Influence of Thermomechanical Treatments and Chemical Composition on the Phase Transformation of Cu-Al-Mn Shape Memory Alloy Thin Sheets

Abstract

This paper investigates the interrelated effects of thermomechanical treatments and chemical composition on the phase transformation capabilities of thin sheets made from Cu-Al-Mn shape memory alloys. The transformation capacity and transition temperatures were determined using DSC and DMA testing, while composition measurements were performed using SEM/EDX analysis. The results demonstrate that applying hot-rolling treatments to alloys of reduced thickness leads to manganese oxidation and modifications in chemical composition, adversely impacting the phase transformation performance. This effect can be mitigated by the use of cold rolling. Additionally, the presence of phosphorus impurities can create inclusions that bind manganese, preventing it from remaining in the solid solution and further affecting phase transformation capabilities.

1. Introduction

The interest in SMA sheets primarily comes from their inherent damping performance, potentially making them a viable structural material intended for vibration damping in lightweight systems. Compared to other materials in which some damping is intrinsically present to a higher or lower degree, shape memory alloys display two additional energy dissipation mechanisms that make them stand out [1]. The first mechanism is related to the twinned martensitic phase which as a consequence of load application goes through de-twinning [1,2]. During load application, the hysteretic movement of martensitic variant interfaces within the alloy causes an increase in internal friction, thus increasing energy dissipation [3,4,5]. The second mechanism of energy dissipation is related to the increased internal friction during temperature-induced direct (cooling) and reverse (heating) martensitic phase transformations of the shape memory alloy, which result in a damping peak [6,7,8]. To date, NiTi remains the most researched and most widely adopted shape memory alloy in the industry. However, despite its reliability and high performance, the high costs of raw materials and fabrication have been motivating researchers to look for alternative solutions for decades. One such alternative is the Cu-based shape memory alloys, which despite cost advantages have yet to see wider application due to issues such as high brittleness, low workability, martensitic stabilization, and more modest shape memory performance in comparison to NiTi [9]. Nonetheless, significant advances in the understanding of Cu-Al-Mn alloys have been made by scholars. These efforts resulted in significantly improved material workability, higher levels of damping performance, better shape recovery, etc. The reduction in aluminum content within the alloy to below 17% at. allowed for workability higher than 60%, where workability was defined as the percentage of thickness reduction before cracking, which is caused by a low degree of order in the parent phase [10]. Improvements in both shape recovery properties and passive damping performance in the martensitic phase were achieved by increasing the relative grain size, which is a consequence of a reduction in the dimensionality of constraints between individual martensitic grains within the alloy [11,12,13]. Specifically, a loss factor of tanδ ≈ 0.07 can be achieved by increasing the relative grain size inside the alloy to d/D ≈ 1, where d and D are the grain diameter and the diameter of a Cu-Al-Mn wire, respectively [6]. Here, the loss factor tanδ is used to express the damping performance, which is defined as the ratio of the loss modulus E″ and the storage modulus E′, and it is measured by dynamic mechanical analysis. Furthermore, at a relative grain size of d/t ≈ 15.4, it was possible to reach values of shape recovery up to 7%, where d and t are the mean grain size and Cu-Al-Mn sheet thickness, respectively [13]. Through the addition of alloying elements, it is possible to increase or decrease the martensitic start temperature Ms, as well as improve the shape memory properties of the Cu-Al-Mn alloy, but at the expense of reducing the cold workability [14]. The achieved improvements in the fabrication and performance of Cu-Al-Mn SMAs, which are making them a more competitive potential alternative to NiTi at lower costs, are being followed by diverse proposals for their application. Energy dissipation improvements via grain size and texture control have led to suggestions for Cu-Al-Mn applications as seismic dampers and isolators for buildings constructed in zones prone to natural disasters such as earthquakes and tsunamis. Proposals for medical guidewires were also made, suggesting that a gradual variation in superelasticity and stiffness from the tip to the end through microstructural control can be achieved, giving them a higher tip flexibility and better pushability and torquability than those of NiTi and stainless steel [15,16]. Initial modeling attempts also indicate that SMAs in the form of thin sheets could find application as structural materials through integration as layers within hybrid composites, where their damping performance could be maximized through the use of pre-strain [17]. The principal goal of the authors in this paper was to identify potential issues in the practical applications of Cu-Al-Mn shape memory alloys (SMAs) in the form of thin sheets and provide an analysis of the effects of the fabrication process which has thus far, to the authors’ awareness, not been investigated. Specifically, this involved examining the thermomechanical treatments necessary for fabricating these sheets from an ingot, including hot rolling, cold rolling, and additional processes like grain growth heat treatments and tensile cycling. The authors discussed how these treatments affected the chemical composition of the SMA sheets and their phase transformation behavior. Furthermore, they highlighted the impact of phosphorus impurities in the alloy on its phase transformation performance, particularly their effect on the concentration of manganese available in solid solution form.

