Nano-materials enhanced protectants for natural stone surfaces

Most heritage buildings and monuments are constructed out of natural stones which suffer irrevocable degradation when undergoing wet weathering, bowing, and dissolution in outdoor conditions. Self-cleaning treatments are effective for stone protecting. Herein, nano-materials which provide enhanced protectants for Marble, Qingshi and Hedishi were prepared. Inherent microscale interstices and holes exist on polished natural stone surfaces. When treated by a commercial protectant, 101S, the surfaces were hydrophobic but not self-cleaning. Colloidal protectants were prepared by dispersion of Al2O3 and SiO2 nano-powder in 101S, respectively. Self-cleaning stone surfaces were achieved after treated by the protectants, meanwhile, the interstices and holes were reserved as much as possible. The principle of the as- prepared protectants is penetrating and crosslinking on the stone surfaces as well as the inner surfaces of the interstices and holes. The reserving of the micro interstices and holes is important since the breathability of the stones is remained. The self-cleaning surfaces showed good thermal stability below 250 °C. Meanwhile, changes of color and gloss of the treated stone surfaces are in the acceptable range.


Introduction
Heritage is one of the most important assets with significant cultural and economic value. Most heritage buildings and monuments are constructed out of natural stones due to their earth-abundance, robustness and harmlessness. In outdoor conditions, natural stones suffer irrevocable degradation when undergoing wet weathering, bowing and dissolution [1,2]. Generally, rain has a pH value varying from 5.6 where acidity results from ambient CO 2 in normal conditions, to 4.5 in areas polluted by SO 2 and NO x [3]. The dissolution rate of salts such as calcite and magnesite increases with decreasing of the pH level of the invaded water. Water containing acidic species invades through the micro interstices and holes are harmful due to the crystallization-dissolution cycles of the soluble salts in the stones which induce tensions inside the porous matrix, thus accelerating its decay [3,4]. Hence, studies on conservation of cultural heritage, especially of natural stones are significant for its future continuation.
Two strategies are employed in the field of stone conservation, i.e., consolidation and surface protection [5][6][7][8][9][10][11]. Consolidating treatments aim at improving the cohesion of degraded stones. Alkoxysilanes, tetraethoxysilane (TEOS) and methyltrimethoxysilane (MTMOS) based organic protectants are dominant in the consolidating practices, mainly due to their ability to penetrate easily into porous materials and the light impact on the permeability and drying properties of the stones. The main issues with such products include cracking during the drying due to gel shrinkage, temporary hydrophobic behavior, influence of the ambient humidity on the gel polymerization and poor chemical affinity between protectants and the stone substrates [12,13]. Inorganic consolidating products, such as calcium hydroxide, barium hydroxide and ammonium oxalate have good durability and compatibility with stone components [14][15][16]. By using these products, lower penetrability and subsequent poor strengthening effects should be considered. In recent years, colloidal dispersions containing silica and calcium, magnesium, strontium hydroxides nanoparticles were used for stone consolidation [17][18][19][20][21]. Although great improvements had been achieved, a high level of risk may be posed, e.g., introduction of the products into the porous network may strongly change the stone characteristics and properties, thus causing unwanted effects or further damage. Other issues may be encompassed in consolidation treatments. Consolidates made of water colloidal suspension of nanosilica were applied on a porous limestone. The stone cohesion was improved by the penetration and consolidation of the consolidates, however, a decrease of durability under a salt crystallization test of the treated stone was found, suggesting that the distribution of the nanoparticle fillers into the stone substrate should be improved [12]. Besides, the efficacy of consolidation with colloidal dispersions strongly depends on humidity and climatic fluctuation [13].
For the surface protection strategy, treatments applied on the cortical portion of the stones and water-repellency had been considered [22][23][24]. Enclosed coatings of acrylic, acryl-siliconic and epoxy resin as well as paraffin were generally adopted [25][26][27][28][29][30][31]. Although these coatings can effectively isolate the pollution sources, the breathability, or the water vapor permeability, of the stones might be greatly weakened especially when a high amount of these coatings are applied. As a consequence, with the variation of the ambient temperature, the sealed moisture inside the stones experience cycles of salts crystallization-dissolution which will also accelerate the decay of the stones. In heavily salt-laden cases, simple application of seal coatings can even increase the deterioration rate of the treated stones, leading eventually to flake, scale, and cracking [32]. Moreover, the durability of this kind of coatings is limited due to the relatively weak adhesion with the stone substrates. Hence, protectants combine water repellence, robust adhesion and breathability maintenance is in need.
Self-cleaning surfaces are effective for natural stone protection due to the excellent water repellence and breathability maintaining possibility [22,33]. A combination of micro-nano roughness and low-surface-energy chemicals derivatization is necessary to self-cleaning surfaces [33,34]. Inherent micro roughness existed on polished natural stone surfaces, and addition of nano particles in protectants is effective for nanoscale roughness construction. Silica nanoparticles were added in polyalkylsiloxane and water static contact angle (SCA) of 160° on treated Marble surface was obtained [18]. Another protectant was prepared by adding nano CaCO 3 nanoparticles in copolymer of epoxy and acrylate [35][36][37]. Performances of robustness, abrasion, contamination resistance, weathering resistance, and durability of natural stones protected by nanomaterials enhanced protectants (NMEPs) have been greatly improved. Meanwhile, the antibacterial and antifungus properties were enhanced as well.
In present work, an approach to NMEPs was developed. Al 2 O 3 and SiO 2 nano-particles were dispersed in a commercial water-resistant coating, 101S, to prepare colloidal protectants. Protectant layers were formed by dip-coating on natural Marble, Qingshi and Hedishi stone surfaces as well as the inner surfaces of interstices and holes beneath the surfaces of several micrometers through pernetrating and crosslinking. Water resistance of the stone surfaces was greatly improved without heavily sacrificing the breathability of the stones. Self-cleaning surfaces were achieved with water SCA of bigger than 150° and hysteresis angles (HAs) less than 20°. The selfcleaning surfaces showed good thermal stability below 250 °C. Meanwhile, changes of color and gloss of the treated stone surfaces are in the acceptable range.

