The hydrophilic-to-hydrophobic transition in glassy silica is ...

The surface reactivity and hydrophilicity of silicate materials are key properties for various industrial applications. However, the structural origin of their affinity for water remains unclear. Here, based on reactive molecular dynamics simulations of a series of artificial glassy silica surfaces annealed at various temperatures and subsequently exposed to water, we show that silica exhibits a hydrophilic-to-hydrophobic transition driven by its silanol surface density. By applying topological constraint theory, we show that the surface reactivity and hydrophilic/hydrophobic character of silica are controlled by the atomic topology of its surface. This suggests that novel silicate materials with tailored reactivity and hydrophilicity could be developed through the topological nanoengineering of their surface.

For more information, please visit our website.

Here, based on reactive molecular dynamics simulations, we generated a series of artificial glassy silica surfaces annealed at different temperatures and, hence, exhibiting different structures. The surfaces were subsequently exposed to water to investigate the relationship between the structure and surface reactivity. Note that similar simulations (although using a different potential parametrization) were performed by Rimsza et al. 28 However, here we use the present simulations to identify the linkages between the surface structure, thermodynamic stability, reactivity, and hydrophilicity. We show that higher annealing temperatures result in more stable and less reactive surfaces. This leads to a hydrophilic-to-hydrophobic transition that is driven by the surface density of silanol groups. By performing a topological analysis of the atoms present at the surface, we show that both reactivity and hydrophilicity are controlled by the atomic topology of the surface of the glass.

In glasses, understanding such linkages between structure and engineering properties is usually challenging due their complex disordered atomic structure. Specially, the complexity of silicate glasses renders it challenging to discriminate the structural features that control reactivity to the first order from those that only show a second order effect. This can be conveniently achieved by topological constraint theory (TCT or rigidity theory), which reduces complex atomic networks into simpler structural trusses, 32–35 thereby capturing the important atomic topology while filtering out less relevant structural details that only weakly impact certain macroscopic properties. In this framework, atomic networks are described as nodes (the atoms) that are connected to each other through mechanical constraints (the chemical bonds). Such constraints comprise the radial two-body bond-stretching (BS) and angular three-body bond-bending (BB) bonds, which maintain the bond distances and bond angles fixed around their average values, respectively. 36,37 As per Maxwell’s criterion, the mechanical stability of a network depends on the balance between the number of constraints and degrees of freedom. 38 Namely, atomic networks can be flexible, stressed–rigid, or isostatic if the number of constraints per atom (n c ) is lower, larger, or equal to 3 (i.e., the number of degrees of freedom per atom in three-dimensional networks). Interestingly, recent studies have suggested that the dissolution rate of silicate glasses and crystals is controlled to the first order by their atomic topology, that is, it exponentially decreases with n c . 5–7,39–44

Although finely characterizing the atomic structure of silica is experimentally challenging, atomistic simulations have been thoroughly used to investigate the hydroxylation of silica. 21–29 However, several questions remain unanswered. 8,30,31 What is the relationship between the atomic structure of silica’s surface and its aqueous reactivity? How does the surface density of silanol groups control silica’s affinity for water? Can the hydrophilicity of silica be tailored by nanoengineering the atomic structure of its surface?

Various factors have been noted to potentially alter the chemical reactivity and hydrophilicity of silica. First, the presence of even small amounts of impurities can greatly affect silica’s surface reactivity. 11,12 The hydrophilicity of silica can also be tailored via polymer deposition. 13 Further, the roughness of silica surfaces can have a significant impact on their degree of hydrophilicity. 14 In addition, although glassy silica is typically hydrophilic under ambient conditions, its surface becomes markedly hydrophobic at elevated temperature (e.g., around 700-800 °C). 15 This has been interpreted from the fact that, at low temperature, glassy silica exhibits a high surface density of Si–OH silanol groups, 16–18 which, in turn, enhances the physical adsorption of water molecules on the surface. 19,20 In contrast, at elevated temperature, the surface silanol groups tend to react with each other to form Si–O–Si siloxane groups, 16,17 which are intrinsically hydrophobic. 15 This suggests that the structure of silica’s surface is closely linked to its hydrophilicity.