2. Materials and Methods

2.1. Experimental Setup

The casting process of the SMA was performed using two furnaces—an arc melting furnace, and an Ages–Galloni induction melting furnace. The rolling of the alloy was performed using two rolling machines with roll diameters of 30 and 15 cm, respectively. Subsequent heat treatments on the obtained alloy sheets were performed using a Carbolite 3236 GPC 12/36 furnace (Hope, Derbyshire, UK). The tensile cycling of alloy sheet samples, described further in the text, was performed on a calibrated MTS Alliance RF/150 testing system equipped with an environmental chamber.

Microstructural and composition analysis was performed using a Nikon ECLIPSE LV150NL optical microscope (Nishioi, Shinagawa-ku, Tokyo, Japan) and a Zeiss Sigma 500 scanning electron microscope (SEM) (Zeiss, Oberkochen, Germany) equipped with an Oxford Instruments Ultim Max 65 energy-dispersive X-ray detector (EDX) (Oxford Instruments, Abingdon, UK).

The phase transformation capacity and transformation temperatures were determined using a TA Instruments Q25 differential scanning calorimeter (DSC) (TA Instruments, New Castle, DE, USA) as well as an Anton Paar MCR 702 MultiDrive (Smart BioMaterials Consortium, Eindhoven, The Netherlands) rheometer equipped with an environmental chamber.

2.2. Alloy Fabrication and Thermomechanical Treatment to Obtain Thin Cu-Al-Mn SMA Sheets

Since the intended final application of the Cu-Al-Mn thin sheets in question is primarily for vibration damping, the target composition during casting and subsequent thermomechanical treatments were chosen to optimize for damping performance while simultaneously allowing enough cold workability of the alloy that would be necessary to go from an as-cast ingot to an SMA sheet with a thickness below 1 mm. Both the alloy damping capacity and material workability were previously shown to be highly dependent on the ratio of alloying elements Al and Mn presenting within the cast [18]. In terms of damping applications, the intention is to study both active and passive damping application potentials of the Cu-Al-Mn SMA thin sheets. For purposes of active damping, this means that it would be necessary to have a practical way of achieving temperature-induced phase transformation of the SMA, during which a peak in damping capacity, i.e., the loss factor, is observed in the material [5]. As the alloy primarily consists of highly conductive copper, the transformation temperatures of the SMA would need to be close enough to room temperature so that a practical method of triggering such transformation would be possible. However, they also need to be sufficiently above room temperature to ensure that the phase transformation would not be accidentally induced during use. It was determined that this would require the A_s and A_f, austenitic and martensitic phase transformation temperatures, respectively, to be within the range of 50–100 °C. Furthermore, once the ingot was cast, it was necessary to perform thermomechanical treatments to obtain thin sheets. Based on a review of the literature, a Cu-Al-Mn alloy with an aluminum content lower than 18% at. was selected due to its high degree of cold workability, which comes from a reduced degree of order within the material, reaching workability rates up to 70% at 16% at. of aluminum [10,18,19,20]. Based on the required material performance criteria described above, the nominal target composition of the Cu-Al-Mn alloy for the casting process was around Cu-16Al-10Mn % at., i.e., Cu-7.6Al-9.67Mn % wt. [18].

The entire casting process, which consisted of two phases, was performed at the facilities of CNR-ICMATE in Lecco, Italy. During the first phase of the casting process, a Cu-Mn pre-alloy was cast using an arc melting furnace under vacuum due to the high reactivity of manganese with oxygen. Combining the obtained Cu-Mn pre-alloy and the necessary additions of the required quantities of copper and aluminum to obtain the final composition, the second phase was performed in an Ages–Galloni induction melting furnace. Composition measurements were performed after each phase of the casting process using EDX analysis. The final products of the casting process were two ingots of Cu-Al-Mn, each with a thickness of approximately 10 mm. The raw materials used in the casting process, their chemical purity, and form can be found in

Table 1. Raw materials used in the casting process of the Cu-Al-Mn shape memory alloy.

