Impregnation of wood with a paraffinic phase change material for increasing heat capacity

ABSTRACT In this contribution, the impregnation of wood with a paraffinic phase change material (PCM) is investigated, specifically to increase the heat capacity of solid wood, which significantly influences the thermal inertia when used in buildings. Four wood species (beech, poplar, oak and spruce) were impregnated with different pressure processes in an autoclave. For poplar, up to 480 kg of PCM per m³ of wood was deposited. For beech and spruce, also more than 200 kg of PCM per m³ of wood was achieved. However, oak was hard to impregnate and only about 100 kg of PCM per m³ was deposited. Leakage, which is undesired, occurred for all the wood species, especially for beech, but could be significantly reduced to less than 10% by increasing the viscosity of the PCM. The heat capacity was increased by one order of magnitude compared to clear wood, as measurements with differential scanning calorimetry showed. Simulations with an analytical model demonstrate the potential for damping temperature amplitudes in buildings in the summer month when applying PCM.


Introduction
A phase change material (PCM) is known to absorb a substantial amount of thermal energy at the transition from solid to liquid phase without an increase in temperature. This process is reversible. Thus, energy is released again during solidification of liquid PCM. This latent heat can be used for damping temperature amplitudes in buildings and is, thus, a method for passive air-conditioning. However, besides the high specific heat capacity, PCMs possess also low heat transfer coefficients, i.e. PCMs are good thermal insulators. Hence, compact blocks of PCM have a high thermal inertia. In order to reduce this inertia, PCMs are usually encapsulated in conductive materials such as metals (Agyenim et al. 2010, Kośny 2015.
As wood has a cellular structure, it can act as encapsulation matrix for PCM. Therefore, the wood can be impregnated with PCM in the liquid state. Literature shows limited studies into the impregnation of wood with PCMs, as indicated in Table 1.
General observations among these studies are that the specific heat capacity usually increased considerably compared to clear wood but leakage of PCM from wood was observed, except with microencapsulated PCM. Leakage is, in general, undesired and needs to be prevented for several reasons, including durability and serviceability. While microencapsulation is an effective means for preventing leakage, it is difficult to apply for impregnation due to the particle size.
In this contribution, results of impregnation tests of different wood species with pressure processes in an autoclave are presented. The application of this composite material is multi-functional construction elements, which bear loads and act as passive air-conditioning without using electric energy and, thus, reducing the energy demand of a building. For this purpose, also the specific heat capacity and the enthalpy as a measure of the storable thermal energy of the wood-PCM composite were determined. Moreover, theoretical investigations were performed to analyse the potential of temperature damping in buildings by applying PCM-impregnated wood.

Material
The paraffinic PCM RT35HC produced by the company Rubitherm, Berlin, Germany was used. This PCM melts at about 35°C and was chosen so as to have the PCM in the solid state at room temperature for more convenient handling. The latent heat in terms of enthalpy is approx. 200 kJ/kg in the range of ±1°C around the melting temperature. In order to reduce leakage, a more viscous version of this PCM was also tested. The increased viscosity resulted from the addition of a thermoplastic powder (6% wt.).
Four wood species were tested: European beech (Fagus sylvatica L.) and poplar (Populus nigra L.), which have a good treatability according to EN 350 as well as English oak (Quercus robur L.) and Norway spruce (Picea abies Karst.), which are known for their poor treatability. Wood samples of 40 mm by 40 mm by 300 mm (width by height by length) were prepared. The end grain was sealed with a lacquer to reduce intrusion of PCM via the end grain surface. About 1 g of a colourant (OIL BLUE 35) per litre of PCM was added to improve the visibility of the PCM within the wood.

