Diffusion Bonding of High-Alloyed Tool Steels with Maraging and Precipitation Hardening Steels

Steel composites for application as protective plates were produced via diffusion bonding. Cold work tool steels were combined with a precipitation hardening steel or a maraging steel using a deformation dilatometer and a vacuum furnace at 1150 °C for 8 h in total. Subsequently, a heat treatment was applied to achieve the final mechanical properties. The microstructure of the interface was investigated by optical microscopy, scanning electron microscopy, electron backscatter diffraction, energy dispersive X-ray spectroscopy and hardness measurements. The results are compared with a simulation of the diffusion of elements performed by MatCalc. Both composites achieve high hardness near the surface of the cold work tool steels, which can have a positive effect on the destruction of projectiles on impact. The influence of carbon diffusion can be observed up to a depth of 3 mm from the interface. As a result of carbon diffusion, soft zones are formed on both sides of the interface, which can be attributed to decarburization, phase transformation and carbide formation. The tough back layer is designed to absorb the remaining energy of the projectile. The combination of a hard front layer and a tough rear layer provides an optimal combination of materials against ballistic threats.


Introduction
Composite materials made of metals and ceramics have been used in the military industry for several decades to increase the protection against ballistic threats [1]. Their structure consists of a hard ceramic front layer, which shatters the projectile on impact and a tough metallic rear layer, which absorbs the remaining energy of the splinters [1,2]. The concept of a hard and a tough layer to improve the protection against ballistic threats can also be realized with two steel sheets with different mechanical properties [3,4]. This has the advantages of better processability and lower costs and could therefore also be used in the automotive industry for civil armored vehicles.
There are various processes for the production of these steel composites, such as explosion welding, diffusion bonding and vacuum hot roll bonding [5,6]. While vacuum hot roll bonding is commonly used for the production of stainless steel clad plates and is therefore well investigated, there are still limited available reports on diffusion bonding [7][8][9]. Although these methods have different approaches to join two sheets, the influence of some process parameters can be compared. To ensure high bonding strength, the interface must be free of oxides or non-metallic particles. Therefore, the sheet metal surfaces are ground or polished before they are joined under a protective atmosphere such as argon, hydrogen or vacuum [10][11][12]. Liu et al. [12] investigated the bonding strength and oxide formation in dependence of the vacuum degree. It was concluded that the size and density of interface oxides can be reduced by increasing the vacuum degree. Furthermore, the absence of oxides at the interface enabled more extensive diffusion of alloying elements, which led to a better bonding strength [12,13]. Diffusion depths can be further increased by increasing the temperature or time of the joining process [14]. This allows diffusion depths of about 10 and several 100 µm for substitutional and interstitial elements, respectively [6,[15][16][17]. Wang et al. [18] investigated the vacuum hot roll bonding of a SUS304 stainless steel clad and a plain carbon steel. Due to carbon diffusion, a zone with chromium carbides on the stainless steel side and a decarburized zone on the carbon steel side has formed. In between a martensitic zone was reported, which formed due to diffusion of Cr and Ni. The formation of these zones is strongly dependent on the chemical differences of the steels used.
Besides composites of metals and ceramics, the alloys which are used today for single layer armour plating are mostly martensitic steels with low alloy content and high hardness. Over the years a better protective effect has been achieved by increasing the hardness. In order to compensate for the increasing brittleness of higher strength plates, it would help to combine them with a layer of tough steel. The difficulty here is to find two steels which can be joined together to form a composite and achieve appropriate mechanical properties with the same heat treatment.
Therefore, in the present work, cold work tool steels were diffusion bonded with maraging and precipitation-hardening steels to produce a composite with a hard and a soft steel layer. The interface characteristics, carbide formation and microstructure influenced by diffusion processes were investigated after an optimized heat treatment.