Materials and sample preparation
Bulk natural stones of Marble, Qingshi and Hedishi were obtained from the local area of Dali prefecture, Yunnan province, China. Commercial water-resistant coating, 101S, was obtained from Solmont Technology (Shenzhen) Co., Ltd. 101 s is a clear and transparent liquid coating with a surface tension of 14 mN/m (at 20 °C), mainly composed by perfluoroalkylpolyether (PFPE) in which -CF 2 and -CF 3 are the water-repellent groups. The coating can easily penetrate into interstices and holes inside natural stones to form a water repellant layer via crosslink at room temperature in air within 48 h. Al 2 O 3 (99.9%, 30 nm) and SiO 2 (99.5%, 15 nm) nano-powders were bought from Shanghai Macklin Biochemical Technology Co., Ltd and Shanghai Aladdin Bio-Chem Technology Co., Ltd, respectively.
The stones were cut into 50 mm × 15 mm × 5 mm coupons. Double faces of each coupon were primarily polished by 400 # grinding wheels and pasted on glass slides by UV curing adhesive. Surfaces of natural stone specimens were prepared by sequentially polishing the coupon surfaces by 1200 # grinding wheels and polishing cloth followed by a 15 min ultrasonic washing in ethanol and DI water (18.2 MΩ, MilliQ) respectively to eliminate organic native oxide and contaminations. The specimens were dried for 24 h under 40 °C in a drying box followed by a 3 min UV-ozone cleaning prior to treatment.
The nano-powders were heated at 100 °C on a heating plate for 1 h to eliminate moistures prior to use. Colloidal dispersion protectants were prepared as follows. Concentrations of 0.1 mg/mL, 0.5 mg/mL, 1.0 mg/mL, 1.5 mg/ mL and 2.0 mg/mL of Al 2 O 3 and SiO 2 nanoparticles were dispersed into the 101S liquid respectively by a five ultrasonic dispersion periods followed by a magnetic stirring for 2.5 h. In each period, alternated ultrasonic of 28 kHz and 40 kHz for 5 min were adopted. The cleaned natural stone specimens were immersed into the protectants for 14 h and dried at 40 °C for 24 h. The preparation diagram of the samples is shown in Fig. 1.

