Hygro-thermal coupling on 4D-printed biocomposites as key for meteosensitive shape-changing materials

ABSTRACT

Hygromorph BioComposites (HBC) exhibit shape-changing actuated by a moisture content (MC) variation due to their bilayered bioinspired architecture. However, their application in daily meteorologic variation is limited due to relative humidity (RH) and thermal couplings. This work aims to investigate the role of temperature variation on the shape changing of 4D printed continuous flax fibres reinforced HBC so that ThermoHygromorph BioComposites (THBC) are introduced.

The responsiveness of the THBC is governed by the MC of the material and the stiffness ratio between the passive and active layer, which are both thermally dependant. An increase in temperature strongly reduces the actuation amplitude while the actuation kinetics is strongly fastened. A 4D-printed HBC structure is designed and submitted to a practical outdoor environmental loading, to understand the impact of daily RH and temperature variations on THBC actuation.

KEYWORDS:

• 4D-printing
• biocomposite
• hygrothermal characterisation

1. Introduction

The growing stakes in terms of access to energy and raw materials are motivating the recent development of bioinspired passive actuators, made from local and renewable materials, as an alternative to complex electro-mechanical systems. The recent concept of 4D printing, consisting in the design of smart materials or structures possessing extra-functionalities such as shape-morphing [1] or self-assembly [2] and manufactured by 3D printing techniques, emerges as a relevant solution for the development of passive actuators.

Several works focus on the development of wood passive actuators for applications such as meteorosensitive building [3] or solar trackers [4]. Other works successfully developed shape changing short wood fibre composites manufactured via 3D printing [5,6]. These materials present interesting actuation properties, but the use of short fibre inevitably presents mechanical limits for the development of structural actuation including both large displacement and stiffness. For several years, Le Duigou et al. worked on the development of hygro-sensitive continuous flax fibre (cFF) reinforced biocomposites called Hygromorph BioComposites (HBC) [7–9]. Their bilayered structure is inspired from several hygromorph biological actuators for seed dispersion such as pinecone [10] or wheat awn [11]. The active and passive layer possess, respectively, a high and low hygroscopic expansion. The differential expansion between the two layers results in the bending of the bilayer strip. The actuation of the slender HBC beam is predictable, thanks to the Timoshenko equations, initially used for the curvature prediction of thermally actuated bimetallic strips [12]. Thus, the curvature of HBC depends on mechanical properties, expansion and thickness of the passive and the active layer. The use of 4D printing allows the description of material syntax by tailoring of the architecture of the bilayer via the printing parameters (e.g. layer height, interfilament distance, fibre orientation) [13,14] to target a specific shape changing.

Recent studies have been done to evaluate the impact of relative humidity and immersion on 3D printed continuous flax fibre HBC [15,16]. De Kergariou et al. [16] showed that 3D printed cFF reinforced PolyLactic Acid (PLA) biocomposites exposed to relative humidity (RH) undergo a decrease of about 64% in longitudinal and transverse elastic moduli between 10% RH and 98% RH. The actuation amplitude (responsiveness) of a HBC during immersion can be enhanced by using a soft matrix, while the use of a stiff matrix improves mechanical performances [15]. In the case of outdoor terrestrial applications, daily variations of relative humidity are always coupled with temperature variations. Thus, according to the climate and the season, the daily variations of temperature can reach several tens of degrees while the relative humidity can vary between 40% and 90% RH.

However, for a given relative humidity, the temperature increase directly impacts the sorption behaviour of a biobased material by reducing its moisture content [17–20] and by increasing the sorption kinetics [21–24]. Furthermore, the evolution of temperature can involve a significant influence on thermoplastic polymer properties, even more so for temperatures close to the polymer glass transition temperature (T_g). The temperature also impacts the interfacial properties between fibres and thermoplastic matrices. A temperature increase leads to a decrease in Interfacial Shear Strength (IFSS) and thus reduction of stress transfer ability [25]. As far as we know, very few studies focus on the influence of hygro-thermal coupling on the behaviour of natural fibre-reinforced biocomposites. Few works have been carried out on the impact of temperature on sorption behaviour and mechanical properties of flax fibre-reinforced composites [22,24]. These studies were carried out under immersion conditions, which implies different degradation phenomena compared with conditioning under relative humidity.

Similarly, most of the work dedicated to smart polymer-based composites is focused on a mono stimulus triggered actuation, temperature or moisture, but rarely both. Few studies investigate the actuation potential of multiresponsive materials which are triggered by moisture variation and temperature, where the temperature is manually controlled by electro-heating or using UV radiation [26,27]. Particularly, Le Duigou et al. [27] have developed an electro-thermo-hygromorph carbon fibre-reinforced polyamide. Nevertheless, it appears essential to study the coupling between relative humidity and temperature in a non-immersed environment to fully understand the role of each stimulus in HBC actuation and to further programme Thermo-Hygromorph BioComposite actuators (THBC) submitted to outdoor daily environmental variation.

