Investigation of Al-B 4 C Metal Matrix Composites Produced by Friction Stir Additive Processing

: Aluminium—boron carbide metal matrix composites (Al-B 4 C MMCs) belong to the class of materials extensively used in the nuclear industry as a thermal neutron absorber in spent fuel casks. This article investigates a novel production method of Al-B 4 C MMCs—Friction Stir Additive Processing (FSAP)—as an alternative production method to casting or sintering. FSAP is derived from friction stir welding, which can be used to local modiﬁcations of microstructure, or it can be used to incorporate the second phase into the processed material. During this study, a variant of FSAP for MMC production was proposed, and its mechanical and thermal neutron absorbing properties have been investigated. Further, the inﬂuence of neutron irradiation on mechanical properties has been studied. Results show that FSAP can successfully produce Al-B 4 C MMCs with 7 mm thickness. Neutron irradiation causes only a slight increase in hardness, while its effect on tensile properties remains inconclusive.


Introduction
Aluminium-boron carbide metal matrix composites (Al-B 4 C MMCs) belong to a class of advanced materials which is extensively used in the nuclear industry. Al-B 4 C MMCs are often a material of choice for the fabrication of baskets in spent fuel casks. The main task of the basket is to maintain subcriticality of the fuel assembly by means of thermal neutron absorption [1][2][3]. Neutron absorption is facilitated by isotope 10 B with a large neutron absorption cross section (3850 barn at 25.3 meV). Isotope 10 B also has relevant natural abundance (19.8%) which makes it attractive for large-scale use [1]. There are some alternatives to B 4 C for use in neutron absorbers; however, each alternative has its own challenges and disadvantages. In [4] they attempted to substitute the B 4 C with Gd 2 O 3 motivation to reduce potential swelling, which can be present in Al-B 4 C MMCs. Nevertheless, Gd 2 O 3 is not chemically inert in the presence of aluminium and forms a multitude of intermetallic phases. Also, Gd 2 O 3 nanoparticles are potentially hazardous substances since they can deposit in the human body [5]. In [6], the use of Al-Mg-Sm alloy in combination with B 4 C was proposed. Samarium can substitute B 4 C to a certain extent, but it is difficult to maintain samarium-aluminium solid solution, due to the large difference between their atomic masses.
There are several ways to produce Al-B 4 C MMCs (e.g., stir casting and sintering). Liquid stir casting, as well as sintering, enables large quantity production of MMCs, but there are several issues with regards to these methods [7]. The production of MMCs using stir casting might lead to non-uniform particle distribution. Porosity [8], interfacial reactions [9] or poor wetting between aluminium and boron carbide [10] can have a negative impact on the performance of the final product. Sintering techniques such as hot pressing do also have some porosity [11], but they can be utilized to produce MMCs with more homogeneous particle distribution. Interfacial reaction can be present even during sintering but choosing suitable technique (such as hot isostatic pressing) both interfacial reaction and porosity can be removed [12]. During sintering aluminium and boron carbide must be in particle form which makes this process more expensive.
Friction Stir Additive Processing (FSAP) is a novel solid-state process for production of MMCs. FSAP is a derivation from Friction Stir Welding (FSW) [13]. The original FSW process uses a special rotating tool composed of a shoulder and a pin. This tool is being plunged into the processed material. Heat generation during welding occurs through friction between the tool and the material, which results in plasticization of the material. Plasticization of metal, mixing action and compression during FSW can be utilized for microstructure modification (Friction stir processing-FSP) or production of MMCs (Friction stir additive processing-FSAP) [13,14]. During FSAP, particles are introduced into pre-machined grooves in base material, and subsequently an FSW-like process is performed along the grooves. Using frictional heat, mixing action and compression, a MMC layer is formed. Prior to processing with a tool composed of pin and shoulder, processing with a pinless tool can be used to deform the uppermost layer of matrix material and to close the grooves, to prevent particle ejection from the grooves during the process itself [14].