Surface characterization
X-ray diffraction (XRD) was employed to analyze the composition of the natural stones. Samples were performed on stainless steel holders in an EMPYREAN, PANalytical, NL with λ = 1.5406 Å of Cu K α radiation. XRD patterns were recorded by scanning from 5° to 90° with a step of 0.02° and a rate of 0.2°/sec in a continuous mode. The working conditions are 40 kV and 40 mA. Scanning electron microscopy (SEM) images were obtained with a SU8020 on gold-sputtered surfaces under 3.0 kV accelerating voltage, 10 -5 Torr vacuum and 3.2 mm working distance. Water contact angles (CAs) were measured with a SDC-200 contact angle measurement (CAM) system. All CAMs were carried out with DI water at ambient temperature of 23-25 °C and relative humidity of 60-75%. The volume and diameter of the DI water droplets used are about 3 μL and 0.9 mm, respectively. On each surface, at least five spots were tested and the averaged value was adopted [38,39]. Thermal stability, breathability and color and gloss changes of the treated surfaces were evaluated and detailed in the following section.

Stone composition
XRD patterns of the three natural stones are shown in Fig. 2 and corresponding main ingredients of the stones are summarized in Table 1. The dominant ingredient of Marble is CaCO 3 , while that of Qingshi and Hedishi is SiO 2 . Graphite and ferric oxides exist in the stones of Qingshi and Hedishi. However, the amount of these species is too small to be detected by XRD.

Water repellence of 101S
In order to evaluate water repellence of 101S, and minimize the affection of surface roughness, water CAs of blank and 101S coated glass surfaces were measured and summarized in Table 2. Water SCA of 62.6 ± 0.3°  [40].
Numerous micro holes and interstices existed on and beneath the natural stone surfaces. Liquid 101S can easily penetrate into the interstices and holes due to its low surface tension of 14 mN/m (20 °C), and then crosslinked on the stone surfaces as well as the inner surfaces of the interstices and holes. The inherent microscale roughness enhanced the water repellence. However, a difference existed from self-cleaning surfaces which require water sliding angle less than 5° (or the HA less than 20°). Self-cleaning surfaces cannot be achieved by 101S treatments and can only be achieved by a combination of low-surface-energy chemicals and surface micro-nano structures.