The present work investigates the role of outdoor key parameters such as temperature and relative humidity variations on the hygroscopic, mechanical and actuation behaviour of 3D/4D-printed continuous Flax Fibre-reinforced PolyButylene Adipate Terephthalate (cFF/PBAT). The first step consists of determining the hygrothermal steady state (i.e. moisture content) of the material at various relative humidities and temperatures corresponding to idealised environmental conditions. The considered humidity range (0−95%RH) as well as the temperature range (20−80°C) are supposed to cover the RH and temperature variation of an oceanic, continental or arid climate, within the limits of experimental constraints. Then, investigation on the influence of relative humidity variations on hygroscopic expansion, thermal expansion and mechanical properties for several temperatures (20°C, 40°C, 60°C, 80°C) are performed. The longitudinal and transversal elastic moduli are then measured at 20°C, 40°C, 60°C and 80°C, for various moisture contents. The influence of the complex coupling between relative humidity and temperature is finally discussed in term of actuation motor for 4D printed [0/90°] asymmetric bilayers THBC actuators.

2. Material and methods

2.1. Material

Continuous flax fibres organised in yarn were provided by Safilin (Sailly-sur-lalys, France). The selected yarn is characterised by a twist of 320 turns/metre and a linear density of 68 Tex (g/km). PolyButylene Adipate Terephthalate (PBAT) Ecoflex® F Blend C1200 was provided by BASF®. PBAT has a glass transition temperature $(T_g)$ below room temperature, measured between −33°C and −24°C [28].

2.2. Filament production and 3D-printing of the samples

The filament used in this work is produced by the co-extrusion of continuous flax fibre’s yarn with PBAT. The flax fibre yarn is inserted in a heated die fed with PBAT. The screw temperature is set at 175°C and the die temperature is set at 190°C. Impregnation of the flax fibre yarn by the melted polymer occurs in the die. The filament production speed is set by a puller at 0.8 mm/min. The impregnated filament is then cooled at room temperature and rolled up. The diameter of the produced filament is constrained by the nozzle diameter of the die (0.6 mm).

The different samples are 3D-printed thanks to a customised Prusa MK3s printer. A tailored nozzle with a 0.8-mm nozzle diameter is used for the printing of continuous flax fibre-reinforced filament. The Gcodes are designed with GrassHopper® since conventional slicers are not adapted to generate continuous paths. During the printing process, the nozzle temperature is set at 145°C and the bed temperature is set at 30°C. The filament is printed at 6 mm/s with an interfilaments distance of 0.6 mm and a layer height of 0.3 mm. The longitudinal and transversal directions are referred to as the in-plan directions oriented at 0° and 90°, respectively, with respect to fibre orientation. The out-of-plane direction corresponds to the orientation defined by the thickness of the biocomposite.

2.3. Microstructure observation

The microstructure observation of 3D-printed cFF/PBAT biocomposites is investigated by scanning electron microscopy (SEM). The samples are polished and sputtered with silver nanoparticles before being inspected using a Jeol JSM-IT500HR SEM at 10 kV. Image analysis is then performed using ImageJ software® (National Institute of Health, USA) to evaluate the fibre volume fraction $V_f$ and porosity content $V_p$. Five transversal cross-sections of unidirectional biocomposite are investigated. Each cross-section is a combination of at least five images because the magnification, which was large enough to observe the microstructure precisely, does not allow the entire cross-section to be observed in a single image. A morphological analysis tool is used to measure the area of flax fibres and porosities. The threshold is adjusted with the same min and max grey values for each image.

2.4. Gravimetric measurement

Relative humidity and temperature conditioning are performed for all the experiments in this study. The moisture content of samples is systematically controlled by gravimetric measurement after conditioning. To do so, samples are dried in an oven at 40°C under vacuum for at least 48 h and until a constant mass is recorded, before each conditioning. A high precision scientific balance (precision of ${10}^{-4}$g) is used to measure samples’ mass. After the drying step, the samples are conditioned in a controlled relative humidity environment. At room temperature, the conditioning is done by using hermetic enclosures where the relative humidity is controlled thanks to saturated salt solutions, going from 9%RH (KOH), 33%RH (MgCL₂), 55%RH (Mg(NO₃)₂) and 76%RH (NaCl) to 98%RH (K₂SO₄). For other temperatures, the conditioning is done by using a WeissTechnik WKL 64/40 climatic chamber able to regulate both temperature and relative humidity. At each relative humidity level, the equilibrium mass of the samples is determined when a constant mass is assumed to be reached. The moisture content is then computed as follows: (1) $\begin{array}{c}\mathrm{MC}(\text{\%})={\displaystyle \frac{M_t-M_0}{M_0}\ast 100}\end{array}$(1) where $M_t$ is the equilibrium mass of the sample after exposure to humidity and $M_0$ is the mass of the dry material before sorption.

