In-situ additive manufacturing of high strength yet ductility titanium composites with gradient layered structure using N2

It has always been challenging work to reconcile the contradiction between the strength and plasticity of titanium materials. Laser powder bed fusion (LPBF) is a convenient method to fabricate innovative composites including those inspired by gradient layered materials. In this work, we used LPBF to selectively prepare TiN/Ti gradient layered structure (GLSTi) composites by using different N2–Ar ratios during the LPBF process. We systematically investigated the mechanisms of in-situ synthesis TiN, high strength and ductility of GLSTi composites using microscopic analysis, TEM characterization, and tensile testing with digital image correlation. Besides, a digital correspondence was established between the N2 concentration and the volume fraction of LPBF in-situ synthesized TiN. Our results show that the GLSTi composites exhibit superior mechanical properties compared to pure titanium fabricated by LPBF under pure Ar. Specifically, the tensile strength of GLSTi was more than 1.5 times higher than that of LPBF-formed pure titanium, reaching up to 1100 MPa, while maintaining a high elongation at fracture of 17%. GLSTi breaks the bottleneck of high strength but low ductility exhibited by conventional nanoceramic particle-strengthened titanium matrix composites, and the hetero-deformation induced strengthening effect formed by the TiN/Ti layered structure explained its strength-plasticity balanced principle. The microhardness exhibits a jagged variation of the relatively low hardness of 245 HV0.2 for the pure titanium layer and a high hardness of 408 HV0.2 for the N2 in-situ synthesis layer. Our study provides a new concept for the structure-performance digital customization of 3D-printed Ti-based composites.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Titanium (Ti) and its alloys are widely used in medical implants [1,2] due to their outstanding biocompatibility, low density, and high specific strength.However, it is a difficult-toprocess material because of its inherent physicochemical properties, such as low hardness [3] and relatively weak thermal conductivity [4].In addition, the low strength of its traditional casting and forging [5] is far from keeping up with the demand of broad applications.Therefore, many researchers have always emphasized and studied the development of highperformance titanium materials.To enhance the strength and hardness of commercial pure titanium (CPTi), adding ceramic particles (such as TiB [6,7], Mo 2 C [8], and B 4 C [9]) or rare earth oxide nanoparticles (such as La 2 O 3 [10]) as reinforcing phases in the Ti matrix were common means.However, the frequently used dispersion techniques for the reinforcement phase (such as ball milling or mechanical mixing [11]) often did not result in a very homogeneous distribution in the matrix, which would seriously degrade the ductility of titanium matrix composites (TMCs) [7,9,12].Particularly, when the size of the reinforcing phase is decreased to the nanoscale, the agglomeration of this phase is inevitable due to the vast specific surface area and the van der Waals forces [13,14].Although this strengthening method could improve the strength and hardness of CPTi, the ductility of TMCs decreases very low compared with CPTi.Furthermore, it might cause side effects such as deterioration of weldability, machinability and increased tendency to crack [13].
Shell nacre consists of hard and soft layers of natural materials varying from the nanoscale to the macroscale, assembled in a complex, layered structure [15].A layered structure could achieve excellent mechanical properties by extending cracks and dissipating energy at the boundary between rigid and soft layers [16].Creating layered structures based on their natural configuration could enhance their overall mechanical properties, including crack resistance, tensile strength, etc.The strength-plasticity balance of several additive manufactured metal-based materials, such as Ti-based [17], Fe-based [18,19], Al-based [20], and high-entropy alloys with submicrostructures [21], has been recently described.In addition, the layered structure of materials is designed through additive manufacturing, which makes the catalyst less prone to agglomeration and can effectively degrade various organic pollutants [22].However, reducing the trade-off between the strength and plasticity of TMCs is challenging by altering the components or microstructure of homogeneous materials using conventional alloy designs [23].Laser powder bed fusion (LPBF) fabrication of CPTi with sub-microstructures requires the ability to predictably tailor the material composition, structure, and performance at arbitrarily spatial locations within the part [24].Therefore, in order to realize material-structure-performance integrated additive manufacturing, it is necessary to further develop current LPBF technology.
LPBF forming composites or gradient materials has been explored.Chen et al [25] designed a gradient material LPBF equipment that enabled Z-direction printing of multi-material structures by letting the powder fall into different grooves of the powder spreading trolley to switch materials and used this equipment to verify the formation of directed energy deposition nozzles [24] for heterogeneous materials.Similarly, Wang et al [26] formed a Ti6Al4V/TiB 2 layer structure material using an LPBF machine with similar principles and obtained improved bending resistance.However, this approach could only achieve a one-dimensional (1D) structure distribution in the Z (building) direction.Achieving a 3D structural gradient material with better performance could be challenging.Therefore, Wei et al [27] prepared a multi-material LPBF setup equipped with an in-situ powder mixing system and an ultrasonic vibration powder feeding system to mix and distribute ductile metal and brittle glass powders horizontally and vertically.However, due to heterogeneous materials' significantly different physical and chemical properties, multimaterial LPBF-ed layered structures could face low bonding strength.Moreover, this way of printing multi-materials would induce additional problems, such as mixed contamination of the printed powder and the high cost or difficulty of separating the as-mixed powder, which contradicts the original intention of multi-material green and sustainable printing for additive manufacturing.Therefore, investigations were performed to introduce a low-concentration active atmosphere to manufacture titanium or TMCs in the LPBF process [28][29][30].Due to the relatively high solubility of O and N in Ti matrix, N and O usually form interstitial solid solutions [31].The solute atoms could hinder the activation of dislocation glide on the prism surface, improving the strength of Ti6Al4V, but the plasticity (<5%) performance has inadequately behaved.Furthermore, modified CPTi-(N) powder in a nitrogen atmosphere and then LPBF forming under Ar also improved strength and allowed considerable elongation [32].Nevertheless, this also adds an energy-consuming step that requires obtaining N elements on the surface of Ti powder under a high-temperature N 2 .Additionally, the thermal conductivity of the gas atmosphere could directly affect the molten pool's temperature gradient and the Marangoni flow [33], and the gas-liquid reactions could hardly be controlled.Therefore, there are few studies on Ti materials that arrange reinforcements at any position of the part or controllable properties during LPBF processing.
Considering the aspects mentioned above, based on the principle of digital materials, preparation of gradient layered structure titanium (GLSTi) composites with a threedimensional (3D) distribution of rigid layers (hard ceramic phase)-ductile layers (soft metal phase) through laser in-situ additive manufacturing was described, injecting different concentrations of N 2 within or between the layers during LPBF processing.The effects of N 2 concentration on the parts' submicrostructure and mechanical properties were investigated.On this basis, in-situ digital image correlation (DIC) tests were performed on tensile deformation to evaluate the formation and interfacial response of TiN/Ti layered structure and the fracture mechanism of the composites.The results highlight the applicability of LPBF to facilitate the composites (multi-materials) with 3D gradient layered structures to balance high strength and plasticity.Moreover, since the in-situ synthesized reinforcement phase nucleates and grows naturally in the Ti matrix, the method also shows the advantages of good compatibility between the Ti matrix and the reinforcement particle with high interfacial bonding strength.