Element	Purity [%]
Cu	99.98 (P~0.04–0.06)
Al	99.99
Mn	99.99

. Within the raw copper, an expected small amount of phosphorus was indicated, since it is commonly used in the copper production processes to avoid copper oxidation.

Once the casting process was finalized and the Cu-Al-Mn ingots were obtained, a set of thermomechanical treatments was needed to reduce the ingots thickness down to a target of ~0.4 mm. These treatments are fully described in the text below. In Figure 1, the isothermal diagram of Cu-Al-Mn that was used as a reference when selecting the temperatures for each thermomechanical treatment that was applied to the material is shown. Additionally, in Figure 1, the point at which the target alloy composition would fall squarely within the secondary β-phase region is around ~900 °C. Hence, this temperature was selected as the betatization and hot-rolling treatment temperature, as marked by the letters HR/BET. Each thermomechanical treatment used during the experimental study will be elaborated on in further text.

According to the type of rolling process applied in the fabrication of the Cu-Al-Mn SMA sheets for this study, two different types of sheets can be distinguished: fully hot-rolled Cu-Al-Mn sheets and Cu-Al-Mn sheets obtained through a combination of both hot- and cold-rolling approaches. In both cases, the first step in the thermomechanical treatment process for the reduction in the Cu-Al-Mn ingot thickness and obtaining the final sheets was a hot-rolling procedure. The procedure involved using a chamber furnace to heat the ingots to a temperature of 900 °C and then subjecting them to lamination using a rolling machine. This was an iterative process used to reduce the thickness of the ingots from ~10 mm to a thickness of ~2 mm. The choice to perform the treatment at 900 °C was to ensure that the rolling was completed while in the presence of the more malleable austenitic β-phase, thus avoiding significant residual stresses and the possible introduction of cracks. Furthermore, this kind of hot-rolling treatment allows for breaking up any dendritic formations remaining from the casting process, which could not be removed by a previous homogenization treatment [21]. At the point of reaching ~2 mm thickness, one piece of the cast alloy was further treated by hot rolling down to ~0.4 mm, using the same iterative process. The second piece of the alloy, however, was further treated by cold rolling. This change in the rolling process was introduced due to a previous study showing that further hot rolling under the presence of material impurities and already low thickness in combination with pressure and high temperatures leads to oxidation of the alloying elements, thus acting on the composition and performance of the SMA [22]. This effect is more deeply evaluated throughout this study and laid out in detail in the Results section. In a similar way to hot rolling, the cold-rolling process was implemented iteratively, and at about 50% thickness reduction from 2 mm toward the final target of 0.4 mm, a 10 min recovery heat treatment at 600 °C was applied to the alloy to alleviate some of the accumulated residual stress and avoid cracking. The cold-rolling stage was finished once the 0.4 mm thickness was achieved. The rolling processes were partially performed at the facilities of ICMATE in Lecco using a rolling mill with a 30 cm roller diameter and partially at the laboratories of Politecnico di Milano using a roller with a 15 cm roll diameter. All further thermomechanical treatments performed on the obtained SMA sheets, such as betatization and quenching, grain growth treatments, and tensile cycling; were performed at the laboratories of Politecnico di Milano, and these are described in detail in further text.

2.3. Experimental Design

The primary goal of this study was to estimate the effects that different types of alloy rolling processes and post-rolling treatments have on the composition of the Cu-Al-Mn SMA sheets, and by extension, on the phase transformation temperatures and overall transformation capacity. In addition to the consideration of the primary elements—copper, aluminum, and manganese, special consideration is given also to the influence of the presence of impurities in the form of phosphorus within the casting. The two rolling approaches used were described in detail in the previous section with the main difference being the use of only hot rolling or a combination of hot and cold rolling. Prior to application, the Cu-Al-Mn SMA sheet samples would be treated by aging to ensure stable phase transformation temperatures [18]. Beyond this post-rolling heat treatment, two additional thermomechanical treatments are considered: abnormal grain growth treatment and tensile cycling treatment (i.e., SMA training).