2.4.1 Dynamic vapour sorption (DVS) method

To finely study the sorption behaviour of 3D-printed cFF/PBAT samples, DVS tests are conducted in addition to gravimetric measurements. The equipment used is an IGASorp-HT system (Hiden Analytical, Warrington, UK), which allows to measurement of the mass uptake and sorption kinetics of a sample thanks to a microbalance, in a temperature and humidity-controlled environment. About 10–30 mg of material is placed in the IGASorp microbalance which has a resolution of 0.1 µg. Before each test, the sample is dried in flowing air (RH < 1%) for one hour at 105°C. The sample is then exposed, at a fixed temperature, to an increase in relative humidity from 0% until 95%RH, in several humidity steps (10%, 30%, 50%, 60%, 70%, 80%, 85%, 90%, 95%). The equilibrium mass $M_t$ of the sample at each step is determined by the extrapolation of a single exponential curve fit to the time-dependent mass response following a step change in RH. The moisture content is then calculated using the same formula (1) as the gravimetric method. The tests are realised at three different temperatures (20°C, 40°C, 80°C)

2.4.2 Diffusion coefficient measurements

Diffusion coefficients have been evaluated from DVS tests to study the impact of temperature and relative humidity on sorption kinetics. Several studies have shown that the moisture adsorption behaviour of natural fibres reinforced biocomposites can be described by Fick’s Law [29–31]. For thin laminates, with a thickness $h$ significantly lower than the width, with uniform initial distribution and equal initial surface concentration, Fick’s Law leads to the following expression [29]: (2) $\begin{array}{c}{\displaystyle \frac{M_t}{M_m}=1-{\displaystyle \frac{8}{{\pi }^2}\sum _{n=0}^{\mathrm{\infty }}{\displaystyle \frac{1}{{(2n+1)}^2}\exp \left({\displaystyle \frac{-{(2n+1)}^2{\pi }^2\mathrm{Dt}}{h^2}}\right)}}}\end{array}$(2) where $M_t$ and $M_m$ are the moisture uptake at time $t$ and at saturation and $D$ is the apparent diffusion coefficient. Short times corresponding to $M_t/M_m\le 0.5$ the previous equation can be estimated by (3) $\begin{array}{c}{\displaystyle \frac{M_t}{M_m}={\displaystyle \frac{4}{h}\sqrt{{\displaystyle \frac{\mathrm{Dt}}{\pi }}}\leftrightarrow D={\displaystyle \frac{\pi }{{(4M_m)}^2}{\left({\displaystyle \frac{M_{t}h}{\sqrt{t}}}\right)}^2}}}\end{array}$(3)

At every relative humidity step of the DVS results, the diffusion coefficient $D$ is extracted on the linear part of the curve $M_t/M_m=f\left(\sqrt{t}\right)$ by minimising the sum of squared errors between experimental data and Equation (3).

2.5. Thermo-expansion measurement

The coefficients of thermal expansion (CTE) of cFF/PBAT and net PBAT were measured by thermo-mechanical analysis (TMA) by using a DMA Q800 TA Instruments and according to the ISO 11359-2 standard. The samples were dried at 40°C for 48 h prior to each test to assume a dry state and only to evaluate the contribution of temperature. The samples were heated from −10°C to 100°C at a rate of 3°C/min. For each measurement, at least three squared samples (5 × 5mm²) were tested by using compressive clamps with a preload of 0.01 N. The coefficient of thermal expansion $\alpha (T)$ is then computed using the following equation: $\alpha (T)={\displaystyle \frac{\epsilon (T)-\epsilon (T-\mathrm{\Delta T})}{\mathrm{\Delta }T}}$where the resolution is $\mathrm{\Delta }T=5^{\circ }C$, and $\epsilon (T)$ is the thermal strain at temperature $T$.

2.6. Hygro-expansion measurement

The swelling of biocomposites exposed to increasing levels of relative humidity has been studied, at several temperatures. The longitudinal and transversal dimensional variations are measured thanks to the KEYENCE VHX-7000 digital microscope and the out-of-plane expansion is measured by using a digital vernier calliper Mitutoyo IP65 0–25 mm with an accuracy of 10⁻³mm. Five 20 mm squared samples with a thickness of 1.8 mm (i.e. 6 printing layers) are analysed and three different measurements per sample are taken for each fibre orientation (longitudinal, transversal, out-of-plane).

2.7. Mechanical characterisation

The longitudinal and transverse tensile properties of 3D-printed reinforced biocomposites are evaluated at different moisture contents and temperatures. Before the tests, samples are conditioned at different relative humidity values (0%, 30%, 50%, 70%, 90%RH) and temperatures (20°C, 40°C, 60°C, 80°C) and the equilibrium moisture content is controlled by gravimetric measurement. During the test, the temperature is controlled by using a thermal enclosure (MATAIR). The temperature in the thermal enclosure is monitored with a thermocouple placed on the specimen and the test is launched when the temperature is stabilised. The relative humidity is not controlled during the test but the desorption that occurred during the testing (a few minutes at 50% RH and room temperature) is negligible (measured to be around 0.2%).

