Renewable and Sustainable Energy Reviews 98 (2018) 239–254

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

Fundamentals of soiling processes on photovoltaic modules T

a,b,c,⁎ d,e a a
Klemens K. Ilse , Benjamin W. Figgis , Volker Naumann , Christian Hagendorf ,
Jörg Bagdahnb
a
Fraunhofer Center for Silicon-Photovoltaics CSP, 06120 Halle (Saale), Germany
b
Anhalt University of Applied Sciences, 06366 Köthen (Anhalt), Germany
c
Department of Physics, Martin Luther University Halle-Wittenberg, 06120 Halle (Saale), Germany
d
Qatar Environment & Energy Research Institute, HBKU, Doha, Qatar
e
ICube Laboratory, Université de Strasbourg–CNRS, Strasbourg, France

ARTICLE INFO ABSTRACT

Keywords: The topic of soiling of photovoltaic module (PV) and concentrated solar power (CSP) collectors has recently
Soiling gained increasing attention due to its impact on solar power production, especially in arid and semi-arid areas
Dust deposition with high concentrations of airborne dust. Soiling describes the deposition of dust and other contaminants on
Solar energy surfaces, reducing solar irradiation by absorbing or reflecting the sunlight, causing energy yield losses which can
Particle adhesion
exceed 1% per day. The amount of soiling is influenced by complex interactions of many factors which can vary
Dew
Cementation
significantly from site to site.
In this study we provide a detailed overview of macroscopic and microscopic factors influencing soiling. This
includes a global analysis of key parameters including airborne dust concentrations, dust characteristics (mineral
composition, size distribution), and particle deposition rates. A theoretical model for relevant particle adhesion
and removal forces is presented to achieve a microscopic understanding of wind cleaning effects. Further, it was
found that dew occurs frequently on PV modules in many soiling affected areas and that this can significantly
increase particle adhesion. Therefore, a detailed analysis and model of the dew-driven soiling mechanisms of
cementation, particle caking, and capillary aging are given on the basis of microstructural material and dust
characterization of outdoor exposed glass samples. Furthermore, we study the complex interplay and dynamics
of different environmental parameters (relative humidity, ambient and module temperature, airborne dust
concentration, wind speed) and their correlation to dust accumulation, and provide explanations with the help of
the developed models. Overall the study aims to provide a useful, in-depth but also global overview and fun-
damental understanding of soiling processes on PV modules down to the microscale, which could be used to
develop appropriate soiling mitigation strategies.

1. Introduction power losses since it reduces solar energy received by the collector by
absorbing or scattering the sunlight. A striking example for soiled PV
Regarding the problem of climate change, great efforts are needed modules compared to a cleaned one can be seen in Fig. 1. For PV, the
to reach the 2015 Paris goals for 2100, i.e. a maximum increase of maximum energy yield loss per day can exceed 1% [1], which accu-
1.5–2.0 °C in global temperature compared to pre-industrialisation by mulates over time if the surface is not cleaned (see also Fig. 3).
significantly reducing CO2 emissions. Renewable energies and espe- Therefore, soiling is one of the major concerns of component and
cially photovoltaics (PV) are regarded as an important contribution for system reliability in areas with high concentrations of airborne parti-
decarbonisation of our energy system. Furthermore, with commercial culates. In contrast to the past, when PV was mainly installed in mod-
offers to supply PV-generated electricity decreasing below 2 ct/kWh, it erate climates, the majority of future markets in coming years will be in
appears that solar power will become the cheapest form of electricity in areas in which high soiling losses are reported in literature, foremost
many regions of the world. However, areas with high irradiation levels China and India [2].
(the global sun-belt) often suffer from high airborne dust loads. Typically, power losses due to soiling can be reduced by frequent
Deposition of dust and dirt on PV modules or mirror surfaces leads to cleaning of the PV modules. Unlike other power degradation

⁎
Corresponding author at: Fraunhofer Center for Silicon-Photovoltaics, Otto-Eißfeldt-Straße 12, 06120 Halle (Saale), Germany.
E-mail address: [email protected] (K.K. Ilse).