Powder materials and LPBF machine
Spherical CPTi powder (AP&C, Canada) manufactured by advanced plasma atomization technique was used in LPBF experiment (figure 1(a)).Figure 1(b) shows the particle size distribution (d10 = 18.48 µm, d50 = 29.76µm, and d90 = 47.34 µm) of the powder measured on the mastersizer 3000 (Malvern P analytical Ltd, UK) laser diffraction particle size analyzer with an average particle size of ∼32 µm.Table 1 lists the chemical composition of the powder quantified by the ICP-OES method (iCAP 7200 Duo, Thermo Fisher Scientific, USA).The CPTi and gradient layered Ti composites were fabricated on a self-developed LPBF device equipped with a 200 W IPG fiber laser with a focused beam diameter of ∼70 µm.The forming chamber was equipped with high-purity (99.999%)Ar and N 2 .The Ar was used to exhaust air or N 2 .The N 2 was selectively introduced to build a mixed-forming atmosphere with controlled N 2 -Ar ratios.

Design of layered structure and LPBF process
The design and preparation of the in-situ self-generated layered structure of the Ti composites are shown in figure 2. Firstly, based on our previous research [30], 10 mm × 10 mm × 5 mm samples were LPBF formed under 0%-20%N 2 (the rest was Ar) to construct the correspondence between N 2 concentration and in-situ synthesized TiN content.The laser parameters used were laser power of 160 W, scanning speed of 1 000 mm s −1 , and hatch space of 0.07 mm.On this basis, we also fabricated uniform nitriding titanium (UNTi) composites (named UNTi5, UNTi10 and UNTi15) with the dimensions as shown in figure 3(a) to investigate the N 2 atmosphere-assisted strengthening of CPTi properties and mechanisms.
Then, 1D layered structure titanium (LSTi) material and gradient layered structure (GLSTi-Z) material with N 2 concentration change in the Z direction was built under N 2 concentrations of 5%, 10%, 15% and 20%, respectively.The twodimensional gradient layered structure material (GLSTi-YZ or GLSTi-XZ) was formed by adjusting the N 2 concentration in the N 2 layer along the Y or X direction, plus the thickness change in the building direction (BD).Finally, the CPTi layer was printed by adjusting the N 2 concentration along the Y or X direction on the first N 2 interlayer and then switching back to Ar. Varying the N 2 concentration along the X or Y direction as the CPTi matrix builds the next nitride layer.Such alternate printing could combine layered materials (GLSTi-XYZ) with gradient changes in XYZ (3D) directions.
The specific design parameters of the layered structure samples are detailed in table 2. L is the length and width of the square specimen, d w is the width of the N 2 in-suit synthesis layer along the X direction, d h is the width of the CPTi layer formed by LPBF in Ar, T Ar is the thickness of the CPTi layer formed by LPBF in Ar, and T N2 is the thickness of the N 2 in-situ synthesis layer.T Ar was set at 0.21 mm, 0.3 mm and 0.42 mm.Then, LPBF in-situ synthesized the T N2 (0.09 mm, 3 layers) thickness.The N 2 in-situ synthesized layer parameters were: laser power = 160 W, scanning speed = 500 mm s −1 , and hatch space = 80 µm.The printing thickness was 30 µm, and the laser scanning strategy adopted a 90 • orthogonal scanning.The laser parameters of CPTi in N 2 were different from those in Ar.The reason was to allow Ti and N 2 to in-situ react entirely.Therefore, the laser energy density could be increased by reducing the scanning speed, and the reaction's contact time could be extended simultaneously [34].The d w and d h of LSTi composites were 10 mm, while d w and d h of GLSTi sample were 3.0 mm and 0.5 mm, respectively.The specific laser parameters are summarized in table 3. Finally, a blank sample was printed on a new Ti alloy substrate under Ar.The sample was named CPTi-Ar, and the forming parameters were consistent with LPBF-ed Ti in 100%Ar (table 3).Additionally, to confirm that the actual N 2 concentration in the forming chamber was the same as the preset, a time-offlight mass spectrometer was also used to quantify the real N 2 -Ar ratios at the building position in the forming chamber.The test equipment and results were detailed in our previous research [30,35].This experiment's airflow, fan frequency, and other related parameters are consistent with the previous study [30].

Microstructure characterizations
The in-situ synthesized layer's surface morphology for pure titanium under different N 2 concentrations was investigated on an ultra-depth field microscope (VHX5000, KEYENCE, Japan).The samples were first wire-cut from the Ti alloy substrate, then ground and polished according to standard procedures, and 0.04 µm silica suspension was applied for final polishing.The microstructure of the cross-section was investigated on a Leica DMI5000M optical microscope.The O/N/H element contents were analyzed using an O/N/H analyzer (EMGA-830 OK; HORIBA).An x-ray diffractometer (X'pert Powder, P analytical) was used to examine the phase composition between 20 • and 80 • at a scanning rate of 0.02 Microstructure and elemental distribution were investigated on a field emission scanning electron microscope (FE-SEM, FEI NOVA Nano430, Netherlands).For texture orientation, electron backscattered diffraction (EBSD) data was collected on the Oxford instrument Symmetry S2 system, and the experimental conditions were voltage 20 kV, WD 17 mm, and step size 0.48 µm.The prior β grain was investigated through the AZtecCrystal software version 2.1.Highresolution transmission electron microscopy (HR-TEM), compositional measurement and selected area electron diffraction (SAED) observations were obtained by a Talos F200S TEM (Thermo Fisher Scientific Co., Ltd, USA).