Surface wettability Wettability of stone surfaces treated with Al 2 O 3 nano-powder added 101S
The concentrations of the Al 2 O 3 nano-powder added in 101S are 0.1 mg/mL, 0.5 mg/mL, 1.0 mg/mL, 1.5 mg/ mL and 2.0 mg/mL. Figure 3a-c present variations of water CAs on the treated stone surfaces with the concentration of the Al 2 O 3 nano-powder. For Marble surfaces (Fig. 2a), when 0.1 mg/mL was dispersed into 101S, water SCA is 166.5°, significantly increased compared with that of 139.3° on the surface treated by pure 101S. Meanwhile, the HA decreases from 34.1° to 11.2°. The SCA continuously increases to the maximum value of 171.2° with the concentration to 1.5 mg/mL. When the concentration of Al 2 O 3 nano-powder reaches to 2.0 mg/mL, the SCA decreases to 166.8°. For the HA, the minimum value obtained when the concentration of Al 2 O 3 nano-powder is 0.5 mg/mL, it increases to 14.6° with the concentration of Al 2 O 3 nano-powder to 1.5 mg/mL, and then decreases to 12.1°. On the treated Qingshi surfaces, the SCA reaches to 161.7° when 0.1 mg/mL Al 2 O 3 nano-powder was added to 101S. The SCA first decreases and then increases with the concentration of Al 2 O 3 nano-powder thereafter. The maximum SCA of 169.2° was obtained when the concentration of Al 2 O 3 nano-powder is 1.5 mg/mL. The HA various in the range of 11.3°-14.6° with the concentration of Al 2 O 3 nano-powder. The treated Hedishi surfaces present excellent water repellence with SCA of 170.3° when 0.5 mg/mL Al 2 O 3 nano-powder was added in 101S, meanwhile, the HA is 16.8°, slightly bigger than those on Marble and Qingshi surfaces. In the range of 0.5-2.0 mg/mL, although the SCA slightly decreases with the concentration of Al 2 O 3 nano-powder, selfcleaning surfaces were achieved.
The optimal SCAs and corresponding concentration of Al 2 O 3 nano-powder are summarized in Fig. 3d. The optimal SCA on treated Qingshi surface is relatively smaller and the corresponding concentration of Al 2 O 3  Table 2 Water SCAs, advancing contact angles (ACAs), receding contact angles (RCAs) on glass and natural stone surfaces and corresponding HAs before and after 101S treatment nano-powder is bigger than those of treated Marble and Hedishi surfaces. The result is related to the difference of the surface pattern between Qingshi and Marble (and Hedishi) which will be depicted later.

Wettability of stone surfaces treated with SiO 2 nano-powder added 101S
Nano SiO 2 powder was used as another additive to 101S for protectants. Variations of water CAs as a function of the concentration of SiO 2 nano-powder are shown in Fig. 4a-c. Water repellence of the treated Marble surface was significantly improved when 0.1 mg/mL SiO 2 nano-powder was added into 101S, i.e., the SCA increased to 166.7° compared with that of 139.3° in the SiO 2 -free case, meanwhile the HA decreased from 34.1° to 11.2°, presenting self-cleaning. This property was maintained in the SiO 2 nano-powder concentration ranging from 0.5 to 2.0 mg/mL with a slightly increase of the HA (Fig. 3a). For the Qingshi surfaces, the variations of the SCA and ACA are similar to those on Marble surfaces. However, the RCA is relatively small in the SiO 2 nano-powder concentration ranging from 0.1 to 0.5 mg/mL. Self-cleaning surfaces were achieved when the concentration reaches to 1.0 mg/mL. The variation of the water CA on treated Hedishi surfaces is similar to that on Marble surfaces. In the whole concentration range of SiO 2 nano-powder surveyed in this work, selfcleaning property is well maintained. The optimal SCAs and corresponding concentration of SiO 2 nano-powder are summarized in Fig. 4d. Similar to the surfaces treated by Al 2 O 3 nano-powder added 101S protectants, the optimal SCA on treated Qingshi surface is relatively smaller and the corresponding concentration of SiO 2 nano-powder is bigger than those on the treated Marble and Hedishi surfaces. We suggest the result is induced by the different patterns of It is demonstrated that water repellence of the treated stone surfaces was significantly improved by dispersion of Al 2 O 3 and SiO 2 nano-powder in 101S. Self-cleaning surfaces were achieved for all stone specimens except for Qingshi surfaces in the cases of the concentration of SiO 2 nano-powder below 1.0 mg/mL. Numerous micro holes and interstices existed on and beneath the natural stone surfaces. The as-prepared liquid protectants penetrated into the holes and interstices and crosslinked to form a water-resistant layer on the inner surfaces as well as on the outside surface. The addition of Al 2 O 3 and SiO 2 nano-powder provides nanoscale roughness on the surfaces. Hence, self-cleaning surfaces were achieved by functions of the 101S and the nanoscale roughness together with the inherent microscale roughness on the stone surfaces.

