Comparative Mechanical Study of Pressure Sensitive Adhesives over Aluminium Substrates for Industrial Applications

The use of adhesives for fixing low-weight elements is showing increasing interest in the industry, as it would reduce the weight of the assembly, costs, and production time. Specifically, the application of pressure-sensitive adhesives (PSAs) to join non-structural naval components to aluminium substrates has not yet been reported. In the present work, a study of the mechanical behaviour of different double-sided PSAs applied on bare aluminium alloy substrates is performed. The influence of surface roughness, surface chemical treatments, and the matrix of the adhesives is studied through different mechanical tests, such as shear, T-peel, and creep. The application of an adhesion promoter improved the mechanical behaviour. Low roughness substrates provided better performance than ground samples. Acrylic foam adhesives were subjected to creep tests, whose results were fitted to a simple mathematical model, predicting the fracture time as a function of the applied load.


Introduction
The industry currently faces different challenges, such as demands for higher performance, lower energy consumption, and the consolidation of more environmentally friendly processes. In this context, one strategy to reduce the emission of pollutants to the ecosystem is the use of lighter and more efficient materials [1][2][3]. This need justifies the rise in the use of aluminium alloys as a substitute for steel, due to their advantageous properties, among them high resistance, low density, resistance to corrosion, resistance to fatigue, and reduction in costs associated with the weight and fuel savings in transport industries [4,5]. However, the reduction in emissions is not limited to the use of lighter materials, but also seeks the use of lighter, simpler, and cheaper joining technologies. One of the most promising alternatives to welding techniques for some specific applications is adhesive bonding. This methodology is advantageous for manufacturing in an energy efficient way since it reduces the weight of the joints and allows the joining of a higher number of surfaces compared to traditional joining methods [6]. Therefore, there has been an increase in the demand for adhesives as an alternative material to traditional bonding processes [1,[7][8][9]. Indeed, the adhesive bonding of aluminium is already widely used in different industries, such as automotive and aeronautics, because of their high performance, low cost, easy application, good damping behaviour, and resistance to corrosion [10][11][12]. There are important potential applications of adhesive joining in other industries, such as naval and offshore, in which the use of cold joining processes and light materials are highly demanded.
At present, the adoption of pressure sensitive adhesives (PSAs) is being frequently used because of their advantages, such as easy and safe application [13]. There are diverse types of PSAs that differ in the adhesive base, support material, as well as in their applications. These adhesives consist of flexible support with a permanent tack adhesive layer that electrical cable supports. Most of the research carried out on aluminium substrates is limited to the use of structural adhesives, especially those of the epoxy and polyurethane type [2,26]. Furthermore, PSA adhesive bonding on aluminium substrates has not been extensively tested. The studies carried out are mainly focused on the optimization of the joints by means of different surface preparations applying different cleaning agents and chemical treatments before assembling. Likewise, in these studies, the response of the PSA aluminium joints was evaluated by modifying some parameters that intervene in the curing process, such as temperature. The efficiency of these treatments is interpreted through the response of the joints when they are subjected to shear stress [23,24]. The creep behaviour at different temperatures of one high-performance acrylic foam PSA applied to aluminium substrates for glazing applications is also investigated in detail in [29,30].
Considering the overall literature reviewed, it can be stated that there is a clear need for knowledge regarding the potential use of PSAs in industry. Concisely, their application to join non-structural components to aluminium substrates is not widely addressed, especially in shipbuilding applications. In order to cover this gap of knowledge, the present work investigates the response of PSA adhesives of different natures when they are applied on aluminium substrates with different roughness and surface chemical treatments. The mechanical properties of the joints are experimentally analysed by means of shear, T-peel, and creep tests.