Mechanical property tests
The hardness changes of samples were measured every 50 µm along the LPBF BD on a micro-Vickers hardness tester (ZHV30, Zwick Roell, Germany), with a load of 200 g and a holding time of 15 s.Each sample was tested at five positions to calculate the mean.According to the ISO 6892-1 standard [36], the dogbone-shaped small size tensile specimen with a width of 6.0 mm and a thickness of 2.5 mm (figure 3(a)), and the loading direction was parallel to the layered structure.The room temperature tensile test was carried out on a CMT5105 series electronic universal testing machine equipped with a 10 mm gauge length extensometer at a 0.5 mm min −1 quasistatic displacement rate.Each parameter was measured three times to obtain the mean value and variance.The deformation mechanism of LSTi composites was examined with DIC systems (Correlated Solutions Inc, USA), as shown in figure 3(b).Some random speckles were sprayed onto the surface of the LPBF-ed Ti samples.In the stretching, the fracture behavior of LSTi composites was recorded by a camera (resolution = 2448 × 2048).Then, these real-time computation data were acquired and analyzed in VIC-3D software.

Element content and phase analysis
The variation of O and N in the cubic samples built by LPBF at different N 2 concentrations was statistically evaluated (table 4), each parameter was tested with three samples to obtain the mean value.Introducing the N 2 , there was a slight increase in O content compared to Ar gas printing because the laser energy density of in-situ nitriding reaction is usually higher than that in pure Ar atmosphere, and the high energy density accelerates oxygen dissolution [37].In addition, the reason for the increase of O content in the sample may also be caused by the simultaneous opening of inlet valve and outlet valve during N 2 -Ar gas exchange (the purpose of opening at the same time is to discharge prior Ar and adjust the N 2 -Ar ratio in the forming chamber.If the outlet is closed, the pressure in the forming chamber will also increase as the N 2 gas intake continues, so the experiment is not safe).However, the N content increased rapidly and approximately linearly, as shown in figure 5(a).It indicated that introducing the N 2 could enhance the absorption and combination by the Ti molten pool after being subjected to laser irradiation.
Figure 4 shows the XRD patterns of CPTi and LSTi composites prepared by LPBF in different gas environments.The samples built using a pure Ar were almost consistent with those of the powder, which are α/α ′ -Ti peaks (JCPDS card 44-1294).In addition, only the Ti phase was found in the environment with a 5%N 2 .No TiN phase was found, indicating that the low N 2 concentration could dissolve more N in the Ti lattice [38].Besides, the particle size or concentration of the TiN particles of 5%N 2 sample was too small to be detected by XRD on the sample's surface, and its presence needs to be confirmed under higher magnification SEM or TEM. Figure 4(b) clearly shows that the (0002) peak of the Ti matrix phase shifted to a smaller diffraction angle.According to the Bragg equation [39] 2d sin θ = nλ, the interstitial solid solution of N in the Ti lattice caused a slight lattice expansion during LPBF [28].However, when the N 2 concentration reached 10%, the XRD pattern showed signs of the TiN phase obviously, and the (111) and (200) peaks showed this point.As the N 2 concentration increased, the TiN (200) peak intensity became more accentuated.
The interplanar spacing and Miller index of the HCP-Ti crystal follow the Bragg formula [39]: where d hkl is the grain plane spacing and the h, k, l are the Mille indexes of the grain plane.a, c are lattice constants.
Based on the XRD patterns of the cube specimen shown in figure 4(b), the crystal plane family {10 10} associated with the lattice constant a and the crystal plane family {0002} associated with the lattice constant c are specially selected to calculate.
Figures 5(b)-(d) show a quasi-linear relationship between the a, c, c/a ratio and N content.As the N content increased, the α/α ′ -Ti lattice constant gradually expanded, and the expansion rate of the c value was significantly higher than that of a.This is consistent with the theory that N atoms could enter into the octahedral gaps of the Ti lattice in the HCP configuration and cause the corresponding lattice expansion, which provides an essential basis for the low N 2 gas to assist in strengthening LPBF Ti composites [32].