Mechanism of the water repellence
Based on the Young's equation, Wenzel and Cassie developed two models to explain the influence of surface roughness on the apparent contact angle of liquid droplets. The Wenzel model suggests that surface roughness will increase the actual contact area of the liquid and amplify the wettability of solid surfaces, and the apparent contact angle can be modified as [41,42]: where θ is the apparent contact angle of droplets on the rough surface; r is the surface roughness factor ( r > 1 ) which is a measure of how surface roughness affects a homogeneous surface and defined as the ratio of actual contact area with the liquid of the solid surface to the geometric area in contact with the droplet; θ 0 is the equilibrium contact angle of the droplets on an ideal smooth (1) cos θ = r cos θ 0 Fig. 4 Variations of water CAs on a Marble, b Qingshi and c Hedishi surfaces treated by the as-prepared protectants with the concentration of SiO 2 nano-powder, and d the optimal water SCAs on the stone surfaces surface of the same material, given by Young's equation as cos θ 0 = (γ sv − γ sl )/γ lv , where γ refers to the interfacial tension and the subscripts s , l , v refer to the solid, liquid, and vapor phases, respectively. While the Cassie model postulates that a hydrophobic rough surface will trap air bubbles in micro-pockets at the solid-liquid interface, leading to a composite interface, and giving the apparent contact angle as [43,44]: where f is the ratio of the solid area in contact with the liquid to the geometric area in contact with the droplet (f < 1).
According to Wenzel model, in order to achieve selfcleaning surfaces, θ 0 must be bigger than 90°, i.e. the solid surface must be hydrophobic. In contrast, the Cassie relation allows for the possibility of θ > 90 • even with θ 0 < 90 • which is generally existed when water droplets contact with hydrophilic solid surfaces. In addition, according to Wenzel model, water will penetrate into the micro-pockets at the solid-liquid interface when contact with solid surfaces. As a consequence, water droplets can hardly sliding or the sliding angle is usually large on the surfaces. Hence, Wenzel model can hardly fulfill the sliding requirement of self-cleaning surfaces, i.e. the sliding angle is less than 5° (or the HA is less than 20°). On the other hand, to achieve Cassie state, low surface tension material is in need. Hence, for self-cleaning surfaces, hierarchical micro/nanometer scale roughness and low surface tension material are necessary. Figure 5 shows SEM images of Marble surfaces and profiles before and after treated by the protectants. Nonuniform micro interstices and holes exist on the polished blank Marble surface. The width of the interstices as well (2) cos θ = f (cos θ 0 + 1) − 1 as the diameter of the holes is in the range of several hundred nanometers, and the length of interstices is mainly several micrometers (Fig. 5a). The depth of these interstices and holes is ranging from nanometers to micrometers (Fig. 5e). The micro interstices and holes almost maintain unchanged after treated with 101S ( Fig. 4b and  f ). During the treatment, the liquid protectants stayed on the Marble surface and penetrated into the interstices and holes, and formed a protective layer after crosslinking of the 101S. The morphology of the Marble surfaces didn't change when covered by the protective layer. That is why the SCA increased from 53° to 139.3° instead of 118°on the 101S covered smooth glass surface. In other word, the microscale roughness on the Marble surface enhanced the water repellence. However, the corresponding HA of 34.1° is relatively big, much below the standard of self-cleaning surfaces. Hence, only microscale roughness on the Marble surface is fall short of the structural request of the self-cleaning surfaces. When 1.5 mg/mL Al 2 O 3 nano-powder was dispersed into 101S, nanoscale embossments formed on the microscale pattern of the Marble surfaces ( Fig. 5c and  g). Self-cleaning surfaces were obtained by combining the micro-nano roughness and the low surface energy of 101S. The interstices and holes were maintained except a little size decrease after treatment. It is important because stones will breathe through the interstices and holes in ambient environment, i.e., moistures outer and inner the stones can exchange through these interstices and holes and the possible breakdown of the stones induced by ambient temperature will be decreased.
The present of 0.5 mg/mL SiO 2 nano-powder addition is similar to that of Al 2 O 3 nano-powder except a little difference of the morphology, i.e., although the size of the embossments is bigger than that induced by Al 2 O 3 nanopowder, the interstices and holes were also maintained ( Fig. 5d and h).
SEM images of Qingshi and Hedishi surfaces are shown in Figs. 6 and 7. The morphology of Hedishi surfaces is similar to that of Marbles while the morphology of the Qingshi surfaces presents step-like or fish scale like morphology in which interstices and holes are not distinct. The unique morphology is the reason of the difference request of the nano-powder concentration to self-cleaning from that of Marble and Qingshi.
Based on the models of Wenzel and Cassie [41][42][43][44] and the morphology of the stone surfaces above-mentioned, a conclusion that Cassie state of water droplets stay on the self-cleaning surfaces can be drawn. The fraction, f , of the solid in contact with the liquid droplet is calculated according to Eq. (2) and summarized in Fig. 8. The average fraction is about 5%, which is a receivable value of droplets in Cassie state when contact with a solid surface [38][39][40]45]. The micro/nanoscale hierarchical structures can trap a large amount of air, which can prevent the penetration of water into the grooves and bestow selfcleaning surfaces.

