Introduction

In relation to conventional carbon-hardened steels, maraging steels are a class of ultra high-strength steels characterized by very little carbon contents [Ref 1, 2]. The high mechanical resistance of these steels is attained by the precipitation hardening mechanism during aging. The traditional maraging steel grades contain about 8-12% cobalt in addition to other alloying elements. The 18 Ni Co-free maraging steels were developed by Inco Alloys Internatonal [Ref 3] in early 1980 due to a sharp rise in cobalt pricing in the late 1970s. As a result, maraging steels are important candidate materials for several critical applications such as rocket motor cases, cryogenic missiles and landing gears. For many of these applications, welding is an important means of fabrication. Electron beam (EB) welding is a process which produces coalescence of metals with extremely high-beam power density (about 105-107 W/cm2) and has tremendous penetrating characteristics, with lower heat input, high-velocity, and the heat-affected zone is much smaller than that of any arc welding process. It is suitable for low-carbon steels like maraging steels. Therefore many research studies have been done on EB welding of 18 Ni Co-free maraging steels [Ref 4-6], but less attention was paid to weld defects such as pits and burn-through holes in weld face.

In the current work, the material of 18Ni Co-free maraging steel in the form of 2-mm-thick sheets was received in the same heat treatment condition: 850 °C1 h solution-annealed and 480 °C4 h aging. The two specimens were prepared from two batches with the same nominal composition Fe-18Ni-3.0Mo-1.4Ti. Microstructures of the base materials and the weld joints were examined. Special attention was paid to the weldability and weld defects of the examined materials.

Experimental Work

The chemical compositions of the examined 18 Ni Co-free maraging specimens are given in Table 1. It is shown that both of the specimens have similar chemical compositions. Two specimens were EB welded with the same welding parameters but their weldability exhibits different characteristics. One of the specimens, recognized as A (Fig. 1a), is weldable, its weld joint structure exhibits integrity from the top surface; while the other, recognized as B (Fig. 1b), has obvious welding defects of pits and burn-through holes in weld face. The corresponding microstructure is investigated in the following section.

Table 1 The main chemical composition and impurity in examined sample (wt.%)
Full size table
Fig. 1
figure 1

Appearances of the welding seam. (a) Specimen A and (b) specimen B

Full size image

Optical microscopy and scanning electron microscopy (SEM) were employed to observe the microstructures of base materials and their weld joints. Both the longitudinal (L-B) and transverse (B-T) sections of the weld joints were etched and examined, and the etchant used was: 100 ml H2O, 8 g (NO2)2C6H2OH (Picric acid), 5 ml H2SO4, 105 ml H2O2, and four drops of HF. X-ray diffraction (XRD) analysis was carried out on base materials without etchant to reveal the phases. XRD was carried out using step scan mode with step size of 0.02, time per step of 3 s and angular interval 62°-92°. Radiation CuKα (1.54056 A) was used with 40 kV and 100 mA.

Results and Discussion

Base Materials

The microstructures of the base materials A and B are shown in Fig. 2. The grain size of A is much larger than that of B, about 10-20 μm and 3-6 μm, respectively. Besides the grain boundaries, the grains consist of randomly dispersed precipitates and bundles of parallel sub-grain boundaries that can be easily seen from Fig. 2(a) in the same etched condition.

Fig. 2
figure 2

Optical micrographs of base materials. (a) Specimen A and (b) specimen B

Full size image

Scanning electron microscopy observation from Fig. 3 indicates that the microstructures of the two materials are lath martensite. From the figures, the grain size of B is remarkably smaller than that of A, which corresponds to the results of optical microscopy. Quite a significant amount of reverted austenite distributes along grain and lath boundaries, and the preferential distribution is affected by alloy-rich segregations during aging [Ref 7]. In addition, a small amount of austenite appears as fine particles within the matrix of A, while much less particles are observed in B, which indicates that the austenite particles in specimen B concentrated in grain boundaries much heavily. A visual representation of information from Fig. 3(a) and (b) shows that the content of austenite especially that distributes along the grain boundaries of B is higher than that of A.