When dealing with as-cast Cu-Al-Mn with a high material thickness, it is possible to obtain a high degree of grain growth through annealing in the single β-phase region [23]. However, in thinner specimens, such as the sheet in this study, it has been observed that once the grain size reaches about two to three times the sheet thickness, there is a stagnation in further grain growth, which has been attributed to the reduction in the dimensionality of the grain structure of the alloy from a three-dimensional to a two-dimensional structure, at which point the tension balance between the free surface and the grain boundaries causes thermal grooving of the boundaries that in turn pins the grain boundary migration [23]. To obtain high relative grain size in the Cu-Al-Mn SMA thin sheets, the authors subjected sheet samples to an abnormal grain growth (AGG) heat treatment [24,25]. The growth in this case occurs by having a few grains engulf their neighboring grains and increase to considerable dimensions. This can be realized by annealing the Cu-Al-Mn thin sheets to a temperature within the dual phase region α (fcc) + β (bcc), see Figure 1, in this case, up to temperatures up to ~700 °C, then following this up by heating the alloy to the single β-phase region, in this case around 900 °C [24]. These steps of the heat treatment process can be seen in Figure 2. The two annealing treatments were denoted by AGG (abnormal grain growth) and BET (betatization) with each lasting for 10 min and being performed at 550 and 900 °C, respectively. After the betatization, the samples were quenched in cold water. As with all samples, an aging heat treatment at 200 °C for 10 min was performed to stabilize the phase transformation temperatures of the SMA [26]. For each sample, the effect of the heat treatment in achieving grain growth was estimated using a Nikon ECLIPSE LV150NL optical microscope. From the captured images, the relative grain size was calculated as r/t, where r and t are defined as the mean grain size and the SMA sheet thickness, respectively.

The final thermomechanical treatment performed during the study was tensile cycling for SMA training [2]. A pseudoelastic loading cycle was employed, whereby the sample was kept at a constant temperature above the austenitic finish temperature A_f, and then a tensile load was applied to induce phase transformation from the austenitic to the de-twinned martensitic phase. The schematic representation of the loading cycle is given in Figure 3, which is based on the so-called Brinson model used to describe the one-dimensional constitutive behavior of shape memory alloys, with the arrows indicating the changing state of the alloy in the σ-T diagram during loading [27,28]. The cycling was performed at a fixed temperature of ~110 °C, which was chosen based on a previous study that suggests that the reverse martensitic transformation is fully complete for values of temperature below 110 °C [29]. The SMA sheets were prepared for tensile cycling using a microcutting machine. The SMA tensile cycling was performed on a calibrated MTS Alliance RF/150 testing system equipped with an environmental chamber used to maintain the constant temperature during cycling. We loaded samples for a series of tensile cycles from 0 to 450 MPa at a rate of 1 mm/min until the stabilization of the SMA hysteretic cycle [30]. The strain during the loading cycles was measured using a strain gauge with a 25 mm gauge length, while the crosshead distance was set to 50 mm.

In summary, based on the thermomechanical treatment path and final sheet thickness, the following set of Cu-Al-Mn thin sheets was produced:

• Hot-rolled (0.4 mm);
• Hot-rolled (2 mm);
• Hot-rolled (2 mm), cold-rolled (0.4 mm);
• Hot-rolled (2 mm), cold-rolled (0.4 mm), grain growth (AGG);
• Hot-rolled (2 mm), cold-rolled (0.4 mm), grain growth (AGG), tensile cycling.

Once all the samples were treated according to the five thermomechanical paths described above, the final betatization was completed by heating to 900 °C with quenching in cold water. This final step in the treatment process was necessary to activate the shape memory properties of the SMA sheets.

The Cu-Al-Mn sheets were analyzed to determine their chemical composition as well as their phase transformation capacity and phase transformation temperatures. Since multiple different testing methods were involved in this process, each requiring samples of specific shape and size, care was taken that the samples were cut from the same bulk sheets of Cu-Al-Mn as close as possible to each other. The cutting of sample Cu-Al-Mn thin sheets was performed using a microcutting machine. Samples for microscopy were additionally prepared by polishing and then chemically etched using a FeCl₃ + HCl solution. Subsequently, measurements of the chemical composition of samples used in the study were performed by scanning electron microscopy (SEM) and an energy-dispersive X-ray detector (EDX) analysis.