Materials and Methods
The chemical composition of the steels used is listed in Table 1. The first composite (designated with A) consists of a cold work tool steel (tool steel 1) and a precipitation hardening (PH) steel. The second composite (designated with B) consists of another cold work tool steel (tool steel 2) and a maraging steel. The two cold work tool steels have the same main alloying elements, with the exception of niobium which is only present in the first and cobalt which is only present in the second. The samples had a cylindrical shape with a diameter of 8 mm and a height of 5 mm and were ground and polished on one side.
The setup of the diffusion bonding process can be seen in Figure 1a. Thermocouples were attached to each steel sample before placing them between the stamps of a deformation dilatometer DIL805A/D (TA Instruments, New Castle, DE, USA) with the polished surfaces at the interface. The samples were heated under a vacuum of 10 −4 to a temperature of 1150 • C and held for 45 min. The retention force of the stamps was 50 N at the beginning and was increased to 250 N after 2 min at temperature. The temperature and retention force during the diffusion bonding process can be seen in Figure 1b. After this bonding treatment the samples were diffusion annealed at 1150 • C for 7.25 h in a vacuum furnace with a vacuum degree of 10 −5 at peak temperature. Subsequently the heat treatment was performed without the use of a vacuum or a protective atmosphere, starting with the austenitisation at 1150 • C/1080 • C (composite A/B) for 10 min followed by an oil quench. The sheets were then tempered twice for 2 h at 540 • C.
For metallographic investigations samples were cut in half and hot mounted in PolyFast from Struers and polished up to 1 µm. Directly after polishing the samples were etched with Kalling 2 [19] to reveal the microstructure or etched with Murakami [19] to reveal the carbides. The etched samples were examined using an optical microscope Axio Imager M1m (Zeiss, Oberkochen, Germany). The carbide and element distribution at the interface was investigated on polished samples using a Clara scanning electron microscope (Tescan, Brünn, Czech Republic) and an X-Max energy-dispersive X-ray spectroscopy (EDX) detector (Oxford Instruments, Abingdon, United Kingdom). The acceleration voltage for these investigations were 15 and 12 kV, respectively. All EDX spot measurements were taken at a distance of a few 10 micrometers from the interface.
To obtain a deformation-free surface the samples were vibropolished with OP-S suspension from Struers for 7 h prior to EBSD measurements. In addition, the samples were plasma cleaned before being measured with an EBSD detector (Ametek, Berwyn, PA, USA) attached to a Versa 3D (FEI Company, Hillsboro, OR, USA). For the hardness mapping a Q60A+ micro hardness tester (Qness, Golling, Austria) was used to indent polished samples with HV 0.1 and a grid spacing of 100 µm.
Diffusion calculations were set up with the program MatCalc 6 (MatCalc Engineering, Vienna, Austria) to estimate the expected element distribution after diffusion bonding and heat treatment of the specimens.

Results
Cold work tool steels were diffusion bonded with a precipitation hardening steel using a deformation dilatometer. Figure 2a-e shows the final microstructure of composite A after diffusion annealing and heat treatment. Since the microstructure of the tool steel 1 did not differ from the heat treated single material, only the side of the PH steel and the interface are presented. In Figure 2b a martensitic microstructure with globular shaped carbides on the left and an austenitic microstructure with elongated carbides on the right can be seen. In between a 20 µm thick martensitic layer can be observed. From the interface to a depth of 750 µm an austenitic structure with twins and strongly attacked grain boundaries can be seen (Figure 2c). Further away from the interface, martensite begins to form (Figure 2d) and the proportion increases with increasing distance up to 1.8 mm, where the microstructure consists mainly of martensite. Adjacent to the area another form of martensite is found, which was attacked more strongly by the etching and expected to be nickel martensite and which becomes coarser towards the end of the sample (Figure 2e). Cold work tool steels were diffusion bonded with a maraging steel using a deformation dilatometer.  In order to investigate the interface in more detail, etching with Murakami [19] of both composites was performed. Figure 4 shows the results of this etching near the interface. Dense carbide networks can be seen in the vicinity of the interface in a distance of up to 1 mm and 650 µm for composite A and B, respectively. While small globular carbides and elongated carbides along austenite or prior austenite grain boundaries are found in both composites, rod-shaped carbides with a length of up to 12 µm are only present in composite B. In order to investigate the present phases near the interface, EBSD measurements were carried out. In Figure 5 the SEM images (a,d), inverse pole figure maps (b,e) and phase maps (c,f) are shown for composite A and B, respectively. Three areas can be seen for composite A, on the left side of Figure 5a the cold work tool steel with a homogeneous distribution of globular carbides, in the middle a martensitic zone without coarse carbides and on the right an austenitic zone with large grains and carbides at the grain boundaries. Compared to composite A, composite B (Figure 5d-f) shows a finer martensite with a lower carbide density on the left side and also a martensitic interface without coarse carbides in the middle. On the right side-the side of the maraging steel-a martensitic matrix with fine carbides near the interface can be seen. Further away from the interface, the carbides become coarser and can be observed in the matrix and at grain boundaries. To determine the chemical composition of the carbides detected, EDX spot measurements were carried out and the related backscattered electron images are shown in Figure 6. Since it is not possible to determine the crystallography with EDX measurements alone, only the chemical composition estimated from the strongest signals besides carbon are given. For composite A, carbides with different amounts of V, W, Mo and Nb were observed on the side of the tool steel 1. Bright and dark carbides were present in the PH steel, containing Nb and Ti, respectively (Figure 6a). The carbides of the tool steel 2 are similar to those of tool steel 1, with the exception that they contain no Nb. In the maraging steel, both the elongated and globular carbides consist of Ti and Mo. Nevertheless, no carbides were found which contained elements of the opposing steel. To observe the distribution of different elements and reveal their diffusion distances in the area of the interface EDX mappings were carried out, which are shown in Figure 7 next to the SEM images. For composite A, Cr is the most abundant element on both sides. While it is evenly distributed on the cold work tool steel side, Cr carbides can be seen on the PH steel side. Nb is present in only a small percentage, but still forms both elongated and globular carbides in the PH steel. V and W, along with Mo, form carbides in the cold work tool steels of both composites, but only small amounts are dissolved in the matrix. On the PH and maraging steel side, little to no V and W was detected. Regarding the maraging steel, Figure 7b shows a dense carbide network near the interface where Ti and Mo are present. The changed microstructure, the formed carbides and the diffusion of the elements influence the hardness. Figure 8 shows microhardness mappings of the two composites. Both achieve the highest hardness on the side of the cold work tool steel and the lowest hardness near the interface on the PH and maraging steel side. The width of the soft zone is approximately 1.8 mm for composite A and 0.6 mm for composite B. Although no change in the microstructure was observed with optical microscopy, a softening of the cold work tool steels can be seen up to 1 mm from the interface.

