Sandblasting Wood as a Technique of Simulated Weathering

Featured Application

Certain customers today demand wood elements that look as if they have been exposed to nature, i.e., simulated weathered wood products. It has been shown that controlled sandblasting can replicate aged structures, which is useful for restoration or aesthetic design. Our study can help manufacturers to select the most suitable wood species and orientation of elements for simulated weathering to fulfil consumer requirements and expectations.

Abstract

While most efforts are aimed at preventing the surface roughening and colour change of wood due to weathering, some customers, mainly for decorative reasons, want wooden objects and elements to give the impression that they have been weathered for a long time. In this study, the simulated weathering of numerous softwoods as well as ring-porous and diffuse-porous woods by sandblasting and greying with iron sulphate was investigated. Calculations of the correlations between wood density, orientation, mass loss and thickness reduction by sandblasting were performed, and the difference between the hardness of late and early wood and the surface profile parameter P_t showed that the surface profiles correlate strongly with mass loss, especially in the tangential orientation. Softwoods appeared to be the most promising for simulated profiling, especially spruce and larch with tangential surfaces. Among the ring-porous woods, oak and sweet chestnut also delivered good results.

1. Introduction

Wood in outdoor areas is exposed to the weather. Kropat et al. [1] defined weathering as a change in the appearance and surface properties of wood when it is exposed outdoors. The process of weathering and its consequences have been the subject of extensive research and numerous reviews and research articles dealing with the effects of weathering on wood. For example, the mechanisms of photodegradation were explained in the review paper by Chen and co-authors [2], and changes in colour, surface profile and hardness are described in a review paper by Sahin et al. [3], which focused on the suitability of wood for landscape applications. Additionally, a review on improving the weathering resistance of wood can also be found in the literature [4], as well as studies on the weathering of tropical hardwood substrates [5]. Further, weathering of aged wood from historical wooden buildings has also been reported [6], and the weathering of wood has been studied using a holistic approach [7], as well. Last but not least, an analysis of chemical changes on the surface of a material similar to wood—moso bamboo—irradiated with UV light has also been carried out [8]. The main factors leading to weathering are solar radiation (ultraviolet and visible light), oxygen, water, heat, particulate matter, environmental pollutants and microorganisms like bacteria and decay fungi [9]. The consequences of the weathering of wood were described in detail in the review by Kropat et al. [1], which cites numerous references on this subject. The most important changes concern the colour of the wood, the surface profile, the appearance of surface cracks and fissures, the change in mechanical properties and the increased wettability of the wood; some of these changes are the result of the effects of weathering on the chemical composition and structure of lignin, cellulose and hemicelluloses.

The most noticeable changes in wood surfaces due to weathering, especially sunlight, are those of the surface colours. According to numerous references cited in the review by Kropat et al. [1], the changes begin with an initial yellowing followed by a deeper brown colouration. A stable grey colour layer then develops, which can be 2 mm to 5 mm thick. This result is probably relatively independent of the original colour of the wood. After a longer period of time, a gradual darkening and blue discolouration occur. The discolouration of the wood is the result of the degradation of lignin [2], which is particularly sensitive to radiation in the UVB range. Phenoxyl radicals and quinone compounds are formed, which are responsible for the colour change. Lignin depolymerises to form certain chromophoric groups, such as o-quinone and p-quinones [6]. In the same review [2], it is also written that the degradation of certain extractives, especially polyphenolic compounds, under UV irradiation causes the formation of quinonoid and carbonyl groups, which leads to discolouration. However, the mostly uneven greying of weathered wood is not only due to the described formation of chromophore groups. The wood surfaces turn grey due to dust particles penetrating the porous structure of the wood and due to fungi and moulds [10,11]. Leached lignin fragments are also displaced to the top layer [8]. Photodegraded and leached wood surfaces consist mainly of cellulose [4,8,12], which is also the reason for the formation of the silver-grey colour.

Another phenomenon related to the weathering of wood is the increase in the profiling of wood surfaces. The erosion rates of wood during natural weathering were studied and reported some time ago [13]. Rain can leach the wood surface and wash out photo-degradation and hydrolysis products, leading to mass loss and erosion of the wood [7]. As erosion occurs faster in early wood than in late wood, a wavy, corrugated surface pattern is created [9].

While wood is typically protected from weathering through various intentional measures, some customers—primarily for decorative reasons— prefer wooden items and components to appear as they have been naturally aged by long-term exposure to weather [1,9]. This can be achieved by simulated weathering, which aims to quickly transform the appearance of fresh wood into weathered wood [1]. According to Kropat et al. [1], simulated weathering can include the treatment of wood surfaces with ferrous compounds to achieve a grey surface colour, treatment with acids that cause discolouration, bleaching, treatment with selected stains, fuming with ammonia, alkali treatments, thermal treatments and various mechanical treatments, including sandblasting.

The focus of our investigation was on the preparation of profiled, i.e., corrugated, and grey wood. Sandblasting was expected to produce a pattern in which the more resistant parts of the wood, such as knots and dense areas such as late wood, protrude outwards in relation to the less resistant parts of the wood [1]. While sandblasting of uncoated wood is very rarely reported in the scientific literature [14], there is much more information on the greying of wood with iron compounds. Jankowska and Kwiatkowski [15], for example, carried out artificial greying of European oak wood with an iron(II) sulphate solution. For the same purpose, surfaces of white oak were treated with iron(III) sulphate solutions [16]. The colouring effect of ferrous sulphate on nine different wood substrates was reported by Hundhausen et al. [17]. Rapid and uniform greying with an additional biocidal effect of solutions containing ferrous sulphate and an additional biocide has also been described [18]. The colour changes of artificially greyed wood were also investigated as a function of the weathering time [19].

