Within this study, the effect of functional additives and processing conditions on the overheating protection performance of thermotropic systems with fixed domains (TSFD) for overheating protection purposes was evaluated. The focus was on improving the overheating protection performance of a prototype TSFD based on a UV curable acrylate resin by optimization of the material constitution (addition of functional additives like surfactants and nucleating agents) and the processing conditions (temperature conditions during manufacturing, annealing). For the evaluated system, an effect of the nucleating agent on the overheating protection performance was ascertained. Furthermore, omission of an annealing step improved the overheating protection performance slightly.
Thermotropic glazings change their optical appearance from transparent to opaque upon exceeding a certain threshold temperature reversibly [1, 2]. Their utilization in the façade of a building can maintain a reduction in the energy consumption for heating, cooling and artificial day-lighting (smart window) [3–5]. Thus, a reduction in the overall energy demand of a building is achievable [3–5]. Furthermore, thermotropic glazings can provide efficient overheating protection for solar thermal collectors and thus can limit the stagnation temperatures to <130°C (smart collector) . Hence, the stagnation control by thermotropic glazings can alleviate the thermal load on the solar thermal systems (especially collectors) and the heat carrier fluid and thus prevent those from deterioration, ageing and failure . Especially for polymeric solar thermal systems, the stagnation control is a prerequisite in order not to exceed the long-term service temperatures of utilized – preferably cost-efficient – polymeric materials [6–9].
Besides other classes of thermotropic glazing materials, thermotropic systems with fixed domains (TSFD) gained interest in recent research due to their specific advantages like ease of adjustment of switching threshold, high long-term stability, low hysteresis, high reversibility and steep switching process [9–30]. TSFD consist of a thermotropic additive (exhibiting a temperature triggered change in refractive index), which is finely dispersed in a matrix material [1, 3]. Below the threshold temperature, the refractive indices of the matrix and the additive are almost equal, yielding a transparent appearance of the TSFD [1, 31]. Upon exceeding the threshold temperature (i.e., the melting temperature of the additive) a steep change in the refractive index of the additive is taking place . The change in the relative refractive index (the ratio of the refractive index of the additive domains to the refractive index of the matrix) yields the onset of intense light-scattering and thus a transmittance reduction [1, 12, 31]. Besides the relative refractive index, the TSFD morphology is of paramount importance for the overheating protection performance [32, 33]. The maximum overheating protection performance is attained by employing spherical scattering domains with a diameter in a range between 200 nm and 400 nm .
Recent studies regarding TSFD revealed a limited overheating protection performance of the layers due to an inappropriate scattering domain size and/or shape [17, 19, 26, 28, 30]. In a preceding study, potential remedies in order to adjust the scattering domain shape and size were outlined . During the manufacturing process, the uncured matrix material (i.e., a mixture of an oligomer, a reactive diluent and a photoinitiator) and the thermotropic additive form a kind of an emulsion. In conventional emulsions the addition of surfactants yields a reduction in the droplet size by changing the interfacial energy between the continuous and the disperse phase [34–37]. The addition of surfactants to mixtures of a matrix material and an additive might also reduce the scattering domain size in an analogous way like it was described for conventional emulsions above. However, surfactants may also induce heterogeneous nucleation in the additive domains. This was observed in emulsions of phase change materials, for example . Usually, the heterogeneous nucleation is induced by intentional addition of nucleating agents to a crystallizable substance [39–43]. The addition of nucleating agents to mixtures of a matrix material and an additive may probably yield a reduction in the size of scattering domains by increasing the number of crystallization loci and by affecting the free energy for crystallization. Furthermore, the temperature conditions applied during the manufacturing process may significantly affect the overheating protection performance of TSFD [14, 29], due to temperature-affected variations in nucleation of crystallization, which probably induce changes in the scattering domain size distribution. Furthermore, an annealing step employed after curing probably enhances the homogeneity of the TSFD (as was observed for other TSFD that were produced in the authors’ laboratory).
Thus, the major objective of the present study is to investigate the effect of surfactants and nucleating agents on the scattering domain parameters and hence the overheating protection performance of TSFD produced from a UV-curable resin. As the overheating protection performance of TSFD is directly related to the TSFD morphology, determining the solar-weighted transmittance of TSFD is an appropriate index in order to quantify the effect of the employed measures on the TSFD morphology. Furthermore, the effect of the processing conditions (thermal treatment, annealing) on the overheating protection performance of TSFD is studied. The effects are investigated by employing tools of factorial design.
2 Materials and methods
2.1 Materials and formulation
To address the aspects discussed above, TSFD M7A1-OTA-p3-RT-0.008 – which was already developed in a preceding publication  – was selected as the prototype system. That specific layer was selected because it exhibited the appropriate scattering domain shape (spherical), but an inappropriate scattering domain diameter for an efficient overheating protection. A reduction in the scattering domain diameter is likely to enhance the overheating protection performance significantly.