Impregnation tests
An autoclave was used for the impregnation tests. Several pressure levels between 0.01 MPa (vacuum) and 0.8 MPa (overpressure) as well as combinations of vacuum and overpressure were tested (see Figure 1). The treatment time was between 1 and 4 h. Three samples were tested for each parameter combination.
The temperature in the autoclave was set to a nominal value of 80°C to have comparable conditions between the clear and the more viscous PCM. The large mass of the vessel and the large amount of PCM resulted in a delayed reaction of the temperature control of the autoclave, which resulted in temperature variations of about ±15°C related to the nominal value. The wood samples and the PCM were heated separately up to 80°C before placing both in an open metal container. The wood was fixed in the container such that it did not float in the PCM. The metal container with the samples was then placed in the pre-heated autoclave and the pressure regime was applied.
After the impregnation was finished, the wood samples were removed from the liquid PCM and allowed to cool in ambient condition. After solidification, excessive PCM was removed mechanically from the wood samples and the mass of the impregnated samples was measured to determine the PCM uptake.
For evaluating the impregnation results, different density or density-like values were determined. At first, the initial density of the wood was calculated as where m wood and V are the mass and the volume of the wood before impregnation. The wood had a moisture content of about 10-12% before impregnation. The density of the impregnated wood is calculated with where m imp is the mass of the impregnated wood. The amount of PCM within the wood volume is calculated with As a measure of efficiency of the impregnation, the theoretically possible maximum mass of PCM m PCM,max within the wood can be calculated with with the density of the PCM ρ PCM = 770 kg/m³ and the density of the cell wall ρ cell wall ≈ 1500 kg/m³ (Matejak and Kozakiewicz 2011).

Leakage tests
In a climate chamber, tests comprising 28 cycles of 12 h at 45°C (with 48% r. h.) and 12 h at 25°C (with 12% r. h.) were performed to investigate leakage. The leached PCM was collected in glass containers below each sample and during the cooling periods the weight of the containers and, thus, the leached PCM was determined. For the samples treated with the viscous PCM, the leached PCM needed to be removed mechanically from the wood surface due to the high viscosity.

Differential scanning calorimetry
The specific heat capacity of the PCM-impregnated wood was determined with dynamic differential scanning calorimetry (DSC) (device: Netzsch DSC 204 F1 Phoenix). For calibration, the sapphire standard was used as performed also by (Nopens et al. 2021). The DSC device provides an electric signal related to the heat energy Q [W] needed to change the temperature T [K] of the sample as output quantity, which is convertible to the specific heat capacity c p [J/(g·K)] of the sample material using the calibration curve of the sapphire standard sample. Small amounts of impregnated wood (approx. 15 mg) were tested in a range of 10°C to 60°C. The samples (discs with a diameter of about 5 mm) were extracted from slices of about 1 mm thickness, which were located at half-length of the impregnation specimens. At first, the samples were heated, then cooled and again heated. Cooling was performed with liquid nitrogen. At the beginning of the cooling cycle, the temperature control acts inertial, which results in a too high cooling rate, which is compensated by the device with heating. This results in a wavy course of the c p -T relation, which is not a material property but attributed to a measurement error.
It is known from literature e.g. (Nazari Sam et al. 2020) that the measured peak temperature and the course of the c p -T relation are depending on the heating rate and the sample size applying dynamic DSC. The smaller the heating rate and the sample size are, the closer the course of c p is to the actual course. With increasing temperature rate and sample size, the melting temperature T melt is shifted to higher values (the fusion temperature T fus to smaller values) and the c p course is widened, which is also associated with smaller maximum c p values. However, decreasing the heating rate results in longer measuring time and a weaker signal with more noise. Decreasing the sample size reduces the representativeness.
Since the main purpose of the DSC investigation was the confirmation of the effectiveness of the PCM in the composite and the determination of the storable thermal energy in the latent range, the temperature rate was chosen quite fast with 5 K/min. Thus, it should be considered for the interpretation of the results that T melt is overestimated (T fus is underestimated) and the course of the c p -T relation is widened, which is associated with underestimated maximum c p values in the latent range.
Interestingly, the latent portion of the enthalpy values of heating (and fusion) are not influenced by the heating rate although the course of the c p -T relation in the latent range is (Nazari Sam et al. 2020). The specific enthalpy values h [J/g], which are a measure of the thermal energy stored in the material, can be calculated for a defined temperature range between T 1 and T 2 as integral of the c p -T relation: h includes both the sensible and the latent energy portion. To separate the latent from the sensible portion a so-called baseline can be constructed from the c p -T relation. The amount of energy below the baseline is the sensible part and above the latent part. For the heating cycles, the baseline is defined as a linear function with supporting points at T 1,base, h = 10°C and T 2,base,h = 60°C. For the cooling cycle, the c p values at T 1,base,c = 10°C and T 2,base,c = 40°C are used since for T > 40°C, the c p -T relation has a wavy course as already mentioned. Thus, the sensible portion of the enthalpy can be calculated as The latent portion of the enthalpy follows as It shall be noted that for calculating h for the cooling cycle above 40°C, the baseline values are also used to avoid the erroneous values in this range. T 1 = 10°C and T 2 = 60°C were used for the calculation of the h values.