While previous studies have focussed on preventing weathering or simulating it chemically or thermally, our work investigates the mechanical simulation of weathering by sandblasting, which is rarely reported in the literature. Our study provides a systematic comparison between different wood species and orientations, supported by quantitative surface profiling and hardness measurements, which represents a novel contribution to this topic. The main objective of this research was thus to investigate sandblasting as part of the simulated weathering of wood and to elucidate this knowledge gap in the field of wood science and technology. With the aim of simulated profiling of wood surfaces, we sandblasted numerous softwoods as well as ring-porous and diffuse-porous hardwoods to assess the differences in efficiency of this mechanical treatment depending on the wood species. We were also interested in the erosion efficiency of sandblasting as a function of wood hardness. Finally, selected sandblasted wood samples were artificially greyed with an iron(II) sulphate solution to produce an aesthetically pleasing artificially aged wood.

2. Materials and Methods

2.1. Materials

The sandblasting experiments were carried out with numerous softwoods as well as with diffuse-porous and ring-porous wood species. The dry wood samples (at equilibrium moisture content (EMC) under laboratory conditions) had the following dimensions: 100 mm × 50 mm × 20 mm. In most cases, two orientations were prepared so that the largest area to be sandblasted was either the radial (labelled (R) in the list of wood species below) or the tangential (T). In most cases, both the radial and tangential surfaces were sandblasted, but in some cases, only the radial or tangential surfaces were sandblasted. All wood samples were from Slovenia, and some are categorised as alien invasive species (A).

The types of wood we used for our experiments and the other materials are listed in

Table 1. Wood species for sandblasting and other materials for the experiments.

Wood	Latin Name	Orientation 1	Alien Invasive Species 2	Xylem Anatomical Type
Norway spruce	Picea abies (L.) H. Karst.	R,T		softwood
Coast Douglas-fir	Pseudotsuga menziesii (Mirb.) Franco	R
Scots pine	Pinus sylvestris L.	R,T
European larch	Larix decidua Mill.	R,T
European ash	Fraxinus excelsior L.	R,T		Ring-porous hardwood
Oak	Quercus sp.	R,T
Sweet chestnut	Castanea sativa Mill.	R,T
Elm	Ulmus sp.	R,T
Honey locust	Gleditsia triacanthos L.	R	A
Tree of heaven	Ailanthus altissima (Mill.) Swingle	R	A
Black locust	Robinia pseudoacacia L.	T	A
Horse chestnut	Aesculus hippocastanum L.	R		Diffuse-porous hardwood
Maple	Acer sp.	R,T
Linden	Tilia sp.	R,T
Willow	Salix sp.	R
Poplar	Populus sp.	R
Boxelder maple	Acer negundo L.	R,T	A
Material	Material characteristics			Provider/producer
Corundum	Mesh: (425–500) μm	product code: 201322172		“Tehnični sistemi d.o.o.” Bizeljsko, Slovenia
Iron(II) sulphate				Carlo Erba, Dasit Group

¹ Orientation of sandblasted surface: R, radial; T, tangential. ² Regarded as alien invasive species in Slovenia: A.

. We chose aluminium oxide corundum as the abrasive for the sandblasting, and iron(II) sulphate solution was used for the artificial greying of the wood. The properties of these two materials are also listed in Table 1.

2.2. Sandblasting and Greying

The wood samples were processed individually in a sandblasting chamber. As the sandblasting was performed by hand, particular attention was paid to maintaining uniform processing conditions. We decided on the test parameters used on the basis of our numerous preliminary tests and some advice from the company that provided the sandblasting chamber and the abrasive. In these preliminary tests, we tested different distances of the nozzle from the wood surface, travelling speeds, angles to the sample surface and the working pressure. So, the nozzle was held at an approximate distance of 5 cm from the surface, with a travelling speed of about 2 cm/s and an angle of 45° to the sample surface. The working pressure was set to 6 bar. Each surface was treated with four passes (from all directions) to ensure uniform surface treatment over the entire area.

Iron(II) sulphate, also known as a green vitriol, is a water-soluble compound that is known to turn wood grey and give it an aged appearance. An approximately 5% solution was prepared by dissolving 10 g of green vitriol in 200 mL of distilled water. This concentration of the iron sulphate solution was based on the information in the cited sources and also on the experience of our employees, who studied the greying of wood in our department [18,19]. The sandblasted samples were immersed in a beaker containing the solution, and in cases where the surface could not be properly wetted, a brush was also used. The objective of this part of our investigation was not to monitor or measure colour changes over time but simply to obtain samples with a greyed or aged appearance. As the greying experiment was only a side-one, i.e., a supplement to our simulated weathering experiments using sandblasting, we did not prepare the unblasted and greyed samples for comparison with sandblasted greyed samples.

2.3. Characterisation and Analyses

To determine the effectiveness of sandblasting, the samples were conditioned under laboratory conditions and characterised before and after sandblasting. For the same purpose, the thicknesses of the samples were measured before and after sandblasting using a digital calliper gauge. The thickness was measured three times on each sample, at both edges (10 mm from the edge) and in the centre. In this way, the average thickness was determined for each sample.