The layers formulated within this study were thus derivatives of the TSFD M7A1-OTA-p3-RT-0.008. The UV-curable resin matrix consisted of 57 wt% polyester acrylate oligomer Ebecryl 800 (Allnex Belgium SA/NV, former Cytec Surface Specialities, Drogenbos, Belgium), 40 wt% reactive diluent OTA-480 (OTA) – a propoxylated glycerol esterified with acrylic acid (Allnex) – and 3 wt% photobleaching photoinitiator Lucirin TPO-L (BASF SE, Ludwigshafen, Germany). The thermotropic additive was paraffin wax Sasolwax 5005 (additive A1; Sasol Wax GmbH, Hamburg, Germany) with a melting point of 55°C . The nomenclature of the thermotropic additive is consistent with previous publications [28–30]. The functional additives were the nonionic surfactant Lutensol AT 11 (BTC Europe GmbH, Cologne, Germany) and the potential nucleating agent synthetic Indigo (Sigma-Aldrich Handels GmbH, Vienna, Austria). Initially, Indigo was chosen as a potential nucleation agent by drawing an analogy between paraffin waxes and polyolefins, which have a similar structure. In polypropylene (a polyolefin), Indigo was recognized as an efficient nucleating agent [44, 45]. Thus, Indigo is supposed to act as a nucleating agent in paraffin waxes also. Table 1 shows the factors and factor levels which were employed upon TSFD formulation. Whereas the factor levels employed for the factor AT 11 concentration were either 0.01 mol kg-1 or 0.02 mol kg-1, the factor levels maintained for the factor Indigo concentration were either 0 mol kg-1 or 0.001 mol kg-1. The thermotropic layers were prepared by dissolving the thermotropic additive and the functional additives in the UV-curable matrix solution. The dissolutions were poured in to the intervening space between two glass panes which were sealed around the edge and stored at either -20°C (factor level “DF”) or room temperature (factor level “RT”) for 10 min prior to and post curing. The thermotropic mixtures were cured with an intensity of 4.6 μW cm-2 yielding a dose of 8.3 mJ cm-2 from 366 nm lamp of a Universal-UV-Lamp (Camag, Muttenz, Switzerland). Free standing layers with a thickness of 900 μm were obtained after removal of the glass panes. The theoretical additive concentration was 5 wt%. The TSFD were either annealed at the mixing temperature (100°C) of the resin matrix and the thermotropic additive (factor level +1) or not annealed (factor level -1). Accordingly, the parental layer M7A1-OTA-p3-RT-0.008 represents the factor combination AT 11: 0 mol kg-1; Indigo: 0 mol kg-1; treatment: RT; annealing: no). The utilized vocabulary that is related to statistical analysis of the factorial designs (e.g., “factor”, “effect”, “interaction”) follows general conventions outlined in the literature [46, 47].
|Factors and levels||[AT 11]|
amol functional additive related to kilograms of thermotropic mixture (matrix+thermotropic additive).
bRT: storage at room temperature for 10 min prior to and post curing; DF: storage at -20°C for 10 min prior to and post curing.
cAt the mixing temperature of resin matrix and thermotropic additive.
2.2 Characterization methodology
2.2.1 Overheating protection performance
The overheating protection performance of the TSFD (the pristine free standing layers) was determined applying UV/visible/near infrared (UV/Vis/NIR) spectrometry. A double beam UV/Vis/NIR spectrophotometer Lambda 950 (Perkin Elmer Inc., Waltham, MA, USA) equipped with an Ulbricht-sphere (diameter 150 mm) was employed. For the given measurement apparatus the radiation passing through (transmittance) the specimen outside a cone of approximately 5° relative to the incident beam direction was defined as the diffuse (scattered) component. The hemispheric and the diffuse transmittance were both recorded at normal incidence in the spectral region from 250 nm to 2500 nm. The solar-weighted transmittance was determined by weighting the recorded spectral data in steps of 5 nm by the AM1.5 global spectrum . The spectrophotometer was adapted by a heating stage to adjust the sample temperature within a range from ambient temperature to a maximum of 115°C . The measurements were performed at room temperature and 70°C. Prior to measurement, the samples were allowed to equilibrate for 5 min at the selected temperature. The heating stage was equipped with a control system consisting of a heating stage internal J-type thermocouple as a temperature sensor and the control unit HS-W-35/M (Heinz Stegmeier Heizelemente HS-Heizelemente GmbH, Fridingen, Germany). Within the heating stage, the sample was positioned in close proximity of the port hole of the Ulbricht-sphere. In situ front- and backside sample surface temperatures as a function of set-point value of the control unit were recorded on a prototype sample with a two-channel temperature measurement instrument T900 (Dostmann electronic GmbH, Wertheim-Reicholzheim, Germany) equipped with a precision K-type thermocouple. The sample temperature was assumed as the average of both recorded surface temperatures. The required set-point values to maintain specific average sample temperatures were calculated from a second order polynomial fit of the temperatures recorded in measurements of the prototype sample. A double determination was carried out.