Assessment of the phase transformation capability of the differently treated Cu-Al-Mn SMA sheets and their phase transformation temperatures was performed via differential scanning calorimetry (DSC) and dynamic mechanical analysis (DMA). During DSC analysis, phase transformations are marked by the measured changes/peaks in heat flow. Conversely, during DMA analysis, phase transformations were measured through changes in the material’s damping capacity, expressed through the loss factor tanδ, which is calculated as the ratio between the loss and storage moduli E″ [MPa] and E′ [MPa], respectively. During phase transformation, SMAs such as Cu-Al-Mn are expected to show a peak in damping capacity. Both in DSC and DMA tests, care was taken to avoid reaching temperatures above 200 °C, as previous studies have shown that temperatures in the range of ~250 to 300 °C cause martensitic stabilization and a loss of transformation capacity [22,31]. DSC testing was performed using samples with mass ranging in the interval between 20 and 100 mg, in a temperature range of −75 to 200 °C with a heating/cooling rate of 10 °C/min. On the other hand, each DMA test was performed in the so-called temperature sweep mode. Parameters such as strain amplitude, tensile preload, and oscillation frequency were fixed during each test, while a temperature ramp was applied in the range of −20 to 120 °C with a heating rate of 2 °C/min. The dynamic test on the samples was performed in both tension and torsion with strain amplitudes of 0.001 and 0.003%, respectively.

In the Results section, the relationship between the chemical composition, thermomechanical history, and the phase transformation performance of the Cu-Al-Mn sheets samples is explored.

3. Results and Discussion

As defined in the prior sections, the nominal target composition for the Cu-Al-Mn shape memory alloy that was developed within this study was Cu-16Al-10Mn % at., which is due to the expectation of transformation temperatures within the desired range of 50–100 °C and a high level of workability due to a low level of aluminum content. A qualitative assessment of the microstructure for each SMA sheet was performed. This assessment was completed using optical microscopy to confirm that the obtained microstructure was indeed martensitic. An example of a successful martensitic transformation is provided in Figure 4. The laminar structure, which is characteristic of martensite, was dominant and observed in all samples within the study. Some composition measurements were also performed by generating EDX maps of the samples, which are laid out further in the text. Along with all the composition measurements performed on the Cu-Al-Mn SMA sheets, DSC analysis was used to evaluate the phase transformation capability and the transformation temperatures of the sheets. Additionally, several measurements were also performed using a DMA machine. Both the measured chemical compositions and the transformation properties of the obtained samples are summarized in

Table 2. Cu-Al-Mn shape memory alloy sheet samples produced, along with chemical compositions estimated by EDX analysis, phase transformation temperatures measured by DSC and DMA, and thermomechanical treatments performed to obtain them.

	Composition [% at.]			Transformation Temperatures [°C]
Sample [%]	Mn	Al	Cu	As	Af	Ms	Mf	Treatment
1	8.7	16.9	74.3	nontransforming				HR1 1
2	8.8	17.1	74.1	nontransforming
3	8.9	17.9	73.3	nontransforming
4	9.5	17.8	72.8	nontransforming
5	9.6	17.6	72.7	nontransforming
6	9.9	17.65	72.5	nontransforming
7	9.9	17.3	72.8	34.4	81.19	/ 6	/
8	10.04	16.9	73.1	27.06	71.21	/	/
9	10.2	16.9	73.1	91.69	109.03	83.23	66.4
10	10.3	17	72.7	59.06	93.12	66.41	−0.47	HR2 2
11	10.8	16.9	72.2	35.83	67.19	59.06	17.85
12	11.1	15.8	73.1	43.1	57.6	30.88	15.21
13	11.1	16.9	72	43.10 7	57.6	30.88	15.21
14	10.4	17.6	72	73.85	99.46	70.08	7.68	HR/CR 3
15	10.5	17.5	72	56.64	89.97	65.44	22.1
16	10.8	18.3	70.9	37.01	48.61	26.91	12.64
17	10.85	18	71.1	41.76	58.82	21.88	2.82
18	10.96	18.2	70.9	44.08	60.38	38.62	26.5
19	11.1	16.4	72.5	42.15	49.4	21.74	16.18
20	11.2	16.7	72.1	32.77	43.74	15.56	3.5
21	10.3	18	71.7	11.01	28.53	15.41	2.71	HR/CR/AGG 4
22	10.3	18.2	71.5	nontransforming
23	10.5	17.8	71.8	nontransforming				HR/CR/AGG/T 5