Water vapor permeability
Porosity and especially the pore space geometry play a great role in the migration of fluids and vapor inside the stones. Small reduction of the pore space in the treated specimens did not have significant influence on the permeability of water vapor [12]. In present work, water vapor permeability was measured by the cup test method. For one kind of stone, two 3 mm thick slices were cut and polished with the same process mentioned above. One slice was treated with the as-prepared protectants, another one was maintained blank as a control specimen. The slices were adhered on cups (with 4 mL   water inside) by glass cement, acting as the lids. The caliber of the cup is 7 mm. After 48 h solidification of the glass cement, the cups were put into a drying box at 30 °C. Mass of the cups was measured by an electronic balance in every period (24 h each). The experimental diagram is shown in Fig. 9a. Relative mass of the cups lidded with stone slices were presented in Fig. 9b-d. All slices of Marble, Qingshi and Hedishi maintained water permeability after treated with protectants, although it is slightly decreased. Water vapor permeability is defined as the mass of water vapor transmitted through a sample per area unit in a time unit (24 h) under defined conditions, and describes the ability of a material to allow water vapor passing through. The following equation was used to calculate the water vapor permeability [12]: where ΔM is the weight change in the steady state (expressed in g), A is the exposed area to water vapor (in m 2 ) and t is the unit time (24 h). In all the cases, the used ΔM was the average of three consequent values of the daily difference in weight.
The calculated permeability to water vapor was found to decrease after treated by protectants. For marble slices, reduction of 14.6% and 32.8%was found after treated with colloidal 101S containing1.5 mg/mL Al 2 O 3 and 0.1 mg/mL SiO 2 nano-powder. The reductions are slightly higher compared with that of the water colloidal suspension of nanosilica treated Marble which is 26%. Reductions of 53.1% and 50.5% was found for Qingshi stone after treated with 1.5 mg/mL Al 2 O 3 and 1.0 mg/mL SiO 2 dispersion in 101S. While that of Hedishi are 55.5% and 40.2% after treated with0.5 mg/ WVP = �M (t × A) mL Al 2 O 3 and 1.0 mg/mL SiO 2 dispersion in 101S. It is evident how different results may be obtained from the interactions between nanoparticle dispersions and porous substrates, as a function of a complex of factors, which include the characteristics of both stones and protectants, the application methods and the environmental conditions of the treatments, as well.