Visual evaluation
The result of the impregnation is evaluated visually. At first, the penetration front is observed based on lateral cuts. In order to avoid smearing of the PCM due to melting by the heat development during sawing, the samples were cooled down to −18°C before cutting. Due to the colourant added to the PCM, the penetration front is directly visible.
Secondly, microscopy is applied to investigate the intrusion of the PCM into the cell structure. Therefore, thin slices (approx. 35 μm thickness) were cut with a microtome. Some of the samples, especially from beech, needed to be stored in water for 72 h to be sufficiently softened for the microtoming.

Simulation
An analytical room climate model (Häupl et al. 2010) was applied to investigate the potential of placing PCM in wood flooring for reducing heat peaks in a building in summer. This model calculates the room temperature depending on the external climate in terms of external air temperature and radiant heat based on the so-called analytical reference year (ARY) (Häupl 2017) as well as the building parameters in terms of geometry, heat transfer resistance of the outer walls and the heat absorption properties of the construction components. Regulation of the interior climate by ventilation, interior heat sources and interior air-regulated heating is taken into account.
Additionally to the original model of Häupl et al. (2010), the effect of PCM is taken into account such that the energy input by solar radiation is reduced until the maximum value of latent heat of the PCM is reached if the room temperature is above the melting point of the PCM. If the room temperature falls below the melting point, the used amount of latent heat is given back to the room.
The considered room has a ground area of 5 m (north and south) by 10 m (east and west) and a height of 3 m and is situated inside a bigger building. Only the western wall, which is designed as a glass façade, is exposed to the outside climate. About 12% of the radiation passes through the windows due to shading and the transmittance coefficient of the glass. The building is assumed to be situated in the city of Dresden (Germany), which has coordinates of 51°N and 13°E. The thermal parameters of the walls and ceilings are chosen such that they correspond to the behaviour of typical solid wood constructions.