The tensile properties are determined by quasi-static tensile tests according to the ISO 527-4 standard, using an Instron 5566A universal testing machine. Two types of samples are tested. The longitudinal samples (unidirectional fibre orientation at 0° with respect to loading orientation) and transverse samples (unidirectional fibre orientation at 90° with respect to loading orientation) are tested with a 10 kN and a 1 kN cell load, respectively. The tests are performed by using a tensile speed of 1 mm/min. During the tests, the strain measurement is done by using an Instron extensometer with a nominal length of 25 mm. According to Shah et al. [32], the longitudinal elastic moduli are determined in the strain range of 0.025–0.100%. Five replicates are tested for each measurement.

2.8. Curvature measurement

After being dried at 40°C during 48 h, the evolution over time of bending curvature of THBCs immersed in a humid environment (90%RH) was studied at several temperatures (20°C, 40°C, 60°C). The THBCs are 4D-printed as [0/90°] asymmetric bilayers with a length of 70 mm and a width of 10 mm. The total thickness of the sample is 0.6 mm and the thickness ratio between the passive (0°) and active (90°) layers is $m=1$. Pictures of the samples are taken at defined intervals, with a higher frequency of measurement at the start of tests and for tests at elevated temperatures. During the tests, the moisture content is also measured by the gravimetric method on a control sample. The measurement of the curvature of the sample is achieved by image analysis using ImageJ software® (National Institute of Health, USA). The curvature of the samples is fitted with a circle function allowing us to measure the radius of curvature.

3. Results

3.1. Microstructure observation

A typical SEM image of a transversal cross-section of a 3D-printed unidirectional cFF/PBAT biocomposite is presented in Figure 1. One can notice that the biocomposite structure is highly heterogeneous. For each printed layer, the flax yarn is found on the upper part of the layer, while the matrix tends to flow downwards. The measured fibre volume fraction is $V_f=30.2\pm 2.4\text{\%}$ and the porosity content is $V_p=2.8\pm 0.5\text{\%}$. These values are in the same range as previous work based on 3D-printed continuous flax fibre-reinforced filament [15,16]. The biocomposite fibre volume fraction is mainly governed by the co-extruded filament fibre volume fraction, which is targeted at around 30% during production. Two types of porosities, which differ in size and location, can be distinguished. The porosities induced by the 3D-printing process, referred to as (1) on Figure 1, have a dimension in the range of 100 µm and are located at the matrix/flax yarn interface. The variability of the flax yarn cross-section promotes the presence of these porosities. The second type of porosities (2) is located inside the flax yarns and exhibits a dimension in the range of 1–10 µm. These porosities are generated during the co-extrusion process and remain present once the filament has been 3D-printed. The twisted structure of the yarn prevents the matrix from penetrating it, limiting impregnation to the filament periphery. Despite the presence of these porosities, the 3D-printing process allow to produce repeatable good-quality biocomposites.

Figure 1. SEM image of a transversal cross section of 3D-printed unidirectional cFF/PBAT biocomposite showing (1) 3D-printing process-induced porosities and (2) porosities located within the flax yarn.

3.2. Influence of temperature on sorption behaviour

Figure 2(a) presents the moisture content evolution of cFF/PBAT samples during a sorption cycle from 0% to 95%RH at 20°C and 40°C and from 0% to 85%RH at 80°C. The sigmoidal shape of the isotherms is characteristic of cellulosic and lignocellulosic materials [33]. The moisture content at saturation reached for cFF/PBAT samples at 95%RH at room temperature is 7.6 wt%. One can notice that temperature has a strong influence on cFF/PBAT sorption behaviour. The temperature increase leads to a consequent decrease in moisture content at saturation. At 95%RH, the moisture content at saturation is 27.7% lower at 40°C compared to 20°C. The decrease in moisture content while increasing temperature has been evidenced in other studies concerning cellulosic material [19,20]. Furthermore, the increase in temperature leads to an increase in water molecule mobility and a decrease in hydrogen bond strength [34,35] which also favours the detachment of water molecules from hydroxyl groups.

Figure 2. (a) Sorption isotherms of cFF/PBAT samples obtained by Dynamic Vapor Sorption (DVS). (b) Sorption isotherms computed with analytical models (GAB model in solid lines and Clapeyron equations in dashed lines) and compared with experimental data.

Figure 2(b) presents sorption isotherms computed with analytical models. The GAB model (Guggenheim-Anderson-de Boer) which is presented in the literature as the most versatile model to fit experimental data of water sorption in cellulosic material [17,36] is used to compute the 20°C and 80°C sorption isotherms. By considering the variation of isosteric heat of sorption $Q_{\mathrm{st}}$ which is evaluated by the Clausius–Clapeyron equations, sorption isotherms can be computed at any temperature between 20°C and 80°C. The 40°C and 60°C sorption isotherms, computed with Clapeyron equations, are presented in dashed line. More information about these models and their implementation can be found in Supplementary Data 1.