Comparable to FSW [15], FSAP has three distinct microstructural regions, namely heat affected zone (HAZ), thermo-mechanically affected zone (TMAZ) and the stir zone. Stir zone is characterized by fine equiaxed grains caused by severe plastic deformation and continuous dynamic recrystallization, geometric dynamic recrystallization or static recrystallization and grain growth. TMAZ is directly adjacent to the stir zone, however it experiences deformation at elevated temperature to a lower extent, and thus no dynamic recrystallization takes place. Higher dislocation density and formation of sub-grains can be observed. HAZ experiences only effects of heat cycles, which can result in local aging or annealing [16].
FSAP and related FSW have been investigated over a few decades, however only little research has been done with regards to thermal neutron shielding or radiation damage. In [22], the effects of radiation (Fe/He ions) have been investigated on friction stir spot welds (FSSW) between oxide dispersion strengthened (ODS) nanostructured steel plates. Nanoindentation tests on steel joints have shown that mainly TMAZ has a certain tendency towards radiation hardening due to formation of dislocation loops. In [22] similar findings have been reported for dissimilar FSSW welds between ODS steel and stainless steel. However, these results cannot be simply transferred to Al-B 4 C system produced by FSAP due to different material properties as well as partially different microstructural features of FSSW and FSAP.

Materials and Methods
Solution treated and artificially aged 8 mm thick plates of EN AW-6060 T66 (Kloeckner Metals Austria GmbH & Co KG, Vienna, Austria) with chemical composition, listed in Table 1, has been chosen as a matrix material. Natural B 4 C particles (at least 20% of 10 B from total boron content, supplied by Alfa Aesar, Ward Hill, MA, USA) with a particle size of 22-59 µm were used as a reinforcement. Chemical composition of B 4 C is shown in Table 2. Tool design and process parameters are summarized in Table 3. Design of processed plate is shown in Figure 1. 1, has been chosen as a matrix material. Natural B4C particles (at least 20% of 10 B from total boron content, supplied by Alfa Aesar, Ward Hill, MA, USA) with a particle size of 22-59 µm were used as a reinforcement. Chemical composition of B4C is shown in Table 2. Tool design and process parameters are summarized in Table 3. Design of processed plate is shown in Figure 1.  Based on extensive DoE investigation, a three-pass FSAP process was selected [25]. DoE investigation was focused on finding optimal distance between process runs stirring the particles into the matrix, as well as the influence of processing direction with respect to Based on extensive DoE investigation, a three-pass FSAP process was selected [25]. DoE investigation was focused on finding optimal distance between process runs stirring the particles into the matrix, as well as the influence of processing direction with respect to particle distribution and number of defects. Since the material flow during FSAP is asymmetrical with respect to the advancing and retreating side, we can expect to have different results of MMC formation on those sides. Advancing side denotes that traverse direction and rotational direction are pointing alongside, whereas on the retreating side traverse direction and rotational direction are oriented inversely, leading to different resulting relative velocities. Putting the advancing side on grooves while keeping 5 mm distance between process runs has been shown as an optimal processing route for defined groove and tool geometry. Exact processing route is depicted in Figure 2.
asymmetrical with respect to the advancing and retreating side, we can expect to have different results of MMC formation on those sides. Advancing side denotes that traverse direction and rotational direction are pointing alongside, whereas on the retreating side traverse direction and rotational direction are oriented inversely, leading to different resulting relative velocities. Putting the advancing side on grooves while keeping 5 mm distance between process runs has been shown as an optimal processing route for defined groove and tool geometry. Exact processing route is depicted in Figure 2. Specimens for metallographic examination have been cut using metallographic precision saw and mounted using cold mounting resin. They have been ground flat for at least 5 min for each grinding stage with pressing force 30 N using grinding papers with grits 120, 320, 600, 1200 and 2500. To enhance the contrast between B4C and Al matrix, etching with 20% NaOH aqueous solution (CRIDA Chemie, Wenden, Germany) was performed for 90 s.