Materials and Methods
AA5083-H111 and AA5754-H111 aluminium alloys, whose compositions are detailed in Table 1, were employed for this investigation. AA5083-H111 aluminium specimens (10 mm × 100 mm × 2 mm) were used to perform shear and creep tests. On the other hand, AA5754-H111 aluminium specimens (25 mm × 200 mm × 0.5 mm) were used for the T-peel test. The specimens of each test were defined according to ISO 13445, ISO 14678, and ISO 11339 standards, respectively. Silicon carbide sandpaper with different grain sizes (P40, P80, P320, P600, and P1200) was used to prepare the surface with different roughness, in which the sandpaper code corresponds to the number of abrasive particles per square inch. The arithmetic mean roughness (R a ) was measured at least 5 times using an adapted roughness tester with a surface probe (Perthometer PGK 120, Mahr, Göttingen, Germany). The bonding area of all samples was cleaned with isopropanol before the assembling process. Furthermore, the influence of the use of a thin layer of an adhesion promoter based on a ciclohexane/xylene solution (Fast Cure Promoter, FCP 60153, TESA, Norderstedt, Germany) was also evaluated after the cleaning procedure. Single-lap joints (SLJ) were assembled with a rectangular area of 10 mm × 20 mm ( Figure 1) using three different types of PSAs: a rigid acrylic foam (RF), a flexible acrylic foam (FF), and a flexible acrylic core (FC). Table 2 shows the thickness and the commercial name of each PSA. It also includes the density values, measured experimentally. FC adhesive is known to contain trimethylolpropane triacrylate (TMPTA) with a high degree of crosslinking. Meanwhile, RF and FF present a high content of methyl methacrylate (MMA).   The joints were subjected to a compression force of 30 N/cm 2 for 60 s, using a Shimadzu universal test equipment (10 kN of maximum load). Then, the joints were cured at room temperature (21 °C) and a controlled relative humidity of 60% for 72 h. After curing, the shear test was carried out at a constant speed of 10 mm/min, in accordance with ISO 13445. Figure 2A shows the experimental setup of the shear tests performed with the Shimadzu universal test equipment. The assembly of the joints was divided into two batches, with and without the application of adhesion promoter to the aluminium specimens prior to assembly. These shear tests allowed the study of the influence of the different surface roughness and chemical treatments on the mechanical performance of the three PSAs. In addition, the T-peel tests ( Figure 2B) were performed in accordance with ISO 11339, employing the best surface conditions measured in the shear tests. All tests were carried out at least in triplicate to assure reproducibility. The failure mode was measured after testing according to ISO 10365, using image analysis software (Image-J, version 1.52q, National Institutes of Health, Bethesda, MD, USA). Ductility (elongation at fracture), and static toughness (overall area below the stress-strain curve, estimated by the 6th degree full polynomial that better fitted to each curve) were extracted from the stress-strain curves, obtained for each adhesive.   The joints were subjected to a compression force of 30 N/cm 2 for 60 s, using a Shimadzu universal test equipment (10 kN of maximum load). Then, the joints were cured at room temperature (21 • C) and a controlled relative humidity of 60% for 72 h. After curing, the shear test was carried out at a constant speed of 10 mm/min, in accordance with ISO 13445. Figure 2A shows the experimental setup of the shear tests performed with the Shimadzu universal test equipment. The assembly of the joints was divided into two batches, with and without the application of adhesion promoter to the aluminium specimens prior to assembly. These shear tests allowed the study of the influence of the different surface roughness and chemical treatments on the mechanical performance of the three PSAs. In addition, the T-peel tests ( Figure 2B) were performed in accordance with ISO 11339, employing the best surface conditions measured in the shear tests. All tests were carried out at least in triplicate to assure reproducibility. The failure mode was measured after testing according to ISO 10365, using image analysis software (Image-J, version 1.52q, National Institutes of Health, Bethesda, MD, USA). Ductility (elongation at fracture), and static toughness (overall area below the stress-strain curve, estimated by the 6th degree full polynomial that better fitted to each curve) were extracted from the stress-strain curves, obtained for each adhesive.   The joints were subjected to a compression force of 30 N/cm 2 for 60 s, using a Shimadzu universal test equipment (10 kN of maximum load). Then, the joints were cured at room temperature (21 °C) and a controlled relative humidity of 60% for 72 h. After curing, the shear test was carried out at a constant speed of 10 mm/min, in accordance with ISO 13445. Figure 2A shows the experimental setup of the shear tests performed with the Shimadzu universal test equipment. The assembly of the joints was divided into two batches, with and without the application of adhesion promoter to the aluminium specimens prior to assembly. These shear tests allowed the study of the influence of the different surface roughness and chemical treatments on the mechanical performance of the three PSAs. In addition, the T-peel tests ( Figure 2B) were performed in accordance with ISO 11339, employing the best surface conditions measured in the shear tests. All tests were carried out at least in triplicate to assure reproducibility. The failure mode was measured after testing according to ISO 10365, using image analysis software (Image-J, version 1.52q, National Institutes of Health, Bethesda, MD, USA). Ductility (elongation at fracture), and static toughness (overall area below the stress-strain curve, estimated by the 6th degree full polynomial that better fitted to each curve) were extracted from the stress-strain curves, obtained for each adhesive.  The adhesive with the best properties in terms of shear and T-peel strength was selected to evaluate its creep resistance on specimens of the same dimensions as those used in the shear tests. The creep tests were performed at 23 • C and 70% of relative humidity, following the indications described in the ASTM D-1780 standard, allowing the measurement of the fracture time as a function of the applied load. The maximum load used in the creep testing was selected according to the peak strength obtained from the shear results (20 kg/cm 2 ), with the other load values being a percentage of it. The applied load values are listed in Table 3 and Figure 3 illustrates the creep test setup. The adhesive with the best properties in terms of shear and T-peel strength was selected to evaluate its creep resistance on specimens of the same dimensions as those used in the shear tests. The creep tests were performed at 23 °C and 70% of relative humidity, following the indications described in the ASTM D-1780 standard, allowing the measurement of the fracture time as a function of the applied load. The maximum load used in the creep testing was selected according to the peak strength obtained from the shear results (20 kg/cm 2 ), with the other load values being a percentage of it. The applied load values are listed in Table 3 and Figure 3 illustrates the creep test setup.