The diffusion coefficients $D$ of cFF/PBAT samples have been computed at every RH step of the DVS tests. More information about the method used to compute the diffusion coefficients are available in Supplementary Data 2. The mean diffusion coefficients over the 0–85%RH range are presented in Table 1. One can see that temperature strongly impacts diffusion kinetics. The mean value of the diffusion coefficient $D$ on the 0–85%RH range is about 8 and 20 times higher at 40°C and 80°C, respectively, compared to 20°C. The diffusion kinetics is accelerated by the increase in water molecules mobility induced by higher temperatures [21].

Table 1. Mean values of cFF/PBAT diffusion coefficients over 0–85%RH range at different temperatures.

	20°C	40°C	80°C
${\mathit{D}}_1$ (10^−5 mm².s^−1)	1.80 ± 0.79	10.93 ± 1.96	34.58 ± 7.72

These results allow the prediction of the hygrothermal steady state of a cFF/PBAT structure subjected to an idealised external temperature and relative humidity. In real outdoor conditions, relative humidity and outside temperature are constantly fluctuating, which means that the material never reaches its hygrothermal steady state.

3.3. Determination of coefficients of thermal expansion (CTE)

Figure 3(a) presents examples of measures of cFF/PBAT and neat PBAT thermal strain obtained by TMA. The computed CTE are presented in Figure 3(b). The results illustrate the strongly anisotropic thermal behaviour of the 3D-printed biocomposite. The measured transversal CTE of cFF/PBAT in the range of 20−80°C (${\alpha }_2=250\pm 10\mathrm{\mu m}/m{/}^{\circ }C$) is similar to the CTE of PBAT (${\alpha }_{\mathrm{PBAT}}=240\pm 10\mathrm{\mu m}/m{/}^{\circ }C$). Flax fibres also exhibit a transversal thermal expansion (${\alpha }_{f1}=82.7\mathrm{\mu m}/m{/}^{\circ }C$) [37], but their contribution is small at the biocomposite scale. In the longitudinal direction, the material does not show any dimensional variations up to around 60°C, a temperature above which the material presents a linear shrinkage. Thus, the material presents a negative longitudinal mean CTE over the range of 20−80°C (${\alpha }_1=-27\pm 12\mathrm{\mu m}/m{/}^{\circ }C$). In the longitudinal direction, the matrix thermal expansion is constrained by the fibres. Several studies have shown that natural fibres such as flax, jute or sisal exhibit a negative thermal expansion which explains the negative CTE of the biocomposite [37–39]. The abrupt change in slope at around 60°C could be explained by a relaxation of the stresses induced during the 3D printing process, as the flax yarn is under tension when the matrix cools down.

Figure 3. (a) Examples of thermal expansion measurement obtained by TMA. (b) Comparison of neat PBAT and cFF/PBAT coefficients of thermal expansion

3.4. Influence of temperature on coefficients of hygroscopic-expansion (CHE)

Figure 4(a) presents the comparison between the swelling of neat PBAT, radial swelling of single flax fibre [40] and transversal expansion of 3D printed cFF/PBAT as a function of relative humidity, at 20°C. Due to its low water absorption capacity, the hygroscopic expansion of PBAT is negligible compared to the radial swelling of a single flax fibre. At the biocomposite scale, the hygroscopic expansion is mainly driven by the swelling of the fibres. However, at 90%RH, the biocomposite’s transversal expansion is 88% lower than the radial swelling of a single flax fibre. This can be partly explained by the constraining effect of the matrix on the fibre swelling [41]. Fruleux et al. [15] have shown that for 3D printed continuous flax fibre-reinforced biocomposites, the stiffer the matrix, the greater the stress imposed by the matrix on the fibres. However, PBAT is a very soft matrix ($E=60\pm 30\mathrm{MPa}$ [15]). Thus, the large difference in hygroscopic expansion between flax fibre and cFF/PBAT is mainly due to the moderate fibre volume fraction ($V_f=30.2\pm 2.4\text{\%}$) of our biocomposites.

Figure 4. (a) Comparison of radial swelling of a single flax fibre [40], transversal expansion of a cFF/PBAT biocomposite, and swelling of neat PBAT as function of relative humidity at 20°C and (b) Evolution of transversal expansion as function of moisture content, for several temperature.

The influence of temperature on the transversal hygroscopic expansion of cFF/PBAT biocomposites is presented in Figure 4(b). The out-of-plane (thickness) expansion is not represented since it exhibits a similar behaviour. Due to the very low longitudinal swelling of flax fibres, the longitudinal expansion of the biocomposite is negligible compared to the transverse and out-of-plane expansions. Longitudinal expansion is therefore less relevant to be studied in the purpose of triggering shape changing. As a first approximation, the relationship between moisture content and hygroscopic expansion can be well described with a linear function (R²>0.98). Thus, the CHE can be identified as the slope of the linear relation between hygroscopic expansion and moisture content and is constant over the whole relative humidity range. The measured CHE and expansion at saturation for transverse and out-of-plane directions, for each temperature, are presented in Table 2. To go further, the hygroscopic strain could also have been interpolated with a sigmoid function as has been proposed in previous studies [8,14]. The influence of temperature on CHE is negligible. However, increasing temperature considerably reduces the biocomposite moisture content, leading to a reduction in hygroscopic expansion. This trend is also observed by Gager et al. [42] for flax fibre-reinforced polypropylene composites.