Fig. 3
figure 3

SEM micrographs of the base materials. (a) Specimen A and (b) specimen B

Full size image

It is interesting to note that the austenite in grain boundaries exhibits the appearance of granular instead of plate-like which was reported previously [Ref 8]; this is mainly due to the different etchant that was used. It demonstrates that different etchants lead to different appearances of reverted austenite.

Figure 4 shows the XRD comparison between two materials. The peaks of (200)α, (211)α and (220)γ can be easily seen from diffraction diagrams, whereas the peaks of intermetallic phase, such as Ni3Ti, Fe2Mo, Ni3Mo are not present, probably because their contents were too low. The austenite content was obtained by the direct comparison method. Quantitative analysis by XRD is based on the intensity of diffraction pattern of austenite or martensite phases in the samples. Without considering carbides, we can obtain the basic equation relating the diffracted intensities and the austenite volume fraction (Cγ) by the direct comparison method in an alloy containing only two phases [Ref 9]:

$$ C_{\upgamma } = \frac{1} {{\frac{{I_{\upalpha } \cdot K_{\upgamma } }} {{I_{\upgamma } \cdot K_{\upalpha } }} + 1}} $$
(1)

Here Iα and Iγ are pattern intensity of martensite and austenite given from XRD results; and Kα and Kγ are their reflection ability, respectively, which can be calculated with crystal structure and other correlative parameters.

Fig. 4
figure 4

XRD of the base materials. (a) Specimen A and (b) specimen B

Full size image

In the present case, (220)γwas compared with (200)α and (211)α, the average value of two compared groups was calculated to minimize the effect of preferential direction, and the austenite content for specimen A and B are 6.8 and 9.6%, respectively.

Weld Joints

Figure 5 shows the microstructures of weld joints. The weld width especially at the bottom of B (Fig. 5b) is wider than that of A (Fig. 5a), which resulted in bigger welding area and larger quantity of molten metals during EB welding process. All welding microstructures of both A and B exhibit a typical dendritic morphology (Fig. 5c, d), and the growth direction of dendritic crystals is basically vertical to the welding direction (Fig. 5a, b). This is because in the high speed of EB welding, the rear part of molten pool is distorted to a tear-drop shape instead of an ellipse. The temperature in the center of welding seam was higher than 1,800 °C, while the weld width was about 2 mm, which resulted in the enormous gradients of temperature and provided the advantageous condition for the growth of dendritic crystals. However, differences in solidification morphology were also observed within the transverse (B-T) section of the weld metal. The grains in specimen A are cystiform-dendritic, which have well-developed secondary dendrites stretched from main dendrites (Fig. 5e), and many of them are in irregular cross shape. In specimen B, grains exhibit more circular microstructure with very few secondary dendrites present (Fig. 5f), presumably owing to the more rapid rate of temperature changing near the crystallization front in the latter specimen, which results in unfavorable growth conditions for secondary dendrites.

Fig. 5
figure 5

Optical micrographs of weld joints. (a) and (b) Weld seam of specimen A and B (L-B section); (c) and (d) Fusion zone of specimen A and B (L-B section); and (e) and (f) Fusion zone of specimen A and B (B-T section), respectively

Full size image

The heat-affected zones of the base materials were also examined by SEM. Nearly similar microstructures are seen in Fig. 6. The region adjoining the weld metal gets heated to high temperatures in the austenite phase field before cooling rapidly. Energy provided by EB welding was high enough to lead to grain recrystallization [Ref 10]. As in this case, crystal grains grew and the grain sizes of specimen A and B in this area are about 45μm and 35μm, respectively, which are 2 and 6 times larger than that before EB welding. The grain boundaries near weld junction exhibit grooving, which is easily seen in Fig. 6(a). It is believed that the presence of inclusions in maraging steel weldments liquified locally in impurity-rich regions and the grain boundaries were markedly attacked by the etchant [Ref 6].