The morphological characterization of the TSFD was carried out applying an optical microscope Olympus BX51 (Olympus Austria Ges. m. b. H., Vienna, Austria) in transmitted light mode. The TSFD were investigated without further preparation. The domain size was evaluated with the measurement tools of the software analySIS (Soft Imaging System GmbH, Münster, Germany). The number weighted mean diameter and the Sauter mean diameter of the scattering domains were evaluated. Furthermore, the fraction of scattering domains that exhibited vacuoles were recorded in optical micrographs by observation of scattering domains that allowed to distinguish clearly whether the domains exhibit or do not exhibit vacuoles. Scattering domains that were not legible properly were not taken into account upon determination of the fraction of scattering domains exhibiting vacuoles.
3 Results and discussion
Table 2 presents the response of the investigated TSFD – the solar hemispheric transmittance detected either at room temperature (τnhRT) or at 70°C (τnh70) – upon application of different factor combinations during the manufacturing process. The factors varied were the concentration of the functional additives AT 11 and Indigo, as well as the thermal treatment and annealing (see Table 1). The solar hemispheric transmittance of the manufactured layers varied between 67.3 and 82.3% at room temperature on the one hand and between 66.1 and 82.3% at 70°C on the other hand. Thus, the obtained layers do not meet the performance requirements for an efficient overheating protection of a solar thermal flat-plate collector, which would require a solar hemispheric transmittance of at least 85% and of maximum 60% below and above the threshold temperature, respectively . The layers formulated with Indigo displayed the lowest solar hemispheric transmittance: their transmittance varied between 67.3 and 76.0% (layers without Indigo 78.8–82.3%) at room temperature and between 66.1 and 76.3% (layers without Indigo 79.6–82.3%) at 70°C. That was attributed to the coloration of TSFD induced by Indigo. In Figure 1, some hemispheric transmittance spectra recorded at room temperature of layers formulated without (solid line; factor combination AT 11: 0.01 mol kg-1; Indigo: 0 mol kg-1; treatment: RT; annealing: no) or with (dash-dotted line; factor combination AT 11: 0.01 mol kg-1; Indigo: 0.001 mol kg-1; treatment: RT; annealing: no) Indigo are presented. Around 600 nm, the layer formulated with Indigo displayed a lower transmittance compared to the layer formulated without Indigo. That was ascribed to the absorption characteristics of Indigo. Furthermore, Indigo introduced inhomogeneities to the TSFD: in optical micrographs, individual pigment particles of Indigo were discernible. Figure 2 shows an optical micrograph of a layer formulated without Indigo (Figure 2A) along with the optical micrograph of a layer formulated with Indigo (Figure 2B). Both layers exhibited spherical scattering domains (bright spots). Apart from the scattering domains, the layer formulated with Indigo showed plate-like structures (black arrows are pointing on several ones) additionally. These structures were Indigo (pigment) particles. These particles are distributed in the matrix and are not nucleating an additive domain. Thus, Indigo seems to be not as effective in nucleating the thermotropic additive A1 as desired. Furthermore, the layers formulated with Indigo exhibited agglomerates of Indigo, yielding heterogeneous coloration.
aReference case M7A1-OTA-p3.RT-0.008 from reference  for comparison.
Summarizing, the addition of Indigo yielded a transmittance reduction of the prototype TSFD which was ascribed to the predominating effect of absorption of the incident radiation. Moreover, the desired nucleating agent-induced reduction of the scattering domain-size by addition of Indigo as potential nucleating agent for the thermotropic additive was not achieved. This may be ascribed to a lack of nucleating potential of Indigo for the utilized paraffin wax. Due to the observed inhomogeneities – caused by Indigo agglomerates – which yielded a potential bias for statistical analysis, subsequent discussion is going to omit TSFD formulated with Indigo.