Table 2. Transversal and out of plane CHE and expansion at saturation of 3D printed cFF/PBAT biocomposites measured on 0–90%RH range.

	20°C	40°C	60°C
${\mathit{\beta }}_{\mathit{y}}(\text{\%}/\mathrm{\Delta }\mathit{m})$	0.30 ± 0.02	0.33 ± 0.04	0.34 ± 0.04
${\mathit{\beta }}_{\mathit{z}}(\text{\%}/\mathrm{\Delta }\mathit{m})$	0.52 ± 0.06	0.45 ± 0.03	0.46 ± 0.02
${\mathit{\epsilon }}_{\mathit{y}}(\text{\%})$	1.59 ± 0.17	1.27 ± 0.22	1.24 ± 0.19
${\mathit{\epsilon }}_{\mathit{z}}(\text{\%})$	2.67 ± 0.38	1.69 ± 0.60	1.85 ± 0.27

Figure 5 presents a comparison of cFF/PBAT expansions induced by stimuli variation in the range of daily RH and temperature variations. Whether it is an increase in temperature or relative humidity, longitudinal expansion is negligible compared to out-of-plane and transverse expansion. Dimensional variations are mainly governed by the hygroscopic state of the material. The thermal out-of-plane expansion induced by a temperature increase of 20°C is 73% lower than out-of-plane hygroscopic expansion induced by an RH increase of 40%RH at 20°C. Thus, the actuation of a THBC actuator will be mainly driven by hygroscopic expansions. It is essential to remember that in outdoor applications, variations in relative humidity and temperature are coupled, and that outdoor temperature can indirectly influence hygroscopic expansion.

Figure 5. Comparison between cFF/PBAT dimensional variations induced by a surrounding relative humidity increase of 40% and a temperature increase of 20°C.

3.5. Hygro-thermo-mechanical properties

Figure 6(a,b) shows the influence of moisture content and temperature on longitudinal and transversal elastic tensile moduli of 3D-printed cFF/PBAT biocomposites. The results obtained at room temperature are similar to those published by Fruleux et al. [15] for the same material. The moisture content has a strong impact on the longitudinal tensile properties of cFF/PBAT biocomposites. At room temperature, the longitudinal tensile elastic modulus reduction is up to 64% (from 12.7 to 4.6 GPa) between the dry state (MC = 0 wt%) and samples stored at 98%RH (MC = 5.3 wt%). The longitudinal tensile properties of the biocomposite are mostly driven by the flax yarn. Masseteau et al. have revealed a reduction in longitudinal elastic modulus of 20.9% between 1.5 and 7.6 wt% of the moisture content for a flax yarn [43]. In the yarn configuration, the plasticisation of the middle lamella which binds single fibres together leads to a reduction of stiffness during water uptake. At the ply scale, the overall plasticisation of fibres, matrix and fibre/matrix interface leads to a reduction of the biocomposite longitudinal stiffness.

Figure 6. (a) Longitudinal and (b) transversal tensile elastic modulus of 3D-printed cFF/PBAT as function of moisture content and for different temperatures.

In the transverse direction, the biocomposite mechanical properties are mainly governed by matrix properties. PBAT is a very soft and extendable matrix (cf Material and Methods). Thus, the transverse tensile elastic modulus of the composites is two orders of magnitude lower than in the longitudinal direction. The transverse tensile elastic modulus decreases with moisture content but the intensity of the decrease between the dry state and 98%RH is quite low (−18%) compared to the longitudinal direction. This decrease in stiffness can be again attributed to the plasticisation effect of water although this effect is less important on the matrix.

Contrary to the effect of moisture content, temperature mainly affects the properties of the matrix and thus the transverse properties of the biocomposite. Thuault et al. [44] have revealed that between 20°C and 80°C, the effect of temperature on a single flax fibre seems to be moderate although a slight decrease in elastic modulus is observed. In the present work, if one considers a dry sample (MC = 0 wt%), the transverse elastic tensile modulus undergoes a decrease of 57% between 20°C and 80°C against 40% for the longitudinal modulus. The strong impact of temperature on longitudinal modulus is partly explained by the moderate fibre volume ratio (∼30vol%) of the biocomposite. For samples conditioned in fixed relative humidity, the increase of the temperature will also cause a reduction of moisture content and a reduction of the plasticising effect of water. An increase in temperature and a decrease in moisture content cause an inverse effect on the longitudinal stiffness of the composite (Figure 6(a)). The three values of longitudinal elastic moduli obtained, respectively, at 40°C, 60°C and 80°C with the highest water content were obtained with the same relative humidity conditioning of 90%RH. The effect of temperature on moisture content is so important that the values of longitudinal elastic modulus are almost equal (between 5.6 and 5.8 GPa) for the three temperatures. For each temperature, a linear fit of the longitudinal elastic modulus as a function of the moisture content has been proposed. The linear fits seem to converge with increasing moisture content. Thus, the effect of temperature on longitudinal elastic modulus is more important at a low relative humidity rate.