Discussion
The aim of the present study was to evaluate the possibility of diffusion bonding of cold work tool steels with precipitation hardening and maraging steels. The microstructural changes were investigated in detail for the composites after the applied treatment. The evolution of the microstructure at and next to the interface can be explained by the diffusion of the alloying elements. Temperature and time are the decisive factors in the diffusion process and therefore the development of the microstructure depends on this parameters during the diffusion bonding. The subsequent heat treatment only has a small influence. In total, the material was treated at 1150 • C for 8 h prior to the heat treatment, resulting in a modified microstructure and the formation of carbides for both composites. To estimate the penetration depth of different elements achieved during diffusion bonding, the simplified approach of Einstein and Smoluchowski can be used. For this purpose, the diffusion length of different elements can be estimated by using Equation (1) [20,21]: where x is the mean diffusion length, D is the diffusion coefficient and t is the time. Using diffusion coefficients for the diffusion in pure iron from the literature [22,23], one obtains mean diffusion lengths between 10-50 µm for substitutional alloying elements. The interstitially dissolved carbon has higher mobility and has a mean diffusion length of 760 µm.
To compare this calculated mean diffusion paths, diffusion simulations were performed with MatCalc. The results of these simulations are shown in Figure 9. The dark grey areas represent the cold work tool steels and the light grey areas represent the PH and maraging steel, respectively. For better visibility of the profiles, only the area near the interface is shown for the substitution elements. At the edge of the diagrams, i.e., at 100 µm distance from the interface, all substitution elements have reached their base composition. Therefore, the initial estimation of the mean free diffusion length with 10-50 µm is in good agreement with the simulation. The different diffusion distances of the various elements can be seen in Figure 9a,b, but the EDX investigations were unable to verify this.  The carbon values of the base alloys are marked on the left and right scale. The simulation of the carbon distribution for both composites shows a smooth transition from the high carbon content on the cold work tool steel side to the low carbon content on the PH or maraging steel side. The carbon content of 0.38 wt% at the edge of the nominally carbon-free steels shows diffusion of the interstitially dissolved carbon over the whole sample length. Compared to that, a change in the microstructure can be seen up to 3 mm from the interface in Figures 2a and 3a. The difference between the simulation and the microstructural changes in the investigated specimen can possibly be attributed to the influence of carbide formation, which was not considered in the simulation.
The microstructure near the interface is therefore mainly influenced by the alloying elements and their diffusion. For composite A, an approximately 20 µm thick martensitic layer has formed at the interface in which only fine carbides are present (Figure 5c). Investigations on stainless steel clad plates by Wang et al. [18] showed martensite formation at the interface. It was concluded that at the interface, due to plastic deformation, austenite locally transforms into martensite. Controversially, Liu et al. [12] reported that the martensite is formed due to the diffusion of ferrite stabilizing alloying elements, especially Cr. Due to the high chromium content of the PH steel in composite A, this may be a possible explanation in the present study for the martensitic layer shown in Figure 5b. In addition, the layer thickness is in good agreement with the simulated diffusion length for substitutional elements.
In composite B the area around the interface is fully martensitic with a transition zone free of coarse carbides (Figure 5f). Furthermore, Figure 3b shows a light etched zone on the side of tool steel 2, which can be attributed to the diffusion of carbon. Chen et al. [13] studied the interfaces of stainless steel plates and found decarburized zones of 75 to 91 µm for temperatures between 1100-1200 • C and a holding time of 2 h. The hardness measurements in Figure 8 revealed that the extent of the decarburized zone is actually larger than it appears in the etched samples. The softening of the cold work tool steel has an approximate width of 1.2 and 0.8 mm for composite A and B, respectively, which is in the same order of magnitude as the calculated distance for the diffusion of carbon.
In the PH steel, coarse grained austenite is stabilized by carbon up to 1.8 mm from the interface. Comparing the microstructure in Figure 2a with the hardness measurement in Figure 8a, it can be seen that the low hardness of austenite is responsible for the soft zone in the PH steel. In a distance of 750 µm from the interface, an increasing amount of martensite was observed. A possible explanation for this is that the amount of carbon is no longer sufficient to stabilize the austenite completely and thus, martensite is formed once the martensite start temperature is reached. With a larger distance to the interface, the carbon content decreases further, resulting in the formation of nickel martensite at a distance of 1.8 mm from the interface, which is typical for a PH steel (Figure 2d) [24]. The carbon diffusion does not lead to any microstructural changes in the maraging steel of composite B, but the attack of the etchant becomes stronger with increasing distance from the interface. This can be explained by the fact that carbon refines the microstructure and thus, a lower etching attack occurs [25].
Since PH steel and maraging steel are almost carbon free, the formation of carbides in the diffusion bonded specimen is due to above mentioned carbon diffusion. The strongest available carbide formers are Nb, followed by Cr for the PH steel and Ti, followed by Mo for the maraging steel [26]. With the occurrence of Cr carbides along the grain boundaries near the interface of composite A, intergranular corrosion can occur locally [6]. The investigated PH 17-4 steel forms intermetallic phases containing Cu and Ni during aging [27]. Since none of these elements has a high affinity to carbon, carbide formation has only a minor influence on precipitation hardening. On the maraging side of composite B, Ti and Mo rich carbides occur, up to a distance of 650 µm from the interface (Figure 6b). With Ti being the stronger carbide former, it is most likely that TiC has formed, in which Mo is dissolved [26,28]. As less Ti is available in the matrix due to carbide formation, there is a reduced effect of precipitation hardening, resulting in a soft zone near the interface (Figure 8b).
Several zones have formed in the two-layered steel composite, which will have positive effects on the ballistic protection properties. The first few millimetres of the surface of the cold work steel remained unaffected by diffusion and could therefore maintain high hardness. This will have a positive effect on the destruction of the projectile on impact. The gradual hardness transition in the decarburized zone towards the well bonded interface should create a stable bond, which is essential for the ballistic testing. Although the carbides in the PH and maraging steel weaken the grain boundaries, the soft zone close to the interface helps to prevent brittle fracture [28,29]. The remaining carbide-free tough zone of the PH and maraging steel should absorbs the remaining energy of the projectile and thus contribute to a positive ballistic test result.

