Development of Passive Fire Protection Mortars

: During a ﬁre event, the stability of steel structures may be compromised, and structural collapse may occur due to the loss of their mechanical resistance as the temperature increases. One of the solutions to reduce this problem is the protection with a coating using enhanced ﬁre-resistant mortars. This paper reports a detailed experimental work aiming to develop gypsum and cement-based mortars for passive ﬁre protection and evaluate their composition’s effect in the ﬁnal thermal performance. Two types of specimens were tested: (i) small specimens composed of a mortar coating (10 mm thick) and one steel plate and (ii) square section short tubular steel columns with 20 mm of coating. The evaluation of the thermal protection was carried out by (a) measuring the thermal gradient between the exposed surface of the protected steel plate under high temperatures and the mortar-steel interface and (b) assessing the ﬁre resistance of the short steel columns. It was concluded that the compositions with gypsum binder present better thermal insulation than the cementitious compositions. Additionally, the introduction of nano- and microparticles of silica still slightly improved the thermal insulation of the tested compositions.


Introduction
There is a growing interest in developing alternative and sustainable construction materials with enhanced properties. As metals are infinitely recyclable, this type of construction is part of a future of "green construction", thus contributing to a sector of the economy with low environmental impact [1][2][3].
However, steel structures show some weaknesses, especially their structural behaviour when subjected to fire [4][5][6][7][8][9]. Due to the high thermal conductivity of the steel, the high section factor of the members, and the rapid degradation of the steel mechanical properties, with the increase of steel temperature, rapid change in the stiffness and mechanical resistance may be noticed in the structure. Its resistance and stability may be compromised, leading to the collapse of some elements or even of the entire structure [10][11][12][13][14][15].
The thermal protection of these structural elements is crucial. The fire protection of structures can be achieved by combining active and passive fire protection systems and management systems (smoke exhaust systems, communication and evacuation procedures, fire detection systems, and compartmentation [16]). Active fire protection can consist of systems or items such as fire extinguishers, standpipes, sprinkler systems, and fire blankets [17]. It is required that these systems have a quick response capacity in extinguishing or controlling the development of a fire in its initial phase [18]. However, implementing these systems has high installation and maintenance costs, and their results can have low functional reliability and unsatisfactory operational results [19]. size, and the addition of silica micro-and nanoparticles on their thermal performance and assessment of fire resistance to the protected short steel columns.
* Water commercial solution ratio in weight %.
To analyze the influence of expanded perlite and expanded vermiculite grain size on the thermal performance of laboratory developed mortars, two different grinding methods were used: Los Angeles (LA) and Industrial Mill (IM). The first method was carried out using the Los Angeles method, which fragmented the aggregate by abrasion and shock using steel balls. The particle size of these raw materials was assessed using a particle size analysis (specification LNEC E 195-1966), and it was observed that their size ranged from 0.075 to 0.85 mm. The second method was carried out using an industrial mill, which by the friction of the aggregate with the drum significantly reduced the size of its particles compared to the LA method. The particle size analysis identified a particle size ranging from 0.025 to 0.40 mm. Finally, different dosages of silica micro and nanoparticles were added and tested in DCM, DGMP, DGMV, and DRCM to assess their influence on the thermal insulation of the respective mortars, as described in the following section of the paper.

Experimental Program
The experimental program included two different types of tests, depending on the type of specimens: steel plate (SP) ( Table 3) and square section short steel columns (SSC) ( Table 4). The experimental program of tests on SP included five different mortar (CM, DCM, DGMP, DGMV, and DRCM), and 45 specimens were produced. Each set of three specimens used in their thermal tests was used to obtain better reliability of results.  3  3  3  3  DGMP  3  3  3  3  DGMV  3  3  3  3  DRCM  3 3 (**) (**) (*) The commercial solution was not modified, so there was no need to test more than three specimens. (**) Since this mortar has the worst results, it has not been tested with the addition of silica micro-and nanoparticles.  Table 3 presents the experimental program defined to evaluate the influence of the size of the aggregates and the addition of silica micro-and nanoparticles on the thermal performance of the developed mortars. The percentage of silica micro-and nanoparticles were the same and equal to 0.5% in weight of binder for each one. All specimens were exposed to high temperatures on one side up to 900 • C. Additionally, the experimental program of tests carried out on 12 SSC under fire conditions included three different fire protection mortars (CM, DCM, and DGMP). DGMV and others new ones will make part of another future experimental campaign, in which different aggregates/additives will be studied.