In order to determine a possible correlation between the hardness of early and late wood and the results of sandblasting, the indentation hardness H_IT was measured on each sample, and the indentation modulus was calculated separately for early and late wood. The measurements were carried out with the instrumented microindenter MCT³ (Anton Paar, Graz, Austria). In instrumented indentation measurements, the deflection of the indenter, the applied force and the resulting indentation depth were recorded simultaneously and in a controlled manner, which enabled a precise evaluation of mechanical properties such as hardness and modulus. The hardness (H_IT) is a measure of the resistance to permanent deformation or damage, and the indentation modulus (E_IT) quantifies the elastic response of the material during indentation and represents the modulus of elasticity resulting from the initial slope of the load–displacement curve. On each sample with dimensions of 10 mm × 10 mm × 20 mm (radial × tangential × longitudinal), six indentations were made on early wood and six indentations on late wood. The indentation tests were performed with a CR6 steel ball with a diameter of 6 mm. Some of the most important parameters of the microindentation measurements are as follows: the acquisition rate, 10 Hz; linear load; maximum load, 5000.00 mN/min; loading and unloading rates, 5000 mN/min; and pause at maximum load before unloading, 10.0 s.

The surfaces were analysed with an Olympus DSX1000 digital microscope. Each sample was placed under the objective of the OLYMPUS DSX10-SXLOB device (1x magnification), and the primary profile was measured perpendicular to the grain direction at two spots (approximately one-third of the sample). Each measurement area consisted of four patches drawn in a straight line with 20% overlap. Exceptions were the samples of larch wood, sweet chestnut wood and honey locust wood, where the overlap was set at 30%, and the robinia (black locust) samples, set at 50% overlap, as these samples were narrower. The DSX software with the MIX (mixed light) light setting was used to observe and record the images. In some cases, surface defects, mainly cracks, were detected both with the naked eye and with a microscope. When this was the case, we chose measurement areas with the lowest possible number of defects, especially cracks. Despite our efforts to avoid the inclusion of defects in the profile measurement areas, some may have still remained. This could be one of the reasons for the considerable variation in the results, including that of P_t values, as described and explained in Section 3.1.1.

The surfaces of some selected samples (Norway spruce wood, linden wood and oak wood) were also analysed with the FEI QUANTA 250 scanning electron microscope (SEM) before and after sandblasting. For the SEM analysis, a thin layer of gold was sputtered onto the samples. The observations were made possible with the xTmicroscope control v6.2.11 software.

2.4. Statistical Analysis

These extensive investigations yielded numerous experimental data (

Table 2. Average indentation hardness (H_IT (MPa)) and indentation modulus (E_IT (GPa)) values ^1,2.

Sample	Radial Surface				Tangential Surface
	Early Wood		Late Wood		Early Wood		Late Wood
	HIT	EIT	HIT	EIT	HIT	EIT	HIT	EIT
Spruce	165(27)	1.98(0.16)	229(58)	2.51(0.32)	62.5(43)	3.27(0.81)	183(39)	1.88(0.30)
Douglas fir	94.9(6)	1.72(0.08)	424(147)	4.91(1.01)
Pine	140(57)	1.94(0.51)	275(67)	2.78(0.53)	210(110)	2.49(0.77)	379(148)	3.35(1.00)
Larch	283(81)	3.47(0.54)	547(170)	5.82(0.95)	329(132)	6.30(2.19)	689(42)	9.29(1.67)
Ash	389(65)	4.68(0.34)	414(203)	6.18(4.2)	448(97)	5.04(0.28)	476(161)	5.60(1.78)
Oak	272(112)	3.42(1.06)	488(116)	4.04(0.69)	776(125)	8.11(1.18)	949(151)	10.8(1.20)
Sweet chestnut	205(84)	3.02(1.46)	374(48)	3.31(0.81)	208(38)	3.59(0.55)	458(84)	5.41(0.45)
Elm	480(136)	7.63(0.67)	661(370)	11.2(2.10)	409(71)	5.31(0.23)	778(249)	9.53(0.71)
Honey locust	504(142)	6.11(0.92)	748(69)	9.30(0.71)
Tree of heaven	353(47)	5.49(0.48)	366(134)	4.81(0.74)
Black locust					419(88)	6.13(1.32)	518(209)	6.69(1.34)
Horse chestnut	225(30)	3.07(0.31)	263(75)	3.22(0.32)
Maple	237(117)	2.94(0.79)	366(45)	4.50(0.52)	519(82)	5.14(0.31)	838(98)	6.89(0.74)
Linden	98(40)	2.79(2.04)	142(48)	1.78(0.25)	113(53)	1.73(0.37)	184(15)	2.10(0.33)
Willow	136(72)	1.96(0.84)	162(23)	2.63(0.30)
Poplar	213(106)	2.62(0.96)	367(82)	3.67(0.47)
Boxelder maple	289(45)	4.88(0.40)	415(134)	6.10(1.06)	860(65)	10.3(0.67)	876(152)	9.65(1.70)

¹ Empty cells mean that samples with this orientation were not available. ² Standard deviations in parentheses.

and Table 3), and we were interested in the possible correlations between the success of profiling by sandblasting and some of the data obtained, such as mass loss or thickness reduction by sandblasting and surface profile parameter P_t, the difference between the indentation hardness of late and early wood and P_t, etc. Therefore, we prepared the raw data for statistical analyses supported by artificial intelligence tools. Thus, some parts of the statistical analysis and interpretation were supported by Microsoft Copilot (GPT-4, July 2025 version), an AI-based assistant developed by OpenAI and integrated into Microsoft services [20]. This tool supported the preparation of the subsequent description of the statistical analysis and also produced a draft description of the results of the statistical analysis. This draft text was then corrected, supplemented and partially changed by the authors and can be found in Section 3.1.3, under Table 3. All statistical analyses were performed using Python 3.12 [21,22] and the libraries pandas [21,22,23], scipy and seaborn. The primary goal was to identify relationships between the selected wood properties using Pearson’s correlation coefficient (r), which measures the strength and direction of linear relationships between pairs of continuous variables. For each pair of variables, the Pearson correlation coefficient and the associated two-tailed p-value were calculated using the pearsonr() function from the scipy.stats module1. The threshold for statistical significance was set at α = 0.05. Degrees of freedom were calculated as N − 2, where N is the number of valid paired observations [24]. To assess the overall strength of interdependence between variables within each group (e.g., radial, tangential, conifers, ring-porous, diffuse-porous), the average absolute correlation (mean |r|) was computed by averaging the absolute values of all off-diagonal elements in the correlation matrix. All correlation matrices and p-values were also calculated separately for radial and tangential surfaces, conifers and ring-porous and diffuse-porous hardwoods. Visualisations and summary tables were generated using matplotlib [25] and seaborn [23].