Understanding the chemical reactivity and hydrophilicity/hydrophobicity of silicate glasses in aqueous conditions is of great concern for several applications, including nuclear waste immobilization, 1,2 laboratory glassware, 3 bioactive glasses for bone regeneration and drug packaging, 4 supplementary cementitious materials, 5,6 or outdoor glasses exposed to atmospheric weathering (e.g., photovoltaic, automotive, and architectural applications). 7,8 In particular, understanding the surface reactivity of glassy silica—an archetypal model for more complex multicomponent silicate glasses—is key for various technological applications. 9,10

Here, we used the ReaxFF potential parametrized by Pitman et al. 51,52 The potential parameters can be found in the supplementary material . Note that an alternative parametrization of ReaxFF from Yeon and van Duin 53 has recently been shown to offer a more accurate prediction of the activation energy for the scission of Si–O bonds. 49 However, we found this parametrization to yield a questionable description of glassy silica. First, we observed that Yeon’s parametrization results in the formation of a high fraction of 5-fold coordinated Si atoms (around 5%, in agreement with previous simulations 28 ). Although the existence of 5-fold coordinated Si defects in silicate glasses has been suggested experimentally, 54 such defects were noted to be extremely rare in most silicate glasses (with a maximum concentration of 0.16% in a fast-quenched potassium silicate glass 54 ). Further, no 5-fold coordinated Si atoms were observed in glassy silica at ambient pressure. 54 Hence, the defect concentration predicted by the parametrization from Yeon et al. appears unrealistically high. In addition, we observed that Yeon’s parametrization yields unrealistic stiffness values—we computed a bulk modulus of 602 GPa for α-quartz, which is about 20 times larger than the experimental value. 55 In contrast, we recently showed that the parametrization from Pitman et al. 51,52 yields an excellent description of glassy silica. 50,56 The atomic structure was found to be in good agreement with the available experimental structure and did not comprise any 5-fold coordinated Si atoms (also see Fig. 1 ). In addition, the predicted stiffness was found to be realistic—with a computed Young’s modulus of 80 GPa (as compared to the experimental value of 73 GPa). 50 Finally, the Pitman parametrization of the ReaxFF potential was found to yield realistic values for the diffusion constant of O atoms and the activation energy of diffusion, which suggests that the dynamics of silica melts are well reproduced. 50

Investigating the chemical reaction between glassy silica and water requires the usage of a reactive potential. To this end, we used the ReaxFF potential 45 implemented through the user-reaxc package in the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS). 46 Unlike conventional classical force-fields, the interatomic energy terms used within ReaxFF depend on a dynamic bond order determined by the local environment of each atom. 47 Hence, ReaxFF is able to capture the dynamics of bond-breaking and bond-forming, on which chemical reactions rely on. Additionally, unlike classical interatomic potentials, the charge of each atom is dynamically assigned via a charge equilibration (QEq) method. 48 For instance, this permits to dynamically adjust the charge of O atoms based on their local environments, e.g., bridging- or non-bridging oxygen (BO or NBO). All of these features render ReaxFF ideal to simulate the chemical reaction between glassy silica and water, while avoiding the computational cost of ab initio simulations. 49 More details about the ReaxFF framework can be found in previous references. 45,50

A bulk silica glass was generated based on the common melt-quench method, 50,57–59 as detailed in the following. First, 3000 atoms were randomly placed in a large cubic simulation box (with a length of 50 Å) while ensuring the absence of any unrealistic overlap. The system was then relaxed at 4000 K for 100 ps to ensure the loss of the memory of the initial configuration. The melt was subsequently cooled from 4000 to 300 K with a cooling rate of 1 K/ps. Finally, the glass was further equilibrated at 300 K for 1 ns. The NPT ensemble, ensuring a zero pressure, was applied throughout the process. We obtained a final density of 2.18 g/cm 3 at 300 K, 50 which compares well with the experimental value (2.2 g/cm 3 ). 60 Note that, although the cooling rate used herein is much higher than those achieved experimentally, the structure of silica has been shown to weakly depend on the cooling rate. 61–63 However, we recently noted that the use of cooling rates larger than 1 K/ps can yield unrealistic structural defects. 61 As shown in Fig. 1 , the structure of glassy silica predicted by ReaxFF is in excellent agreement with neutron diffraction data. 50,64