An analysis of variance (ANOVA) of the data presented in Table 2 revealed the most important factors with regard to the solar hemispheric transmittance. The ANOVA was carried out according to the scheme presented by Montgomery  for evaluation of factorial designs applying a so-called fixed effects model. For the solar hemispheric transmittance at room temperature, the effect of the factor annealing was significant only (see Table 3). The other factors and the factor interactions displayed insignificant effects. On the contrary, for the solar hemispheric transmittance at 70°C, the effect of the factor treatment was significant only (see Table 4). Except for the factor AT 11 concentration – its effect was indifferent – all the other factors and all the factor interactions exhibited insignificant effects. The effect significance levels were defined according to Kleppmann’s  definition for effect significance assessment with regard to the α-error: A p value >0.05 denotes an insignificant effect, whereas a p value between 0.05 and 0.01 corresponds to an indifferent effect. A p value between 0.01 and 0.001 denotes a significant effect, whereas a value <0.001 is ascribed to a highly significant effect.
|Source of variation||Effects and interactions||pa for test regarding factor effect on τnhRT||Effect significance assessmentb|
aTest statistics according to Montgomery .
bAccording to Kleppmann .“-” insignificant, “+” significant.
|Source of variation||Effects and interactions||pa for test regarding factor effect on τnh70||Effect significance assessmentb|
aTest statistics according to Montgomery .
bAccording to Kleppmann .“-” insignificant, “o” indifferent, “+” significant.
However, an ANOVA does not convey information about the direction and magnitude of the factor effects (or the interactions of factors). In order to obtain this information, the calculation of the main effects (briefly: effects) of the factors and the interactions of the factors was carried out according to Montgomery  (the coded factor levels according to Table 1 were employed upon calculation of the effects and interactions). That way, the so-called effects and the interactions are interpreted to be parts of the fitting parameters in a regression model in order to obtain a least-squares fit of the response of the investigated design. More precisely, the effects and interactions are exactly twice the coefficients of the corresponding variables in the regression model, whereby the applied factor levels represent specifically set values of the variables [46, 47].
Figure 3 displays a plot regarding the factor effects of the factors concentration of the functional additive AT 11, thermal treatment and annealing on the solar hemispheric transmittance at room temperature (RT) and at 70°C of the layers formulated without Indigo. The individual data points are presented for room temperature (squares) and for 70°C (circles). Furthermore, the change in the mean value of the solar hemispheric transmittance upon changing a factor level is presented either with a solid line (for RT) or a dashed line (for 70°C). The results were derived from the data presented in Table 2. The mean solar hemispheric transmittances at room temperature and 70°C were both lower for the factor level 0.01 mol kg-1 than for the factor level 0.02 mol kg-1 of the factor AT 11 concentration. The nearly parallel slope of the mean plots for the factor AT 11 concentration indicated that this factor did not interact significantly with the measurement temperature. Furthermore, for both factor levels of the factor AT 11 concentration, the mean solar hemispheric transmittance only changed to a minor extent when exceeding the switching threshold. Probably, an increase in the AT 11 concentration improved the overall transmittance of the TSFD by affecting the scattering domain size distribution. Imagine two scattering domain size distributions: whereas one size distribution displays scattering domains with sizes close to the optimum, another size distribution displays bigger domains. Assuming a slight difference in refractive indices of matrix and additive already existing at room temperature, a TSFD exhibiting the latter scattering domain size distribution will show pronounced forward-scattering and reduced back-scattering compared to a TSFD exhibiting optimally sized scattering domains. Thus, solar hemispheric transmittance will also be higher. However, the factor effects of the factor AT 11 concentration with regard to the solar hemispheric transmittance at room temperature and at 70°C were small. Actually, the factor effects were insignificant and indifferent with regard to the solar hemispheric transmittance detected at room temperature and 70°C, respectively (see Tables 3 and 4). Notwithstanding, the obtained information with regard to the TSFD morphology might provide more insight: Table 5 presents the number weighted and Sauter mean diameters of the scattering domains of the layers formulated without Indigo employing different factor combinations (two replicates each). An ANOVA revealed all the factor effects and their interactions to be insignificant with regard to both mean diameters. This was attributed to the low homogeneity of the layers. The low homogeneity is indicated by the rather different mean diameters that were obtained for the replicates 1 and 2 at each factor combination. Consequently, the additive AT 11 seems to be inappropriate in order to reduce the scattering domain size and thus to yield a significant improvement of the overheating protection performance. This finding is supported by the observation that the addition of AT 11 was not effectual in order to enhance the difference between solar hemispheric transmittance at room temperature and at 70°C. For a layer lacking any AT 11 (specifically if that was layer M7A1-OTA-p3-RT-0.008 , which represents a treatment combination of AT 11: 0 mol kg-1; Indigo: 0 mol kg-1; treatment: RT; annealing: no; see Table 2), a solar hemispheric transmittance of 81.2 and 78.5% was achieved below and above the threshold temperature, respectively. Upon the application of an AT 11 concentration of 0.01 mol kg-1, the solar hemispheric transmittance was recorded to be 81.9 and 81.1% below and above the threshold temperature, respectively. For an AT11 concentration of 0.02 mol kg-1, the solar hemispheric transmittance was 81.6 and 81.8% below and above the threshold temperature, respectively. These mean values were calculated from respective data presented in Table 2 – rows 1 and 2. Thus, AT 11 did not act in the supposed manner (like a surfactant in an emulsion) in the investigated resin systems. However, the two other factors, treatment and annealing, showed more distinct effects, as already demonstrated by ANOVA.
sample 1 (μm)
sample 2 (μm)
For both factor levels of the factor treatment, the mean solar hemispheric transmittances at room temperature were almost the same. For the factor level RT, the mean solar hemispheric transmittance was higher at 70°C than at room temperature. By contrast, for the factor level DF, the mean solar hemispheric transmittance was lower at 70°C than at room temperature.