Vickers hardness measurements (1 kg load was applied for 15s) were performed on metallographic cross-sections. Hardness profile was measured along three lines (each 30 points, 1 mm distance between points) in different depths from the top surface (1 mm, 3.5 mm, and 6 mm). Top surface is defined as the surface where grooves are milled. Hardness was measured prior to irradiation (machine from EMCO GmbH, Salzburg, Austria) as well as after irradiation (machine from Bruker Corp., Billerica, MA, USA). Tensile samples were manufactured according to Figure 3 using electrical discharge machining. Total 3 samples were produced where 1 was tested without irradiation and 2 were tested after irradiation. Uniaxial tensile tests were performed on universal tensile testing machine (Instron, Norwood, MA, USA) Crosshead speed during testing was set to 1.3 mm/min.
Absorption of thermal neutrons has been measured at horizontal channel HK1 of the research reactor LVR-15 in research facility CV Řež, Ltd., Husinec, Czech Republic. Detailed parameters of HK1 are listed in [26]. Detection of incident neutron fluence has been facilitated by in alloyed aluminium (Al + 1% In), while for detection of transmitted neutron fluence pure In was used. Both detectors had circular shape (0.1 mm thick, diameter 10 mm) and were placed in the middle of the samples. Indium detectors utilize reaction 115 In(n,γ) 116m In to convert neutron fluence into γ rays. Gamma rays can be detected by means of high-purity germanium (HPGe) detectors and from γ count rates neutron flux can be calculated. To eliminate effects of reflected neutrons, Cd shielding was placed behind the samples. Samples were placed in the channel for 1 h. Measurement of induced activity of detectors started during 1 h after retrieving from channel and was performed for 2000 s using HPGe detectors. From induced activity we can determine incident and transmitted neutron fluences.
Irradiation of tensile samples was performed in the research reactor LVR-15 in CV Řež. Samples were irradiated for 74 s at reactor power output of approximately 100 kW. Specimens for metallographic examination have been cut using metallographic precision saw and mounted using cold mounting resin. They have been ground flat for at least 5 min for each grinding stage with pressing force 30 N using grinding papers with grits 120, 320, 600, 1200 and 2500. To enhance the contrast between B 4 C and Al matrix, etching with 20% NaOH aqueous solution (CRIDA Chemie, Wenden, Germany) was performed for 90 s.
Vickers hardness measurements (1 kg load was applied for 15s) were performed on metallographic cross-sections. Hardness profile was measured along three lines (each 30 points, 1 mm distance between points) in different depths from the top surface (1 mm, 3.5 mm, and 6 mm). Top surface is defined as the surface where grooves are milled. Hardness was measured prior to irradiation (machine from EMCO GmbH, Salzburg, Austria) as well as after irradiation (machine from Bruker Corp., Billerica, MA, USA). Tensile samples were manufactured according to Figure 3 using electrical discharge machining. Total 3 samples were produced where 1 was tested without irradiation and 2 were tested after irradiation. Uniaxial tensile tests were performed on universal tensile testing machine (Instron, Norwood, MA, USA) Crosshead speed during testing was set to 1.3 mm/min. Temperature of samples did not rise during irradiation above the ambient temperature (approx. 20 °C). Goal was to achieve a fluence of 1.5 to 2 × 10 13 n/cm 2 in energies above 100 keV [27] (cumulative total fluence after approx. 300 years in dry cask [1]). Due to the emission spectrum of the neutron source, irradiation by lower energy neutrons was inevitable. Reached neutron fluences are shown in Table 4.

Metallography and Hardness Measurement
Metallographic cross section of Al-B4C MMC after three pass processing is shown in Figure 4 below.  neutron fluence pure In was used. Both detectors had circular shape (0.1 mm thick, diameter 10 mm) and were placed in the middle of the samples. Indium detectors utilize reaction 115 In(n,γ) 116m In to convert neutron fluence into γ rays. Gamma rays can be detected by means of high-purity germanium (HPGe) detectors and from γ count rates neutron flux can be calculated. To eliminate effects of reflected neutrons, Cd shielding was placed behind the samples. Samples were placed in the channel for 1 h. Measurement of induced activity of detectors started during 1 h after retrieving from channel and was performed for 2000 s using HPGe detectors. From induced activity we can determine incident and transmitted neutron fluences.