Starting from the bulk glassy silica sample previously generated, two surfaces were created by instantly elongating the simulation box along the z-axis by 15 Å on both sides, thereby resulting in the creation of void spaces above and below the simulated system. Such a distance ensures that the two surfaces do not “see” each other. The system was subsequently equilibrated at 300 K for 1 ns in the NVT ensemble (i.e., constant number of atoms N, volume V, and temperature T). No significant variations in the density of silica were observed upon equilibration. Next, the system was independently annealed at 700, 1000, 1300, and 1600 K for 1 ns in the NVT ensemble—to preserve the void space in between the two surfaces. Note that these temperatures remain low as compared to the simulated glass transition temperature (around 3000 K 58 ). In all cases, this duration was found to be long enough to achieve the convergence of the pressure and energy of the system.

For each annealed silica system, a silica–water interface was generated by inserting some water molecules into the void space with a density of 1 g/cm 3 (see Fig. 2 ). The silica–water system was first subjected to an energy minimization and subsequently allowed to react at 300 K for 1 ns. During this process, the lateral dimensions (x and y) were kept fixed, while the NPT ensemble was used on the z-direction to keep a zero pressure.

The surface energy of the silica samples was estimated from the variation in the potential energy of the system upon the creation of the surface and subsequent thermal annealing (divided by the created surface area). In all cases, the potential energy was determined after energy minimization—to remove any thermal contribution to the computed energies. The obtained surface energy values are found to be within the range of available experimental 66–68 and simulation results. 28 As shown in Fig. 4 , we observe that the surface energy decreases with increasing annealing temperature. This confirms that, when subjected to thermal annealing with increasing temperatures, the silica surface becomes more stable and achieves a more energetically favorable structure. Such relaxation-induced decrease in the energy is similar to that observed upon the thermal annealing of bulk samples. 69,70 Such relaxation arises from the fact that glasses are by nature out-of-equilibrium materials. 71,72 The degree of relaxation of a glass can be estimated from its fictive temperature, which is defined as the temperature of the metastable equilibrium supercooled liquid exhibiting an energy comparable to that of the glass. 58,73 The fictive temperature of a glass typically decreases with decreasing cooling rate as the system is able to stay at the supercooled liquid metastable equilibrium down to lower temperatures. 61 Upon annealing, bulk glasses are able to partially adjust their atomic structure in order to resemble glasses formed with lower cooling rates, that is, with lower fictive temperatures. As such, here, an increase in the annealing temperature results can be effectively described as a decrease in the fictive temperature of the glass surface.

We observe here an equal concentration of NBO and Si iii defects, as the breakage of a Si–O bond results in the simultaneous formation of both of these defects (i.e., one on the upper surface and the other on the lower surface). Note that, since no structural defects are observed in pristine bulk silica, the surface density of defects was calculated here from the total number of defects divided by the lateral cross section of the simulation box and, hence, does not depend on any arbitrary definition of the thickness of the surface. As shown in Fig. 3(a) , the surface density of NBO decreases monotonically with increasing annealing temperature. We observe that the surface densities of NBO and 3-fold coordinated Si atoms remain equal to each other upon annealing. This suggests that thermal annealing partially allows the atoms to locally reorganize and relax some of the structural defects by mutual recombination of NBO and Si iii defects, in agreement with previous results. 20,26,28,65 This reorganization partially induces the formation of ES units. We observe that such ES units may form when a 3-fold coordinated Si atom is initially directly connected to a Si atom comprising a NBO. The NBO then tends to create a bond with the undercoordinated Si, thereby resulting in the formation of a 2-membered ring structure. As shown in Fig. 3(b) , the surface concentration of ES units is found to increase with increasing annealing temperature.