Preparation of the Specimens
The manufacturing process, shape, and dimensions of the SP and moulds were defined to measure the thermal gradient generated between the inner surface of the steel plate exposed to high temperatures and the exposed surface of the fire protection mortar. Thus, a suitable mould was designed and manufactured for these tests ( Figure 1). Appl

Preparation of the Specimens
The manufacturing process, shape, and dimensions of the SP and moulds were defined to measure the thermal gradient generated between the inner surface of the steel plate exposed to high temperatures and the exposed surface of the fire protection mortar. Thus, a suitable mould was designed and manufactured for these tests ( This mould assigns the desired geometric shape to the specimen, was easy to transport and clean, was reusable for many experimental tests, and was easy to assemble and disassemble when concreting and removing the specimen. During the fabrication of mortars, a balance accurate to 0.1 gr, a graduated beaker, Hobart N50 mixer with 5 L of capacity and a stainless-steel lab spatula were used. The procedure adopted in the fabrication of mortars was as follows: 1. The raw materials were weighed and placed inside the mixer container. 2. Then, the mixer was put into operation for 5 min at a slow speed (136 rotations per minute). At the same time, the corresponding amount of water was added, with a This mould assigns the desired geometric shape to the specimen, was easy to transport and clean, was reusable for many experimental tests, and was easy to assemble and disassemble when concreting and removing the specimen. During the fabrication of mortars, a balance accurate to 0.1 g, a graduated beaker, Hobart N50 mixer with 5 L 1.
The raw materials were weighed and placed inside the mixer container.

2.
Then, the mixer was put into operation for 5 min at a slow speed (136 rotations per minute). At the same time, the corresponding amount of water was added, with a constant flow rate to guarantee the homogeneous addition of water in the whole mortar.

3.
After this procedure, the mortar was manually kneaded with a spatula to remove parts of the mortar that were on the walls of the container and thus homogenize the mixture, then returning the container to the mixer for another 2 min. 4.
The mortar was placed inside the mould ( Figure 2).  About 48 h after casting the SP in the moulds described above, they were demoulded and placed in the curing process ( Figure 3a) for 28 days in the laboratory environment with controlled temperature (25 °C) and a relative humidity (RH) of 55%. The specimens were tested with 6 months of age.  The specimens comprised a S355 steel plate with a square section of 250 mm of edge and a thickness of 5 mm; and 10 mm thick fire protection mortar on one side of the steel plate ( Figure 3b). The temperature measurement in the SP was carried out by placing 4 type K thermocouples. The thermocouples were placed at different depths across the specimen ( Figure 4). In the production of mortars, the same procedure was followed to ensure that the different properties of the mortars were only dependent on their composition. To minimize possible effects that temperature and humidity might have on the properties of each mortar composition, all mixtures of each composition were manufactured on the same day and placed in a room with controlled environmental conditions. It is well known that the moisture content greatly influences the fire behavior of mortars at elevated temperatures [39,40], in the same way as in the concretes.
About 48 h after casting the SP in the moulds described above, they were demoulded and placed in the curing process ( Figure 3a) for 28 days in the laboratory environment with controlled temperature (25 • C) and a relative humidity (RH) of 55%. The specimens were tested with 6 months of age. About 48 h after casting the SP in the moulds described above, they were demoulded and placed in the curing process ( Figure 3a) for 28 days in the laboratory environment with controlled temperature (25 °C) and a relative humidity (RH) of 55%. The specimens were tested with 6 months of age. The specimens comprised a S355 steel plate with a square section of 250 mm of edge and a thickness of 5 mm; and 10 mm thick fire protection mortar on one side of the steel plate ( Figure 3b). The temperature measurement in the SP was carried out by placing 4 type K thermocouples. The thermocouples were placed at different depths across the specimen ( Figure 4). The specimens comprised a S355 steel plate with a square section of 250 mm of edge and a thickness of 5 mm; and 10 mm thick fire protection mortar on one side of the steel plate ( Figure 3b). The temperature measurement in the SP was carried out by placing The specimens comprised a S355 steel plate with a square section of 250 mm of edge and a thickness of 5 mm; and 10 mm thick fire protection mortar on one side of the steel plate ( Figure 3b). The temperature measurement in the SP was carried out by placing 4 type K thermocouples. The thermocouples were placed at different depths across the specimen ( Figure 4). With this distribution of thermocouples, it was possible to determine the thermal gradient between the surface of the mortar exposed to high temperatures (ESMHT-thermocouple 3) and the unexposed surface (USMHT-thermocouple 2), as well as the temperature on the inner surface of the steel plate (ISSP-thermocouple 1) and its external surface (ESSP-thermocouple 4). In Figure 5, it is possible to identify the thermocouples on the specimen following Figure 4. With this distribution of thermocouples, it was possible to determine the thermal gradient between the surface of the mortar exposed to high temperatures (ESMHTthermocouple 3) and the unexposed surface (USMHT-thermocouple 2), as well as the temperature on the inner surface of the steel plate (ISSP-thermocouple 1) and its external surface (ESSP-thermocouple 4). In Figure 5, it is possible to identify the thermocouples on the specimen following Concerning the SSCs tests, specimens were defined by a hollow square section 150 × 150 × 8 mm, with a height of 1250 mm, and the steel grade was S355. At the column ends, it was centered and welded a steel plate (section 300 × 300 × 20 mm), as shown in Figure  8. To evaluate the temperature evolution on the external surfaces of the steel columns during the test, 12 type K thermocouples were welded, equidistant from each other on all the specimen's surfaces, applied in 3 groups of 4 thermocouples at different heights. These termocouples were welded in the middle of the surfaces of the steel tubular columns. To guarantee a constant and uniform mortar thickness of 20 mm along the steel columns, a modular formwork was developed with the ability to assign the desired geometric shapewith easy assembly and disassembly while concreting the specimen. Figure 6 depicts the location of the three groups of thermocouples and the different concreting steps of the steel columns. The specimens were tested after curing for 6 months. Concerning the SSCs tests, specimens were defined by a hollow square section 150 × 150 × 8 mm, with a height of 1250 mm, and the steel grade was S355. At the column ends, it was centered and welded a steel plate (section 300 × 300 × 20 mm), as shown in Figure 8. To evaluate the temperature evolution on the external surfaces of the steel columns during the test, 12 type K thermocouples were welded, equidistant from each other on all the specimen's surfaces, applied in 3 groups of 4 thermocouples at different heights. These termocouples were welded in the middle of the surfaces of the steel tubular columns. To guarantee a constant and uniform mortar thickness of 20 mm along the steel columns, a modular formwork was developed with the ability to assign the desired geo- metric shapewith easy assembly and disassembly while concreting the specimen. Figure 6 depicts the location of the three groups of thermocouples and the different concreting steps of the steel columns. The specimens were tested after curing for 6 months. Figure 6. Distribution of thermocouples (a) and fabrication of specimens, preconcreting (b), concreting on the short steel columns (c), and concrete specimen without formwork (d).