Irradiation of tensile samples was performed in the research reactor LVR-15 in CV Rež. Samples were irradiated for 74 s at reactor power output of approximately 100 kW. Temperature of samples did not rise during irradiation above the ambient temperature (approx. 20 • C). Goal was to achieve a fluence of 1.5 to 2 × 10 13 n/cm 2 in energies above 100 keV [27] (cumulative total fluence after approx. 300 years in dry cask [1]). Due to the emission spectrum of the neutron source, irradiation by lower energy neutrons was inevitable. Reached neutron fluences are shown in Table 4.

Metallography and Hardness Measurement
Metallographic cross section of Al-B 4 C MMC after three pass processing is shown in Figure 4 below.   Figure 2. Groove geometry is depicted black, with higher particle concentration circled yellow. Area with lower particle concentration is circled green.
In Figure 4 we can see that the chosen process is able to successfully incorporate particles into the matrix. Frictional heat generated by tool and mixing action was sufficient to create MMC reinforced volume. No remnants of grooves or tunnel defects are visible. Chosen tool and process parameters (rotational speed; processing speed) promote a downward directed material flow since one can observe particles even around 7 mm beneath the surface, while grooves are only 3.5 mm deep. The width of the MMC layer is not constant since material in the upper area experiences additional shoulder driven material flow which enlarges the TMAZ size. Particle distribution is not entirely homogeneous, but it occurs in typical FSW/FSAP structure-so called onion rings. We can also observe differences in particle  Figure 2. Groove geometry is depicted black, area with higher particle concentration circled yellow. Area with lower particle concentration is circled green.
In Figure 4 we can see that the chosen process is able to successfully incorporate particles into the matrix. Frictional heat generated by tool and mixing action was sufficient to create MMC reinforced volume. No remnants of grooves or tunnel defects are visible. Chosen tool and process parameters (rotational speed; processing speed) promote a downward directed material flow since one can observe particles even around 7 mm beneath the surface, while grooves are only 3.5 mm deep. The width of the MMC layer is not constant since material in the upper area experiences additional shoulder driven material flow which enlarges the TMAZ size. Particle distribution is not entirely homogeneous, but it occurs in typical FSW/FSAP structure-so called onion rings. We can also observe differences in particle density in the processed volume. In the upper layers we can see darker areas (circled yellow), which suggests higher particle density than in the root area (circled green). Hardness measurements in Figure 5a show a similar trend.  Table 5). x: distance from the axis of the R3 tool (see Figure 2). At 1 mm depth, the non-irradiated cross section shows a slight hardness increase on the overlap of advancing sides of process run 1 and process run 3, which can be again assigned to locally higher particle concentrations as we can see in Figure 5.
Such hardness increase is even observable at 3.5 mm depth of non-irradiated cross section, again due to higher particle concentration caused by overlapping advancing sides. The average hardness of each line can be seen in Table 5. We can see that average hardness can be correlated to particle density which can be observed in Figure 4. Upper layers have a darker appearance, which suggests higher particle density. Across the height of the sample, particle density is getting gradually lower, and the average hardness is also following this trend.
Hardness measurement after irradiation shows quite homogeneous profile with slightly increased values across the whole depth profile.  Table 5). x: distance from the axis of the R3 tool (see Figure 2). At 1 mm depth, the non-irradiated cross section shows a slight hardness increase on the overlap of advancing sides of process run 1 and process run 3, which can be again assigned to locally higher particle concentrations as we can see in Figure 5.
Such hardness increase is even observable at 3.5 mm depth of non-irradiated cross section, again due to higher particle concentration caused by overlapping advancing sides. The average hardness of each line can be seen in Table 5. We can see that average hardness can be correlated to particle density which can be observed in Figure 4. Upper layers have a darker appearance, which suggests higher particle density. Across the height of the sample, particle density is getting gradually lower, and the average hardness is also following this trend.