3. Results

3.1. Sandblasting Results

As described in the “Materials” section, our sandblasting experiments were carried out with numerous types of wood. We sandblasted the surfaces of four softwood species, seven ring-porous hardwood species and five diffuse-porous species. Therefore, in this section, we present only the most striking and meaningful results, either with a very good or excellent visual appearance of the sandblasted samples or, on the other hand, some selected results where sandblasting only led to a barely visible sandblasted surface profile.

3.1.1. Visual Appearance and Profiles of Sandblasted Surfaces

In the softwood samples, very good visual results of sandblasting were obtained in the wood samples of Douglas fir and larch (Figure 1). The radial surfaces of the Douglas fir also all contained a coloured heartwood. The early wood was sanded faster than the late wood in all samples, with the latter having a flat surface at the top of the profile. The wood samples were coarse and had an open, fibrous surface, especially in the early wood. The early wood was sanded so that the cross-section of the samples resembled a sawtooth pattern. Very similar results were achieved when sandblasting larch wood. Radial sandblasting reveals a very clear texture of the wood. When sandblasting tangentially orientated larch wood surfaces, a very pronounced texture of the wood was also achieved. Less rough, profiled surfaces were achieved when sandblasting Scots pine wood. Nevertheless, quite well-profiled, corrugated surfaces were also achieved in this case. It should be noted that in all Scots pine wood samples, abrasive particles were embedded in the damaged wood, which were clearly visible to the naked eye (Figure 2). In fact, some abrasive grains were visible in almost all sandblasted samples, but only sometimes, as in the case of the Scots pine wood, the abrasive particles were more abundant.

In the group of ring-porous woods, the best sandblasting results were achieved with oak (Figure 3). The large pores of the early wood were eroded more intensively than the harder late wood. The obtained surface is rough and has deep channels in the sanded early wood. Numerous large sanding particles have settled in the area of the early wood. Smaller knots are visible in all samples that are less sanded (same height as the late wood). On the other hand, in this group, the elm wood samples were at the bottom of the list in terms of visually assessed sanding efficiency. All samples have a surface that resembles tissue distressing but which is the result of sandblasting the wood. The transitions between early and late wood are noticeable on tangential surfaces, so that the open pores are clearly visible and gradually merge into the less pronounced late wood (Figure 4).

The results of sandblasting diffuse-porous hardwood were similar in efficiency to those of sandblasting ring-porous wood. However, our visual impression was that the profile achieved by sandblasting was even less pronounced than with sandblasted ring-porous hardwood samples. With sandblasted boxelder maple, for example, the discolouration significantly stands out. The patterns are similar. Fine lines can be seen on the surfaces, and this is the only result of sandblasting apart from the rough surface. When sandblasting tangentially aligned boxelder maple surfaces, the surface shows slightly accentuated transitions between early and late wood, which is the only result of sandblasting apart from a slightly rougher surface (Figure 5).

Analysing the surfaces with an Olympus DSX1000 digital microscope also revealed different values for surface primary profile parameters, which were the average values of 12 parallel measurements per wood species. To complement the visual impression of the effectiveness of sandblasting, we decided to compare the primary profiles of the sandblasted surfaces using the parameter P_t (Figure 6). P_t is defined as the total height of the profile, i.e., the vertical distance between the maximum profile peak height and the maximum profile depth along the assessment length [26].

As can be seen in Figure 6, the measured P_t parameters roughly correspond to our visual assessment of the waviness of surfaces as a result of sandblasting. Firstly, it should be noted that the error bars, which show the spread of the results, are relatively large in some cases. There may be several reasons for this phenomenon. As described in Section 2.3, cracks on surfaces were detected in some cases. Despite our efforts to avoid the inclusion of defects in the profile measurement areas, some may have remained, and this could be one of the reasons for the considerable scatter. Furthermore, the measurements with the indenter were automatically performed in a predetermined line following the early or late wood, and it was impossible to ensure that there was always exclusively late or early wood under the tip. Finally, wood is an extremely heterogeneous material, even in the same piece, which always contributes to a considerable variation in the value of the measured property. High P_t values were observed in spruce, Douglas fir and larch, for example, as well as in ring-porous woods in oak and sweet chestnut. In contrast, the P_t of maple and boxelder maple was lower. We also calculated the average values of P_t for radial and tangential surfaces for three groups of wood samples: softwood, ring-porous wood and diffuse-porous wood. These values measured on radial surfaces were 623 µm, 586 µm and 503 µm, respectively, and on tangential surfaces were 711 µm, 525 µm and 434 µm, respectively. Finally, the average values of P_t for radial and tangential surfaces together were 667 µm for softwoods, 556 µm for ring-porous woods and 468 µm for diffuse-porous woods. Thus, based on the visual impression and surface profile values, it appears that the best wood species for sandblasting that promise good results are softwood, followed by ring-porous and diffuse-porous hardwood. Nevertheless, the values shown in Figure 6 should not be considered uncritically: some surface defects such as cracks or knots can have an effect on the measured values. Figure 7, for example, shows an image of the tangential surface of a linden wood sample. The image was created using the DSX software of the Olympus DSX1000 digital microscope. The model does not show a clearly grooved surface, but a larger crack can be seen on the left side of the image. It is assumed that this crack was made more pronounced by the sandblasting process. It was included in the primary profile measurements and the average value of P_t, which, in this case, reached an unrealistically high value of 1733 µm. This would be the highest value of all P_t values determined, although the surface is obviously quite smooth (Figure 7). This is also the reason why we did not include information about the tangential P_t parameter for linden wood in the graph in Figure 6.