Experimental Testing System and Procedure
The experimental testing system ( Figure 7) used in thermal analysis of steel plates consisted of a cylindrical oven with internal dimensions 400 mm in height and 250 mm in diameter, capable of reaching a maximum temperature of 1200 °C (a) and the respective oven controller (b). A Datalogger TDS-530 was used as a data acquisition system (c) to record the temperature readings. Regarding the test procedure, after sealing all the existing unions and holes of the oven with rock wool, a 8 cm thick rock wool blanket with a circular opening of 210 mm in diameter was also placed on the top of the oven (e), which allows the passage of heat from the interior of the oven to the specimen. Subsequently, the specimen (d) was placed on the top of the oven, i.e., on the rock wool blanket and centred with the barycentric axis of the oven. In all tests, the specimens were placed in a suitable position to ensure perfect accommodation with the rock wool and thus avoid heat losses between the specimen and the oven. The specimen was heated at a heating rate of 15 °C/minute until reaching the desired temperature level (900 °C). Temperatures inside the specimen and the oven were

Experimental Testing System and Procedure
The experimental testing system (Figure 7) used in thermal analysis of steel plates consisted of a cylindrical oven with internal dimensions 400 mm in height and 250 mm in diameter, capable of reaching a maximum temperature of 1200 • C (a) and the respective oven controller (b). A Datalogger TDS-530 was used as a data acquisition system (c) to record the temperature readings. Regarding the test procedure, after sealing all the existing unions and holes of the oven with rock wool, a 8 cm thick rock wool blanket with a circular opening of 210 mm in diameter was also placed on the top of the oven (e), which allows the passage of heat from the interior of the oven to the specimen. Subsequently, the specimen (d) was placed on the top of the oven, i.e., on the rock wool blanket and centred with the barycentric axis of the oven. Appl Figure 6. Distribution of thermocouples (a) and fabrication of specimens, preconcreting (b), concreting on the short steel columns (c), and concrete specimen without formwork (d).