Roughness
The surface roughness of the different sanded/ground aluminium surfaces was measured at least five times per condition, in different positions of the bonding area, to assure that measurements were representative of each ground level. The roughness values reported in the paper correspond to the mean arithmetic roughness parameter (Ra). Table 4 shows the average and standard deviation of the arithmetic mean roughness values (Ra). As expected, the surface treatments with larger sandpaper grain sizes (P40 and P80) implied higher surface roughness. Thus, Ra clearly decreases as the number of particles of the sandpapers increases.

Roughness
The surface roughness of the different sanded/ground aluminium surfaces was measured at least five times per condition, in different positions of the bonding area, to assure that measurements were representative of each ground level. The roughness values reported in the paper correspond to the mean arithmetic roughness parameter (R a ). Table 4 shows the average and standard deviation of the arithmetic mean roughness values (R a ). As expected, the surface treatments with larger sandpaper grain sizes (P40 and P80) implied higher surface roughness. Thus, R a clearly decreases as the number of particles of the sandpapers increases.

Shear Testing
The typical behaviour of the three PSAs studied when subjected to shear stress is shown in Figure 4. The stress-strain curves represented in this figure correspond to joints using aluminium specimens ground with a sandpaper size of P1200 with the adhesion promoter. In these curves, a different shape is observed for each adhesive matrix (acrylic foam/acrylic core). Regarding the acrylic foam adhesives (RF and FF), the RF showed higher elastic modulus (stiffness) than the FF, whereas the FF presented a greater elastic deformation and strength. In these curves, the slope is slightly decreasing up to the maximum stress value, meaning that the stiffness slowly diminishes as these adhesives elongate. In the case of the acrylic core (FC), the curve exhibited a non-linear behaviour at the elastic zone, its slope gradually increasing up to the maximum strength. The different nature of the cores seems to be the reason for this stiffness change tendency. Concerning the strength, measured by the maximum shear stress values, the FF was the adhesive providing the highest values, whereas the FC was the one presenting the lowest resistance. The elongation of the adhesives prior to bond failure was also analysed. A higher elongation is directly related to a higher elasticity of the adhesive and, therefore, a better capacity to recover its original shape after the load release. According to the maximum ultimate shear strength (USS) and elongation, the FF adhesive provided the highest strength and ductility.

Shear Testing
The typical behaviour of the three PSAs studied when subjected to shear stress is shown in Figure 4. The stress-strain curves represented in this figure correspond to joints using aluminium specimens ground with a sandpaper size of P1200 with the adhesion promoter. In these curves, a different shape is observed for each adhesive matrix (acrylic foam/acrylic core). Regarding the acrylic foam adhesives (RF and FF), the RF showed higher elastic modulus (stiffness) than the FF, whereas the FF presented a greater elastic deformation and strength. In these curves, the slope is slightly decreasing up to the maximum stress value, meaning that the stiffness slowly diminishes as these adhesives elongate. In the case of the acrylic core (FC), the curve exhibited a non-linear behaviour at the elastic zone, its slope gradually increasing up to the maximum strength. The different nature of the cores seems to be the reason for this stiffness change tendency. Concerning the strength, measured by the maximum shear stress values, the FF was the adhesive providing the highest values, whereas the FC was the one presenting the lowest resistance. The elongation of the adhesives prior to bond failure was also analysed. A higher elongation is directly related to a higher elasticity of the adhesive and, therefore, a better capacity to recover its original shape after the load release. According to the maximum ultimate shear strength (USS) and elongation, the FF adhesive provided the highest strength and ductility.

Shear Results without Adhesion Promoter
The influence of the grinding surface process on the PSA's shear resistance was analysed as follows. Figure 5 reports the peak values (USS) obtained from the samples without the adhesion promoter. In the three PSAs, no clear relationship between USS and the