Results and discussion
Impregnation tests Figure 1 shows the results of the impregnation tests. It can be seen that for each wood species depending on its density according to Equation 1 a different maximum amount ρ PCM/wood,max of PCM is possible per wood volume, which was calculated with Equation 4. The values of ρ PCM/wood,max for beech and oak were approximately 400 kg/m³ and for poplar and spruce approximately 500-600 kg/m³. Thus, for beech and oak a considerable lower charging with PCM was possible than for poplar and spruce due to the higher density of the bulk material. Figure 1 also shows that the impregnation result in terms of the amount of PCM stored in the wood ρ PCM/wood depends on the wood species, the viscosity of the PCM and the applied pressure treatment. As already known from EN350, the wood species has a large impact on impregnability due to different anatomical characteristics of the cells.
For spruce, only a fraction of the theoretically maximum possible amount of PCM was achieved with low process pressure. However, with increasing pressure the amount of PCM could be substantially increased to values of more than 200 kg/m³ of wood. It was noted that there was significant variability in the results of spruce. With the viscous PCM, spruce was hard to impregnate and only small PCM deposition was achieved.
Oak was also hard to impregnate and the applied pressure treatments resulted in only low PCM deposition. Thus, oak was excluded from some of the further investigations. In contrast, high deposition of PCM was achieved for beech for both the clear and the viscous PCM. Moreover, the applied pressure treatment had only minor influence and, in principle, simple soaking was deemed sufficient. Due to the high density of beech, the maximum deposited amount of PCM was 280 kg/m³, which was not much higher than that achieved for spruce.
The highest deposition was achieved for poplar where in some cases the theoretically possible maximum deposition of PCM (480 kg/m³) was almost achieved for clear PCM. Even for the viscous PCM, an uptake of 270 kg/m³ was still obtained. This demonstrated that pressure treatment supported high deposition for poplar.

Visual evaluation
The results of the impregnation were visually evaluated. Due to the colourant, the position of the PCM was clearly identifiable. At first, lateral cuts were observed. It can be seen in Figure 1(a,c,e) that using clear PCM resulted in a more homogeneous distribution in the sample. With the viscous PCM (Figure 1(b,d,f)), a concentration of PCM in clusters often along the annual rings was observed. Moreover, there was a stronger difference of PCM concentration between the end grain and the centre. This was especially the case in spruce and poplar samples, where wide regions without any PCM were visible.
The microscopic investigations show where the PCM was deposited in the cell structure. Figure 3 shows photomicrographs after impregnation with clear PCM at 0.2 MPa. In beech and poplar, impregnation was primarily within the vessels. In spruce, the PCM was situated in the tracheid. Figure 4 shows the results of cyclic leakage tests. In general, the samples impregnated with the unmodified PCM showed more leakage than with the viscous PCM. However, the results of the samples with clear PCM possessed substantial differences between the wood species. While the leakage of the poplar samples impregnated at 0.4 MPa ceased after 14% of the PCM had left the wood, leakage of spruce and beech continued after 28 cycles when already 63% or 52% respectively had been leached. These results indicate that poplar is a good candidate for being impregnated with PCM.

Leakage tests
The leakage of the more viscous PCM was substantially reduced compared to the results with clear PCM, with any leached material having to be removed mechanically and not just flowing from the samples. After 28 heating and cooling cycles, the leakage was 15% for beech but did not stop at this stage. For spruce and poplar, the leakage had stopped and a loss of 4% and 10% of PCM respectively was observed. This showed that the thermoplastic additive was an effective means for reducing leakage. However, further investigations have to show whether it is possible to reduce leakage close to zero with this approach. Figure 5 shows the specific heat capacity of different samples as determined with differential scanning calorimetry (DSC). The c p -T relations showed constant energy uptake until about 30°C during the heating cycles. Afterwards, the energy increased rapidly and reached a maximum at about 45°C. Subsequently, the energy uptake decreased again rapidly and merged with the initial linear slope. It can be seen that the temperature of the maximum c p value is smaller in the second heating cycle. This might result from redistribution of the PCM within the wood and possibly leaching of PCM in the sample holder, which leads to increasing conductivity and less inertia.