For the factor annealing, the findings were quite similar. The layers lacking annealing displayed a higher mean solar hemispheric transmittance at room temperature than the annealed layers. For the layers lacking annealing, the solar hemispheric transmittance was lower at 70°C than at room temperature. For the annealed layers, the temperature-dependent behavior was inverted. Such kind of behavior was observed already in earlier studies [28–30]: TSFD without defects (like vacuoles or voids) showed a decrease in the solar hemispheric transmittance upon exceeding the threshold temperature. This behavior is consistent with the observed change in the relative refractive index (an increasing deviation from unity) between matrix and thermotropic additive upon exceeding the threshold temperature [28, 30]. By contrast, TSFD displaying defects (especially vacuoles) exhibited an increase in the solar hemispheric transmittance upon exceeding the threshold temperature [28–30]. The low solar hemispheric transmittance at room temperature was attributed to the relative refractive index established between matrix/additive (n∼1.5) and vacuoles (n=1) which was significantly different from unity. Upon melting of the additive, the vacuoles disappeared and the subsequent change in the relative refractive index closer towards unity at the scattering interfaces yielded a higher solar hemispheric transmittance than at room temperature. Taking these previous findings into account, the annealed layers were suspected to exhibit defects. Furthermore, the insignificance of all the factors and their interactions with regard to the mean diameters of the scattering domains suggested that the effect significances of the factors treatment and annealing, with regard to solar transmittances, were not ascribable to differences in the mean scattering domain size rather than to other morphological features.
Thus, the findings for the factors treatment and annealing were an indication for possible changes in the internal material structure upon changing the factor levels. Thus, in Figure 4, optical light micrographs of some layers formulated upon application of different factor combinations are displayed. The focus is on the different factor levels of the factors treatment and annealing. Whereas the layers lacking annealing displayed a low number of vacuoles inside the spherical scattering domains (Figure 4A and 4C), the annealed layers exhibited a higher number of vacuoles (dark areas) inside the scattering domains (Figure 4B and 4D). Nevertheless, the vacuole concentration was low compared to the layers which were not optimized with regard to the curing procedure . It can be concluded that in spite of enhancing the homogeneity of TSFD – which was initially supposed – the annealing procedure introduced additional vacuoles to the TSFD. This is confirmed by the high fraction of scattering domains exhibiting vacuoles (details see Table 5) in the layers which were annealed (75–88%). The layers lacking annealing exhibited a smaller fraction of scattering domains with vacuoles (32–59%). An interesting finding was that the effects of the factors treatment and annealing on the fraction of scattering domains that were displaying vacuoles, were indifferent and highly significant, respectively (ANOVA see Table 6). The negative sign of the effect of the factor treatment (-0.141) indicates that lowering the storage temperature from room temperature (factor level RT) to -20°C (factor level DF) reduced the vacuole fraction. This finding gives rise to the assumption – which was already postulated earlier, but no clear evidence was found  – that the differences in coefficient of thermal expansion (CTE) of the matrix material (e.g., 0.5×10-4 °C-1–1.5×10-4 °C-1 for an UV-cured polyester acrylate layer ) and the paraffin wax (0.7×10-3 °C-1–1.1×10-3 °C-1 ) together with different storage temperatures affected the number of the vacuoles that were formed. The vacuole formation process and its implications on the solar hemispheric transmittance are extensively discussed in the references [29, 30]. Nevertheless, the effect of the factor treatment is indifferent and thus represents no clear evidence. By contrast, the positive sign of the (highly significant) effect of the factor annealing (0.328) indicates that employing an annealing step increased the vacuole fraction. This observation led to the conclusion, that the significant effect of the factor annealing on the solar hemispheric transmittance at room temperature is associated with these vacuoles. The vacuoles were probably formed during the annealing process. This was ascribed to a thermally-driven diffusion process of the additive in its molten state, leaving less thermotropic additive within the cavities provided by the matrix material after the annealing process. This hypothesis is supported by the observation of a thermally-induced diffusion processes of a thermotropic additive in a preceding study , yielding a loss of thermotropic additive inside the matrix cavities. Vacuoles have a refractive index equal to unity, yielding a relative refractive index at the scattering interfaces (boundaries to matrix or additive, which both have an RI around 1.5 at room temperature) that is far from unity. Consequently, high scattering efficiency is achieved, yielding a low solar hemispheric transmittance at room temperature. Upon exceeding the threshold temperature, the remaining additive fills the cavity completely due to its higher CTE compared to the matrix (e.g., 0.5×10-4 °C-1–1.5×10-4 °C-1 for a UV-cured polyester acrylate layer  vs. 0.7×10-3 °C-1–1.1×10-3 °C-1 for paraffin ) and vacuoles vanish. Thus, a relative refractive index closer to unity is established at the scattering interfaces (only matrix/additive; with an RI around 1.5 and 1.44 for matrix and additive, respectively). Consequently, the achieved lower scattering efficiency yields a higher solar hemispheric transmittance at 70°C than at room temperature. For layers lacking vacuoles, a change in relative refractive index between matrix and scattering domain away from unity is achieved upon exceeding the threshold temperature. Consequently, an increase in scattering efficiency and hence, a reduction in solar hemispheric transmittance upon exceeding the threshold temperature is achieved for these layers. For the layers that were not annealed, their lower vacuole fraction made the impact of the vacuoles on the overheating protection performance less dominant compared to the annealed layers, which exhibited a high vacuole fraction.