https://doi.org/10.1016/j.rser.2018.09.015
Received 26 January 2018; Received in revised form 2 September 2018; Accepted 8 September 2018
1364-0321/ © 2018 Elsevier Ltd. All rights reserved.
K.K. Ilse et al. Renewable and Sustainable Energy Reviews 98 (2018) 239–254

Nomenclature µ Dynamic viscosity of air

Kinematic viscosity
A Hamaker constant PM2.5/10 Particulate matter with an aerodynamic diameter smaller
a Contact radius of separation than 2.5 µm or 10 µm
d Particle diameter q Electric charge
0 Vacuum permittivity qBE Mean particle charge at Boltzmann equilibrium
E1/2 Young modulus rms1/2 Root mean square of surface roughness
FC Capillary force Density
FD Drag force s Distance of separation
FG Gravity BE Surface charge density at Boltzmann equilibrium
FL Lift force max Maximum surface charge density
FVdW Van der Waals force 1/2 Water contact angle
Fadh Sum of adhesion forces u* Friction/shear velocity
Fel Electrostatic force v Wind speed/free stream velocity
g Gravitational acceleration vStokes Stokes settling velocity
w Surface tension Wa Surface energy /work of adhesion
1/2 Characteristic length of roughness feature 1/2 Poisson ratio
MD Drag moment

mechanisms of PV modules, such as potential induced and light induced different time scales (e.g. nanoseconds of particle impact or seasonal
degradation (PID and LID), the power output degradation due to soiling changes of environmental factors or climate change over years) play a
can usually be reversed and the light transmission restored to 100%. On role, as well as dynamics and history of particle adhesion (e.g. initial
the other hand, frequent cleaning will result in high operation and state, aging, cementation). In addition, one has to consider a combi-
maintenance (O&M) costs, increasing the levelized cost of electricity nation of processes and influencing factors at the macro, system scale
(LCOE), especially for regions where water and labour are scarce or (e.g. wind turbulences in a PV field, partial shading by inhomogeneous
expensive. Hence, there are great efforts to optimize cleaning strategies, soiling) and processes which take place at the micro- and nanoscales
including different approaches such as wet or dry cleaning, automated (e.g. particle size, surface adhesion mechanisms). Fig. 2 summarizes
or manual cleaning methods, different brush or cloth types or chemical some of these influencing factors in terms of relevant size and time
additives. Besides cleaning, another mitigation strategy is to reduce the scales. Bringing together all the influencing factors, size scales and time
dust accumulation rate using the natural cleaning mechanisms of wind frames is ambitious, since this also requires a large variety of char-
and rain. To increase their cleaning effect, much work is being under- acterization methods. Accordingly, most of the scientific studies can
taken to develop anti-soiling coatings [3–9]. only address some part of the whole picture.
In regions with low rainfall and high soiling rates, such as arid and In this paper we provide a comprehensive study of the link between
semi-arid climates, dust removal by wind can be utilized as a natural macroscopic/system scale environmental parameters and the micro-
cleaning mechanism. The effective soiling rate and anti-soiling func- scale properties of dust particles, particle adhesion and removal. These
tionality of surfaces will be the result of particle deposition versus factors are examined theoretically and by experimental approaches.
particle removal/resuspension, which are determined by particle ad- First of all, we provide a short overview of soiling field studies from the
hesion forces and removal forces. The complex interplay of relevant literature, which cover further details of the important influencing
forces and their multiple influencing factors is discussed in chapter 3. factors (Section 2). This is followed by an in-depth analysis of the key-
Detailed evaluation of the soiling processes is a truly complex pro-
blem because of the many influencing factors such as environmental
parameters (e.g. airborne dust concentration, relative humidity, wind
speed and direction, ambient temperature), PV module properties (e.g.
surface roughness and chemistry, module frame, temperature), site and
installation characteristics (e.g. tilt angle, orientation) and others.
Furthermore, dynamic factors such as changes in these parameters over

Fig. 2. Complexity of the soiling problem due to many influencing factors with
huge variations in size and time scales. The macroscopic environmental para-
meters are coloured in blue whereas the microscopic soiling processes are
Fig. 1. Soiled and cleaned PV modules after one month of soiling at the QEERI shown in green. Controllable influencing factors such as the module or plant
Solar Test Facility in Doha, Qatar. design as well as operation and maintenance strategies are marked in orange.