Heat storage
After reaching 60°C, the samples were cooled back to 10°C. The energy release was found to be constant until about 35°C. Subsequently, the energy release increased again rapidly and reached a peak at about 30°C and for beech and oak another one at a somewhat lower temperature. At    Figure 5. Specific heat capacity and latent portion of enthalpy versus temperature as determined of differential scanning calorimetry with heating/cooling rate of 5 K/min. about 20°C, the energy release was back at the initial value. This indicated that there was a hysteresis.
As already mentioned before, the maximum values of c p are at a considerably higher temperature than the nominal value of the PCM and the c p -T course is widened in the latent range resulting in underestimated maximum c p values. This results from the quite fast applied temperature rate and has to be considered as measurement error. Still, the c p values are one magnitude larger than for clear wood, which has a value of about 1.7 J/(g·K) (chain line in Figure  5) for a moisture content of 10% for all common wood species (e.g. Niemz 1993). Considering the measurement error due to the high temperature rate, the actual maximum c p values of the impregnated is even higher but on a smaller temperature range. Figure 5 shows also the latent portion of the enthalpy h lat as determined with Equation 7. The maximum value of h lat is not influenced by the temperature rate of the DSC and can be considered as objective value. The enthalpy of fusion and solidification has for each sample approximately the same value. The values |h lat,m | in Figure 5 are the mean absolute maximum values of the two heating and the cooling cycles rounded to whole 10 J/g. It can be seen that the maximum values of h lat are proportional to the amount of the PCM in the sample, whilst the effect of wood species was rather insignificant. An exception was somewhat the values for spruce. It shall be noted the spruce showed high variability of PCM uptake between the specimens and a quite high non-uniformity in the PCM distribution within a specimen (Figure 2(e)) as well as a higher uptake in the outer regions while the DSC samples were taken from a section in the centre. Thus, the nominal ρ PCM/wood value might not coincide with the local value in the DSC sample. The enthalpy results show that for the following simulations the amount of PCM within the wood can be directly used to estimate the latent storage capacity of thermal energy.

Simulation
In order to estimate the potential benefit for applying PCMimpregnated wood for construction purposes, an exemplary simulation was performed. The used PCM has a specific latent heat of about 200 kJ/kg in the range of ±1°C around its melting point. It was assumed that the outer 8 mm of the flooring in the room was impregnated with PCM. With the ground area of 50 m², this results in an impregnated volume of 0.4 m³. As previously shown, it is realistic that 200 kg/m³ of PCM was deposited in the wood, which leads to a total amount of 80 kg of PCM, which has a total latent heat of 16,000 kJ in the range of ±1°C around the melting point. The melting point is assumed to be at a temperature of 23°C. Respective PCM products are also commercially available and have comparable properties to the one used in the current investigations. Figure 6(left) shows the simulated outside temperature according to ARY and the room temperature without using PCM. It can be seen that in the winter it is actively heated, which keeps the temperature approximately constant in this time. In spring and summer, the inside temperature increases continuously and starts to decrease again at the end of the summer.
Figure 6(right) shows the internal temperature with using PCM and without. It can be seen that after reaching the melting temperature of the PCM at 23°C in spring, the temperature peaks are reduced for almost 1°C. In summer, the solar gain is stronger and the amount of PCM is not sufficient to keep the temperature over the entire day at a low value. Thus, a higher amount of PCM would be necessary to reduce the temperature peaks further. Additionally, the external temperature does not decrease in the hottest nights below the melting temperature of the PCM and the thermal energy cannot be released in this time. Although this is an arbitrary example, it gives an idea how the room temperature and, thus, the amount of energy potentially necessary for cooling can be reduced. In this example, 78 kWh of energy by solar radiation has been compensated by the PCM, which reduces the need for active cooling by air-conditioning.

Conclusions
The investigations presented show that a substantial amount of PCM can be deposited depending on wood species within Figure 6. Distribution of outside and room temperature over one year with and without using PCM. the cell structure by a pressure treatment. The considerable increase of specific heat capacity and enthalpy of the PCMtreated wood was demonstrated with differential scanning calorimetry. Still, leakage of PCM needs to be further reduced. The PCM-treated wood can be used for multi-functional structural elements for heat storage and temperature buffering as well as for load bearing in engineered timber structures like walls and ceilings of dowel-laminated or cross-laminated timber. With this regard, the performed room climate simulations show that using PCM-impregnated wood in buildings can improve climatic serviceability and reduce the demand of electric energy for active cooling.