Hardness measurement after irradiation shows quite homogeneous profile with slightly increased values across the whole depth profile. On the other hand, we must consider the hardness of the received base material (74 HV); we were not able to reach it even at areas with increased particle concentration. The T66 condition of base material is characterized by very fine nm-sized Mg-Si precipitates. Distribution and size of Mg-Si precipitates is usually changed, which results in loss of hardness [28]. Reinforcement particles were in our case in µm range and thus they were not able to substitute T66 heat treatment.

Thermal Neutron Absorption and Tensile Tests
Thermal neutron absorption can be described using the following Equation [12] where I 0 and I denote intensity of incident and transmitted neutron flux, respectively, Σ is the total macroscopic absorption cross section (material property) and d is the thickness of the material. Results of measurement are shown in Table 6. Table 6. Summary of thermal neutron absorption measurement. In order to set the benchmarks for the FSAP produced MMC, the determined cross section needs to be compared with different available materials. Such comparison is shown in Table 7 below. One can see that similar absorption properties to stainless steel-based materials or NiGd alloys are possible, but it has a lower absorption cross section than commercially available Al-B 4 C MMC. Tensile tests were conducted to measure the ultimate tensile strength and elongation at fracture of Al-B 4 C MMCs produced by FSAP. Al-B 4 C MMCs produced by FSAP are candidate for structural application in spent fuel casks, where high strength as well as high elongation at fracture are required. Stability of those properties under neutron irradiation (resistance to radiation hardening) is also required.
Specimens were taken out from the center of the stir zone of the process run 3 according to Figure 3. One specimen was tested in as-processed state, and two specimens were tested in irradiated state. Results of tensile tests are shown in Figure 6.
Specimens were taken out from the center of the stir zone of the process run 3 according to Figure 3. One specimen was tested in as-processed state, and two specimens were tested in irradiated state. Results of tensile tests are shown in Figure 6.
Tensile tests show generally ductile behavior, with elongation at fracture above 15% and ultimate tensile strength slightly below 200 MPa. These properties can be assigned to recrystallized microstructure which is a typical feature of the stir zone.
Results of tensile testing do not show clear effects of radiation hardening. Set of tensile specimens showed similar ultimate tensile strength, Rm, and elongation at fracture A, regardless of the irradiation state (Table 8).

Discussion
In [14] an overview of common MMC layer thickness produced by FSAP was discussed. Layer thickness produced by FSAP is mostly below 5 mm, while using techniques described in this paper, production of FSAP MMCs with thickness up to 7 mm is possible.
FSAP is often intended to improve the wear properties of components [29]. Wear is a surface phenomenon; one does not need to introduce reinforcement into the bulk of the components. However, for applications in the nuclear industry (e.g., neutron absorbers, baskets for spent nuclear fuel), bulk MMCs with thickness up to 10 mm are used [2,30]. Moreover, basket structures for spent nuclear fuel use aluminium cladded MMCs to prevent corrosion and boron loss [31]. MMCs produced by the FSAP method show a similar featurefrom below there is a layer of aluminium base material (approximately 1 mm thick) which remained unaffected during processing, and from above there is a layer of aluminium with low particle content which was formed by the tool shoulder.
Hardness measurement and tensile testing were performed in order to investigate the mechanical properties of the FSAP MMCs. Hardness values of the base material could not be Tensile tests show generally ductile behavior, with elongation at fracture above 15% and ultimate tensile strength slightly below 200 MPa. These properties can be assigned to recrystallized microstructure which is a typical feature of the stir zone.
Results of tensile testing do not show clear effects of radiation hardening. Set of tensile specimens showed similar ultimate tensile strength, R m , and elongation at fracture A, regardless of the irradiation state (Table 8).

Discussion
In [14] an overview of common MMC layer thickness produced by FSAP was discussed. Layer thickness produced by FSAP is mostly below 5 mm, while using techniques described in this paper, production of FSAP MMCs with thickness up to 7 mm is possible.