Apparent color variation
Usually, color changes induced by protectant treatments are evaluated by a colorimetric analysis using a reflectance colorimeter with a CIE standard illuminant [12]. The measured parameters are L*, a* and b*, where L* accounts for luminosity, a* and b* are coordinates of red-green and blue-yellow. Total color difference (ΔE*) can be calculated by ΔE* = ((ΔL*) 2 + (Δa*) 2 + (Δb*) 2 ) 1/2 . Inherent rich colors and texture patterns exist in the natural stone surfaces. Different parts of the surfaces appear different colors. Hence, it is difficult to evaluate color change by a colorimetric analysis. In present work, the as-prepared colloidal protectants are clear and transparent. The color variation of treated stone surfaces with the protectants is reliably small. Figure 10 shows the optical photos of the blank and treated stone surfaces taken on a same background. Except the inherent difference of the texture patterns of the natural stones, the color of the surfaces shown in Fig. 10 is almost unchanged. Although a colorimetric analysis is not done, we believe the color changes are within the acceptable range, i.e., the total difference is lower than five [13] or lower than three according to other authors [46,47] who coated similar colloidal dispersions on stone surfaces. The gloss of the

Thermal stability and durability
In order to survey the thermal stability of the as-prepared self-cleaning surfaces, specimens of Marble (treated  by 101S containing 1.5 mg/mL Al 2 O 3 and 1.5 mg/mL SiO 2 , respectively), Qingshi (treated by 101S containing 0.5 mg/mL Al 2 O 3 and 1.0 mg/mL SiO 2 , respectively) and Hedishi (treated by 101S containing 2.0 mg/mL Al 2 O 3 and 2.0 mg/mL SiO 2 , respectively) surfaces were annealed at 100, 150, 200, 250 and 300 °C for 15 min on a hot plate in air, respectively. After naturally cooled to room temperature, the surfaces were subjected to water CAMs. Figure 11 shows water CAs of the control (without annealing) and the annealed surfaces. The ACAs and SCAs of the surfaces maintained almost unchanged below 250 °C and decreased sharply when the temperature is higher than 250 °C. The RCAs experienced a period of increase below 200 °C and then decrease thereafter. The biggest RCAs rose in the period of 150-200 °C. The surfaces became hydrophilic when annealed at 300 °C, a reflection of decomposition of the 101S. The results agree well with the thermal stability of surfaces coated by fluorinated low surface tension chemicals, indicating the good thermal stability of the as-prepared selfcleaning surfaces [38][39][40].
The treated stone specimens were placed in outdoor conditions for 9 months (from October 2020 to July 2021), and water CAs were measured. The results are summarized in Table 3 and 4. Compared to the initial data, the SCAs are almost unchanged, while the HAs are slightly increased. The results shown in Table 3 and 4 indicate that the durability of the treated stone surfaces is good and the process developed in this work has actual application value.

Conclusions
Inherent microscale interstices and holes exist on the polished natural Marble, Qingshi and Hedishi surfaces. When coated by the commercial protectant, 101S, the surfaces were hydrophobic but not self-cleaning. After being treated by colloidal dispersions of Al 2 O 3 and  SiO 2 nano-powder in 101S, self-cleaning surfaces were achieved. Meanwhile, the interstices and holes were reserved.
The nanoparticles provided nanoscale embossments on the inherent microscale structures on the natural stone surfaces to construct micro-nano hierarchical roughness which is the structural base of the self-cleaning surfaces. The principle of the protectants prepared in this work is penetration and crosslinking on the stone surfaces as well as the inner surfaces of the interstices and holes. The reservation of the micro interstices and holes is important since the breathability of the stones can be remained as much as possible. The self-cleaning surfaces showed good thermal stability below 250 °C and damaged with the annealing temperature elevation thereafter.
For the optimal self-cleaning surfaces, the water vapor permeability decreased about 24% for Marble, while that for Qingshi and Hedishi are about 52% and 48%, respectively. The color of the treated surfaces is almost unchanged compared to the blank ones. Little decreases of water CA of the treated stone surfaces were observed after 9 months aging in outdoor conditions, indicating good durability of the self-cleaning surfaces.
Nano-powder addition into water repellence protectants or nanomaterial enhanced protectants provides an easy and effective approach to self-cleaning surfaces of natural stones which will provide an avenue to industrial production for stone surface protection.