Starting from the basic structure of bulk silica—wherein tetrahedral Si units are connected to each other through their four corners, i.e., through BO atoms—the artificial dry silica surfaces simulated herein exhibit various structural defects, including 3-fold coordinated Si atoms (Si iii ), terminating NBO, and edge-sharing (ES) Si units (i.e., two-membered Si ring), 9,18,53 in agreement with previous simulations. 28 It should be noted that, although they might exist in high-vacuum conditions, such completely dry surfaces (i.e., non-hydroxylated) are unlikely to be observed in realistic conditions of humidity since the silica surface will almost instantly get hydroxylated.

We now study the effect of thermal annealing on the reactivity of silica surfaces. Note that the thermal annealing phase is initially performed on the dry silica surfaces to generate a series of surfaces with different structures and degrees of stability. However, the hydration of these surfaces is subsequently performed at 300 K. Our simulations do not intend to reproduce the effect of heating on the de-hydroxylation of silica surfaces. As shown in Fig. 5(b) , we observe that the surface density of silanol groups decreases with increasing annealing temperature. This demonstrates that the reactivity of the artificial silica surfaces generated herein decreases upon annealing, which results from the relaxation of high-energy surface defects. This observation is in agreement with the outcomes of experiments performed on annealed silica samples. 16,17,19,75

(a) Density profile of 4-fold Si (pristine Si, denoted Si iv ), 3-fold Si (defected Si, denoted Si iii ), and H atoms as a function of the distance from the silica–water interface (in the absence of annealing). (b) Silanol surface density as a function of the annealing temperature. The line is guide to the eye.

(a) Density profile of 4-fold Si (pristine Si, denoted Si iv ), 3-fold Si (defected Si, denoted Si iii ), and H atoms as a function of the distance from the silica–water interface (in the absence of annealing). (b) Silanol surface density as a function of the annealing temperature. The line is guide to the eye.

After contact with water, we observe that some H atoms diffuse within the silica glass and form a partially hydrated layer with a depth of around 4 Å [see Fig. 5(a) ]. This induces a structural reorganization of the silica surface, in agreement with previous simulations. 22,25,49,52 In detail, we observe that all free terminating NBO, 3-fold coordinated Si, and 2-membered ring ES units that are initially present in the dry silica surfaces disappear upon the first picoseconds of hydration and convert into Si–OH silanol groups. This suggests that these structural defects are highly energetically unfavorable. In the case of the as-cut silica, i.e., with no thermal annealing, the silanol surface density is found to plateau at around 5.8 nm −2 . This agrees well with the range of values (from 4 to 6 nm −2 ) observed experimentally 16,17,74 and in previous simulations. 21,52

To assess the affinity of the hydroxylated silica surfaces with water, the undissociated water molecules were manually separated from the surface by inserting a void space (20 Å) between the silica surface and the water. This spacing was chosen to be larger than the ReaxFF potential cutoff (i.e., 10 Å) so that the atoms of silica and water effectively do not “see” each other any longer. Note that the silanol groups formed during hydration were still attached to the surface. The difference of potential energy, before and after separation of the water from the silica surface, was used to calculate the silica–water interface energy (or binding energy), which offers a direct estimation of the level of hydrophilicity of the silica surfaces. As shown in Fig. 6 , we note that the interface energy decreases with increasing annealing temperature. This shows that, although it is largely hydrophilic without any thermal treatment, silica becomes significantly more hydrophobic as its surface becomes more stable and exhibits fewer structural defects.

We now investigate the relations between structural defects, silanol density, and hydrophilicity. First, as shown in Fig. 7, we observe a positive linear correlation between the silica–water interface energy and the silanol surface density. This strongly supports the idea that the hydrophilicity of silica is primarily controlled by the number of silanol groups per unit of surface. This can be understood from the fact that silanol groups can interact with polar water molecules through the formation of hydrogen bonds (H-bonds), whereas BO (siloxane groups) are intrinsically hydrophobic.19,20,26 Namely, if the silanol groups are, on average, close enough to each other, each water molecule will create two H-bonds with the silica surface (hydrophilic surface), whereas, if the silanol groups are far from each other, only one H-bond per water molecule will be created (hydrophobic surface).76 By extrapolation, we find that silica would become fully hydrophobic (i.e., negative silica–water interface energy) for a silanol surface density of around 0.6 nm−2. This is in agreement with previous simulation data,20 suggesting a hydrophilic-to-hydrophobic transition occurring between 0 and 3.7 OH nm−2.