Shear Results without Adhesion Promoter
The influence of the grinding surface process on the PSA's shear resistance was analysed as follows. Figure 5 reports the peak values (USS) obtained from the samples without the adhesion promoter. In the three PSAs, no clear relationship between USS and the grinding level was found in the ground samples. However, the PSAs displayed a different behaviour when the aluminium substrates were not ground (NG). Thus, the FC NG samples provided an USS value of 0.79 MPa. The values for the ground FC samples ranged between 0.84 and 1.12 MPa, with the peak (1.12 MPa) in the P40 condition. Regardless the surface treatment, FC always provided lower USS than RF and FF. Meanwhile, the joint shear resistance of both the RF and the FF notably increased when the NG condition was applied to the aluminium substrates. Concisely, the maximum values of the RF adhesive subjected to grinding processes ranged between 1.11 and 1.28 MPa, with 1.56 MPa for the NG condition. Similarly, the FF adhesive values ranged between 1.18 and 1.28 MPa for the ground samples, presenting the highest USS (2.13 MPa) when the substrates were not ground (NG). The FF was the PSA that provided the highest USS value among the three adhesives studied.
shear resistance of both the RF and the FF notably increased when the NG condition was applied to the aluminium substrates. Concisely, the maximum values of the RF adhesive subjected to grinding processes ranged between 1.11 and 1.28 MPa, with 1.56 MPa for the NG condition. Similarly, the FF adhesive values ranged between 1.18 and 1.28 MPa for the ground samples, presenting the highest USS (2.13 MPa) when the substrates were not ground (NG). The FF was the PSA that provided the highest USS value among the three adhesives studied. The fracture mode was studied in the samples, characterising the adhesive/cohesive rupture in terms of the percentage of cohesive failure. Figure 6 plots this percentage for joints subjected to different surface treatments without adhesion promoter. The percentage of cohesive failure of the joints assembled with the FF adhesive improved as the roughness reduced. This behaviour may be related to the decrease in the amount of air bubbles that were trapped in the interface, and consequently to the increase in the active surface of the joint. The strength of the joints assembled with the FC adhesive did not follow any clear trend as the surface roughness varied. The FC displayed the highest cohesive percentages in ground samples, although it showed the lowest shear resistance ( Figure 6). This may be because the acrylic core adhesive (FC) exhibits higher formability than acrylic foam adhesives (RF and FF). This difference may lie in the fact that the FC adhesive has greater malleability than acrylic foam adhesives (RF and FF), presenting a better grip on rough substrates than on smooth substrates [44]. In addition, the cohesion of the FC joints was slightly favoured when the substrates had a roughness lower than 1 μm (obtained with sandpapers of P320, P600, P1200, and NG). The behaviour of this adhesive was similar to that described by Cui et al. [4], in which the adhesive based on epoxy and polyurethane did not present a linear trend between grain size sanding and USS. The fracture mode was studied in the samples, characterising the adhesive/cohesive rupture in terms of the percentage of cohesive failure. Figure 6 plots this percentage for joints subjected to different surface treatments without adhesion promoter. The percentage of cohesive failure of the joints assembled with the FF adhesive improved as the roughness reduced. This behaviour may be related to the decrease in the amount of air bubbles that were trapped in the interface, and consequently to the increase in the active surface of the joint. The strength of the joints assembled with the FC adhesive did not follow any clear trend as the surface roughness varied. The FC displayed the highest cohesive percentages in ground samples, although it showed the lowest shear resistance ( Figure 6). This may be because the acrylic core adhesive (FC) exhibits higher formability than acrylic foam adhesives (RF and FF). This difference may lie in the fact that the FC adhesive has greater malleability than acrylic foam adhesives (RF and FF), presenting a better grip on rough substrates than on smooth substrates [44]. In addition, the cohesion of the FC joints was slightly favoured when the substrates had a roughness lower than 1 µm (obtained with sandpapers of P320, P600, P1200, and NG). The behaviour of this adhesive was similar to that described by Cui et al. [4], in which the adhesive based on epoxy and polyurethane did not present a linear trend between grain size sanding and USS. when changing the grinding process. For both parameters, the FF adhesive showed higher values than the others, regardless of the grinding condition. The NG condition visibly led to the best ductility and toughness results for the three PSAs, especially in the case of the FF adhesive. Thus, the NG samples provided ductility values of 22.0%, 23.8%, and 46.6%, for the FC, RF, and FF, respectively. Meanwhile, the toughness of NG samples was 1.6 MJ/m 3 , 4.0 MJ/m 3 , and 13.3 MJ/m 3 for the FC, RF, and FF, respectively.  Figures 7 and 8 show the ductility and toughness results of joints assembled without the adhesion promoter, respectively. Both properties did not seem to follow a clear trend when changing the grinding process. For both parameters, the FF adhesive showed higher values than the others, regardless of the grinding condition. The NG condition visibly led to the best ductility and toughness results for the three PSAs, especially in the case of the FF adhesive. Thus, the NG samples provided ductility values of 22.0%, 23.8%, and 46.6%, for the FC, RF, and FF, respectively. Meanwhile, the toughness of NG samples was 1.6 MJ/m 3 , 4.0 MJ/m 3 , and 13.3 MJ/m 3 for the FC, RF, and FF, respectively.  Taking into account the overall results reported in this section, it is clear that FC adhesive present similar mechanical behaviour regardless the surface roughness. This is related its composition and production method, presenting a higher gelation than RF and FF adhesives. This property provides to FC a high flexibility and adaptability to the different roughness conditions tested, reporting similar USS, ductility, and toughness results. FC adhesive is known to contain trimethylolpropane triacrylate (TMPTA) with a high degree of crosslinking, providing the gelled texture [45].
Note that RF and FF adhesives do not have gel structures, having therefore less flexibility and adaptability to rough surfaces. Thus, RF and FF present a lower aluminiumadhesive contact area as the roughness is higher. This means that both adhesives show their best response when assembled on NG substrates. This behaviour is related to a high content of methyl methacrylate (MMA) in their compositions, providing a greater adhesion to substrates [45]. As the MMA polymer presents a lower crosslinking degree than TMPTA, the structure of the RF and FF adhesives does not present gel texture, reducing their adaptability to rough substrates, decreasing therefore the effective area of the union [45]. This explains why RF and FF (adhesives without gel texture) show their best mechanical behaviour in NG condition.
In addition, according to the results obtained by other authors [46,47], an increase in the molecular weight of acrylic PSAs leads to a decrease in resistance. The density values reported in Table 2 show that FC presents a higher density than RF and FF, which is re- Taking into account the overall results reported in this section, it is clear that FC adhesive present similar mechanical behaviour regardless the surface roughness. This is related its composition and production method, presenting a higher gelation than RF and FF adhesives. This property provides to FC a high flexibility and adaptability to the different roughness conditions tested, reporting similar USS, ductility, and toughness results. FC adhesive is known to contain trimethylolpropane triacrylate (TMPTA) with a high degree of crosslinking, providing the gelled texture [45].
Note that RF and FF adhesives do not have gel structures, having therefore less flexibility and adaptability to rough surfaces. Thus, RF and FF present a lower aluminiumadhesive contact area as the roughness is higher. This means that both adhesives show their best response when assembled on NG substrates. This behaviour is related to a high content of methyl methacrylate (MMA) in their compositions, providing a greater adhesion to substrates [45]. As the MMA polymer presents a lower crosslinking degree than TMPTA, the structure of the RF and FF adhesives does not present gel texture, reducing their adaptability to rough substrates, decreasing therefore the effective area of the union [45]. This explains why RF and FF (adhesives without gel texture) show their best mechanical behaviour in NG condition.
In addition, according to the results obtained by other authors [46,47], an increase in the molecular weight of acrylic PSAs leads to a decrease in resistance. The density values reported in Table 2 show that FC presents a higher density than RF and FF, which is related to the higher molecular weight of TMPTA (C 15 H 20 O 6 ) when compared to MMA (C 5 H 8 O 2 ). The results obtained in the present study are in good agreement with previous investigations [46,47], as the highest molecular weight PSA (FC) leads to the lowest USS values.
The deformability of crosslinked adhesives is also related to the molecular weight; an increase in molecular weight leads to an increase in PSA deformation. The recovery capacity increases with the cross-linking of the polymeric chains [48]. The crosslinking of the polymeric chains increases the flexibility and the elastic limit of the PSAs. As commented before, TMPTA has a more crosslinking degree and higher molecular weight than MMA. This is the reason why FC shows higher flexibility (lower slope at elastic region, observable in Figure 4) than RF and FF. The obtained results are therefore in good agreement with those reported by [48].