Fig. 6
figure 6

SEM micrographs of the heat-affected zones near weld junction (L-B section). (a) specimen A and (b) specimen B

Full size image

Analysis on Weld Defects

Both the specimens A and B are 18 Ni Co-free maraging steels with the same nominal chemical composition and EB welded with the same welding parameters. However, the weld joints of specimen B get burn-through problem. An explanation is offered in terms of different forces on molten pool in the process of EB welding. Figure 7 shows the forces on molten pool before the weld defects appear.

Fig. 7
figure 7

The force on molten pool

Full size image

In Fig. 7, fG is the pressure duo to gravity of melting metals, fE is stress of high-energy EB and fS is surface tension of molten pool. The value of fG depends on quantity of molten metal during EB welding, fE depends on welding voltage and current, and fS can be determined by the category of metal and the surface condition of molten pool. Besides, the adsorption between solid and liquid state and the cohesion in molten metal have a negligible contribution in keeping the stability of molten pool from burn-through [Ref 11]. Obviously, fG and fE are the main forces that lead to weld penetration, while fS gives a negative force. When fS cannot counteract the composition forces of fG and fE, weld defects arise.

In this study, EB welding was carried out without filler, the metals in molten pool have the same chemical composition as base materials. Therefore, the microstructures of the two materials with different composing phases and their relative contents have distinct contributions to thermal conduction performance, which resulted in diverse heat accumulation and quantity of melting metals, and have a great influence on weldability.

For 18 Ni Co-free maraging steel, electron heat conduction is a dominant mechanism of heat transport. Comparing specimen A with B, the contrast of grain size is remarkable; the grain size of specimen A is about two times bigger than that of B in base materials, and the grain size is also bigger in heat-affected zones when the grains recrystallize. Because of the smaller grain size, specimen B has more grain boundaries, which scatter electrons more strongly, and this inhibits heat conduction. In addition, there is larger quantity of reverted austenite in specimen B, especially in grain boundaries. Unfortunately, the thermal conductivity of austenite is poorer than that of martensite [Ref 12]. As discussed above, grain boundaries and reverted austenite are weak link to heat conduction, which makes a poorer thermal conductivity of specimen B than that of A.

From the longitudinal (L-B) section of molten pool, specimen B shows a larger welding area, which illustrates specimen B has more quantity of molten metals during EB welding process and an increasing value of fG. Besides, from the microstructure of weld joints in transverse (B-T) section, the temperature gradient in the area near the crystallization edge of specimen B must be greater than that of specimen A, which is consistent with the above conclusion of poorer thermal conductivity of specimen B. During EB welding, the temperature in the region adjoining the weld metal was nearly the same in two specimens, and approached fusion point. Specimen B had a greater temperature gradient and a wider weld width, which led to distinct heat accumulation and larger value of fG than that in specimen A. Meanwhile, with the increase of molten metal, the width at the bottom of molten pool increased, the weld line tended to become parallel to the direction of thickness and shortened, and the fS decreased. The stress fE remained constant in the same welding condition. As a result, the value of fG plus fE exceeded fS, and burn-through problems happened. Because of poor thermal conduction performance in specimen B, the energy input should be reduced when EB welding in contrast with specimen A, which could give advantageous effect to avoid the welding defects.

Conclusions

(1) The microstructure of the two 18 Ni Co-free maraging steels investigated in the study is lath martensite. The austenite distributes in plate form along granular and lath boundaries while some as fine particles within the matrix. Compared with specimen B, the grain size of specimen A is about two times bigger and has much less content of reverted austenite.

(2) All welding microstructures of both specimen A and B exhibit a typical dendritic morphology. However, the microstructure of specimen A weld in transverse (B-T) section is cystiform-dendritic with well-developed secondary dendrites, while B exhibits a more circular microstructure with few secondary dendrites. Compared with the change in specimen A, the grains near weld junction in specimen B recrystallized and grew markedly.

(3) In comparing the weldability of the two materials, it is noted that welding defects of burn through are contributed greatly by small grain size and high volume of austenite content, which induces poor thermal conduction performance.