Experimental Testing System and Procedure
The experimental testing system ( Figure 7) used in thermal analysis of steel plates consisted of a cylindrical oven with internal dimensions 400 mm in height and 250 mm in diameter, capable of reaching a maximum temperature of 1200 °C (a) and the respective oven controller (b). A Datalogger TDS-530 was used as a data acquisition system (c) to record the temperature readings. Regarding the test procedure, after sealing all the existing unions and holes of the oven with rock wool, a 8 cm thick rock wool blanket with a circular opening of 210 mm in diameter was also placed on the top of the oven (e), which allows the passage of heat from the interior of the oven to the specimen. Subsequently, the specimen (d) was placed on the top of the oven, i.e., on the rock wool blanket and centred with the barycentric axis of the oven. In all tests, the specimens were placed in a suitable position to ensure perfect accommodation with the rock wool and thus avoid heat losses between the specimen and the oven. The specimen was heated at a heating rate of 15 °C/minute until reaching the desired temperature level (900 °C). Temperatures inside the specimen and the oven were In all tests, the specimens were placed in a suitable position to ensure perfect accommodation with the rock wool and thus avoid heat losses between the specimen and the oven. The specimen was heated at a heating rate of 15 • C/minute until reaching the desired temperature level (900 • C). Temperatures inside the specimen and the oven were measured every 5 s. When the target temperature in the specimen was reached, it was maintained uniform during 3 h and after, the test was given as concluded.
The experimental layout for the short steel columns (SSC) under fire conditions (Figure 8) consisted essentially of a reaction steel frame (A) to apply the serviceability load on the specimen, a support steel frame (B), a hydraulic jack (C), and an electric furnace (H).
Appl. Sci. 2021, 11, x FOR PEER REVIEW 10 of 22 measured every 5 s. When the target temperature in the specimen was reached, it was maintained uniform during 3 h and after, the test was given as concluded.
The experimental layout for the short steel columns (SSC) under fire conditions (Figure 8) consisted essentially of a reaction steel frame (A) to apply the serviceability load on the specimen, a support steel frame (B), a hydraulic jack (C), and an electric furnace (H). This steel frame was defined by HEB 500 columns and a HEB 600 beam (A), with a stiffness capable of minimizing possible displacements of this steel structure during the tests. Additionally, a support 3D steel frame consisting of two frames (B) with HEB 300 columns and HEB 400 beams accommodate the testing specimen to similar actual boundary conditions. Regarding the test equipment, a 3 MN hydraulic jack (C) and its controller (J), a 3 MN load cell (D), and a 1 MN load cell (E) were used for measuring the compression forces. Ten linear variable displacement transducers were used for displacements measurements (F), a Datalogger (G) for data acquisition, and an electric furnace (H) to heat up the steel columns (I).
A hydraulic jack controlled by a servo-controlled central was used, and a preload of 50% of the design value of the loadbearing capacity of the columns at ambient temperature (ULS) was applied (727.8 kN) to simulate a service load on the specimen. After stabilising this loading in the specimen, the furnace was switched on and the specimen heated according to the temperature evolution established by the ISO 834 standard fire curve [41]. This steel frame was defined by HEB 500 columns and a HEB 600 beam (A), with a stiffness capable of minimizing possible displacements of this steel structure during the tests. Additionally, a support 3D steel frame consisting of two frames (B) with HEB 300 columns and HEB 400 beams accommodate the testing specimen to similar actual boundary conditions. Regarding the test equipment, a 3 MN hydraulic jack (C) and its controller (J), a 3 MN load cell (D), and a 1 MN load cell (E) were used for measuring the compression forces. Ten linear variable displacement transducers were used for displacements measurements (F), a Datalogger (G) for data acquisition, and an electric furnace (H) to heat up the steel columns (I).
A hydraulic jack controlled by a servo-controlled central was used, and a preload of 50% of the design value of the loadbearing capacity of the columns at ambient temperature (ULS) was applied (727.8 kN) to simulate a service load on the specimen. After stabilising this loading in the specimen, the furnace was switched on and the specimen heated according to the temperature evolution established by the ISO 834 standard fire curve [41].
Due to the increase in temperature during the test, significant thermal elongation was generated in the specimen. The test was stopped based on the axial contraction criterion defined by ISO 834-1:1999 [42], which defines a limit shortening of the specimen according to its initial length, i.e., when the specimen could no longer support the initial applied load subjected to high temperatures. Thus, the final value of vertical deformation considered was 12.5 mm. Finally, the degree of detachment of the protective mortar and the failure mode in each specimen were observed.