Morphology and microstructures of LPBF UNTi composites
Figure 6 shows morpho-structural peculiarities of the surface melt tracks of typical LPBF processed samples fabricated in a diluted N 2 atmosphere.The UNTi5 sample (figure 6(d)) exhibited a bright silver metallic luster similar to CPTi-Ar (figure 6(a)), with only a tiny number of spherical particles attached.However, with a slight increase in N 2 concentration, the sample obtained using 10%N 2 was still similar to 5%N 2 , slightly rougher, and more unfused powder sticks were attached to the surface.As the N 2 concentration was increased to 15%, a widened scanning track could be observed in the upper surface morphology.It is also confirmed by the singletrack results (figure S1) in the supplementary material.This is due to the increased number of in-situ chemical reactions of Ti with N during LPBF at high N 2 concentrations [30,40].Simultaneously, the high temperature gradients between the molten pool and the solidified track led to more vigorous Marangoni convection [4], expanding the molten pool size.In addition, the morphology of the molten pool shows a more golden color of TiN [41], and there were strips of nitrides in the overlapping area of adjacent molten channels with high N 2 concentration (figure 6(m) of 20%N 2 samples).
Figures 6(b) and (c) shows that the cross-section microstructure of CPTi-Ar sample was mainly laths α-Ti.The microstructure of low N 2 presented elongated grains, and higher N 2 concentrations formed a mixed structure, including acicular grains, lamellar Ti and precipitated grains.It was also shown that the rapid heating and cooling rate of LPBF introduced a non-equilibrium phase (β → α ′ ).Higher N 2 concentration could promote in-situ reaction between molten Ti and active N atoms/ions, leading to an increase in TiN content [40].Therefore, the distribution of TiN particles in the samples was improved.In addition, the high-magnification SEM micrographs showed that the lamellar β grain (red arrow) appeared near the TiN and α grain boundary.It revealed that introducing N into the building gas during the LPBF process could cause the Ti molten pool to experience a more intense thermal history process [42].As a result, the LPBF process effectively influenced the heat around the molten pool and changed the microstructure of the CPTi.
Figure 7 shows the EBSD texture and grain structure of LPBF samples formed by LPBF under Ar + 0/5/10/15%N 2 , and figure 8 shows the corresponding inverse pole figures.The results indicated that the CPTi-Ar samples exhibited coarse-grain epitaxial growth in the BD.With the increase of N 2 concentration, the grain morphology changed gradually.Compared with the grains formed by LPBF under Ar, the finer acicular α ′ -Ti grains with a more pronounced hierarchical structure were observed at low (5%) N 2 concentration.The final grain morphology was distributed in the Widmanstätten   structure [43,44].This process continued until the prior β grains consisted entirely of thin laths α/α ′ , as observed in UNTi15 sample, where random orientation was observed when the concentration of N 2 was increased (figure 7(d)).
Furthermore, a large percentage of low-angle grain boundaries (LAGBs) was observed in the 10%N 2 and 15%N 2 sample (figures 7(c) and (d)), in contrast to CPTi fabricated by LPBF in 5%N 2 or conventional techniques [3].The parent (prior) β grains were reconstructed according to the as-built EBSD results, as shown in figures 7(e)-(h).According to the Burgers orientation relations in Ti [45]: {110} β //(0001) α and <111 > β //< 11−20 > α , it could be affirmed that for the CPTi built by LPBF in Ar, the shape of the prior β grains obtained was more regular (figures 7(a) and (e)).In contrast, for the 10%N 2 sample, the α/α ′ lath grains into parallel platelets colonies due to the shape of the prior β grains became irregular (figure 7(g)), which affected the epitaxial growth process.Under a 15%N 2 , the growth of prior β grain was also influenced.Prior β grains grew more rapidly, continuous columnar grain growth was disrupted, and coarser columnar grains appeared in the lap region (figure 7(h)).This could indicate that introducing N 2 in the forming atmosphere could show a more significant thermal influence around the melt pool and TiN were serving as nucleation sites and thus resulting in less epitaxial growth of the prior β grains.Overall, the EBSD results indicated that the CPTi built by LPBF using an N 2 atmosphere showed continuous orientation changes, and partial cross-orientation, indicating the formation of a submicrostructure (as shown in figures 7(b)-(d)).Figures 7(i)-(l) shows the phase maps of LPBF-ed CPTi.It could be seen that there was only a Ti phase in sample formed by LPBF in Ar and 5%N 2 .TiN was able to be found above 10%N 2 , and the volume fractions of TiN under 10%N 2 and 15%N 2 were 2.2% and 2.7%, respectively.In addition, the grain size (figures 7(m)-(p)) showed that the average grain size of LPBFed CPTi under Ar was 7.45 µm.With the increase of N 2 concentration, the grain size decreases gradually.However, the grain sizes were similar at 10%N 2 and 15%N 2 , which were 2.91 µm and 2.89 µm, respectively.
The inverse pole figure also shows the relationship between the orientation distribution of α/α ′ grains and the prior β grains with the increase of N 2 concentration.As shown in figure 8, the α-Ti texture coefficient decreases from 5.5 for the CPTi-Ar sample to 2.15 for the 15%N 2 sample, which corresponds to the weakening of orientation with the increase of N 2 concentration.Moreover, CPTi-Ar had a partial (0001) growth orientation, and there was an obvious {110}β//(0001)α relationship between the reconstructed prior β grains and α-Ti.The 5%N 2 sample displayed lamellar α ′ -Ti variant selection with 60 • //< 11 20 > orientation when the α grain was precipitated from the prior β grain.However, when the N 2 concentration increased to 15%, a large number of first-type clusters were generated inside the nearly equiaxed prior β grains [46], and the variants had a certain random orientation.

Mechanical properties of LPBF UNTi composites
Micro-Vickers hardness could be an important basis for characterizing material hardness and wear resistance [37].The XOZ plane of the LPBF sample was intercepted for micro-Vickers hardness testing.The microhardness value of LPBF-ed CPTi with different N 2 concentration is shown in figure 9(a).The average microhardness of CPTi-Ar sample was (239 ± 9) HV0.2.And it gradually increased with increasing N 2 concentration, reaching a maximum value of (409 ± 10) HV0.2 at 20%N 2 .
Figure 9(b) shows the stress-strain curves of LPBF-ed CPTi and UNTi composites.The corresponding results are summarized in table 5.In general, for LPBF pure Ti and its alloys, solid solution and brittle precipitation could be caused by high interstitial element (N, O) content, resulting in undesired degradation of mechanical properties (plasticity) [5].Therefore, in this work, the LPBF-ed CPTi in low-concentrations N 2 performed well.Especially the UNTi5 sample showed a decent combination of high strength and ductility.According to table 5, the ultimate tensile strength (UTS), yield strength, and plasticity of the LPBF-ed UNTi5 samples were (958.83 ± 9.78) MPa, (886.17 ± 7.34) MPa, and (17.27 ± 0.89) %, respectively.The tensile strength and yield strengths were 37.7% and 47.9% higher than those of CPTi-Ar sample at (696.48 ± 2.95) MPa and (599.19 ± 9.57) MPa, respectively.Compared with the UNTi10 formed by LPBF in medium concentrations N 2 , the mechanical properties of the UNTi15 with high N 2 concentrations were only slightly enhanced.However, the plasticity was almost nonexistent (figure 9(b)) because N could behave as an α-stabilizer [31].Interstitial N could enormously increase the transformation temperature of the high-temperature prior β grains, expanding the α-grain region, and promoting martensite formation [31], thereby improving the strength of the UNTi composites.When the N 2 concentration reached 10% and above, the in-situ synthesis between the low-viscosity Ti and N atoms increased [29,38].UNTi10 and UNTi15 samples produced more N-solid solution, and even excessive TiN particles (figures 7(k) and (l)) made the Ti matrix low ductility, decreasing their mechanical properties.
To investigate the stress-strain distribution of the layered structure during tensile deformation more visually, in-situ monitoring was performed using the DIC system, and the   surface morphology.The fracture was flat, almost without a neck, which conformed to the fracture characteristics of brittle materials.Moreover, the UNTi15 sample showed more ubiquitous cracks, nitride-smooth fracture surfaces, and noticeable cleavage features than the CPTi-Ar sample.It could be observed that an acicular martensitic microstructure was formed in the N 2 layer due to a higher cooling rate during the LPBF process [48].In contrast, the LSTi sample showed a coarse block or lath α/α ′ -Ti microstructure in the LPBF forming Ti in Ar, as shown in figure 11(c).