3.1.2. SEM Images of Spruce and Oak Wood Samples

SEM micrographs of untreated and sandblasted spruce and oak wood are shown in Figure 8 and Figure 9. In the SEM micrograph of untreated spruce wood (50-fold magnification, Figure 8A), we can see two transitions between annual growth rings, with one annual ring in the centre (yellow arrows). The early wood with larger pores is clearly distinguishable from the late wood (red circle). The surface of the wood is slightly damaged, especially in the early wood area, where the cells have thinner walls and are consequently softer. The wood is otherwise free from growth anomalies. The surface of a sandblasted spruce wood sample (Figure 8B) shows a distinctly damaged wood surface. There are large protrusions throughout the observed profile.

SEM images of untreated oak wood (50x and 500x magnification) are shown in Figure 9A,B, and images of sandblasted oak wood are shown in Figure 9C. In one of the vessels, the tylose vessel elements of the early wood are clearly visible (Figure 9A, blue arrow), while the others are not tylosed. At 500x magnification (Figure 9B), the surface created by planing can be seen even more clearly. The tissue is only slightly torn, mainly in the area of the early wood vessels (yellow arrow). The image clearly shows where one vessel element ends and another begins (blue arrow). Finally, in Figure 9C is a SEM image (500x magnification) of a sandblasted oak surface. A rough, torn surface can be recognised. A sanding particle can also be clearly recognised (blue circle). It has characteristic straight and sharp edges that distinguish it from wood tissue. In addition, several grooves can be recognised, which were created by the impact of the blasting agent on the wood surface.

3.1.3. Indentation Hardness and Reduction of Mass and Thickness by Sandblasting

Figure 10 shows typical curves from the tests with the microindenter, which were used to calculate the indentation hardness and the indentation modulus. The load (Fn in N) versus indentation depth (Pd) curves consist of a load and unload range with a pause of 10 s.

Indentation hardness (H_IT in MPa) is calculated as the ratio of the maximum applied load to the contact area at that load:

$$H_{IT}={\displaystyle \frac{F_{max}}{A_p\left(h_c\right)}}$$

(1)

where F_max is the maximum applied load in N and A_p(h_c) (in µm²) is the projected contact area at the contact depth h_c (µm) [27].

The indentation modulus (E_IT) is derived from the slope of the unloading curve at the point of maximum load, considering the elastic response of the material and the geometry of the indenter (in our case, a 6 mm steel ball):

$$E_{IT}={\displaystyle \frac{\sqrt{\pi }}{2\mathsf{\beta }}}\cdot {\displaystyle \frac{S}{\sqrt{A_p\left(h_c\right)}}}$$

(2)

where S is the slope of the unloading curve, β is a correction factor (~1 for a spherical indenter) and the other symbols have the same meanings as described in Equation (1) [27].

The average values (calculated from six measurements on early wood and six measurements on late wood per surface orientation) for the indentation hardness and the indentation modulus are listed in Table 2.

As described in the literature [28], the indentation hardness values cannot be easily converted into other, more conventional hardness values, such as Brinell hardness. For this reason, the hardness values from the literature were not included in Table 2. However, woods with high hardness values, expressed by H_IT, are also described as hard in the literature (e.g., ash, elm or black locust) [29], and the same can be said for soft woods (e.g., willow, pine or poplar) [29]. If we look at the H_IT and E_IT data in Table 2, we can first see in some cases a very large scatter of the measured values expressed by the standard deviations, such as in the cases of H_IT for oak, maple, willow or poplar wood (radial surface, early wood). As previously reported [30], this could be due to the high variability of wood properties, even within a single sample, but also to the low spatial resolution. It is quite difficult to place a steel ball exactly in the centre of early wood or late wood tissues, especially if the annual rings of the samples are narrow. Nevertheless, the hardness values of early wood were determined to be lower than those of late wood, which was the expected result [30,31,32,33], and the most important reason for this is the difference between the densities of early and late wood [33,34].

Table 3. Densities, mass loss, thickness reduction, ΔH_IT and P_t for different wood species.

Wood	Orientation	Density (g/cm3) 1	Mass Loss (%)	Decrease of Thickness (%)	ΔHIT 2 (MPa)	Pt (µm)
Spruce	radial	430 [35]	1.868	0.518	64	708
	tangential		2.514	0.768	120	887
Douglas fir	radial	530 [35]	1.454	0.504	329	707
Pine	data	510 [35]	1.727	0.541	135	472
	data		1.411	0.508	360	554
Larch	radial	530 [35]	1.786	0.539	264	604
	tangential		1.548	0.182	360	692
Ash	radial	710 [35]	1.098	0.646	25	481
	tangential		1.091	0.387	28	546
Oak	radial	740 [35]	1.316	0.393	216	761
	tangential		1.036	0.185	173	638
Sweet chestnut	radial	560 [35]	1.817	0.682	169	789
	tangential		1.505	0.556	250	521
Elm	radial	570 [35]	1.225	0.481	181	386
	tangential		1.237	0.512	369	458
Honey locust	radial	670 [35]	1.245	0.366	244	596
Tree of heaven	radial	540 [36]	0.825	0.241	13	504
Black locust	tangential	710 [36]	1.091	0.341	99	464
Horse chestnut	radial	500 [35]	1.763	1.093	38	457
Maple	radial	620 [35]	1.250	0.624	129	433
	tangential		1.213	0.591	319	429
Linden	radial	560 [35]	2.482	1.319	44	518
	tangential		1.548	0.905	71	3
Willow	radial	500 [35]	1.796	0.640	26	655
Poplar	radial	450 [35]	1.131	0.363	154	524
Boxelder maple	radial	420 [36]	0.825	0.241	126	428
	tangential		1.091	0.341	16	441

¹ Average values were taken for the density; if the tree species for the wood was not exactly known, average values were taken for similar species from the same genus. ² ΔH_IT is the difference between H_IT for late and early wood. ³ The value was not measured.