Conclusions
In the present study, cold work tool steels were diffusion bonded with PH and maraging steels using a deformation dilatometer. The two composites were characterized in terms of microstructure, carbides, diffusion of alloying elements and mechanical properties. The following conclusions can be drawn: (1) By combining diffusion bonding with application of pressure and subsequent annealing, metallically bonded steel composites were successfully produced. (2) The combination of cold work tool steel with PH steel resulted in the formation of a martensitic layer at the interface due to the diffusion of substitutional elements. Furthermore, the diffusion of carbon into the PH steel led to the stabilisation of austenite, the formation of carbon martensite and the formation of Cr and Nb carbides. (3) With the combination of cold work tool steel and maraging steel, a decarburized zone could be observed in the micrographs. While no influence by diffusion of the substitutional alloying elements could be observed, the diffusion of carbon on the side of the maraging steel led to a weak etching attack and the formation of Ti and Mo rich carbides near the interface. (4) For both compounds, a softening of the cold work tool steel was found by hardness measurements near the interface, which can be explained by the diffusion of carbon. In the PH steel close to the interface, a soft zone of 1.8 mm was observed, which can be explained by the low hardness of the stabilized austenite. In the maraging steel, the formation of Ti and Mo rich carbides lowered the alloy content in the matrix. This reduced the formation of intermetallic phases and also resulted in a soft zone of 0.6 mm near the interface. (5) The diffusion simulation showed good agreement with the observed diffusion lengths of the substitutional alloying elements. After diffusion bonding at 1150 • C and 8h holding time a mean free diffusion length of 10-50 µm was calculated for substitutional elements. The higher mobility of the carbon leads to a diffusion over the whole sample length. (6) The presented composites appear to have great potential for application as protective plates.
With a hard layer to destroy the projectile and a tough layer to absorb the remaining energy, optimal properties were achieved. The well bonded interface with the surrounding soft zone should prevent brittle failure or delamination. Funding: Funding of the Austrian BMK (846933) in the framework of the program "Production of the future" and the "BMK Professorship for Industry" is gratefully acknowledged.

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