FIG. 7.

View largeDownload slide

Silica–water interface energy as a function of the surface density of silanol groups. The grey area indicates the range of surface density of silanol groups for which a hydrophobic silica surface is expected. The line is a linear fit.

FIG. 7.

View largeDownload slide

Silica–water interface energy as a function of the surface density of silanol groups. The grey area indicates the range of surface density of silanol groups for which a hydrophobic silica surface is expected. The line is a linear fit.

Close modal

Finally, we investigate how the silanol surface density and, consequently, silica’s hydrophilicity are controlled by the atomic topology of the surface. In the absence of any defects, Si and O atoms create 2 and 1 BS constraints, respectively [note that each BS constraint is shared by two atoms so that the number of BS constraints created by a given atom is equal to half of its coordination number (CN)]. In addition, Si atoms create 5 BB constraints, which correspond to the 5 independent angles that need to be fixed to define the Si tetrahedra. In contrast, O atoms do not create any BB constraints due to the large angular flexibility of the Si–O–Si bond angles.37,70,77,78 This leads to an isostatic structure, i.e., nc = 3 (see enumeration in Table I). In turn, the presence of structural defects (e.g., Siiii or NBO) tends to decrease the rigidity of the surface as compared to that of the pristine bulk.

TABLE I.

Summary of the constraint enumeration for Q4 (i.e., 4-fold coordinated Si atoms comprising no terminal oxygen atom), Siiii (i.e., 3-fold coordinated Si atoms comprising 3 BO atoms), and Q3 units (i.e., 4-fold coordinated Si atoms comprising one terminal oxygen atom). For each type of unit, the table summarizes the coordination number (CN) of Si, the number of bond-stretching (BS) and bond-bending (BB) constraints per unit, the total number of constraints (BS+BB), and the number of silanol groups per Si unit. The last row indicates the variation in these quantities when a Si–O bond is broken, which results in the transformation of two Q4 units into a Siiii and a Q3 unit (see text).

Unit

. 

CN

. 

BS

. 

BB

. 

If you are looking for more details, kindly visit Hebei Silicon Research Electronic Materials Co., L.

BS+BB

. 

No. of silanol

. 

Q4 4 4 5 9 0 Si

iii

 3 3 3 6 1 Q3 4 4 5 9 1 2Q4 → Si

iii

+ Q3  −1 −2 −3 +2 Unit

. 

CN

. 

BS

. 

BB

. 

BS+BB

. 

No. of silanol

. 

Q4 4 4 5 9 0 Si

iii

 3 3 3 6 1 Q3 4 4 5 9 1 2Q4 → Si

iii

+ Q3  −1 −2 −3 +2 View Large

To understand the linkages between surface topology and reactivity, we computed a “surface number of topological constraint per atom” nc*⁠, that is, the average number of topological constraints per atom of the surface before exposure to water. Note that the surface is defined here as a slab ranging from z = 0 to 4 Å, where z is the distance from the silica–water interface [see Fig. 5(a)]. Rather than only considering the rigidity of the “skeleton network,” that is, wherein terminating atoms are ignored, we explicitly accounted for all the atoms of the network herein (i.e., by including terminating O atoms in the constraint enumeration). Indeed, it was shown that, although the two enumeration schemes predict the same isostatic threshold (nc = 3), the explicit incorporation of 1-fold coordinated atoms into the constraints enumeration offers a more accurate description of the glass rigidity in the flexible (nc < 3) and stressed–rigid domains (nc > 3).79,80

Figure 8 shows the surface number of topological constraints per atom, nc*⁠, as a function of the annealing temperature. Overall, the obtained nc* values are found to be in the 2.25–2.75 range. This range of values is located between two notable extrema, namely, nc = 2, which is the minimum number of constraints per atom needed to obtain a rigid 1-dimensional structure, and nc = 3, which is the minimum number of constraints needed to obtain a rigid 3-dimensional structure.34 We observe that the rigidity of the surface (i.e., as captured by nc*⁠) increases upon annealing. This arises from the transformation of undercoordinated 3-fold Si and NBO atoms, which results in an increase in the overall connectivity of the surface.