Shear Results with Adhesion Promoter
The shear performance developed by the adhesives when applying the adhesion promoter is presented in Figure 9. As observed, the application of this chemical treatment developed a higher mechanical performance of the three PSAs in most of the conditions. This behaviour was more pronounced when assembling aluminium substrates without sanding.

Shear Results with Adhesion Promoter
The shear performance developed by the adhesives when applying the adhesion promoter is presented in Figure 9. As observed, the application of this chemical treatment developed a higher mechanical performance of the three PSAs in most of the conditions. This behaviour was more pronounced when assembling aluminium substrates without sanding. Generally, the joints manufactured with the FF adhesive displayed higher strength values compared with the other adhesives, independently of the condition used. For this adhesive, a significant increase in ultimate shear strength (USS) was observed after the application of the adhesion promoter, especially for ground substrates. The NG condition for this adhesive presented a similar behaviour with and without promoter. In the case of the RF adhesive, the use of the adhesion promoter also increased the USS. The highest result was observed for the NG condition. Regarding the FC adhesive, an important increase in USS was obtained with the promoter in all surface conditions, where the peak value was reached in the NG condition. Figure 10 compares the effect of using adhesion promoter in the USS values for the three PSAs studied. The figure clearly shows that the USS values of joints with promoter (P) are significantly higher than without promoter (NP) for the three PSAs. Generally, the joints manufactured with the FF adhesive displayed higher strength values compared with the other adhesives, independently of the condition used. For this adhesive, a significant increase in ultimate shear strength (USS) was observed after the application of the adhesion promoter, especially for ground substrates. The NG condition for this adhesive presented a similar behaviour with and without promoter. In the case of the RF adhesive, the use of the adhesion promoter also increased the USS. The highest result was observed for the NG condition. Regarding the FC adhesive, an important increase in USS was obtained with the promoter in all surface conditions, where the peak value was reached in the NG condition. Figure 10 compares the effect of using adhesion promoter in the USS values for the three PSAs studied. The figure clearly shows that the USS values of joints with promoter (P) are significantly higher than without promoter (NP) for the three PSAs.  Table 5 shows the improvement percentage (IP) data of shear resistance after the application of the adhesion promoter (SR P ) with respect to the values obtained without the promoter (SR NP ). These values were calculated for each adhesive and each grinding condition according to Equation (1): The results clearly demonstrate that the adhesion promoter enhanced the shear resistance of all adhesives. Note that the statistical error of these values is around 5%; therefore, the values below this percentage in this table are considered negligible. Figure 11 reports the cohesive failure percentage values obtained from the analysis of the fractured surfaces of adhesives with the adhesion promoter. It can be seen that the application of the promoter in sanded substrates supported a notable enhancement in the cohesion of the joints manufactured with the FF and the FC adhesives. This improvement was especially notable in the FC, where the use of the adhesion promoter developed a practically cohesive fracture (average of 99.55%) and only the P40 condition generated a percentage below 100%. These values indicate that the adhesive was properly adhered to the entire substrate area, effectively resisting to the shear forces imposed by the test. However, the RF adhesive hardly showed significant improvements in the cohesive failure percentage after the application of the promoter. The cohesion fracture of this adhesive displayed low percentages, similar to those reached without promoter, indicating that the adhesion between this PSA and the substrates was not appropriate, leading to poor mechanical resistance and adhesion fracture. In short, the FF and the FC developed cohesive fracture percentages close to 100%, with this value being always below 5% for the RF. The improvements observed for the FF and the FC adhesives were related to the chemical activation of the surface and the moldability capacity. Despite the application of the promoter, the RF adhesive did not seem to develop surface chemical activation. Its low moldability, related to its relatively high stiffness (Figure 4), may be the reason for this poor cohesion.  Figure 12 shows the ductility results of joints assembled with the adhesion promoter. Similar values were obtained for all grinding conditions. The promoter led to a clear ductility enhancement, especially for the FF adhesive. For the RF adhesive, the results obtained for the joints without the adhesion promoter reached an average of 24.8%, which increased to 28.2% when the adhesion promoter was applied. For joints assembled with the FF adhesive, an increase in ductility values from 35.4% to 52.8% was observed when the adhesion promoter was applied over the sanded substrates. For the FF joints assembled with NG condition, the ductility was similar with and without adhesion promoter (around 47%). In the case of the FC joints, the application of adhesion promoter did not improve significantly the ductility values (22-24%).  