¹ Hot-rolled (0.4 mm). ² Hot-rolled (2 mm). ³ Hot-rolled (2 mm), cold-rolled (0.4 mm). ⁴ Hot-rolled (2 mm), cold-rolled (0.4 mm), abnormal grain growth treatment. ⁵ Hot-rolled (2 mm), cold-rolled (0.4 mm), abnormal grain growth treatment, and tensile cycling (“training“). ⁶ DMA tests were performed with a heating ramp, so only the austenitic transformation was recorded. ⁷ Two samples for composition measurement from the same bulk and one sample for the DSC.

along with the corresponding thermomechanical treatment for each sample. The DSC analysis was performed within the range of −75 and 200 °C; hence, any samples that did not display signs of phase transformation within this range were considered “nontransforming”. The lower temperature limit during DSC testing was based on the machine limitations. However, the upper limit temperature was selected because prior studies suggest that heating above the temperature range of about 240 to 300 °C results in martensitic stabilization and complete loss of the shape memory effect (SME) [7,22,23]. In Figure 5, an example of a DSC test performed on a specimen of Cu-Al-Mn sheet is given.

Looking at the data presented in Table 2, where the samples are sorted according to their respective manganese content and treatment histories, it is possible to see that the phase transformation temperatures of the Cu-Al-Mn SMA sheets are highly sensitive to small changes in the concentrations of alloying elements. The presence of the shape memory effect, or lack thereof, indicated a dependence on the ratio between the measured levels of manganese and aluminum within the Cu-Al-Mn alloy samples, respectively. Based on this observation, in Figure 6, it was possible to group samples according to their transformation capacity which was estimated through DSC/DMA testing. As a rule of thumb, it appears that samples with a ratio of Mn/Al which was above ~0.62 retained a persistent shape memory effect across multiple DSC cycles within the range of −75 and 200 °C. Conversely, samples for which the ratio of Mn/Al was below ~0.57 showed no transformation capacity within the above-mentioned range of testing temperatures. The Mn/Al ratio range between 0.57 and 0.62 of the was indicated separately as the range of “inconsistent” phase transformation performance. Testing of the samples within this composition ratio range has resulted in samples with both persistent phase transformation capability as well as those without any clear signs of transformation. Furthermore, three samples initially showed phase transformation behavior; however, it did not persist with further DSC cycling. These samples are separately highlighted as “inconsistent” in Figure 6. Additionally, an example of the inconsistent transformation behavior described above is given in Figure 7.

The sensitivity of manganese to oxidation at high temperatures has presented an issue for the Cu-Al-Mn SMA fabrication since the initial casting process, as mentioned in the section on alloy fabrication.

In Figure 8, the Cu-Al-Mn sheet samples were presented according to their manganese content and the applied thermomechanical treatment path. The first group of samples, from 1 to 9, present with the highest reduction in manganese content. This group of samples was obtained by fully hot rolling the initial ingot of ~10 mm thickness down to a thickness of 0.4 mm. The loss of manganese, and consequently the loss of SME, within this group of samples was attributed to a combination of the very low sample thickness and high temperatures (900 °C) between each rolling pass along with the rolling pressure itself. In contrast, the second group of samples, 10 to 13, which like the previous was obtained from the same ingot, was treated by hot rolling from the initial ~10 mm thickness down to 2 mm. In this case, the reduction in manganese content was drastically lower than in the previous group with the largest difference being almost ~2.25% at. of Mn. Furthermore, the effect of high temperatures in the later phases of rolling, between 2 and 0.4 mm, is further highlighted by the manganese content and SME effect found in the third group of samples, 14 to 20, which was treated with hot rolling down to 2 mm and subsequently cold-rolled the rest of the way to 0.4 mm. In the case of this group, the reduction, or the lack thereof, is comparable to that of the second group of samples. The final two groups of samples were both subjected to a combination of hot and cold rolling in the same way as the third group. However, in addition to these thermomechanical treatments, they were also thermally treated to increase the relative grain size within the sheets. Another significant difference is that while these samples were obtained using the same thermomechanical process as the previous sample groups, with the exception of the grain growth treatment, they came from the second ingot of the Cu-Al-Mn alloy. The samples within these two groups show inconsistent shape memory effect performance, which is in line with their estimated Mn/Al ratios of 0.57, 0.57, and 0.59, respectively. The samples can be found in Figure 6, where they are numbered as samples 21, 14, and 11, respectively. However, the second ingot from which the samples were derived had an estimated average manganese concentration of 10.43% at and a Mn/Al ratio of 0.59. Since these initial values are quite similar to the manganese concentration and Mn/Al ratio reported for the final samples in Figure 6 and Figure 8, it is challenging to assess the impact of the grain growth thermomechanical treatment on the shape memory effect. Therefore, a more in-depth investigation is necessary for samples with this thermomechanical treatment history.