Microstructure of LPBF gradient layered TiN/Ti composites
Figures 11(d)-(g) show the interface between the Ti layer and the Ti-N 2 layer.In N 2 in-situ synthesis layer, as the laser scanning speed decreased, more N solid solutions were melting into the Ti matrix (figure 11(e)).Therefore, many acicular martensites appeared in the case of the LPBF samples processed with N 2 .Most notably, many TiN particles (bright second-phase) were found in the microstructure of LSTi composites, as shown in figures 11(c) and (f), and (i).In the LPBF process, N 2 could be decomposed by high-energy laser irradiation to produce N atoms and the absorbed nearby hightemperature molten pool.Then in-situ reacting with Ti atoms to produce TiN enhancers (N + Ti → TiN), which favorably precipitate and grow from the molten pool during rapid cooling due to the higher homogeneous melting point of TiN (2 950 • C) than Ti (1668 • C) [49].Moreover, the N and O elements EDS distribution of figure 11(h) can be seen in the supplementary material (figure S3), N is enriched in the N 2 region, and O is uniformly distributed in the N 2 and Ar atmospheres regions.Besides, keyhole defects were also found in the GLSTi-XYZ sample (figure 11(h)).The reason was that the laser energy density of the N 2 layer was higher than that of the Ti layer formed by LPBF in Ar to make Ti molten pool absorb more heat.Part of the operating gas or vaporized metal gas was trapped in the sample, forming pores [50].That is detrimental to part performance and should be avoided as much as possible.Fortunately, figure S2 shows that there is no significant increase in porosity with increasing N 2 concentration.Consequently, a compromise parameter was chosen that allows in-situ synthesis of TiN with a few keyholes.
Figure 12 shows the EBSD at the interface between the Ti layer and the TiN containing layer of LSTi20-0.3and GLSTI-XYZ samples.At 20%N 2 , Ti matrix of LSTi20 obtained acicular α/α ′ -Ti grains, which were parallel to the prior β grain.After introducing N 2 , a higher cooling rate could lead to a faster nucleation rate, thereby refining α ′ -Ti.When the GLSTi-XYZ was converted from N 2 layer to the next LPBFed Ti layer in Ar, a small part of TiN at the top of the interface layer (that is, the bottom of the Ti molten pool) changed into N and solid dissolved in the Ti matrix, forming compact and fine acicular α/α ′ -Ti.There was no significant difference in the grain boundary angle between LSTi20-0.3and GLSTi-XYZ.The phase distribution showed that the volume fraction of TiN on the LSTi20-0.3sample in 20%N 2 layer was 2.1%.
Figures 12(b) and (e) show the IPF map of the corresponding reconstructed prior β grains of the LSTi composites.It could be found that the Ti layer was very similar to the CPTi-Ar (figure 7(a)), which was a coarse grain.Turning to the N 2 in-situ synthesis layer, the continuous growth of columnar grain was interrupted, forming a finer acicular α ′ -Ti.Furthermore, due to the scanning strategy of 90 • rotation layer-by-layer, the direction of heat input changed continuously, interrupting the epitaxial growth of prior β grains [26,35].Figures 12(f) and (g) show that the grain sizes of the fine-grained (FGZ) and coarse-grained (CGZ) zones of the LSTi20 sample were 2.7 µm and 5.6 µm, respectively.The grain sizes of FGZ and CGZ for GLSTi-XYZ sample were 1.8 µm and 3.5 µm.Obviously, the N 2 in-situ synthesized composite with more gradient layered structure has a more obvious grain refinement effect than that of the single layered structure.
Figure 13(a) shows the HR-TEM micrographs and SAED pattern of the acicular α ′ -Ti phase in the BF-TEM micrograph (figure 13(b)) of the GLSTi-XYZ sample.As could be observed, the lattice fringe spacing of the (110) plane in the Ti matrix was 1.493 Å.In the second phase precipitate, figure 13(c) shows the HR-TEM micrograph and SAED pattern of the dark precipitated phase in BF-TEM.The lattice fringe spacing of the (200) plane was 2.149 Å.These lattice fringe spacings and precipitates correspond to α-Ti and TiN, respectively, suggesting that TiN was precipitated from the Ti molten pool [47].The tightly bound interface shows a clean interface and strong interfacial bonding (figure 13(c)).Figures 13(e) and (f) present the EDS and the large dark area also shows that TiN was uniformly combined.