Figure 12 shows the ductility results of joints assembled with the adhesion promoter. Similar values were obtained for all grinding conditions. The promoter led to a clear ductility enhancement, especially for the FF adhesive. For the RF adhesive, the results obtained for the joints without the adhesion promoter reached an average of 24.8%, which increased to 28.2% when the adhesion promoter was applied. For joints assembled with the FF adhesive, an increase in ductility values from 35.4% to 52.8% was observed when the adhesion promoter was applied over the sanded substrates. For the FF joints assembled with NG condition, the ductility was similar with and without adhesion promoter (around 47%). In the case of the FC joints, the application of adhesion promoter did not improve significantly the ductility values (22-24%).  Figure 12 shows the ductility results of joints assembled with the adhesion promoter. Similar values were obtained for all grinding conditions. The promoter led to a clear ductility enhancement, especially for the FF adhesive. For the RF adhesive, the results obtained for the joints without the adhesion promoter reached an average of 24.8%, which increased to 28.2% when the adhesion promoter was applied. For joints assembled with the FF adhesive, an increase in ductility values from 35.4% to 52.8% was observed when the adhesion promoter was applied over the sanded substrates. For the FF joints assembled with NG condition, the ductility was similar with and without adhesion promoter (around 47%). In the case of the FC joints, the application of adhesion promoter did not improve significantly the ductility values (22-24%). The average toughness values for the three adhesive joints assembled using the adhesion promoter are shown in Figure 13. As observed, the values were not highly influenced by the surface preparation. Comparing the results of Figures 9 and 13, it is clear that The average toughness values for the three adhesive joints assembled using the adhesion promoter are shown in Figure 13. As observed, the values were not highly influenced by the surface preparation. Comparing the results of Figures 9 and 13, it is clear that the use of the adhesion promoter improved the toughness of the three adhesives. On average, it increased from 3.8 to 4.4 MJ/m 3 in the RF, from 1.6 MJ/m 3 to 2.1 MJ/m 3 in the FC joints, and from 6.0 MJ/m 3 to 12.1 MJ/m 3 in the ground FF joints. The most evident toughness increment was observed in the FF adhesive. For this PSA, the highest toughness was reported in the NG condition, regardless of the application of the adhesion promoter (NG condition without promoter provided relatively high toughness values).  The reported results demonstrate that the shear behaviour of the three PSAs was notably increased when the adhesion promoter was employed. The surface treatment did not lead to a high variation of mechanical behaviour when the adhesion promoter was applied, with the NG condition being generally better than ground surfaces. These outcomes are interesting for industrial applications, as the adhesion promoter involves an improvement in mechanical behaviour regardless of the surface roughness. This improvement was due to the increase in the Van der Waals bonds and electrostatic forces of the The reported results demonstrate that the shear behaviour of the three PSAs was notably increased when the adhesion promoter was employed. The surface treatment did not lead to a high variation of mechanical behaviour when the adhesion promoter was applied, with the NG condition being generally better than ground surfaces. These outcomes are interesting for industrial applications, as the adhesion promoter involves an improvement in mechanical behaviour regardless of the surface roughness. This improvement was due to the increase in the Van der Waals bonds and electrostatic forces of the joints, as the use of the adhesion promoter notably improves the mechanical properties (USS, ductility and toughness) and delays the failure [49]. Additionally, it is also known that the adhesion promoter increases the surface energy of substrates [39].
Generally, acrylic adhesives have a weight composition of 3-10% pure acrylic acid and 90-97% alkyl acrylate [50,51]. The polymer crosslinking in the manufacture of PSA is determined by the chemical reaction between the crosslinking agent and the adhesive. Crosslinking process increases the yield strength and the resilience capacity of the adhesive, decreasing the deformation capacity. Likewise, an excessive degree of crosslinking leads to polymer gelation, which deteriorates the adhesive properties such as stickiness and ductility [52,53]. The acrylic core adhesive (FC) used in the present study presents a more gelled texture than foam acrylic adhesives (FF and RF). This may be due to the higher crosslinking processes of FC, which leads to lower deformability compared to the other two acrylic foam adhesives ( Figure 4). The different structures of these acrylic backing adhesives seem to be the reason for the different adhesion patterns as the surface roughness is modified (Figures 5 and 9). Thus, while the adhesion of FC is not affected by roughness, both FF and RF show a resistance decrease as the roughness increases (Figures 7 and 12).