|Source of variation||Effects and interactions||pa for test regarding factor effect on vacuole fraction||Effect significance assessmentb|
aTest statistics according to Montgomery .
bAccording to Kleppmann .“-” insignificant, “o” indifferent, “++” highly significant.
The considerations made earlier indicated that vacuoles are supposed to be absent at 70°C. This hypothesis is supported by the assignment of an insignificant effect to the factor annealing with regard to solar hemispheric transmittance at 70°C. If vacuoles would have been present at 70°C, the established relative refractive index far from unity at the scattering interfaces would have yielded a high scattering efficiency and thus, a lower solar hemispheric transmittance compared to the layers without vacuoles. The absence of vacuoles is hence also the reason why the layers formulated upon application of the different factor levels of factor annealing displayed rather similar mean solar hemispheric transmittance at 70°C.
However, the studies on TSFD morphology did not give a guiding line for the reason why the factor treatment is the only significant factor with regard to solar hemispheric transmittance at 70°C. As already pointed out, no significant effects on the mean scattering domain diameter were evident for the different factor combinations. This was ascribed to the small effects of the different applied factors, probably yielding only small changes in scattering domain size, which were probably not resolvable with the light microscopy applied within this study on the one hand and on the other hand, the effects were probably biased by the inhomogeneity of the TSFD.
A key result of this study is that the employed factor combinations did not improve the scattering domain size of the TSFD and hence, did not improve the overheating protection performance of the TSFD reasonably. However, the experiments also showed that the overheating protection performance of a TSFD is significantly affected by vacuoles. The vacuole formation has to be avoided under any circumstances. A preceding study  pointed out that the vacuole formation can be prevented by careful adjustment (i.e., reduction) of the radiation intensity and dose employed upon curing of the TSFD. Now it is proven that selection of disadvantageous processing conditions other than the radiation intensity and dose can also yield the undesirable vacuole formation and thus, a reduction of overheating protection performance of the TSFD investigated.
4 Summary and outlook
For the investigated TSFD system, the addition of Indigo as a potential nucleating agent reduced the solar hemispheric transmittance at room temperature significantly, due to coloration. Indigo also introduced undesirable inhomogeneities into the TSFD, due to its insolubility and its lack of dispersibility in the resin matrix. Furthermore, it was not able to nucleate the thermotropic additive in the desired way and thus, to reduce the scattering domain size. This was ascribed to a lack of nucleation potential of Indigo with regard to the employed thermotropic additive. Altogether, the addition of Indigo was not beneficial with regard to the TSFD performance. Thus, a statistical data analysis was conducted for the layers formulated without Indigo. For these layers, the application of an annealing step yielded a lower solar hemispheric transmittance at room temperature than for the layers lacking annealing. Furthermore, the annealed layers showed an increase in solar hemispheric transmittance upon exceeding the threshold temperature. The observed behavior of the annealed layers was ascribed to a high fraction of scattering domains exhibiting vacuoles which were formed upon annealing. The vacuole formation was attributed to the thermally-induced diffusion of the molten additive, yielding an additive loss inside the cavities provided by the surrounding matrix material. The vacuoles introduced a relative refractive index at the scattering interfaces at room temperature that was far from unity, yielding intense scattering. Upon melting of the thermotropic additive, the vacuoles disappeared and the relative refractive index at the scattering interfaces got closer to unity, along with a reduction in scattering efficiency. In contrast, the layers lacking annealing displayed significantly fewer vacuoles. Thus, the effect of the vacuoles in these layers was less prominent compared to the annealed layers. Consequently, a high solar