Figure 9(a) shows the number of silanol groups per Si atom (limited to the Si atoms belonging to the surface) as a function of the surface number of constraints per atom nc*⁠. We observe that the number of silanol groups per Si atom decreases fairly linearly with nc* and eventually becomes zero when nc*=3⁠. This can be understood as follows. The nc*=3 threshold corresponds to the value obtained for pristine silica, that is, when the atomic network is fully polymerized and only comprise hydrophobic siloxane groups. In this situation, silica does not exhibit any “open” chemisorption site for water molecules to dissociate into hydroxyl groups. Starting from this ideal network, the breakage of a Si–O bond can be expressed as the following reaction 2Q4 → Siiii + Q3, where a Qn unit is defined as a 4-fold coordinated Si atom connected to n other Si units (i.e., comprising 4 − n terminating O atoms) and Siiii being a 3-fold coordinated Si (E′ center) connected to 3 other Si units (i.e., with 3 BO atoms). As summarized in Table I, this reaction induces the loss of 3 constraints (1 BS and 2 BB). In turn, the resulting Siiii and Q3 defects offer two potential additional chemisorption sites for silanol groups. After normalization by the number of atoms in the surface, one gets the following expression for the number of potential silanol groups per Si atom in the surface OH/Si:

OH/Si=2(3−nc*).

(1)

As shown in Fig. 9(a), the computed number of silanol groups (after hydration) agrees very well with the prediction based on the surface number of constraints per atom offered by Eq. (1) (i.e., relying on the atomic topology of the surface before hydration). This suggests that the short-term reactivity of silica (as measured by the surface density of silanol groups) can be fully predicted from the atomic topology of the unreacted surface (with no fitting parameter).

FIG. 9.

View largeDownload slide

(a) Number of silanol groups per surface Si atom as a function of the surface number of constraints per atom nc*⁠. The line is a theoretical prediction given by Eq. (1) (no fitting parameters, see text). (b) Silica–water interface energy as a function of the surface number of constraints per atom nc*⁠. The line is a linear fit. The grey area indicates the range of surface number of constraints per atom for which a hydrophobic silica surface is expected.

FIG. 9.

View largeDownload slide

(a) Number of silanol groups per surface Si atom as a function of the surface number of constraints per atom nc*⁠. The line is a theoretical prediction given by Eq. (1) (no fitting parameters, see text). (b) Silica–water interface energy as a function of the surface number of constraints per atom nc*⁠. The line is a linear fit. The grey area indicates the range of surface number of constraints per atom for which a hydrophobic silica surface is expected.

Close modal

Since the silanol surface density offers a good estimation of silica’s affinity for water (see Fig. 7), the correlation between silanol surface density and hydrophilicity suggests that the atomic topology of the surface controls the hydrophilicity of silica. The close relationship between surface hydrophilicity and surface topology is demonstrated in Fig. 9(b), which shows the water–silica interface (binding) energy as a function of the number of surface topological constraints per atom, nc*⁠. We observe that the affinity of the silica surface for water (i.e., as captured by the water–silica interface energy) decreases linearly with nc*⁠. This suggests that the more rigid the surface is (i.e., more connected, with higher nc* values), the more hydrophobic it becomes. Namely, flexible (low nc* values) and more rigid (⁠nc* values close to 3) surfaces are associated with hydrophilic and hydrophobic behaviors, respectively. Based on our results, we identify that the hydrophilic-to-hydrophobic transition occurs around nc*=2.82⁠.