T-Peel Test
Taking into account that the NG condition was the easiest surface preparation and provided the best shear performance, this was the condition employed for the T-peel tests. Thus, T-peel tests were performed employing not sanded AA5754 H111 aluminium substrates of 0.5 mm thickness treated with the adhesion promoter. Representative Tpeel stress-strain curves of the three adhesives are included in Figure 14

T-Peel Test
Taking into account that the NG condition was the easiest surface preparation and provided the best shear performance, this was the condition employed for the T-peel tests. Thus, T-peel tests were performed employing not sanded AA5754 H111 aluminium substrates of 0.5 mm thickness treated with the adhesion promoter. Representative T-peel stress-strain curves of the three adhesives are included in Figure 14 Table 6 reports the fracture mode expressed as the percentage of cohesive failure of adhesive joints subjected to T-peel tests. Both the RF and the FF displayed fully adhesive fractures. This fracture type changed for the FC joints, showing high T-peel cohesion. This disparity in the failure mode may be associated with the adhesive nature. Acrylic foam matrix adhesives (RF and FF) can present non-linear viscoelastic properties, assisted by  Table 6 reports the fracture mode expressed as the percentage of cohesive failure of adhesive joints subjected to T-peel tests. Both the RF and the FF displayed fully adhesive fractures. This fracture type changed for the FC joints, showing high T-peel cohesion. This disparity in the failure mode may be associated with the adhesive nature. Acrylic foam matrix adhesives (RF and FF) can present non-linear viscoelastic properties, assisted by the presence of bubbles in the foam. When these adhesives were subjected to stress, these bubbles started to deform, stretching and collapsing, causing a variation in the viscosity of the adhesive [54,55]. The different cohesion values for the shear and the T-peel tests were related to the direction of the adhesive matrix strain (note that shear and T-peel tests were performed at the same deformation rate). For the shear test, the strain direction was quasiequatorial, while for the T-peel test the deformation occurred in the axial direction [56]. When the joints were subjected to shear, the movement occurred in a direction that did not allow the bubbles to grow, leading to a cohesive failure. In the T-peel tests, the strain was in the axial direction, permitting the bubbles' growth. This led to a damping behaviour of the adhesive matrix, resulting in an adhesive failure. In the case of the FC adhesive, both the shear and the T-peel tests provoked cohesive behaviour, related to the homogenous matrix free of bubbles.