Mechanical properties of LPBF gradient layered TiN/Ti composites
Figure 14(a) shows the microhardness variation of the LSTi15-0.30sample from the Ar region (Ti) through an N 2 layer (TiN) to the next Ar region.The microhardness values initially increased as the Ti matrix was switched to the N 2 layer and then decreased with the transition to the Ti matrix.In the N 2 layer, the highest value of 373 HV0.2 was reached.This is because the introducing N 2 in the LPBF process could form more acicular α ′ -Ti, which could improve hardness [51].Besides, under the high N 2 concentration, the in-situ synthesis between molten Ti and N atom/ion generates TiN hard particles, which is one of the explanations for the increase in hardness [41].Then, as the N 2 layer was switched back to the Ar layer, a transition layer was affected by the synthesis of TiN in the previous layer.Therefore, the interval of ∼90 µm (thickness of the T N2 ) first showed a rapid drop in the curve from the highest to the low hardness interval of the Ti matrix.After continuing to print 3-5 layers by LPBF in Ar (each layer thickness of 30 µm), it gradually reached the lowest hardness value.
The layered structure Ti composites werebuilt through the LPBF process to improve the performance of strengthplasticity.Tensile properties of LSTi composites were summarized in table 6. Figure 14(b) shows that when the T Ar was 0.21 mm, the LSTi composites could be stretched and plastically deformed at an N 2 concentration ranging from 5% to 20%.Especially under the 15%N 2 , the strength was able to increase by 36% while the plasticity could still maintain an elongation of 20%.Even if the UTS of LSTi20-0.21sample was above 1100 MPa, the elongation rate was still 9.1%.This performance was better than UNTi10 sample and comparable to most LPBF-ed Ti6Al4V after heat treatment (UTS of 1094 MPa, elongation of 11%) [52].These properties could greatly develop the application scenarios of CPTi.Moreover, when there was a soft Ti layer with the same thickness, the strength increased, and plasticity decreased as the N 2 concentration increased.
Additionally, when the thickness of the T Ar increased from 0.21 mm to 0.42 mm, it was found that choosing an appropriate thickness for the soft Ti layer could improve the balance of strength-plasticity.For example, LSTi20-0.30could achieve nearly the same high UTS as LSTi20-0.21,but the elongation could be doubled simultaneously (figures 14(c) and (d)).Therefore, further reducing the proportion of the N 2 layer (TiN) was able to achieve the same performance regulation by increasing the N 2 concentration.Just like LSTi20-0.42 had similar mechanical properties to LSTi15-0.21.This means that there is an additional means of controlling the performance of in-situ synthesis CPTi using N 2 atmosphere, which can optimize the printing efficiency.Moreover, the GLSTi composite has improved the mechanical properties and the ability to maintain plasticity, enhancing strength and ductility.For example, the thickness of the T Ar is 0.30 mm, and the ductility of GLSTi-YZ prepared by LPBF under the 15%N 2 could reach more than 25%, but the strength could still be increased by 32%.Furthermore, GLSTi composites showed increased strength, but the elongation rate could be nearly the same as that of the pure Ti matrix (as shown in figure 14(f)).
In order to better observe the strain of the layered structure Ti composites, DIC monitoring was also conducted on the cross-sectional view.As seen from figures 15(a) and (b), after in-situ synthesis of TiN layered structures in CPTi [26,53], enhanced deformation occurred, increasing the hardening strain rate during tension.This resulted from an improved strain hardening rate by forming back stress reinforcement in the soft Ti layer [54].The strain hardening rate capability was related to the amount and distribution of TiN.With the TiN arrangement forming the XYZ-3D distribution, the crossdistribution of soft and hard zones leads to the GLSTi-XYZ samples (figure 15(d)) exhibiting more plastic deformation.It was mainly the layer where the TiN gradient distribution led to the appearance of different flow stresses.These rigid zones were mechanically incompatible with the elastic-plastic and plastic deformation phases.Due to mechanical incompatibility, deformation strain partitioned, with the softer layers maintained higher plastic strains and the rigidity layers impeded dislocations and increased strength again [55].Thus, GLSTi-XYZ samples exhibited a higher elongation at break.
Figure 16 shows the fracture morphology of typical LSTi and GLSTi samples, which behaved differently from the UNTi composites.The fracture characteristics of LSTi samples displayed a transition from ductile fracture to brittle fracture with the increase of N 2 concentration (from 5% to 20%), and the 'cleavage step' with a smoother surface replaced 'dimple' as the main fracture feature.Moreover, the LSTi composites produced under the same N 2 concentration first showed an increase in plasticity and then converged with the increase in the thickness of the T Ar (figures 16(e) and (f)).However, compared with the UNTi sample, the dents of GLSTi were larger and deeper, and the fracture dents were very shallow.The small number of spherical hole defects in GLSTi-XYZ did not become the source of crack propagation, which showed the excellent plasticity of TMCs with a gradient layered structure.In addition, a small area of plastic deformation was also observed in figure 16(h).This type of region showed the characteristics of tear ridges, crack surfaces and micro-cracks (figure 16(j)).It could be seen that most of the GLSTi samples exhibited a mixed mode of ductile and quasi-cleavage fracture.The fracture morphology of the sample was consistent with its plastic evolution.The GLSTi composite fabricated by LPBF N 2 in-situ synthesis considerably improved the balance control of the strength and plasticity of CPTi.This experiment could also be extended to other atmosphere-strengthened alloys.For example, carbon-containing gas laser in-situ synthesis of carbides strengthens alloys' hardness and strength achieves high-performance design and rapid prototyping.

LPBF in-situ synthesis mechanism of TiN using N 2
According to the XRD patterns (figure 4), the partial molten Ti reacted with N 2 to synthesize the TiN compound with more than 10%N 2 .Figure 17 shows a schematic diagram of the in-situ synthesis mechanism of Ti under N 2 + Ar.As shown in figure 17(a), after introducing a certain amount of N 2 gas into the forming chamber under high-energy laser irradiation, the N 2 gas becomes plasma N. Then it was adsorbed by the high-temperature (>2000 • C) molten pool.Considering the Ti-N binary phase diagram [56], the reaction Ti + 1/2N 2 ⇋ TiN shows a negative Gibbs free energy [38], and TiN will form and exist in crystalline form in the Ti molten pool (figure 17(b)) [30].A decreasing diffusion rate of N atoms occurred during cooling and solidification (figure 17(c)), and some N atoms remained in the lowtemperature phase.However, TiN cannot be melted in this process [38].Therefore, TiN particles could precipitate and behave as hetero-nucleation sites, resulting in grain refinement [30].It is worth noting that the LPBF sample has undergone several subsequent thermal cycles (figures 17(e) and (f)).The small undissolved particles of TiN could be re-melted into the molten pool in the next layer of laser scanning and decomposed in subsequent thermal cycles.However, it is a small amount of solid dissolved N, and the Marangoni flow in the molten pool enhances the uniformity of N diffusion.Compared with the N 2 layer, due to the lower laser energy density of CPTi formed by LPBF under Ar, a large amount of TiN can be retained in the Ti matrix.