As previously mentioned, the raw materials used in the casting process of the Cu-Al-Mn SMA ingot, specifically copper, contained trace amounts of phosphorus. During the EDX analysis of the Cu-Al-Mn sheets obtained at the end of their respective thermomechanical treatment paths, in a number of them, traces of phosphorus were detected, ranging from 0.04 up to 0.27% at. The specific samples and their corresponding compositions, with phosphorus also considered, are given in

Table 3. Average chemical composition of samples in which the presence of phosphorus was detected.

Sample Compositions Considering Phosphorus
Sample [#]	Mn [% at.]	Al [% at.]	Cu [% at.]	P [% at.]
1	8.8	17.9	73.2	0.2
2	8.9	16.6	74.3	0.3
3	9.5	17.8	72.6	0.1
4	9.7	17.4	72.8	0.2
5	9.9	17.6	72.4	0.1
6	9.9	17.2	72.6	0.2
7	10.2	17.3	72.3	0.1
8	10.5	17.5	71.9	0.1
9	11.0	16.4	72.4	0.1
10	11.0	15.8	73.0	0.1
11	11.1	16.9	71.9	0.1
12	11.2	16.7	72.0	0.1

.

The presence of phosphorus at these concentrations would not necessarily be an issue and might have been overlooked in other cases. However, SEM and EDX analyses revealed numerous inclusions within the samples, which were primarily composed of phosphorus and manganese, with some containing more copper and very little aluminum. Figure 9 shows an example of the SEM/EDX analysis, where the inclusions are visible in the electron image (Figure 9a), and the concentrations of manganese and phosphorus are highlighted in the EDX maps (Figure 9c,d). After identifying these inclusions, multiple samples were analyzed to estimate the average composition of the particles, and the results are shown in Figure 10. The main concern is that the high manganese content in these particles means it is not present in the solid solution form of the shape memory alloy, potentially affecting its performance.

For the purposes of making the most conservative estimate on the effect that the presence of such inclusions could potentially have on the amount of manganese that is trapped within them rather than being inside the solid solution, a couple of assumptions were made. The first assumption supposes that the ratio between the atomic percentage of a given element inside the inclusions and the atomic percentage of that same element in the entire sample inclusions is equal across all the elements. This assumption was given as (Equation (1)), where P_inc and P_bound stand for the atomic percentage of phosphorus within the inclusion itself and the atomic percentage of phosphorus within the entire samples, which is bound to the inclusion, respectively. The same notation is applied to the remaining elements. The second assumption was that the entirety of the phosphorus measured within the samples was there in the form of the mentioned inclusions. This was based on observations via SEM/EDX, which did not show measurable levels of phosphorus in the areas outside of the mentioned inclusions. This assumption was given as (Equation (2)), where P_tot refers to the total amount of phosphorus measured in the sample. Combining these two assumptions, it was possible to roughly estimate the atomic percentage of Cu, Al, and Mn bound within the inclusions according to the equations (Equations (3)–(5)) given below.

$${\displaystyle \frac{{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}{{\mathrm{P}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}}}={\displaystyle \frac{{\mathrm{M}\mathrm{n}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}{{\mathrm{M}\mathrm{n}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}}}={\displaystyle \frac{{\mathrm{A}\mathrm{l}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}{{\mathrm{A}\mathrm{l}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}}}={\displaystyle \frac{{\mathrm{C}\mathrm{u}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}{{\mathrm{C}\mathrm{u}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}}}$$

(1)

$${\mathrm{P}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}={\mathrm{P}}_{\mathrm{t}\mathrm{o}\mathrm{t}}$$

(2)

$${\mathrm{M}\mathrm{n}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}={\displaystyle \frac{{\mathrm{M}\mathrm{n}}_{\mathrm{i}\mathrm{n}\mathrm{c}}·{\mathrm{P}}_{\mathrm{t}\mathrm{o}\mathrm{t}}}{{\mathrm{P}}_{\mathrm{p}\mathrm{a}\mathrm{r}\mathrm{t}}}}$$