hemispheric transmittance at room temperature and a transmittance reduction upon exceeding the threshold temperature were ascertained. Nevertheless, due to absence of vacuoles, the morphology of the layers which were subject to annealing and which were not subject to annealing was likely to be similar above the switching threshold of the TSFD. Consequently, the factor annealing had no significant effect on solar hemispheric transmittance at 70°C. By contrast, the variation of the thermal treatment of the TSFD prior to and post the curing process (storage at -20°C for 10 min prior to and post curing instead of storage at room temperature) had a significant effect on the solar hemispheric transmittance at 70°C (but not on transmittance at room temperature). However, a morphological analysis by optical light microscopy revealed no significant differences with regard to the scattering domain size of the layers stored at -20°C or room temperature. The employed measures revealed no breakthrough in attempting a significant improvement of the overheating protection performance of the TSFD. Although several of the applied measures slightly improved the overheating protection performance of the investigated TSFD, the attained solar-weighted transmittance change was not appropriate in order to maintain an efficient overheating protection for a solar thermal collector. This was primarily attributed to an inappropriate scattering domain size. Hence, other factors than those investigated might have a more important effect on the scattering domain diameter and thus, the solar-weighted transmittance change. Some of these factors might be the deformation employed upon mixing, the interfacial tension matrix/additive, the viscosity ratio of the molten additive and the matrix resin, or the type of the flow field (laminar or turbulent), which seem to play a more important role than initially expected [35, 52]. Thus, future work has to focus on investigating other factors or entirely different approaches in order to improve the scattering domain size.
Nevertheless, the formation of vacuoles via a thermally-induced diffusion of the molten thermotropic additive yielding a deterioration of the overheating protection performance of TSFD might have severe implications with regard to the application of a TSFD as an overheating protection glazing. In order to preserve the long-term overheating protection performance of TSFD, such diffusion has to be suppressed, probably by an encapsulation of the thermotropic additive with a barrier layer. Otherwise, the efficiency of a solar thermal collector equipped with a TSFD exhibiting a thermally-induced additive diffusion as an overheating protection glazing will suffer significantly from the reduced solar hemispheric transmittance of the TSFD. Accordingly, investigations dealing with the adjustment of the scattering domain size on the one hand and the encapsulation of the additive in order to prevent the additive diffusion on the other hand and a subsequent assessment of the long-term stability of the obtained TSFD, are currently under way.
This research project is funded by the State Government of Styria, Department Zukunftsfonds (Project number 5019). The efforts in determination of solar-optical properties of parts of the formulated TSFD by Alexander Klutz (Department of Polymer Engineering and Science, University of Leoben, Leoben, Austria) are gratefully acknowledged. Furthermore, the authors want to acknowledge the contributions of Allnex Belgium SA/NV (former Cytec Surface Specialities Inc., Drogenbos, Belgium), Sasol Wax GmbH (Hamburg, Germany) and HDS-Chemie HandelsgesmbH (Vienna, Austria), BASF SE (Ludwigshafen, Germany) and BTC Europe GmbH (Cologne, Germany).
 Inoue T. Energy Build. 2003, 35, 463–471.Search in Google Scholar
 Yao J, Zhu N. Build. Environ. 2012, 49, 283–290.Search in Google Scholar
 Resch K, Hausner R, Wallner GM, Lang RW. In Polymeric Materials for Solar Thermal Applications, 1st ed., Köhl M, Meir MG, Papillon P, Wallner GM, Saile S, Eds., Wiley-VCH: Weinheim, 2012, p 255.10.1002/9783527659609.ch15Search in Google Scholar
 Resch K, Weber A. Berg- Huettenmaenn. Monatsh. 2011, 156, 429–433.Search in Google Scholar
 Bühler FS, Hewel M. Reversibly thermotropic transparent molding material, useful e.g. in glazing or covers for shading and light-heat regulation in houses and cars etc., DE19841234C1, German Patent, 1998.Search in Google Scholar