240
K.K. Ilse et al. Renewable and Sustainable Energy Reviews 98 (2018) 239–254

parameter of airborne dust, its impact on soiling rates and physical/ mechanical cleaning systems, electrodynamic screens (EDS) as well as
chemical properties (Section 3). Then, we address the topic of adhesion super hydrophobic and super hydrophilic coatings.
and removal forces relevant for loosely bound particles on planar sur- Costa et al. [16] continue their work [14] and update the literature
faces. For this, a simplified physical model of spherical particles is used review on solar energy related soiling for publications of the year 2016
to investigate the correlation between influencing factors and particle including 184 references.
adhesion and removal (Section 4). In addition to that, a focus is made Figgis et al. [17] provide a review of particle mechanics with focus
on dew driven soiling mechanisms that increase particle adhesion in- on deposition, rebound, resuspension as well as discussion of influen-
cluding cementation, particle caking and capillary aging. Based on cing factors such as surface orientation and moisture.
microstructural investigations of soiled glass samples, which have been Javed et al. [18] study the influences of environmental parameters
exposed to outdoor conditions in the deserts of Qatar and Chile, con- on daily soiling losses and their modelling.
clusions are drawn on the potential importance of dew formation for Picotti et al. [19] review the different steps of dust generation, dust
soiling rates and its relevance for desert environments (Section 5). Fi- deposition, adhesion and removal, including important physical rela-
nally, the theoretical correlations between dust accumulation and en- tions for all stages.
vironmental factors are validated by the statistical evaluation of out- Hammad et al. [20] review soiling literature especially concerning
door experiments in Qatar using an outdoor soiling microscope (Section the Middle East and North Africa (MENA region) and study perfor-
6). mance system modelling including cleaning frequency optimization.
With that, this study aims to provide different perspectives, theo- Figgis et al. [21] investigate factors affecting condensation on soiled
retical considerations and experimental results on the topic of soiling to PV modules, including hygroscopic dust content, relative humidity,
reduce its complexity and help to identify and summarize the important surface/dew point temperature difference, and surface wettability.
aspects which should be considered for development of suitable stra- Said et al. [22] review the effects of environmental factors on PV
tegies to prevent or mitigate soiling. The study focuses mainly on module performance and dust accumulation as well as dust mitigation
soiling of PV modules in arid and semi-arid environments and soiling by strategies.
mineral dusts. However, the basic conclusions should not be limited to
these areas and possibly also apply to other solar power technologies
3. Key parameter airborne dust and its properties
and different contaminant types such as engine exhaust, industrial or
agricultural emissions, pollen, bird droppings and biofilms.
3.1. Airborne dust concentration