(3)

$${\mathrm{A}\mathrm{l}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}={\displaystyle \frac{{\mathrm{A}\mathrm{l}}_{\mathrm{i}\mathrm{n}\mathrm{c}}·{\mathrm{P}}_{\mathrm{t}\mathrm{o}\mathrm{t}}}{{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}}$$

(4)

$${\mathrm{C}\mathrm{u}}_{\mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}}={\displaystyle \frac{{\mathrm{C}\mathrm{u}}_{\mathrm{i}\mathrm{n}\mathrm{c}}·{\mathrm{P}}_{\mathrm{t}\mathrm{o}\mathrm{t}}}{{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{c}}}}$$

(5)

Based on the EDX measurements of the overall sample compositions and the estimated average compositions of the inclusions, given in Figure 10, applying the approach laid out above, an estimate of the concentrations of Al, Mn, and Cu bound inside the inclusions was made as well as the difference with respect to the measurements in which the presence of phosphorus was simply neglected. The resulting estimates can be found in

Table 4. Estimates of aluminum, manganese, and copper bound inside the inclusions, as well as in solid solution, and the difference in composition with respect to measurements that do not consider phosphorus.

Sample	Average Percentage Bound to P			Average Al, Mn, Cu in Solid Solution			Average Composition Change
[#]	Al	Mn	Cu	Al	Mn	Cu	Al	Mn	Cu
1	0.011	0.4	0.04	18.0	8.4	74	0.1	−0.4	1
2	0.031	0.59	0.165	16.735	8.39	74.88	0.135	−0.51	0.38
3	0.0050	0.184	0.022	17.8505	9.345	72.804	0.0505	−0.155	0.104
4	0.021	0.41	0.09	17.487	10.84	73.16	0.087	1.14	0.26
5	0.18	0.4	1.4	16.6	10.8	72.6	0.2	−0.2	0.0
6	0.01250	0.221	0.079	16.757	11.025	72.220	0.057	−0.175	0.120
Sample	St. dev. of Percentage Bound to P			St. dev. of Al, Mn, Cu in Solid Solution			St. dev. of the Composition Change
[#]	Al	Mn	Cu	Al	Mn	Cu	Al	Mn	Cu
1	0.006	0.1	0.03	0.5	0.4	1	0.9	0.7	2
2	0.001	0.01	0.008	0.004	0.01	0.02	0.004	0.01	0.02
3	0.0003	0.002	0.001	0.0008	0.002	0.003	0.0008	0.002	0.003
4	0.002	0.01	0.01	0.006	0.01	0.02	0.006	0.01	0.02
5	0.04	0.1	0.3	0.1	0.1	0.6	0.1	0.1	0.6
6	0.0005	0.004	0.003	0.002	0.005	0.009	0.002	0.005	0.009

.

Looking specifically at the difference in aluminum and manganese concentrations between the two cases, which vary from 0.05 to 0.2% at., and from 0.155 up to as much as 1.14% at., respectively, indicates that the presence of phosphorus inside the Cu-Al-Mn SMA sheets, even in such low quantities as 0.1% at., might have a significant impact on their phase transformation temperatures and/or transformation capacity due to the alloy’s high sensitivity to small compositional variations.

4. Conclusions

• Hot-rolling thermomechanical treatment for Cu-Al-Mn sheets introduces manganese depletion if it is applied at low sheet thicknesses (below 2 mm) and is caused by oxidation.
• The increase in manganese oxidation with thickness reduction is attributed to an increase in the surface area of the alloy placed into contact with the open air during the hot-rolling process as the sheet thickness decreases.
• Changes in the Mn/Al ratio in the alloy composition due to Mn oxidation result in the loss of the SME once the ratio drops below approximately 0.57.
• The application of cold rolling after sheets reach lower levels of thickness ensures preservation of the chemical composition, thus preserving the SME.
• The presence of phosphorus impurities, even in low amounts of 0.04–0.06 wt.%, suggests a possibility of significantly altering the expected alloy performance by binding large amounts of manganese and thus preventing them from being in the solid solution.

Future research efforts are planned to investigate more precisely the point at which thermomechanical treatment steps can cause manganese depletion inside the alloy for both hot rolling and grain growth thermal treatments. Furthermore, a more detailed study to determine the extent to which the phosphorus inclusions alter the composition of the solid solution is warranted.