FSAP is often intended to improve the wear properties of components [29]. Wear is a surface phenomenon; one does not need to introduce reinforcement into the bulk of the components. However, for applications in the nuclear industry (e.g., neutron absorbers, baskets for spent nuclear fuel), bulk MMCs with thickness up to 10 mm are used [2,30]. Moreover, basket structures for spent nuclear fuel use aluminium cladded MMCs to prevent corrosion and boron loss [31]. MMCs produced by the FSAP method show a similar feature-from below there is a layer of aluminium base material (approximately 1 mm thick) which remained unaffected during processing, and from above there is a layer of aluminium with low particle content which was formed by the tool shoulder.
Hardness measurement and tensile testing were performed in order to investigate the mechanical properties of the FSAP MMCs. Hardness values of the base material could not be reached, but slight peaks in hardness could be correlated with increased particle concentration in the cross section. Hardness values vary slightly across the depth of processed volume, which can also be attributed to changing particle concentration.
Irradiated samples show a slight increment in the average hardness. Hardness increase can be accounted for by changes in microstructure such as the formation of dislocation loops, similarly as mentioned in [22]. Tensile tests show that FSAP MMCs have ductile behavior with elongation at fracture over 15%, which can be explained by recrystallized microstructure of the FSAP stir zone. Tensile tests do not show clear signs of radiation hardening at a given irradiation scenario, which is a positive signal with regards to intended application in the nuclear industry. Table 9 sets results from FSAP MMCs in comparison with different processing techniques. FSAP competes against liquid phase processes such as stir casting or powder metallurgy. References mentioned below, can be used for comparison, since Al 6060 and Al 6063 have very similar chemical composition and reinforcement type as well vol.% of the reinforcement are comparable to results presented in this article. As mentioned above, stir casting is prone to defects associated with liquid phase processing (particle agglomerations), thus it has impaired mechanical properties. Powder metallurgy processes suffer less from such defects, but their cost is quite high. As shown in this work, FSAP can be an interesting alternative to other processing techniques. One can achieve higher ultimate tensile strength than stir casted MMCs while retaining excellent elongation at fracture. Although strength is lower, compared to the powder metallurgically produced parts, elongation at fracture of FSAP MMCs surpass MMCs produced by the more costly route. High elongation at fracture of FSAP microstructure can be pushed even further to superplastic behavior. Using modified multipass friction stir processing, grain sizes down to 100 nm can be produced and elongation at fracture up to 2150% can be reached [34].
Comparison with other materials used in neutron shielding is shown in Table 7. Absorption cross section of Al-B 4 C MMCs produced by FSAP reached values lower than commercially available Al-B 4 C MMCs but remained slightly higher than boron infused stainless steel or NiGd alloys. FSAP composites seem to be inferior commercially available composites with regards to absorption cross section. However, measurement of macroscopic absorption cross section is dependent on sample thickness [12]. Boron absorbs more low energy neutrons, which results in a change of the shape of the neutron spectrum (beam hardening) [35]. Neutron detectors show energy dependence caused by reaction 115 In(n,γ) 116m In, and thus the neutron absorption cross section is also thickness dependent.
Therefore, it has been proven that the sensitivity of FSAP composites to neutron irradiation is low, so they could be considered as an alternative production method for applications in the nuclear industry. As mentioned above, FSAP could produce ultra-finegrained microstructure, which could have influence with regards to radiation damage. Grain boundaries can act as sinks of clusters and reduce swelling. In addition, grain boundary sliding could reduce radiation embrittlement by bypassing dislocation induced plasticity [36].

Conclusions
The main goal of this work was to produce Al-B 4 C MMCs by multipass friction stir additive processing and investigate its properties with regards to application in the nuclear industry. It has been shown that by means of FSAP, Al-B 4 C MMCs with a thickness of up to 7 mm can be produced. Additionally, thermal neutron absorption cross section was found to be comparable with alternative materials and mechanical properties did not deteriorate after irradiation. Radiation hardening was not observed by means of tensile tests. Neutron absorption properties of FSAP MMCs lie within the range for commonly used materials.

Funding:
The presented work has been realized within Institutional Support by the Ministry of Industry and Trade and was obtained with the use of infrastructure Reactors LVR-15 and LR-0, which is financially supported by the Ministry of Education, Youth and Sports-project LM2018120.
Institutional Review Board Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.