2. Soiling literature overview

Intuitively, the most important factor for soiling is the concentration
of airborne particulates. Micheli and Muller [23] recently performed a
Within this paragraph, we present a selection of insightful papers
systematic correlation analysis for more than 100 possible influencing
and review articles on the topic of soiling. Subsequently, we review
parameters with data from 20 different soiling stations installed in the
chosen important aspects in detail which are needed for better under-
USA. They found that the annual average of the daily mean values for
standing of soiling processes. A synopsis is provided for papers in
airborne dust concentration recorded by monitoring stations deployed
chronological order:
near the PV systems (both particulate matter with an aerodynamic
Cuddihy [10] addresses theoretical considerations of soil retention
diameter of less than 10 µm or 2.5 µm, PM10 and PM2.5 respectively)
and provides a guidance for the development of advanced photovoltaic
show a strong correlation to the soiling rate with a coefficient of de-
encapsulation systems to minimize soil retention. Chemical, physical
termination R2 of 0.82. Together with rainfall frequency/average
and mechanical criteria required for low soil retention were established
length of dry periods, the airborne dust concentration is the best soiling
on the basis of the limited published literature and theoretical con-
predictor.
siderations. The discussion of soiling mechanisms includes 1) ce-
This correlation can also be seen in Fig. 3 and Fig. 4. Fig. 3 shows
mentation by water soluble salts, 2) deposition of organic materials, 3)
two different comprehensive data sets for maximum daily soiling rates
surface tension and 4) particle energetics.
and annual mean concentration of PM10, which have been compiled
Mani and Pillai [11] review the research from 1940 to 2010 in-
from literature [24,25]. The used regions for data sorting are marked in
cluding a summary of influencing factors such as dust properties, wind
Fig. 4, which shows a global map of modelled annual median airborne
velocity, ambient temperature and humidity, tilt angle of PV system,
PM2.5 dust concentration [26].
glazing and site characteristics as well as general recommendations of
The results indicate a good correlation between high maximum
mitigation measures for different climatic zones.
Sarver et al. [12] provide an extensive and comprehensive review
on history, investigations, results, literature and mitigation approaches
for soiling, including very detailed discussions on most relevant aspects
of soiling.
Sayyah et al. [1] review energy yield losses caused by dust de-
position on solar collectors as well as results of laboratory soiling stu-
dies and analyse advantages of different cleaning processes including
natural, manual, automatic, and passive methods.
Zaihidee et al. [13] provide a review on optical effects of dust de-
posited on PV modules and electrical parameters, including partial
shading, and influence of factors of dust properties, weather conditions,
module properties and PV installation design.
Costa et al. [14] summarize more than 230 publications on soiling
issues related to solar energy system between 2012 and 2015, also
showing a strong increase in the number of annual publications on the Fig. 3. Left: Maximum soiling rates (power/efficiency/transmission loss per
topic of soiling with main focus on PV starting 2009 driven by China day) for different regions of the world, deducted from review of data from 63
market. publications on soiling [24]. Right: annual mean airborne dust concentrations
Jamil et al. [15] provide a review of soiling mitigation methods of PM10 for different regions of the world deducted from the Ambient Air
including manual cleaning, natural cleaning, water cleaning, Pollution Database, WHO, May 2016 [25].

241
K.K. Ilse et al. Renewable and Sustainable Energy Reviews 98 (2018) 239–254

Fig. 4. Global map of modelled annual median airborne PM2.5 dust concentration, showing selected regions for analysis and data sorting in Fig. 3 and Fig. 5: Middle
East (ME), East- and Southeast Asia (E&SE-As.), North- and West Africa (N&W Afr.), South Asia (S-As.), Europe (Eur.), North America (N-Amr.), South America (S-
Amr.). © [2016] WHO, modified and reprinted with permission from [26].

soiling rates for world regions with high dust concentrations (Pearson Desert in Chile). Besides mineral dusts emitted from desert lands, there
correlation coefficient R = 0.84 between median PM10 and median are many other possible sources for airborne dust including industry
soiling rate in Fig. 3). The main reason for the high airborne dust (process dusts, exhaust), traffic (carbon particles, soot), agriculture (soil
concentrations and dust deposition rates in regions such as the Middle emissions, animal feed dusts, emissions from cattle breeding) and or-
East, East- and Southeast Asia, North- and West Africa as well as South ganics (pollen, seeds, bird droppings, leaves, lice, lichen, algae, moss).
Asia is their proximity to local dust sources such as desert areas char- Their proportion of airborne dust and corresponding soiling can vary
acterized by arid or semiarid climate conditions and very low pre- strongly for different locations. Within the scope of this study, we are
cipitation [27]. The dominant global dust sources are situated in the so primarily focusing on the contribution of mineral dusts.
called ‘dust belt’ between 15°N and 45°N, which stretches from the west It should be noted that the airborne PM10 or PM2.5 dust con-
coast of Africa through the Arabian Peninsula and the Middle East to centrations seem to be a good explanation for variation in soiling rates
Mongolia/China (Gobi–Taklamakan region) [27–29], see also Fig. 4. over large distances (i.e. between geographical locations), but have
Further (secondary) sources of dust can be found in North America been reported to be an inadequate explanation of day-to-day variations
(Great Basins in Western United States and Mexico), central Australia, in soiling rates in a particular location, because other local weather
South Africa (Namib, Kalahari Desert), and South America (Atacama conditions such as wind speed and relative humidity also play an

Fig. 5. Mineral composition from XRD analysis of sieved dusts (< 38 µm) collected from the ground of different regions throughout the world [29,33]. The normative
mineral concentrations (mass %) from supplementary material S4.2 [33] have been normalized to 100% and sorted by region: Middle East (ME), East- and Southeast
Asia (E&SE-As.), North- and West Africa (N&W Afr.), South Asia (S-As.), Europe (Eur.), North America (N-Amr.), South America (S-Amr.).