3.6 Towards a design of meteosensitive 4D THBC-printed structure

The influence of the complex coupling between relative humidity and temperature is finally discussed in term of the actuation motor for THBC actuators, 4D-printed as [0/90°] asymmetric bilayers. The curvature variation as a function of time of the THBC, stored at 90%RH at different temperatures are represented in Figure 7(a). The responsiveness (i.e. maximum curvature variation) and reactivity (i.e. curvature speed) at each temperature are presented in Table 3. One can see that the responsiveness is strongly affected by the increase in temperature (−25% and −48% at 40°C and 60°C, respectively, compared to 20°C). However, the reactivity is boosted by the temperature increase (1.8 and 4.4 times higher at 40°C and 60°C, compared to 20°C). The curvature variation of the THBC as a function of moisture content is represented in Figure 7(b). The relation between curvature variation and moisture content can be estimated as linear. Thus, the increase in reactivity can be directly related to the increase in diffusion kinetics for higher temperatures. The decrease in responsiveness while temperature increases is mainly due to the decrease in moisture content at saturation. However, a slight decrease in the slope of the linear fit is also observed with increasing temperature. Timoshenko’s equation, which estimates the curvature variation $\mathrm{\Delta }k$ of a bilayer subjected to a moisture content variation $\mathrm{\Delta }\mathrm{MC}$, is helpful for understanding the decrease of the slope with increasing temperature: (6) ${\displaystyle \frac{\mathrm{\Delta }k}{\mathrm{\Delta }\mathrm{MC}}={\displaystyle \frac{\mathrm{\Delta }\beta }{h}.f(m,n)}}$(6) (7) $\mathrm{with}f(m,n)={\displaystyle \frac{6{(1+m)}^2}{3{(1+m)}^2+(1+\mathrm{mn})\left(m^2+{\displaystyle \frac{1}{\mathrm{mn}}}\right)}}$(7) where $h$ is the bilayer thickness and $\mathrm{\Delta }\beta ={\beta }_a-{\beta }_p$ is the differential CHE between the active (90°) and passive (0°) layers. The active CHE ${\beta }_a$ corresponds here to the transverse CHE ${\beta }_y$ of the biocomposite, and the passive CHE ${\beta }_p$, corresponding to the longitudinal CHE of the biocomposite, is considered negligible compared to ${\beta }_a$. The structural parameters $h$, $m=h_p/h_a$ and $n=E_p/E_a$ correspond, respectively, to the bilayer total thickness, the thickness ratio and the stiffness ratio between the passive and active layers. The thickness ratio is equal to $m=1$ given that the active and passive layers have the same thickness. Although the bilayer swells in the out-of-plane direction with moisture absorption, the total thickness $h$ can be estimated as constant since the out-of-plane expansion is low (${\epsilon }_z<2.67\text{\%}$). The results presented earlier have shown that temperature has a very low effect on the transverse CHE ${\beta }_y$, and so on $\mathrm{\Delta }\beta$. However, temperature affects the stiffness of the material. Thus, the function $f(m,n)$, which is dependent on the stiffness ratio $n$ between the passive and active layer, is thermally dependent. The bilayer curvature prediction computed with Timoshenko’s equation is presented in Figure 7(b). The computation of the parameter $n$ has been made using the linear fits obtained by the mechanical characterisation presented above (Figure 6).

Figure 7. (a) Curvature variation over time of bilayers conditioned at 90%RH after being dried. (b) Curvature variation of the bilayers as function of moisture content. The linear fits of experimental data are represented in dashed line and the curvature prediction computed with Timoshenko equations are represented in solid line.

Table 3. Actuation properties of 4D-printed THBC at different temperature.