Table 3. Densities, mass loss, thickness reduction, ΔH_IT and P_t for different wood species.

Wood	Orientation	Density (g/cm3) 1	Mass Loss (%)	Decrease of Thickness (%)	ΔHIT 2 (MPa)	Pt (µm)
Spruce	radial	430 [35]	1.868	0.518	64	708
	tangential		2.514	0.768	120	887
Douglas fir	radial	530 [35]	1.454	0.504	329	707
Pine	data	510 [35]	1.727	0.541	135	472
	data		1.411	0.508	360	554
Larch	radial	530 [35]	1.786	0.539	264	604
	tangential		1.548	0.182	360	692
Ash	radial	710 [35]	1.098	0.646	25	481
	tangential		1.091	0.387	28	546
Oak	radial	740 [35]	1.316	0.393	216	761
	tangential		1.036	0.185	173	638
Sweet chestnut	radial	560 [35]	1.817	0.682	169	789
	tangential		1.505	0.556	250	521
Elm	radial	570 [35]	1.225	0.481	181	386
	tangential		1.237	0.512	369	458
Honey locust	radial	670 [35]	1.245	0.366	244	596
Tree of heaven	radial	540 [36]	0.825	0.241	13	504
Black locust	tangential	710 [36]	1.091	0.341	99	464
Horse chestnut	radial	500 [35]	1.763	1.093	38	457
Maple	radial	620 [35]	1.250	0.624	129	433
	tangential		1.213	0.591	319	429
Linden	radial	560 [35]	2.482	1.319	44	518
	tangential		1.548	0.905	71	3
Willow	radial	500 [35]	1.796	0.640	26	655
Poplar	radial	450 [35]	1.131	0.363	154	524
Boxelder maple	radial	420 [36]	0.825	0.241	126	428
	tangential		1.091	0.341	16	441

¹ Average values were taken for the density; if the tree species for the wood was not exactly known, average values were taken for similar species from the same genus. ² ΔH_IT is the difference between H_IT for late and early wood. ³ The value was not measured.

In Table 3, experimental and literature data are presented that were used together with some data from Table 2 for the analysis of possible correlations between the hardness, difference of hardness between early and late wood, decrease in thickness, mass loss because of sandblasting and profile, as expressed by the primary profile parameter P_t.

To investigate potential dependencies between the measured wood properties, an artificial intelligence tool [20] performed a series of Pearson correlation analyses using sources cited in the Materials and Methods [21,22,23,24,25]. The analysis was also performed separately for radial and tangential surfaces and for specific anatomical wood groups: softwoods, ring-porous hardwoods and diffuse-porous hardwoods. Variables included density, mass loss (Δm), thickness reduction (Δt), surface profile (P_t) and indentation hardness (H_IT), as well as indentation modulus (E_IT) for both early and late wood. Firstly, it was found that both the indentation hardness of the early wood and the late wood were very strongly correlated with the respective indentation modulus values (r = 0.91 and r = 0.87; both p < 0.00000001), confirming the mechanical consistency of the indentation measurements. As expected, a strong positive correlation was found between thickness reduction (Δt) and mass loss (Δm) (r = 0.72, p < 0.001), suggesting that samples that were thinned more also lost more mass. This relationship was particularly strong for radial alignment (r = 0.79, p < 0.001). We were interested in the surface primary profile as a result of the sandblasting process, and it was calculated that mass loss was strongly correlated with P_t (r = 0.61, p = 0.006), particularly in the tangential orientation (r = 0.82, p = 0.004), suggesting that more degraded samples had more profiled surfaces. This relationship is shown in Figure 11. On the other hand, a moderately negative correlation (r = −0.34, p = 0.083) between density and thickness reduction suggests that denser wood samples are more likely to resist thickness reduction (Figure 11). Contrary to expectations, several pairs of properties showed very weak or no correlation: density − P_t (r = −0.03, p = 0.874), Δt − P_t (r = +0.00, p = 0.983), P_t − ΔH_IT (r = +0.14, p = 0.492). These results indicate that surface profile (P_t) is not related to wood density, nor to thickness reduction, and especially not to ΔH_IT. We thought that a large difference between the hardness of the late and early wood would lead to a stronger waviness of the sandblasted wood (more intense blasting of the softer early wood). The lack of this correlation suggests that other factors, such as microstructural breakdown, may play a more important role.

To assess the overall strength of the relationships between wood properties, the programme calculated the average absolute Pearson correlation coefficient (|r|) for all pairwise combinations of Δt, Δm, P_t, ΔH_IT and density for each group (sandblasting of radial or tangential surfaces, softwood and ring-porous and diffuse-porous hardwoods). These values reflect the general tendency of properties to vary within each group. For example, softwoods and diffuse-porous hardwoods showed stronger dependencies between physical and mechanical properties, possibly due to a more uniform anatomical structure or degradation behaviour. In contrast, radial samples showed weaker correlations, suggesting that the property changes are less synchronised in this orientation.