242
K.K. Ilse et al. Renewable and Sustainable Energy Reviews 98 (2018) 239–254

important role [18,30,31]. considered the net result of three processes: particles from the atmo-
sphere impacting the surface (“deposition”), immediately rebounding
3.2. Dust composition from the surface without adhering (“rebound”), or later being re-
suspended by wind (“resuspension”) [17]. Each process can be ex-
The properties of airborne dust, including size distribution, shape pressed as a dust flux rate (mass of dust per surface unit and time unit),
and chemical/mineralogy composition, can vary greatly by location. In so that they arithmetically balance:
desert areas, airborne dust is mainly composed of minerals such as
Accumulation = Deposition Rebound Resuspension
quartz, illite, kaolinite, smectite, carbonates, gypsum, feldspar and iron
oxides [32]. It can be expected that the composition of airborne dust is For PV systems, the projected area and mass of the soiling layer is
strongly linked to the ground dust composition from local terrain [27]. dominated by silt-sized particles (several to tens of microns in dia-
Therefore, the investigation of Engelbrecht et al. [29,33] on resuspen- meter). The deposition of silt particles is mostly controlled by gravity
sions of dust collected from over 60 different ground sites throughout (sedimentation) and turbulence of the airflow (inertial deposition). The
the world provides a good overview of expected chemical and mineral importance of inertial deposition to ambient soiling varies in the lit-
compositions. For a clear presentation, the data of mineral composition erature. In the Negev desert, Biryukov [37] found that sedimentation
from XRD analysis of sieved dusts (< 38 µm) provided by Engelbrecht (gravity settling) alone explains observed soiling of solar collectors. At a
et al. has been compiled and sorted for the different regions of the field test site in Qatar, it was reported that 96% of deposition mass on a
world, see Fig. 5. From Fig. 5, one can deduce that there are large greased collector occurred in low wind speed (< 3 ms-1), i.e. sedi-
differences in the mineral dust composition of different locations mentation dominated [17]. On the other hand, on the roof of a building
throughout the world, especially concerning the individual proportions in Chicago it was found that there was significant soiling on the un-
of the main constituents such as quartz, calcite, dolomite, kaolinite and derside of a greased horizontal collector, indicating the presence of
hematite. The mineral composition seems to be similar for the whole inertial deposition [38].
Middle East, whereas North and West Africa or North America show Overall it seems that unless conditions are windy (wind speed above
great variations between sample sites. ~ 3 m/s), the particle deposition rate for PV can be adequately ac-
The chemical/mineralogical composition of particles can strongly counted for by sedimentation alone. In this case, the deposition flux
influence the optical behavior and corresponding transmission losses rate of particles in the relevant size range can be roughly described by
caused by accumulated dust on glass surfaces, as has been shown in their Stokes settling velocity, vStokes :
several studies [12,29,34–36]. Furthermore, the chemical/miner- vStokes = d 2g /18µ
alogical composition of dust becomes important concerning dust ad-
hesion, especially when condensation of water and corresponding ce- where is particle density, d is particle diameter, g is gravitational
mentation processes take place, see Section 5.2. acceleration, and μ is dynamic viscosity of air. The deposition flux rate
is obtained by multiplying vStokes by the airborne concentration of
3.3. Particle size distribution particles of that size. Further details and further considerations on the
particle settling velocity and its influencing factors including correc-