We report on the optical and morphological properties of silica thin layers deposited by reactive RF magnetron sputtering of a SiO 2 target under different oxygen to total flow ratios [r(O 2 ) = O 2 /Ar, ranging from 0 to 25%]. The refractive index (n), extinction coefficient, total transmission, and total reflectance were systematically investigated, while field-emission scanning electron microscopy, atomic force microscopy, and three-dimensional (3D) average roughness data construction measurements were carried out to probe the surface morphology. Contact angle measurements were performed to assess the hydrophilicity of our coatings as a function of the oxygen content. We performed a thorough numerical analysis using 1D-solar cell capacitance simulator (SCAPS-1D) based on the measured experimental optical properties to simulate the photovoltaic (PV) device performance, where a clear improvement in the photoconversion efficiency from 25 to 26.5% was clearly observed with respect to r(O 2 ). Finally, a computational analysis using OptiLayer confirmed a minimum total reflectance of less than 0.4% by coupling a silica layer with n = 1.415 with another high-refractive-index (i.e., >2) oxide layer. These promising results pave the way for optimization of silica thin films as efficient antireflection and self-cleaning coatings to display better PV performance in a variety of locations including a desert environment.

1. Introduction

The front cover glass required for photovoltaic (PV) module insulation is the first surface to receive irradiation toward the solar cell, and the first surface to limit the photon flux impinging due to optical losses, which can be counteracted by means of antireflective (AR) coatings. The soiling adherence inherently disrupts the intended function of the AR coatings, thus reducing the power output of the PV plants.1−4 This issue is hence motivating the increased research attention toward AR and antisoiling coatings for photovoltaic solar panels, lasers, automobiles, optical instruments, and solar collectors.5−9

More specifically, the performance of PV modules has been found to be reduced by up to 22% due to dust encapsulation.10 A real-time measurement tool has recorded a performance drop of about 25%/μm for natural dust layer-encapsulated c-Si PV modules.11 Hence, it becomes critical to meet the conjunction requirements of enhanced antisoiling as well as antireflection properties. The improvement in optical and electrical efficiency is noticeable by integrating self-cleaning/antireflection coatings in PV modules.10−12 To develop an optimized AR layer, the film has to satisfy the destructive interference conditions for light waves reflected from the glass/coating and coating/air surfaces.8,9 However, control of the coating thickness “t” plays a central role in developing such films as it has to be one-fourth of the incident spectrum wavelength (λ/4 × nARC) for an optimized AR property. Additionally, the refractive index nARC of the AR layers should be (nS × nA)1/2, where nS is the refractive index of the substrate and na is the refractive index of air.10 The calculation showed that values around 1.22 and 120 nm of refractive index and the layer thickness, respectively, are required to meet the need for nS of around 1.5.

However, no material exists in nature with such a low refractive index value, whereas the lowest reported values are 1.39 for magnesium fluoride and 1.46 for silica.13 It has been demonstrated that introducing porosity in the coating can lead to the reduction of the refractive index. A three-dimensional (3D) crossed nanoporous layer has been found to be capable of Fresnel light reflection control with improved transmission.14 The physical vapor deposition process was used to grow such films on a low-alkali borosilicate glass with surface patterning to introduce superhydrophobicity.14 The contact angle (CA) was measured at 160° with a transmission range of 95%.14 Later, a report by Zhang et al.15 confirmed that transmission can be improved further to achieve ∼97% through a composition of silica–silica nanoparticle coating using the sol–gel technique. A remarkable refractive index of 1.21 has been successfully reported through controlled growth.15 However, the results showed superhydrophilicity with a CA of 5° with antifogging capacity.15 Such hydrophilic properties have been attributed to the impact of larger surface roughness. Hence, such results show a tradeoff between higher transmissivity and wettability. A lotus-leaf-like hierarchical structure results in measuring transmittance and water contact angle of 95% and 162°, respectively.16 Söz et al.17 demonstrated that a thickness of 125–150 nm is a prerequisite in developing superhydrophobic AR layers through chemical composition modification, resulting in fine-tuning of the topography and mean roughness of the surface. In addition, a report by Kintaka and co-workers18 studied the usage of methyl tri-ethoxysilane (MTES) and tri-methylethoxysilane (TMES) as precursors to developing hydrophobic AR layers, where the CA increases from 22 to 108°. The stacking layers of SiO2/TiO2 have also been investigated and have resulted in higher transmission, where titania nanoparticles resulted in a higher refractive index (n = 2.45) beyond the optimum value.19 However, although such layers demonstrated improved optical properties due to a minimal reflectance value, they were not qualified to be used as self-cleaning assemblies.