Steel Plate Specimens
The typical temperature evolution in the tested SP (CM, DCM, DGMP, DGMV, and DRCM) subjected up to 900 • C as a function of time, is shown in Figure 9.
Appl. Sci. 2021, 11, x FOR PEER REVIEW 11 of 22 Due to the increase in temperature during the test, significant thermal elongation was generated in the specimen. The test was stopped based on the axial contraction criterion defined by ISO 834-1: 1999 [42], which defines a limit shortening of the specimen according to its initial length, i.e., when the specimen could no longer support the initial applied load subjected to high temperatures. Thus, the final value of vertical deformation considered was 12.5 mm. Finally, the degree of detachment of the protective mortar and the failure mode in each specimen were observed.

Steel Plate Specimens
The typical temperature evolution in the tested SP (CM, DCM, DGMP, DGMV, and DRCM) subjected up to 900 °C as a function of time, is shown in Figure 9. It is possible to verify that the oven used in this experimental layout generated a heating curve (oven temperature curve) close to the desired heating curve (setpoint temperature). Thus, all specimens were exposed to a uniform and identical thermal action, making the tests comparable to each other. It is possible to verify that the oven used in this experimental layout generated a heating curve (oven temperature curve) close to the desired heating curve (setpoint temperature). Thus, all specimens were exposed to a uniform and identical thermal action, making the tests comparable to each other.
Between 0 and 35 min, there was a slow and gradual increase in temperature of the specimens (CM, DCM, DGMP, DGMV, and DRCM) up to 100 • C. This temperature evolution was influenced by the water content in the mortars, i.e., by the energy required to heat and evaporate the liquid/crystallized water existing in the mortar. After this process, there was a sharp increase in temperature between 35 and 75 min, depending on the thermal conductivity of the mortar and the heating rate. Finally, until the end of the test, there was a slight increase in temperature in the specimen before a steady state temperature distribution was reached. This type of thermal development in the specimens was common for all compositions tests.
Based on the values obtained from the various tests carried out, it was possible to measure the average temperature on the inner surface of the steel plate of the specimen and thus establishing a direct relationship between the developed composition and its thermal protection capacity. The average values of the maximum temperature on the inner surface of the specimen's steel plate (thermocouple 1) are shown in Table 5.  From the results presented in Table 5, it is possible to evaluate the thermal efficiency of the developed mortars: DCM, DGMP, DGMV, and DRCM, when compared with the best commercial solution tested (CM). Furthermore, it allows evaluating the influence of the size of the aggregates and the addition of silica micro-and nanoparticles on their thermal performance. The second column of Table 4 refers to mortars developed without NS and MS, in which the raw materials were obtained using the Los Angeles method (LA method). The temperature of CM was used as a reference for comparison (311 • C). The average temperature on the inner surface of the steel plate (ISSP) in the DCM, DGMP, DGMV, and DRCM mortars were 293, 310, 278, and 323 • C, respectively. These results show that the thermal protection provided by DCM, DGMP, and DGMV mortars were 5.8%, 0.3%, and 10.6% thermally more efficient than CM, respectively. However, the DRCM is thermally worse (3.9%) when compared to the CM. Finally, the compositions with vermiculite (DCM and DGMV) have better thermal efficiency.