In arid and semi-arid climates, the main proportion of soiling is tions especially for small particle sizes (< 1 µm) can be found in
attributed to dust particles in the silt size-range (particle diameter be- [17,39,40].
tween 2 and 63 µm), since larger sand particles (diameter > 63 µm) are Particles are predicted to rebound from the surface if their kinetic
too heavy to be lifted high enough to settle on PV modules except in energy on impact exceeds the work of adhesion between the particle
very strong wind events like desert storms. Lawrence et al. [27] re- and surface (allowing for energy dissipation due to inelastic collision)
viewed and summarized different size distributions for dust deposits [41]. Therefore, rebound fraction is expected to be enhanced in situa-
from local, regional and global sources and found the main fraction of tions of large particles, high wind speed, and low humidity. Given the
particles to be in the silt size range (50%, 60% and 70%, respectively) diversity of particle and surface characteristics that are encountered in
and minor fractions of clay (< 2 µm) and sand. This similarity of the outdoor settings, there are no widely-used predictive models. Previous
results for dust size distributions from different locations agrees with research indicates that ambient rebound fractions range widely, from
the brittle fragmentation theory, which states that the emitted size 14% for flow parallel to a collector over an unspecified period in wind
distribution of dusts is essentially independent of wind speed and soil speed < 4 ms-1 , to almost 100% over a 10-min period in wind
properties and dominant emission is in the range 2–20 µm [28,32]. speed > 3 ms-1 for horizontal glass coupons in Qatar [17,37]. Re-
Furthermore, it can be assumed that there is a correlation between suspension of dust particles by wind is addressed in Section 4.2.
airborne size distribution and particle size distribution of dust deposited Regarding the effects of environmental parameters, high PM and RH
on PV modules. For example, Javed et al. [18] reported mean and exacerbate soiling by increasing the deposition rate of dust and in-
median size ranges on PV modules in Qatar to be between 9 and 25 µm hibiting its detachment. High wind speed has competing effects, how-
including monthly changes as well as lower size ranges for longer ever, because it increases both deposition of dust particles (due to the
durations of exposure of the surfaces. A literature overview of particles inertial mechanism) and the probability of rebound (due to their
accumulating on outdoor surfaces provided by Figgis et al. [17] shows greater kinetic energy). The effect of wind speed on the amount of net
dominant particle sizes of 10–32 µm. deposition (deposition minus rebound) is therefore not obvious and is
Besides dust composition, the light transmission losses caused by further complicated by resuspension. It is therefore difficult to accu-
accumulated dust on glass surfaces are strongly influenced by the rately predict a dust accumulation rate theoretically at a candidate PV
particle size distribution with significantly increased losses per accu- site, and field measurements are necessary.
mulated dust mass reported for smaller particle sizes [12,13]. Since
there are many different characterization methods for the determina- 4. Particle adhesion and removal
tion of particle size distribution as well as different normalization ap-
proaches (commonly particle number or volume size), care should be 4.1. Particle adhesion forces
taken when results of different studies are compared.
As previously described, the dust accumulation and corresponding
3.4. Particle deposition rate soiling rate is determined by the interaction of particle deposition, re-
bound and resuspension, which are strongly influenced by particle
The rate at which dust particles accumulate on the PV surface can be adhesion and removal forces. Fig. 6 provides an overview of typical