Influences of nitrogen on mechanical properties
Compared with LPBF CPTi in Ar, the strength of UNTi composites is considerably improved.During tensile deformation, the UNTi5 sample resulted from the interaction between dislocations and N/O atoms.The N/O atoms maintained a stable state at low strain.Attributable to the reason, dislocation movement was limited by the N/O atoms' pinning effect on dislocation sliding and grains [31].Therefore, UNTi sample formed by LPBF under 5%N 2 exhibited excellent work hardening characteristics.Using solid solution strengthening and grain refinement, the strengthening mechanism of CPTi at low N 2 concentration could be quantified.The overall enhancement (∆σ YS ) results are as follows: ( As shown below, when solid solution atoms of N-O were used to examine the strengthening effect, the strengthening occurs as follows [57]: where ∆σ sN is the strength enhanced by solid-solution strengthening, S F is the Schmid factor obtained by EBSD data, c is the concentration of solute, F m is the maximum interaction force, G is the shear modulus, b is the Burgers vector, ω (∼5b) is the width of the edge dislocation.( 1 3 with the slope value estimated experimentally (∆σ SN and c 2 3 have positively correlated linear relationships).Considering the effect of O/N equivalent, it was defined as: Moreover, based on Hall-Petch formula [58], fine grain strengthening (∆σ GB ): where k is the yield constant (15.7 MPa mm −2 ) [32], σ 0 is the lattice friction stress, d is the average grain size determined by EBSD results.The comparison yield strength of the measured and calculated results for the UNTi composites are shown in figure 18(a).However, under relatively high N 2 concentration (>10%), there were more hard and brittle TiN particles that were not melted away, which played a role in precipitation strengthening in UNTi10 and LSTi composites.Therefore, equation (2) was not applicable, so it was explained by the traditional composite material strengthening mechanism, expressed as: where σ Y is the total yield strengthening, and formula (6) can be used to calculate its value.∆σ CTE is dislocation strengthening, which generates geometrically necessary dislocations (GND) due to the mismatch of the coefficient of thermal expansion (CTE) between the reinforcing phase and the matrix.M is the Taylor factor, with a value of 2.5; α is a constant (0.1-0.5); G shear modulus, Poisson's ratio υ is 0.33, which is converted following the E of CPTi-Ar sample in table 5 according to the equation ( 9); ρ represents the dislocation density.∆σ Load is the load transfer enhancement, and the load applied during the deformation process could be transferred from the Ti matrix (soft) to the TiN particle (hard).Where p is the particle shape coefficient of the reinforcing phase, that is, the aspect ratio; V ω is the volume fraction of TiN (figures 7(i)-(l)), σ ys is the yield strength of CPTi-Ar sample.The as-calculated σ Y of UNTi5 and UNTi10 was very close to the measured values.Figure 18(b) shows the comparison of GLSTi composites with other TMCs [59][60][61] and partially additively manufactured Ti6Al4V [62], and enhanced performance could be found with various distributions.

Effects of TiN/Ti gradient layered structure on mechanical properties
The strengthening mechanism of layered composites could be discussed from the rule of mixture (ROM) and the laminate theory.In general, the strength of heterogeneous layered composites is related to the strength of individual constitutive materials, estimated by ROM [54]: where V i and σ i are the volume fractions and strengths of component i (the UNTi and Ti).The UTS of UNTi20 was calculated according to the relationship between UTS and microhardness in figure 9(c).However, it could be observed that using the ROM rule to calculate the strength of the LSTi composites (figure 19(a)), the UTS calculation value was 5% lower than the average test value.In addition, the strength of the LSTi composite was much higher than that of CPTi-Ar sample [63,64] while having excellent ductility.This proved that other mechanisms that were not discussed could affect the strength, and the root cause may be related to the unique spatial heterostructure and hetero-deformation induced (HDI) strengthening [65].
Materials with an excellent strength-ductility combination typically delay necking and shear band formation during tensile deformation, allowing the plastic strain to occur sequentially in wider dimensions and suppressing crack initiation and propagation [21].Figure 19(b) shows the deformation process of hetero-layered materials, which can be categorized into three stages as the strain increases [66].Stage I is the elastic deformation stage of the soft and hard layers, which is similar to the traditional homogeneous materials.Stage II is the elastic-plastic deformation, when the soft layer first starts the dislocation slip to produce plastic deformation, while the hard layer will remain elastic, which results in the mechanical incompatibility of the soft-hard layers.At this stage, the GND in the soft layer may be blocked by the hard layer, causing GND to accumulate at the interface and generate strain gradients near the boundary (figure 19(c)).This results in forward stresses in the TiN region and back stresses in the soft Ti layer, synergistically strengthening to increase the overall yield strength of Ti composites.Stage III is the stage in which both the soft and hard layers undergo plastic deformation.Since the soft layer is subjected to much higher strains than the hard layer, a so-called strain distribution occurs.When adjacent soft and hard layers are subjected to different plastic strains, the strain gradient near the interface will become larger as the strain distribution increases, resulting in back-stress strain hardening.In addition, as shown in figure 11(c), due to the good interfacial bonding, cracks are more likely to propagate through the TiN region rather than along the Ti molten pool boundary.Therefore, the combination of HDI enhancement and good interface synergistically improves strength of LSTi composites.
For CPTi formed by LPBF in Ar, the local necking easily extended to the whole sample when the plastic deformation turned unstable.However, for the LSTi composites, the GNDs in the softer region accumulated near the boundary of the harder region, leading to back stress in the soft layer and forward stress in the hard layer.As shown in figures 19(d)-(f), the GND of the CPTi-Ar sample is 3.92 × 10 14 m −2 , and the UNTi10 sample is 4.8 × 10 14 m −2 .While in the soft Ti layer and hard TiN region of GLSTi composites, the GNDs are 4.28 × 10 14 m −2 and 4.93 × 10 14 m −2 , respectively.Thus, the difference in mechanical strength and GNDs distribution converted the uniaxial tensile stress into complex, mutually confined 3D stresses in the GLSTi composites.Dislocations sliding on intersecting slip planes may intersect, leading to dislocation entanglement and accumulation of dislocations to increase HDI strain hardening [26,54,65].This helps to disperse the plastic strain of GLSTi composites throughout the entire gauge length and improve elongation.
From the current understanding of the mechanical property studies of hetero-layered materials, qualitative guidance can be provided for the design of TiN/Ti layered composites with gradient structure.As discussed previously, HDI strengthening and HDI strain hardening are the main reasons for the excellent mechanical properties observed for LSTi composites.Therefore, it was hoped to maximize the effect of back stresses in the Ti layer and minimize the effect of forward stresses in order to optimize the HDI strengthening.Firstly, in terms of the geometry of the soft and hard region structures, since the boundary area and volume of the soft layer in a planar layered structure is higher than that of the equiaxial structure, a planar soft-hard distribution structure is more effective in promoting the HDI effect.Secondly, in terms of the boundary space between the soft and hard layers.If the spacing (d h in figure 2) is too large, other regions in the hard layer will not be effective in promoting HDI strengthening and strain hardening [67].As the thickness of T Ar (figure 2) decreases, the density of the regional interface increases, which contributes to the development of GND stacking and thus increases the back stresses.Finally, it was found that the optimum volume fraction for soft Ti layers for layered composites is in the range of 20%-30% [68].For the highest HDI strengthening and strain hardening, the soft layer should be fully constrained by the hard layer.Therefore, the cross-gradient layered distribution of TiN/Ti layers can exhibit a higher HDI strengthening effect than single-layered structure, leading to superior strengthductility synergy.