Creep Test
The previous shear and T-peel tests reported the best mechanical behaviour for the FF adhesive, assembled over substrates that were not ground and treated with the adhesion promoter. Thus, the joints manufactured under this condition were selected to develop creep tests. After curing time, the joints were subjected to different loads. These loads were selected based on the maximum shear resistance (Section 3.2). The creep behaviour was characterised by the fracture time and the failure mode. These results are reported in Table 7. Figure 16 displays a typical creep behaviour curve obtained with a load of 20 kg/cm 2 (50% of the maximum shear resistance). An inverse correlation between the failure time and the applied load was noted, with a significant lifetime increment with low loads. The adhesive demonstrated slight creep periods when loads of 1 to 10 Kg/cm 2 were applied, indicating a maximum failure time of 115 h for the 1 Kg/cm 2 load. To increase the creep fracture time, it is necessary to reduce the applied load to values below 0.5 Kg/cm 2 . This load value represents 2.5% of the maximum shear load. For the minimum load applied (0.25 Kg/cm 2 ), the joints did not fracture after 8760 h (1 year).  Regarding the fracture analysis, the maximum load supported by the FF adhesive (20 Kg/cm 2 ) led to a total cohesive fracture. The results demonstrate that, as the load decreased, the type of fracture became more adhesive. This evolution in the adhesion properties was related to the critical stress required to initiate the cavitation process. Generally, cavity growth takes place as the joints are subjected to higher stress values, which provokes higher creep rates and shorter fracture times. This reduction in fracture time is due to a faster release of hydrostatic stress after the cavitation process [26,41,[56][57][58][59]. Therefore, higher stress around the cavities promoted a reduction in the damping behaviour, increasing the cohesive fracture.
Based on all data acquired for the FF joints subjected to the static loads, a simple modelling of the failure time was developed. Figure 17 plots the creep data, representing the load applied versus failure time in a double logarithmic scale. The linear equation better fitting by linear regression to these data is depicted in the figure. The high coefficient of determination value (R 2 ) confirmed the high data fitting. Note that the result was an exponential equation type, provided by the double logarithmic scales. Using this model, the failure time of the FF adhesive joints can be estimated as a function of the applied load. Equation (2) depicts this fitting model.

=
(2) where y is log (applied load), x is the log (fracture time), and a and b are the fitting parameters, 0.913 Kg/cm 2 and −0.405 Kg/(cm 2 ·h), respectively. Regarding the fracture analysis, the maximum load supported by the FF adhesive (20 Kg/cm 2 ) led to a total cohesive fracture. The results demonstrate that, as the load decreased, the type of fracture became more adhesive. This evolution in the adhesion properties was related to the critical stress required to initiate the cavitation process. Generally, cavity growth takes place as the joints are subjected to higher stress values, which provokes higher creep rates and shorter fracture times. This reduction in fracture time is due to a faster release of hydrostatic stress after the cavitation process [26,41,[56][57][58][59]. Therefore, higher stress around the cavities promoted a reduction in the damping behaviour, increasing the cohesive fracture.
Based on all data acquired for the FF joints subjected to the static loads, a simple modelling of the failure time was developed. Figure 17 plots the creep data, representing the load applied versus failure time in a double logarithmic scale. The linear equation better fitting by linear regression to these data is depicted in the figure. The high coefficient of determination value (R 2 ) confirmed the high data fitting. Note that the result was an exponential equation type, provided by the double logarithmic scales. Using this model, the failure time of the FF adhesive joints can be estimated as a function of the applied load. Equation (2) depicts this fitting model.
where y is log (applied load), x is the log (fracture time), and a and b are the fitting parameters, 0.913 Kg/cm 2 and −0.405 Kg/(cm 2 ·h), respectively.

Conclusions
The present work reported the mechanical behaviour of three acrylic pressure-sensitive adhesives (PSAs) assembled over aluminium specimens with different surface and chemical treatments. This study is in high demanded in some industries, such as the shipbuilding and offshore sectors, as the application of PSAs to join non-structural components to aluminium substrates has not been widely investigated. In this sense, shear, Tpeel, and creep tests were performed, followed by surface fracture analysis. Based on these mechanical results, the following conclusions could be drawn:

•
Low roughness values (with Ra values equal to lower than 0.21 μm) notably improved the USS of the acrylic foam adhesives (RF and FF). Acrylic core adhesive (FC) presented similar behaviour regardless of the roughness.

•
The application of adhesion promoter before assembling resulted in a significant improvement in USS and cohesion for the three adhesives.

•
The FF adhesive displayed the best shear and T-peel behaviour.

Conclusions
The present work reported the mechanical behaviour of three acrylic pressure-sensitive adhesives (PSAs) assembled over aluminium specimens with different surface and chemical treatments. This study is in high demanded in some industries, such as the shipbuilding and offshore sectors, as the application of PSAs to join non-structural components to aluminium substrates has not been widely investigated. In this sense, shear, T-peel, and creep tests were performed, followed by surface fracture analysis. Based on these mechanical results, the following conclusions could be drawn:

•
Low roughness values (with R a values equal to lower than 0.21 µm) notably improved the USS of the acrylic foam adhesives (RF and FF). Acrylic core adhesive (FC) presented similar behaviour regardless of the roughness.

•
The application of adhesion promoter before assembling resulted in a significant improvement in USS and cohesion for the three adhesives.

•
The FF adhesive displayed the best shear and T-peel behaviour.