243
K.K. Ilse et al. Renewable and Sustainable Energy Reviews 98 (2018) 239–254

crevices [21]. Therefore, capillary forces can be present already at low

humidity levels like 30% or 40% [19,21,44,45]. It is reported that ca-
pillary forces increase with increasing relative humidity levels since
more water will be present at the surface, filling more and more as-
perities between the (rough) surface and (rough) particles [21,43–45].
As soon as the relative humidity rises to values close to 100%, the
surface may be flooded by water and capillary forces decrease [19].
Besides relative humidity, the surface energy and the corresponding
water contact angle (WCA) have a strong impact on the amplitude of
capillary forces as is shown in Fig. 6b) for a hydrophilic glass surface
(WCA = 0°) and a hydrophobic surface (WCA = 150°). There will be a
strong decrease in the capillary adhesion force for hydrophobic, water
repellent surfaces, because no stable meniscus or continuous capillary
bridges can form [44]. The same applies to an increased surface
roughness, since particle and surface are further apart than for the
smooth case and asperities cannot be filled by liquid anymore. For the
case of rough surfaces, possible decreases of capillary forces by orders
of magnitude are reported in literature [44].
Van der Waals (VdW) forces FVdW are always present between par-
ticles and surfaces. They act on very short ranges since they originate
from interacting dipoles at two surfaces which are in contact.
Consequently, they scale with the effective contact area between the
particle and the surface (described by the contact radius a, see Fig. 6a)
as has been vividly demonstrated by Kazmerski et al. [46]. Accordingly,
when a surface roughness is introduced, this could significantly reduce
the effective contact area and corresponding VdW forces [19,44] as can
be seen from the calculated forces in Fig. 6b). For this, it should be
noted that the size scale of roughness is very important: asperities
should be significantly smaller than particle diameter (about three or-
ders of magnitude) to reduce adhesion, since a roughness in comparable
Fig. 6. a) Schematic illustration and simulation of particle adhesion and re- size scale can act as a trap for the particles [44]. In addition to
moval for a SiO2 sphere on solar glass. b) Adhesion forces and threshold con- roughness, material properties strongly impact van der Waals adhesion,
ditions for particle removal are shown in dependency of particle diameter: van such as the surface energy, which is strongly correlated to the Hamaker
der Waals (VdW) force for a smooth and a rough surface, capillary force for constant (see equation in Table 1), or the material stiffness (Young
hydrophilic (0°) and hydrophobic (150°) surface, electrostatic forces for parti- modulus), which determines the effective contact area between particle
cles with mean charge according to Boltzmann equilibrium and maximum and surface. In wet environments, van der Waals forces are con-
surface charge density, gravity (FG) due to weight of sphere. The dotted lines siderably decreased because of the reduction of the Hamaker constant
provide threshold values of maximum sum of adhesion forces which is allowed
by the interstitial water [42] as well as electrical double layer repulsion
for detachment of particles at a given wind speed of 10 m/s due to rolling,
as known from DLVO theory [43], which is an important factor re-
sliding or lift-off. The formulas and parameters used for the calculation of the
force ranges are summarized in Table 1.
garding wet cleaning processes for soiled surfaces.
Electrostatic forces Fel between particles and surfaces can be either
attractive or repulsive. Their strength depend on many influencing
particle adhesion forces as well as possible detachment modes by wind factors, such as the presence of electric fields and their field strength,
(see Section 4.2) occurring for a spherical SiO2 particle on a glass charge state of particles and surfaces including the charge distribution;
surface in dependence of the particle size. This includes van der Waals the electrical properties of particles and surface (e.g. conductive, non-
forces for a smooth and a rough glass surface, capillary forces for hy- conductive), the dielectric properties of intervening medium as well as
drophilic and hydrophobic glass surface properties, electrostatic forces separation distance [19,43,47]. Table 1 shows a simple approximation
assuming mean particle charging at the Boltzmann equilibrium as well for electrostatic forces acting on a dielectric particle at a non-con-
as gravity (weight of sphere). The physical relationships and parameters ductive surface [42,48]. This formula was used for the calculation
used for the calculation of the and forces and threshold values shown in shown in Fig. 6b), assuming a particle charge distribution at Boltzmann
Fig. 6b) are summarized in Table 1. From Fig. 6b), it can be deducted equilibrium [49] and a maximum possible surface charge density for
that if present, capillary forces are the dominating forces for all particle the PV module glass which is determined by the electric field limit for
sizes [42], followed by van der Waals forces. Influences of gravity and discharging [42,48,50]. It can be concluded, that for the assumptions
electrostatic forces are comparably low and can be neglected for the made and in the relevant particle size range, the electrostatic forces
relevant particle size range of dust and assumed charge conditions. have much lower impact on particle adhesion than capillary or van der
According to the formulas used, the adhesion forces significantly in- Waals forces (see Fig. 6b)). This conclusion is also valid assuming also a
crease with increasing particle size except for rough surface conditions. maximum surface-charge density for the particles [42,47,50]. Further-
In the following, we will take a closer look at the most important ad- more, electrostatic forces due to charged surfaces will further be low-
hesion forces for soiling and their influencing factors. More detailed ered as soon as thin films of water are present causing charge com-
information and literature reviews are presented in [19,43]. pensation. Therefore, electrostatic forces are typically not the dominant
Capillary forces FC arise mainly from liquid bridges which are pre- adhesion forces for particle-surface separations smaller than 20 nm
sent between particles and the surface, being a combination of surface [43]. Nevertheless, if there are electrical fields present above the PV
tension force and the (dominant) capillary pressure force [43]. Under module surfaces, which could be the case for dry environmental con-
typical environmental conditions, water can condense and forms nano- ditions (e.g. due to potential differences between solar cells and
or micro-scale menisci already at relative humidity levels well below grounded module frames, which can also cause PID [51]), the long
100% due to capillary condensation, e.g. in surface scratches or small range of electrostatic attraction may significantly change the deposition

244