Conclusions
In the present paper, UNTi, LSTi and GLSTi composites were successfully prepared by in-situ synthesis of N 2 with Ti at different locations of LPBF-ed parts to generate TiN.The main findings were as follows.
(1) With different N 2 concentrations, the microstructure of LPBF-ed CPTi evolved from coarse α-Ti grains under Ar to fine acicular martensite α ′ -Ti.TiN was not synthesized below 10%N 2 concentration, and the in-situ synthesized TiN fractions for 10%N 2 and 15%N 2 were 2.2% and 2.7%, respectively.(2) The hardness and strength of UNTi composites with an overall distribution of TiN increased with increasing N 2 concentration, while the plasticity decreased.However, when the in-situ synthesized TiN is distributed in layers, the LSTi composites achieve a balanced modulation of strength and plasticity at different ratios of N 2 concentration.The UTS was between 830-1100 MPa, and the plasticity could be regulated at 9%-30%, far more reasonable than UNTi composites.(3) Using N 2 to in-situ synthesize TiN layers via LPBF in a cross-gradient layered arrangement permits the establishment of higher back stresses in the soft Ti layer than in single-layered structure before the hard TiN layer begins to yield, leading to superior strength-ductility performance of the GLSTi composites.This work significantly extends the use of CPTi materials and the LPBF's ability to form gradient Ti matrix composites freely.

Figure 1 .
Figure 1.Pure Ti powder used in the experiment: (a) SEM morphology and (b) particle size distribution.

Figure 2 .
Figure 2. Distribution and construction diagram of LPBF UNTi, and in-situ synthesis of LSTi composites.

Figure 3 .
Figure 3. Mechanical testing and characterization.(a) 3D Model and dimensions of tensile samples with layered structure, (b) DIC testing devices.

Figure 5 .
Figure 5. Analysis of nitrogen and oxygen content and lattice constant.(a) Variation of N and O content with N 2 concentration; (b)-(d) are the relationship of α/α ′ -Ti lattice constant a, c and c/a between and N content, respectively.

Figure 11
Figure 11  illustrates the microstructure of typical LPBF layered TiN/Ti composites.Figures11(a) and (b) show the OM microstructures of the LSTi10-0.3and LSTi20-0.3samples, while figure11(c) shows high magnification SEM images of the LSTi20-0.3sample.It could be observed that an acicular martensitic microstructure was formed in the N 2 layer due to a higher cooling rate during the LPBF process[48].In contrast, the LSTi sample showed a coarse block or lath α/α ′ -Ti microstructure in the LPBF forming Ti in Ar, as shown in figure11(c).Figures11(d)-(g) show the interface between the Ti layer and the Ti-N 2 layer.In N 2 in-situ synthesis layer, as the laser scanning speed decreased, more N solid solutions were melting into the Ti matrix (figure11(e)).Therefore, many acicular martensites appeared in the case of the LPBF samples

Figure 13 (
Figure 13(d) presents the HAADF image of the LSTi-YZ sample.The light-colored lath precipitates are α ′ -Ti matrix phases, and the dark-colored block precipitates are TiN.Figures13(e) and (f) present the EDS and the large dark area also shows that TiN was uniformly combined.

Figure 13 .
Figure 13.The TEM results of the GLSTi-XYZ sample: (a) HR-TEM and SAED of α-Ti, (b) BF-TEM images, (c) HR-TEM and SAED of TiN, (d) HAADF images; (e) and (f) are EDS mapping results of the wireframe in (d).

Figure 17 .
Figure 17.Principle of LPBF in-situ synthesis with N 2 in Ti molten pool.(a) N 2 is decomposed into active N atoms/ions by laser irradiation, (b) Ti undergoes a chemical reaction with active N atoms/ions, (c) localized formation of TiN/Ti composites, (d) in-situ synthesis of TiN/Ti composites layer, (e) atmosphere changes back to 100% Ar and powder is re-laid, and (f) Ti matrix layer built by LPBF in Ar.

Figure 18 .
Figure 18.Strengthening mechanisms and performance comparison.(a) Comparison of measured and calculated values of yield strength with different N 2 contents.(b) Performance comparison of LSTi composites with other LPBF CPTi or TMCs.

Figure 19 .
Figure 19.Mechanism analysis of the strengthening and toughening mechanism in LSTi composites.(a) Comparison of ROM rule calculations, (b) three deformation stages of heterostructured materials, and (c) mechanical incompatibility deformation of TiN and Ti layers; GND maps for (d) CPTi-Ar, (e) UNTi10 and (f) GLSTi-XYZ sample, respectively.

Table 2 .
The specific design parameters of the LSTi composites.

Table 4 .
O/N element content results (mean value) of LPBF formed samples.

Table 5 .
Summary for tensile properties of LPBF-ed CPTi and UNTi composites.

Table 6 .
Summary for tensile properties of LPBF in-situ synthesized LSTi composites.