 DeArmitt C, McKee GE. Moulded article with temperature dependent transparency, EP1985663A1, German Patent, 2008.Search in Google Scholar
 Knausz M. Untersuchung des Schaltverhaltens von Thermotropen Gießharzen, Bachelor thesis, University of Leoben, 2009.Search in Google Scholar
 Mikl MJ. Herstellung und Charakterisierung von Thermotropen Systemen mit Fixierten Domänen, Diploma thesis, University of Leoben, 2010.Search in Google Scholar
 Resch K, Wallner GM. In Proceedings of ISES Solar World Congress 2007: Solar Energy and Human Settlement, Goswami DY, Zhao Y, Eds., Springer: Berlin, 2007, p 541.Search in Google Scholar
 Resch K, Wallner GM. Polym. Adv. Technol. 2009, 20, 1163–1167.Search in Google Scholar
 Resch K, Wallner GM, Lang RW. Macromol. Symp. 2008, 265, 49–60.Search in Google Scholar
 Ruhmann R, Seeboth A, Muehling O, Loetzsch D. AST 2012, 77, 124–131.Search in Google Scholar
 Schmid A. Untersuchung der Zusammenhänge zwischen Morphologie und den Überhitzungsschutzeigenschaften von Thermotropen Systemen mit Fixierten Domänen, Bachelor thesis, University of Leoben, 2012.Search in Google Scholar
 Seemann M. Charakterisierung des Lichtstreuverhaltens und der Morphologie von Thermotropen Verglasungen auf Thermoplastbasis, Master thesis, University of Leoben, 2010.Search in Google Scholar
 Weber A. Analyse der Morphologie und des Schaltvorganges von Thermotropen Polymeren mittels Rasterkraftmikroskopie, Master thesis, University of Leoben, 2010.Search in Google Scholar
 Weber A, Resch K. In Solar Building Skins: Conference Proceedings of the 6th ENERGY FORUM, Economic Forum, Ed.: Munich, 2011, p 73.Search in Google Scholar
 Weber A, Resch K. J. Polym. Res. 2012, 19:9888, 1–8.Search in Google Scholar
 Weber A, Schlögl S, Resch K. J. Appl. Polym. Sci. 2013, 130, 3299–3310.Search in Google Scholar
 Nitz P. Optical Modelling and Characterisation of Thermotropic Systems, Dissertation, Albert-Ludwigs-University, 1999.Search in Google Scholar
 Resch K, Weber A, Gruber DP, Schnetzinger K, Kern W. Smart Windows – Smart Collectors: Entwicklung, Modellierung und Vermessung von Überhitzungsschutzverglasungen für Fassaden- und Kollektoranwendungen: Endbericht WPR-NKP.09.014-03. Polymer Competence Center Leoben GmbH: Leoben, 2013.Search in Google Scholar
 Walstra P. Chem. Eng. Sci. 1993, 48, 333–349.Search in Google Scholar
 Tadros T, Izquierdo P, Esquena J, Solans C. Adv. Colloid Interface Sci. 2004, 108–109, 303–318.Search in Google Scholar
 Dalmazzone C. Oil Gas Sci. Technol. 2000, 55, 281–305.Search in Google Scholar
 Kurja J, Mehl NA. In Plastics Additives Handbook, 5th ed., Zweifel H, Ed., Hanser: München, 2001, p. 949.Search in Google Scholar
 Horn D, Rieger J. Angew. Chem. Int. Ed. 2001, 40, 4330.Search in Google Scholar
 Ehrenstein GW, Riedel G, Trawiel P. Praxis der Thermischen Analyse von Kunststoffen, 2nd ed., Hanser: München, 2003.Search in Google Scholar
 Fairgrieve S. Nucleating Agents, Rapra Technology Ltd.: Shawbury, UK, 2005.Search in Google Scholar
 Rieger J. In Polymer Crystallization: Observations, Concepts, and Illustrations, Reiter G, Sommer J, Eds., Springer: Berlin, 2003, p. 7.Search in Google Scholar
 Galeski A. In Polypropylene: Structure, Blends and Composite: Structure and Morphology, Karger-Kocsis J, Ed., Chapman & Hall: London, 1995, Vol. 1, p. 116.Search in Google Scholar
 Di Marco G, Pieruccini M. In Polymer Crystallization: Observations, Concepts, and Illustrations, Reiter G, Sommer J, Eds., Springer: Berlin, 2003, p. 366.Search in Google Scholar
 Montgomery DC. Design and Analysis of Experiments, 7th ed., Wiley: Hoboken, NJ, 2009.Search in Google Scholar
 Kleppmann W. Taschenbuch Versuchsplanung: Produkte und Prozesse Optimieren, 6th ed., Hanser: München, 2009.Search in Google Scholar
 Resch K, Weber A, Gruber D, Schnetzinger K, Kern W. Smart Windows – Smart Collectors: Entwicklung, Modellierung und Vermessung von Überhitzungsschutzverglasungen für Fassaden- und Kollektoranwendungen: Zwischenbericht 2, WPR-NKP.09.014-02, Polymer Competence Center Leoben GmbH: Leoben, 2012.Search in Google Scholar
 Schimmelpfennig M, Weber K, Kalb F, Feller K, Butz T, Matthäi M. In Jahrbuch für den Praktiker 2007: Handbuch für Rohstoff und Formulierung Industrie, Gewerbe und Haushalt, Ziolkowsky B, Ed., Verlag für chemische Industrie: Augsburg, 2007, Vol. 50, p. 417.Search in Google Scholar
 Grace HP. Chem. Eng. Commun. 1982, 14, 225–277.Search in Google Scholar
©2014 by Walter de Gruyter Berlin/Boston