakteristik feindisperser KörnerschwärmeInvestigating the Triboelectric 557

Charging Behaviour of Fine Powders. BHM Berg- und Hüttenmännische Monatshefte 2010, 155, 558

565–570, doi:10.1007/s00501-010-0616-1. 559

28. Shen, X.; Wang, A.E.; Sankaran, R.M.; Lacks, D.J. First-principles calculation of contact 560

electrification and validation by experiment. J. Electrostat. 2016, 82, 11–16, 561

doi:10.1016/j.elstat.2016.04.006. 562

29. Wang, A.E. Driving forces for the triboelectric charging of well-defined insulated material 563

surfaces, Case Western Reserve University, Ohio, 2020. 564

30. Matsusaka, S. Control of particle tribocharging. KONA Powder Part. J. 2011, 29, 27–38, 565
Minerals 2021, 11, x FOR PEER REVIEW 22 of 22

doi:10.14356/kona.2011007. 566

31. Liu, B.Y.H.; Agarwal, J.K. Experimental observation of aerosol deposition in turbulent flow. 567

J. Aerosol Sci. 1974, 5, doi:10.1016/0021-8502(74)90046-9. 568

32. Kallio, G.A.; Reeks, M.W. A numerical simulation of particle deposition in turbulent 569

boundary layers. Int. J. Multiph. Flow 1989, 15, 433–446, doi:10.1016/0301-9322(89)90012-8. 570

33. Grosshans, H.; Papalexandris, M. V. Numerical study of the influence of the powder and pipe 571

properties on electrical charging during pneumatic conveying. Powder Technol. 2017, 315, 227– 572

235, doi:10.1016/j.powtec.2017.04.012. 573

34. John, W.; Reischl, G.; Devor, W. Charge transfer to metal surfaces from bouncing aerosol 574

particles. J. Aerosol Sci. 1980, 11, 115–138, doi:10.1016/0021-8502(80)90029-4. 575

35. Li, J.; Yao, J.; Zhao, Y.; Wang, C.H. Large eddy simulation of electrostatic effect on particle 576

transport in particle-laden turbulent pipe flows. J. Electrostat. 2021, 109, 103542, 577

doi:10.1016/j.elstat.2020.103542. 578

36. Friedlander, S.K. Smoke, Dust, and Haze: Fundamentals of Aerosol Dynamics; 2nd ed.; Oxford 579

University Press, 2000; ISBN 0-19-512999-7. 580

37. Noorani, A.; Sardina, G.; Brandt, L.; Schlatter, P. Particle transport in turbulent curved pipe 581

flow. J. Fluid Mech. 2016, 793, 248–279, doi:10.1017/jfm.2016.136. 582

38. Wang, H.; Fotovat, F.; Bi, X.T.; Grace, J.R. Tribo-charging of binary mixtures composed of 583

coarse and fine particles in gas–solid pipe flow. Particuology 2019, 43, 101–109, 584

doi:10.1016/j.partic.2018.07.001. 585

39. Guha, A. Transport and deposition of particles in turbulent and laminar flow. Annu. Rev. Fluid 586

Mech. 2008, 40, 311–341, doi:10.1146/annurev.fluid.40.111406.102220. 587

40. Röhrbein, J.; Weber, A.P. A system for on-line characterization of gas-borne particle surface 588

properties based on their photoemission. J. Aerosol Sci. 2018, 120, 82–91, 589

doi:10.1016/j.jaerosci.2018.03.002. 590

41. Wood, N.B. A simple method for the calculation of turbulent deposition to smooth and rough 591

surfaces. J. Aerosol Sci. 1981, 12, 275–290, doi:10.1016/0021-8502(81)90127-0. 592

42. H. Schlichting , K. Gersten; 9th ed.; Springer-Verlag Berlin Heidelberg: New York, 1997; ISBN 593

978-3-662-07555-5. 594

43. Fan, F.G.; Ahmadi, G. A sublayer model for turbulent deposition of particles in vertical ducts 595

with smooth and rough surfaces. J. Aerosol Sci. 1993, 24, 45–64, doi:10.1016/0021- 596

8502(93)90084-M. 597

598