The fourth column of Table 5 refers to mortars developed without NS and MS, in which the raw materials used were obtained using the industrial mill method (IM method). The average temperature in the ISSP in the DCM, DGMP, DGMV, and DRCM mortars were 309, 325, 275, and 366 • C, respectively. When comparing the ISSP temperatures of the DCM, DGMP, and DRCM mortals made with raw material from Los Angeles milling with the ISSP temperatures of the same mortars made with raw material from industrial milling, there was respectively a loss of insulation capacity by 5.5%, 4.6%, and 13.3% on average. As for the thermal performance of the DGMV, there is a slight improvement in thermal capacity of 1.1%. Overall, these results demonstrate that the reduction in grain size of the raw materials used (perlite and vermiculite) did not benefit the thermal performance of the tested compositions. Note that the reduction in the porosity of mortars may have an adverse effect in the fire insulation, but only neglecting the thermomechanical effect of mortars and the thermal deformation of the structural members, as it was the case in the thermal tests of the protected steel plates. Finally, compared to the commercial mortar with the developed mortals, DCM (cementitious mortar) and DGMV (gypsum mortar with vermiculite) had greater thermal insulation capacity, while mortars DGMP (gypsum mortar with vermiculite) and DRCM (refractory cementitious mortar) had lower thermal insulation capacity.
The sixth column of Table 5 refers to mortars developed with NS and MS, and the raw materials used were obtained by the LA method. This column only presents the results of the tests carried out with DCM, DGMP, and DGMV since the results of DRCM never presented a thermal insulation capacity higher than the commercial solution. The average temperature on the ISSP in the DCM, DGMP, and DGMV were 308, 275, and 268 • C, respectively. It appears that the introduction of NS and MS in the tested dosages slightly improved the thermal performance of the mortars by 11.3% for DGMP and 3.6% for DGMV. Comparing to CM (commercial mortal), both developed mortars with NS and MS (A2, A3, and A4) had greater thermal insulation capacity than the mortars without such particles.
Values for mortars with NS and MS and raw materials obtained by the IM method are presented in the eighth column. The average temperature in the ISSP in the DCM, DGMP, and DGMV were 297, 274, and 266 • C, respectively. The results demonstrate that the thermal protection provided by DCM, DGMP, and DGMV were 4.5%, 11.9%, and 14.5% thermally more efficient than CM, respectively. Comparing these results with those of the fourth column, the addition of NS and MS in DCM, DGMP, and DGMV improved their thermal performance by 3.9%, 15.7%, and 3.3%, respectively.
In addition, comparing these results with those of the sixth column, the manufacture of DCM, DGMP, and DGMV with raw materials processed with the IM method gives only a slight benefit on the thermal performance of these compositions, by 3.6%, 0.4%, and 0.8%, since nanoparticles tend to increase the shrinkage of materials. This is due to stronger retention of water, and capillary forces developed.
Briefly, the differences in the results presented may be due to the higher porosity of the materials obtained through the LA method than the IM method, i.e., in general the higher the porosity, the higher the thermal insulation is. However when micro-and nanoparticles of silica were added, the thermal responce of fire protection materials might be different (as it was the case) since those particles can work as fire barriers, i.e., close more efficiently some porous in the materials.
The average crack width values of the mortars after carrying out the tests at high temperatures are presented in Table 6, and specimens after testing and the respective type of cracking are illustrated in Table 7. Based on the crack withs presented in the column of the Table 6, the CM has the highest average crack width compared to the DCM, DGMP, DGMV, and DRCM. Compared with CM, mortars DCM, DGMP, DGMV, and DRCM have lower average crack widths by 55.5%, 80.5%, 63.4% and 76.6%.  In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. In general terms, the addition of NS and MS tends to increase the cracking of the developed compositions, except for the DGMV, as shown in Table 7. Furthermore, it could be observed that the crack width was worse than the number of cracks (with smaller widths). To sum up, crack widths higher than 0.6 mm might compromise their thermal protection. DGMV DRCM (**) (**) (*) The commercial solution was not modified, so there was no need to test more than three specimens. (**) Since this mortar has the worst results, it has not been tested with the addition of silica micro-and nanoparticles.