3.2. Greying

Although greying with the aim of simulated weathering was not in the focus of our investigations, we nevertheless carried out some greying of sandblasted samples by immersing them in an iron(II) sulphate solution. The aim of this part was not to analyse the greying process in detail, such as, for example, to examine the colour of the greyed samples in relation to their chemical composition or to measure colour differences as a result of greying. Rather, the aim was to prepare sandblasted samples that visually resemble wood that has been exposed to weathering for a long time. In this section, therefore, only a few examples of greyed samples are shown for illustrative purposes. We did not grey untreated–unblasted samples to isolate the sandblasting effect from natural variability. This is what should be continued in the future: further studies should focus on the detailed investigation of combined sandblasting and greying treatments as a realistic simulated weathering process of wood.

Figure 12 below shows an image of the surface of spruce samples in radial orientation. After exposure, the surfaces of the treated samples look old and worn. The newly formed colour is fairly homogeneous over the entire surface of the samples, with the exception of some areas of the late wood where the colour is slightly lighter and the greying is less pronounced. The embedded sanding particles are still present and are even more visible on the darker surface of the wood. Figure 12 also shows an image of the top surface of spruce samples in tangential orientation. The newly formed colour is fairly homogeneous over the entire surface of the samples, with the exception of the areas in the late wood, where the colour tone is slightly lighter and the greying is less pronounced. In order to investigate this phenomenon in the future, unblasted samples should be greyed under identical conditions to isolate the sandblasting effect from the natural variability. In the first sample counted from the left in Figure 12B, swelling occurred after thorough wetting and subsequent drying in sunlight. This led to a relaxation of the internal stresses, which caused a crack in the early wood of one of the knots, resulting in pronounced flaking of the treated surface.

Figure 13 below shows an image of the surfaces of oak samples in a radial orientation. After exposure, the surfaces of the treated samples look old and worn, similar to the softwood samples. The surfaces are darkened and bluish (indigo) in colour, which is characteristic of the reaction between tannins and iron ions. This reaction leads to the formation of ferric pyrogallate complexes, resulting in immediate and intense darkening. Oak heartwood, rich in hydrolysable tannins, exhibits this effect prominently. The anatomical structure of oak, especially the wide rays composed of parenchyma cells, facilitates the accumulation and surface migration of extractives, enhancing the staining effect [17]. On our samples, the abrasive inclusions are also recognisable. They are visible in the early part of the wood as light-coloured grains. Areas of poor wetting are also visible, although a brush was used to aid immersion. On the right side of this figure (Figure 13B), an image of the surface of oak samples in a tangential orientation is shown. The number of embedded abrasives is lower in the tangential orientation than in the radially orientated examples. Areas of poor wetting with the iron sulphate solution are also visible.

The third illustrative example of greying is elm with radial and tangential sandblasted and greyed surfaces (Figure 14). After exposure, the surfaces of the treated samples look old, but not as worn as oak or chestnut, for example. The surfaces are greyed and, despite their darker colour, the sandblasting effect is no more pronounced on these samples than on ash samples, for example. In samples with tangential sandblasted and greyed surfaces, the patterns appear more worn and aged than in the radial orientation. The surfaces are effectively greyed, which emphasises the transition between early and late wood.

4. Discussion

If we critically analyse the obtained measurement data and the correlations between them, it seems that softwoods are the most promising species if sandblasting is expected to be as efficient as possible, especially spruce and larch and tangential surfaces. Among the ring-porous woods, oak and sweet chestnut also achieved good results, even in a radial orientation. In the group of diffuse-porous woods, willow was surprisingly profiled. In general, the tangential surfaces are more profiled than the radial ones after sandblasting. All in all, softwoods are the better choice for simulated weathering by sandblasting and greying.

Our results complement the information from the literature that the erosion and weathering behaviour of wood is closely related to its anatomical structure. Lemaster et al. [14] demonstrated that sandblasting with sodium bicarbonate or plastic media preferentially removes early wood, reinforcing the natural structure of the annual rings. This selective erosion mimics natural weathering and can be used to artificially age wood surfaces. Our results show that tangential surfaces, which have more early wood, tend to have more profile after sandblasting, which is consistent with the erosion patterns observed by Lemaster et al. [14]. Laskowska et al. [37] investigated the influence of sandblasting on the surface properties of spruce and larch wood as well as oak and Norway maple wood [37]. They concluded that their results emphasise the need to consider the wood species and anatomical variability when selecting surface preparation methods and finishing systems. Although the sandblasting process differs from conventional sanding, we believe that these conclusions confirm our findings. Sahin et al. [3] investigated surface profile changes as a result of natural weathering. They wrote that the characteristics of surface roughness in most wood species are very complicated and not well understood. To determine the causes of roughness, one must understand factors such as the wood’s cell wall structure, moisture conditions, annual ring orientation and density. They claim that the susceptibility to changes in surface appearance (including surface profile) is variable and highly dependent on the botanical origin of the wood species as well as the nature of the wood itself. This was also evident in our tests and should be considered when selecting wood for simulated weathering by sandblasting.

We have compared our sandblasted samples with wood that has been left outdoors for a longer period of time. Figure 15B shows a photograph of a radial surface of Scots pine wood that has been exposed to the weather in the Dolenjska region of Slovenia, on a wooden object from 1936—i.e., 89 years old. The wood has never been painted. The year of construction of this object can be seen on the wooden element of the building in Figure 15A. Figure 15C shows two samples with radially sandblasted and greyed surfaces for comparison. We believe that the profile of the sandblasted Scots pine wood samples we have prepared and the profile of pine wood that has been left outdoors for 89 years look similar. The difference is that the 89-year-old sample has different and more uneven colour, which is due to blue stains and stains from all kinds of possible airborne pollutants.