A silica thin film has the properties to show simultaneously high optical transparency and low refractive index values, suitable to be used as optical films from the near-ultraviolet to the near-infrared spectral range.20−23 This silicon oxide layer suppresses any parasitic light absorption in the visible range due to its suitable optical band gap and transmission range.19−22 Also, low refractive index values allow these layers to be coupled with high-refractive-index materials for antireflection coating applications by minimizing the total light reflectance.20−22 In addition, due to their high optical transparency, high hydrophobicity property, and the uniformity of their surface, silica layers are extensively used in various industrial and general-purpose applications including flexible displays, optics, bioengineering, ophthalmic, and energy generation.10−15,19−40

Many techniques have been already employed to grow silica films, including electron beam evaporation, ion-assisted deposition, ion beam sputtering, magnetron sputtering (MS), sol–gel, and atomic layer deposition.19−27 Among all of these techniques, magnetron sputtering (MS) demonstrated the best deposition competency due to the fine control over the growth parameters, film-uniformity, homogeneity, the low density of pinholes to resist defects, and stable optical properties.23,27−29 Hence, nowadays, MS has become a state-of-the-art technique to grow pristine silica films even with large-scale capabilities. Moreover, having the ability to perform the reactive growth of these films under oxygen or any other reactive atmosphere help in controlling the film’s stoichiometry. The stable control over the growth process using the MS technique has eventually enabled many photovoltaic (PV) industries to grow thin layers of silica layers as antireflection coatings for PV modules.23,27−29 In general, growth parameters play a tremendous role in understanding film quality, which becomes critical even with the stoichiometry ratio.23,27−29 Reactive MS deposition adds an advantage to grow oxygen-rich or oxygen-poor silica films with nanoporous morphology, where process gas has a direct effect on altering the layer morphology.24,28−30

In this work, we have investigated the optical and microstructural properties of 100 nm silica thin films grown on a glass substrate by the radio-frequency (RF) sputtering technique. The main goal was to develop an optimized deposition recipe of silicon oxide thin films (SiOx) with a controllable roughness using the sputtering system for antireflection and antidust coating applications. More specifically, we aim at pointing out the multiple correlations existing between the deposition parameters (i.e., mainly oxygen content), the morphological and structural properties of the grown films, their optical characteristics, their surface wettability properties, and the associated output PV performance. In addition, the effect of oxygen plasma treatment of glass is studied and results are compared with deposited silica film properties. Furthermore, we simulated the effect of multistack layers with various refractive indexes against PV performance. The thin films were grown at a constant process temperature of 200° C under different oxygen to total flow ratios [r(O2) = O2/Ar, ranging from 0 to 25%]. A maximum oxygen flow of r(O2) of 25% has been set based on the chamber’s set pressure tolerance level to prevent any arcing issue. The associated oxygen flows bring the deposition pressure to 1.1 × 10–3 Torr, considering the sputtering deposition chamber size. Hence, it has been advised to not reach high-pressure (>1 × 10–3 Torr) reactive sputtering, which provides a high degree of ionization of the sputtered species to control the arc of the target. Ellipsometry and UV–vis spectroscopy were used to extract the optical properties, while contact angle measurements were performed for hydrophilicity properties. Furthermore, 3D roughness measurements by Dektak Stylus and contact-angle atomic force microscopy (AFM) measurements were applied to probe the surface topology and field-emission scanning electron microscopy (FESEM) was used for determining the film’s microstructure. Finally, numerical simulations were carried out using SCAPS-1D and OptiLayer simulators to implement the experimentally measured optical properties (especially the coupling with various obtained refractive indexes) to determine their effects on the solar cell device performance. Our simulated results demonstrated a clear improvement in the photoconversion efficiency from 25 to 26.5% with respect to r(O2) in the silica films and enabled us to computationally screen oxide materials and assess their potentials before testing them experimentally with energy conversion devices.

If you are looking for more details, kindly visit hydrophilic silica.