Short Steel Columns Specimens
Regarding the protected square-section short steel columns (SSC), the assessment of the column temperature was based on the arithmetic average of the temperatures obtained from the 12 thermocouples welded to the steel column, which were distributed into three sections with four thermocouples each one (Figures 6 and 10). As an example, Figure 11 shows the temperature evolution in steel at the lower section of the column where temperatures were recorded ( Figure 6).
(*) The commercial solution was not modified, so there was no need to test more than three specimens. (**) Since this mortar has the worst results, it has not been tested with the addition of silica micro-and nanoparticles.

Short Steel Columns Specimens
Regarding the protected square-section short steel columns (SSC), the assessment of the column temperature was based on the arithmetic average of the temperatures obtained from the 12 thermocouples welded to the steel column, which were distributed into three sections with four thermocouples each one (Figures 6 and 10). As an example, Figure  11 shows the temperature evolution in steel at the lower section of the column where temperatures were recorded ( Figure 6). Based on Figure 10, it is possible to observe that the furnace temperatures slightly delayed the initial minutes concerning the ISO 834 standard fire curve. This part of the curve is difficult to reproduce in an electric furnace, and this becomes worse for larger furnaces (high initial thermal inertia). However, near 9 min after the beginning of the heating, the furnace temperatures followed close to the ISO 834 standard fire curve (ISO  834-1, 1999). Nevertheless, the evolution of temperatures inside the furnace over time was uniform in all fire tests, meaning that the tests were comparable.
Based on Figure 11, it is possible to observe that the temperature development recorded in the four thermocouples of group 3 was similar. There is only a difference of 3.3°C between the thermocouple with the highest (TH_12) and the lowest temperature (TH_9) at column failure.
Until the specimens became unstable, graphs were generated with the evolution of the vertical deformation of the specimen as a function of time ( Figure 12) to identify the critical experimental temperature, as well as graphs of temperature evolution as a function of time ( Figure 13). Based on Figure 10, it is possible to observe that the furnace temperatures slightly delayed the initial minutes concerning the ISO 834 standard fire curve. This part of the curve is difficult to reproduce in an electric furnace, and this becomes worse for larger furnaces (high initial thermal inertia). However, near 9 min after the beginning of the heating, the furnace temperatures followed close to the ISO 834 standard fire curve (ISO 834-1, 1999). Nevertheless, the evolution of temperatures inside the furnace over time was uniform in all fire tests, meaning that the tests were comparable.
Based on Figure 11, it is possible to observe that the temperature development recorded in the four thermocouples of group 3 was similar. There is only a difference of 3.3 • C between the thermocouple with the highest (TH_12) and the lowest temperature (TH_9) at column failure.
Until the specimens became unstable, graphs were generated with the evolution of the vertical deformation of the specimen as a function of time ( Figure 12) to identify the critical experimental temperature, as well as graphs of temperature evolution as a function of time ( Figure 13).
heating, the furnace temperatures followed close to the ISO 834 standard fire curve (ISO 834-1, 1999). Nevertheless, the evolution of temperatures inside the furnace over time was uniform in all fire tests, meaning that the tests were comparable.
Based on Figure 11, it is possible to observe that the temperature development recorded in the four thermocouples of group 3 was similar. There is only a difference of 3.3°C between the thermocouple with the highest (TH_12) and the lowest temperature (TH_9) at column failure.
Until the specimens became unstable, graphs were generated with the evolution of the vertical deformation of the specimen as a function of time ( Figure 12) to identify the critical experimental temperature, as well as graphs of temperature evolution as a function of time ( Figure 13).   [41] for this short steel column was 586.7°C.
In Figure 12, the tested SSC2 failed at 17 min, SSC4 at 77 min, SSC6 at 84 min, SSC8 at 95 min, SSC12 at 102 min, and SSC14 at 91 min, corresponding to the temperature of 617, 531, 585, 572, 566, and 559°C, respectively (see also Table 8). In Figure 13 and Table 8, it can be seen that the commercial mortar has a more effective thermal protection for lower temperatures (the highest delay in the temperature rise at the beginning of the test). However, for temperatures higher than 400°C, its thermal capacity tends to decrease due to the degradation of the mortar (the highest temperature rise rate at the ending of the test). Figure 14 depicts the specimen before and after being tested. Regarding the instability modes of the steel columns, local instability was observed despite the column being a class 1 cross-section under fire conditions.   [41] for this short steel column was 586.7 • C.
In Figure 12, the tested SSC2 failed at 17 min, SSC4 at 77 min, SSC6 at 84 min, SSC8 at 95 min, SSC12 at 102 min, and SSC14 at 91 min, corresponding to the temperature of 617, 531, 585, 572, 566, and 559 • C, respectively (see also Table 8). In Figure 13 and Table 8, it can be seen that the commercial mortar has a more effective thermal protection for lower temperatures (the highest delay in the temperature rise at the beginning of the test). However, for temperatures higher than 400 • C, its thermal capacity tends to decrease due to the degradation of the mortar (the highest temperature rise rate at the ending of the test). Figure 14 depicts the specimen before and after being tested. Regarding the instability modes of the steel columns, local instability was observed despite the column being a class 1 cross-section under fire conditions.  Figure 13. Evolution of temperature in the specimens of SSC as a function of time. Table 8 shows the average temperature values acquired in the specimens after 15, 30, 60, and 90 min and the average temperature values for the failure instants of each specimen. The critical design temperature calculated according to EN 1993-1-2: 2005 [41] for this short steel column was 586.7°C.
In Figure 12, the tested SSC2 failed at 17 min, SSC4 at 77 min, SSC6 at 84 min, SSC8 at 95 min, SSC12 at 102 min, and SSC14 at 91 min, corresponding to the temperature of 617, 531, 585, 572, 566, and 559°C, respectively (see also Table 8). In Figure 13 and Table 8, it can be seen that the commercial mortar has a more effective thermal protection for lower temperatures (the highest delay in the temperature rise at the beginning of the test). However, for temperatures higher than 400°C, its thermal capacity tends to decrease due to the degradation of the mortar (the highest temperature rise rate at the ending of the test). Figure 14 depicts the specimen before and after being tested. Regarding the instability modes of the steel columns, local instability was observed despite the column being a class 1 cross-section under fire conditions. The results from Table 8 clearly show that the application of mortars works as a good thermal barrier to fire, since columns without passive fire protection failed after 17 min of testing, whereas the protected columns failed on average, beyond 81 min. Furthermore, the results also show that the thermal protection of mortars developed in the laboratory was more efficient than that provided by the commercial mortars. Finally, it appears that the introduction of the NS and MS slightly improved the thermal performance of the mortars developed in the laboratory. Table 8 shows that the specimen without passive fire protection presented a fire resistance rating (FRR) of 15 min (R15), and the protected specimens with the CM allowed a FRR of R60. Concerning the thermal performance of the mortars developed in the laboratory, it was found that the specimens protected with the DCM had a FRR of R60. When NS and MS were added to its composition, a FRR increased to R90. The specimens protected with the DGMP, with and without NS and MS, presented a FRR of R90.
The findings results obtained from the short steel columns protected with the different mortars types were in agreement with the ones obtained in the tests carried out on the protected steel plates, which makes these exploratory tests useful for preliminary selection of promising fire protection materials.