	20°C	40°C	60°C
$\mathrm{\Delta }{\mathit{k}}_{\mathit{m}\mathrm{ax}}({10}^{-2}\mathit{m}{\mathit{m}}^{-1})$	2.00 ± 0.10	1.51 ± 0.05	1.04 ± 0.11
$\mathit{d}k/\mathit{dt}({10}^{-6}\mathit{m}{\mathit{m}}^{-1}.{\mathit{s}}^{-1})$	1.53 ± 0.12	2.76 ± 0.06	6.78 ± 0.75

One can notice that Timoshenko’s equation does not accurately predict the variation in bilayer curvature. This may be due to the different assumptions made when calculating the curvature variation. Firstly, the hygroscopic expansion of the active layer is assumed to be linear as a function of water content (i.e. ${\beta }_y$ is a constant). Other studies have proposed a sigmoid function to interpolate the hygroscopic strain of flax fibre-reinforced biocomposite with moisture content variation [8,14]. Thus, the curvature variation computation could be more accurate if the evolution of the CHE with moisture content were taken into account. Secondly, the elastic properties used for the computation of the parameter n are those of an equivalent homogeneous ply obtained from the mechanical characterisation of the unidirectional biocomposite. In the case of the bilayer, the passive layer, which is printed directly on the printing bed, has a different microstructure from the active layer. Thus, the calculation of the parameter n cannot be exact. However, the predictions obtained using Timoshenko’s equations highlight the impact of temperature on the slope of the function $\mathrm{\Delta }k=f(\mathrm{\Delta MC})$.

In practical outdoor applications, THBCs are subjected to cyclic variations in temperature and relative humidity. These environmental conditions depend on numerous parameters, such as the type of climate, the region and the season considered. However, a coupling between temperature and relative humidity variations is systematically observed. At the start of the day, relative humidity is at its highest, while temperature is minimal. Then, under the effect of the sun’ radiations, temperature rises while relative humidity falls. These variations are reversed in the second half of the day as the sun moves lower in the sky. These daily variations are highly variable and are particularly pronounced on sunny days. THBC are subjected to constantly fluctuating environmental conditions, and thus, are operated in a non-equilibrium state which results in a complex behaviour.

In order to observe the response of a THBC subjected to a real environmental loading case, a 4D-printed bilayer assembly structure is designed. The printing path of this structure is shown in Figure 8(a). The actuation potential of this THBC over the entire relative humidity range (from 0 to 90%RH) at room temperature is shown in Figure 8(b). However, the relative humidity variations of a real environmental loading only cover a part of the relative humidity range. Thus, morphing response of the same THBC, submitted to typical relative humidity variations on a sunny summer’s day in Lorient, France (temperate climate), is represented in Figure 8(c). In this case, the daily temperature variations are not considered. Relative humidity varies by only about 40% of the entire humidity range (from 50 to 90%RH). Thus, the total actuation of the THBC is inevitably limited by the environmental load case. Figure 8(d) presents a second environmental loading case, in which the daily temperature variation is considered. The temperature variation considers the change in outside temperature, to which is added the heating of the material induced by the photothermal effect. Increasing temperature during the day increases actuation kinetics. It also helps to reduce the water content of the material, thereby increasing the response of the THBC. The responsiveness of the THBC is increased by 13% in the case where temperature variation is considered compared to the case where it remains constant. This result confirms that relative humidity is the main stimuli that drive the actuation of THBC. However, the impact of temperature cannot be neglected, as it has a significant influence on the kinetics of actuation and the water content at saturation.

Figure 8. (a) Printing path of THBC bilayer assembly. (b) Representation of the actuation potential of the THBC over the full range of relative humidity at 20°C. (c) Representation of the actuation of a THBC subjected to daily relative humidity variations at a constant temperature (not representative of a real environmental loading scenario). (d) Actuation of a THBC subjected to the same daily relative humidity variations, but with temperature variation considered. The temperature considered includes the outside temperature as well as heating induced by solar irradiance.

4 Conclusion

In recent years, numerous studies have focused on 4D printing to develop new intelligent materials or adaptative structures. Hygromorph biocomposites are of particular interest for the development of passive/autonomous actuators exhibiting shape-morphing responses to surrounding relative humidity variations of the environment. However, relative humidity variations are naturally coupled with temperature variations. Thus, this article aimed to experimentally investigate the influence of hygrothermal coupling on sorption behaviour, dimensional variations and mechanical properties of 3D-printed cFF/PBAT. It has been shown that temperature increases strongly  fasten the sorption kinetics while it reduces the moisture content at saturation. The dimensional variations of a unidirectional biocomposite are mainly governed by the material hygroscopic state. The temperature has a low influence on the coefficient of hygroscopic expansion. However, the decrease of moisture content while increasing temperature leads to a diminution of hygroscopic expansion. The stiffness of the material is strongly affected by both temperature and moisture content variations. The impact of temperature on 4D-printed THBC bilayers has been investigated. The curvature amplitude of a bilayer is partly governed by the moisture content of the material and the stiffness ratio between the passive and active layers of the bilayer, which are both thermally dependant. Thus, an increase in temperature strongly reduces the curvature amplitude (−84% at 60°C compared to 20°C). At the same time, the actuation kinetics is strongly fastened with a temperature rise. Finally, we designed a meteosensitive shape-changing THBC, able to fold itself as an origami to highlight the design possibilities offered by 4D printing.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that supports the present study are available from the corresponding author upon reasonable request.

Additional information

Funding

The authors wish to acknowledge AID/DSTL (FR/UK) bilateral Grant No. 65 0040.