In addition, we obtained some wood samples of Scots pine and spruce exposed to the weather for 3 or 5 years in the yard of the Department of Wood Science and Technology in Ljubljana, Slovenia (Figure 16). Here, too, their colour is different, namely greyer than the colour of the samples that were artificially greyed with iron(II) sulphate solution immediately after sandblasting. We measured the parameters of the surface profile on these three naturally weathered samples and compared the values of the P_t parameter with the P_t values determined on the sandblasted radial surfaces of pine and spruce samples (

Table 4. Average ¹ P_t values (μm), obtained on radial surfaces of pine and spruce wood samples that were exposed outdoors for either 3 or 5 years and on sandblasted samples.

	Scots Pine, 3 Years	Sandblasted Scots Pine	Spruce, 3 Years	Spruce, 5 Years	Sandblasted Spruce
Pt	535	471	622	750	708
Std	63	98	81	115	281

¹ Nine values were used for calculation of the average values.

). From the data in Table 4, it can be seen that the sandblasted pine sample was slightly less profiled than the three-year naturally weathered pine. On the other hand, the profile of the sandblasted radial surface of spruce wood was as if it had been exposed to the open air for between 3 and 5 years. It should be noted that only one piece with an exposure time of 3 years as well as only one piece with an exposure of 5 years were used for the P_t measurements, and the results would perhaps be more realistic if a larger number of samples were used. Nevertheless, it can be seen that our sandblasting process results in profiles that are similar to those of outdoor wood.

The greying of sandblasted wood by treatment with iron(II) sulphate solution was only an additional experiment to gain a brief impression of how realistic simulated weathering results are that can be achieved by profiling using sandblasting in combination with artificial greying. However, this part should be investigated in more detail in the future. For example, sandblasted greyed samples should be compared with non-blasted greyed samples. In addition, moisture cycling and dimensional stability tests should be carried out on sandblasted greyed samples. This would provide information about possible dimensional instability after sandblasting and greying treatments, especially as there are reports of cracking occurring in greyed wood during outdoor exposure [15], but also that the number of cracks was lower in exposed greyed wood than in the samples that were not treated with iron [18].

The use of aluminium oxide (Al₂O₃) as an abrasive in sandblasting and of iron(II) sulphate (FeSO₄) for the artificial greying of wood raises important environmental and occupational health and safety issues. Therefore, the last paragraph of this section briefly discusses the waste treatment and occupational safety aspects of sandblasting with Al₂O₃ dust and greying with iron(II) sulphate solution. Although aluminium oxide is well suited for surface profiling, it produces fine dust during blasting, which can pose a risk to the respiratory tract if inhaled. To minimise these risks, appropriate ventilation systems, personal protective equipment (PPE) and respiratory protection must be used—for instance, in accordance with OSHA guidelines [38] or some other national legislation. In addition, spent abrasives may contain embedded contaminants and should be analysed and disposed of according to hazardous waste protocols to prevent soil and water contamination [39,40]. Iron(II) sulphate is commonly used in wood treatment but can cause skin and eye irritation on contact. Safe handling includes the use of gloves, goggles and protective clothing. Spilled iron sulphate treatment solutions should be treated properly [41]. However, it should be stressed that according to EPA assessments, iron salts used in low concentrations are not expected to pose significant risks to aquatic ecosystems [42]. These considerations emphasise the importance of responsible materials and waste management in the laboratory and in industry.

5. Conclusions

We have found that the sandblasting process can only be effectively profiled on certain types of wood under the parameters used. Sandblasting proved to be effective on all tested softwood species and some ring-porous hardwoods, but only to a limited extent on diffuse-porous hardwoods.

A comparison of the treated samples of ring-porous and diffuse-porous hardwoods shows that sandblasting is quite effective in profiling the surfaces of certain ring-porous wood species. In contrast, the success and intensity of surface profiling were not as high for diffuse-porous hardwoods.

It was calculated that the relative mass loss (in %) correlates with the primary profile, especially in the tangential direction. Therefore, the average values of primary profile parameters (e.g., P_t) were higher on tangential surfaces than on radial surfaces. Although sandblasting on radial surfaces seemed to be generally more successful upon visual inspection, the measured values could not confirm this.

We found that sometimes, abrasive particles of different sizes were embedded in the wood samples and could not be removed by blowing them off with compressed air.

Although greying with the aim of simulated weathering was not the focus of our investigations, we nevertheless carried out some successful greying of sandblasted samples by immersion in iron(II) sulphate solution.

Since the surface profiling of wood by sandblasting in the simulated weathering process in combination with artificial greying, especially for various softwoods, ring-porous woods and diffuse-porous woods, has been described extremely rarely in the literature so far, further investigations should be carried out in continuation of this work. Combined sandblasting and greying treatments as a realistic simulated weathering process of wood should be investigated in more detail in the future. In order to isolate the sandblasting effect from natural variability, unblasted greyed wood samples should also be prepared and investigated. Other experiments, such as moisture cycling tests, outdoor weathering performance of simulated weathered wood, etc., should be conducted as well, and other abrasive types and blasting parameters, including additional wood species, should be tested.

Nevertheless, it is believed that even this limited study can help manufacturers to select the most suitable wood species and orientation of elements for simulated weathering to meet consumer requirements and expectations.

Finally, the surface profiling of wooden elements to make them wavy or corrugated, as if they had been exposed to the weather for a long time, is usually achieved today by brushing with wire brushes. However, this method is limited to flat components. In contrast, sandblasting allows for surface profiling even on various three-dimensional wooden objects. We, therefore, believe that the results of this study may also be of interest to manufacturers of three-dimensional wooden elements with profiled, corrugated surfaces.