Conclusions
The objective of this work was to develop gypsum or cement-based mortars for passive fire protection and to evaluate the influence of aggregate size and the addition of silica micro-and nanoparticles on its thermal performance. The thermal performance of these mortars in short steel columns under a compression service load and subjected to high temperatures was assessed. The following conclusions can be drawn:

•
The furnace, the dimensions of the specimens, and the test procedure adopted in the tests at high temperatures allowed an adequate thermal exposure of the specimens. It allowed the evaluation of the thermal performance of the various mortars tested. • Some mortars developed in the laboratory (DCM, DGMP, and DGMV) have better thermal performance when compared with the best commercial solution tested (CM). Furthermore, in this set of mortars without nano-and microparticles of silica, the mortars with vermiculite in their constitution were those with the best thermal performance.

•
Most mortars with raw material milled through the Los Angeles method had better thermal performance than mortars developed with raw material processed through the Industrial Mill method. However, if nano-and microparticles of silica are added in their composition, the mortars developed with raw material obtained through the Industrial Mill method may have a slightly superior thermal performance.

•
The addition of nano-and microparticles of silica improves the insulating capacity of the mortars.

•
Overall, the results demonstrate that the reduction in grain size of the raw materials used (perlite and vermiculite) did not benefit the thermal performance of the tested compositions.

•
Regarding the cracking of the mortars, it was concluded that perlite (DGMP) contributes to its low value. It was also concluded that, in general terms, the addition of NS and MS tends to increase the cracking degree of the developed compositions.

•
The compositions that use gypsum as a binder (DGMP and DGMV) had the best thermal insulation capacity. Under the tested conditions, it was found that 10 mm of mortar coating was sufficient to form an efficient thermal barrier, reducing the ISSP temperature by approximately 70% of the temperature recorded inside the oven (900 • C).

•
The thermal protection level of columns with the mortar developed in the laboratory with the best overall thermal performance (DGMP with nano-and microparticles of silica) was 19% more efficient than the commercial solution and increases by 5.9 times the fire resistance of an unprotected short steel column. These results demonstrated the actual impact that the application of such mortars can have as passive fire protection of steels structures.
Bearing in mind the experimental findings obtained in this research study, further experimental tests on short steel columns with the developed mortars and new ones by using different additives as well as with different loading conditions will be carried out in the near future.

Acknowledgments:
The authors gratefully acknowledge the Portuguese Foundation for Science and Technology (FCT) for its support under the framework of research project PTDC/ECI-EGC/31850/2017 (NANOFIRE-Thermal and Mechanical behaviour of Nano Cements and their application in steel construction as fire protection) and also to the University of Coimbra (UC) for their support under the Scientific Employment Stimulus Programme given to the first author, as well as to the European Regional Development Fund, the European Social Fund, and European Structural and Investment Funds. This work was also financed by FEDER funds through the Competitivity Factors Operational Programme-COMPETE and by national funds through FCT within the scope of the project POCI-01-0145-FEDER-007633 and through the Regional Operational Programme CENTRO2020 within the scope of the project CENTRO-01-0145-FEDER-000006.

Conflicts of Interest:
The authors declare no conflict of interest.