Ni Interlayer to Improve Low-Pressure Diffusion Bonding of 316L SS Press Fit Tube-to-Tubesheet Joints for Coiled Tube Gas Heaters

The Department of Nuclear Engineering at the University of California, Berkeley (UCB) is developing a coiled tube gas heater (CTGH) design, depicted in Fig. 1, that supports the use of Brayton-cycle power conversion for advanced nuclear reactors. UCB has developed an initial main design to transfer heat from a low-pressure 700 °C, fluoride-salt coolant (at approximately 0.1 MPa) to compressed air (at approximately 1.8 MPa), which is then expanded through a gas turbine to produce power [1–3]. UCB has also developed an alternative heat-exchanger design that uses supercritical carbon dioxide (S-CO₂) instead of air, with low-pressure (0.1 MPa) liquid sodium metal as the primary coolant; this design requires operation at higher pressures (up to 20 MPa and 550 °C) [4]. Both of these designs thus require that the heat-exchanger tubes withstand the challenging combination of high-temperature and a high-pressure differential between the inside and the outside of the tubes. Furthermore, in contrast to pressurized-water-reactor steam-generator tubes that experience a net internal pressure and are loaded in radial tension, the tubes in a CTGH experience a net external pressure and are loaded in radial compression. This net external pressure works to weaken the tube-to-tubesheet joint. Due to these unusual conditions and concerns about conventional-fusion-welding defects [5] in the material, UCB has investigated diffusion-bonding (DB) technique for making the tube-to-tubesheet joints. UCB has focused specifically on joining tubes and tubesheets made out of 316L stainless steel (316L SS), because it is one of the candidate materials for service in the elevated temperatures, high-pressure differentials, and corrosive atmospheres of the CTGH.

DB joints have been extensively investigated, and the main parameters are well known [6]: the roughness of the surfaces being joined and the temperature, time, atmosphere, and pressure of the heat-treatment process (i.e., the diffusion-bonding process). In conventional DB, the two pieces of the joint are pressed together with constant pressure during the heat treatment, but the geometric configuration of the tubes and tube manifold (tubesheet) in the CTGH (Fig. 1, left) makes it difficult to apply such pressure on those DB joints. Therefore, UCB has developed a novel joint procedure for low-pressure DB of these joints: Matching tapers are machined on the end of the tube and in the hole in the manifold, and the two components are press-fit together with low pressure prior to the DB heat treatment. The inset on the left part of Fig. 1 shows a mock-up of one of these tube-to-tubesheet joints, fabricated to simulate this specific area of the heat exchanger.

Several different DB procedures for making these DB joints have been investigated at UCB. Experiments with tapered interference fits, tube expansion, and hybrid friction diffusion bonding (HFDB) have been successfully conducted at UCB with thin-walled 316L SS tubes (OD: 6.35 mm (1/4 in.) and ID: 4.57 mm (0.18 in.)) [7–10]. This paper specifically describes the experiments on the procedure of using commercial plating techniques to apply a nickel (Ni) interlayer of various thicknesses onto the tubes to support the tube-to-tubesheet tapered interference fits. Such an interlayer, made from a “soft” material, was suggested as an improvement for the DB process of stainless steel by Kazakov [6] and Yeh and Chuang [11] and was described by many others [12–16]. Ni was chosen here as the interlayer material since it is one of the major components of the 316L SS. The joint quality was evaluated quantitatively by pull-out testing as well as qualitatively by optical microscopy (OM) and scanning electron microscopy (SEM).

In UCB's previous studies [7–10] of DB joining of thin-walled tubes to tubesheets, the pull-out force during testing was limited, by the small tube cross-sectional area, to approximately 8.7 kN. In the present study, rods were used instead of tubes to allow higher force to be developed on the joint. The tube was simulated by a rod (OD: 6.35 mm (1/4 in.)), and the tubesheet was simulated by a 12.7 mm (1/2 in.) thick collar. Figure 2 shows a drawing and a picture of a fabricated sample. Commercial 316L SS was used to fabricate the tube-to-tube sheet samples. The chemical composition of the 316L SS as provided by the vendor is given in Table 1. The composition was subsequently verified using SEM/energy dispersive X-ray spectroscopy (EDS).

The samples were fabricated on a computer numerical control machine or on a manual lathe, utilizing standard cutting tools. Three sample sets were fabricated (three samples in each group): group 1 samples with a 5 μm Ni interlayer, group 2 samples with a wood's nickel strike (WNS) interlayer, and group 3 reference samples without an interlayer. The 5 μm Ni interlayer was applied with a commercial electroless plating kit, following the manual instruction [17]. WNS is an electroplating process during which the protective oxide layer on the stainless steel is etched, while a thin Ni layer (approx. 0.25 μm [18]) is deposited at the same time. The literature contains several WNS procedures/parameters [19–23], of which this study used the following:

1. (1)

   24 g nickel chloride.

2. (2)

   12.5 ml HCl.

3. (3)

   Distilled water, to 100 ml solution.

4. (4)

   Ni anode.

5. (5)

   Cathodic process: The positive pole connects to the plating part; the negative pole connects to the Ni anode; current density of 5.4 × 10⁻⁴ A/mm² for 2 min and then current density of 1.55 × 10⁻⁴ A/mm² for an additional 2 min.

where P is the interference pressure as a result of diametrical interference; D_t is the tube diameter, where o indicates the outer diameter and i the inner diameter; Δ_t+tol is the upper tolerance of the tube outer diameter; D_s is the tubesheet diameter, where o indicates the disk diameter and i the hole diameter; Δ_s−tol is the lower tolerance in the hole diameter; E is the Young's modulus, where t indicates the tube and s the tubesheet; and ν is the Poisson's ratio.

The heat treatment was conducted in a vacuum furnace at 1000 °C for 1000 min, as recommended by Haneklaus et al. [9]. A moderate vacuum (approx. 10⁻¹ mBar) was chosen here to allow slight oxidation of the surfaces, which served to highlight the interfacing surfaces in the postmortem micrographs and thereby to make easier the evaluation of the DB quality.

The quality of the DB joint between the tapered tube-to-tubesheet fitting surfaces was determined primarily by measuring the load required to pull the rod out of the tubesheet. Quantitative pull-out testing was described by Massey and Jones [25] as an “established method” to determine the quality of heat-exchanger joints. Kikuchi et al. [26] used this criterion to determine the quality of the tube-to-tubesheet joints of the Japan sodium-cooled fast reactor (JSFR) heat-exchanger tubes, which have to withstand similar pressure differentials as the tubes in the UCB CTGH do. In contrast to the traditional configuration of pull-out testing [25,26] though, the pull-out testing in this study was conducted with a computer-controlled tensile/compression testing machine, made by MTS Systems Corporation, Eden Prairie, MN. The sample was mounted in an adaptor, and the test was run as a tensile test. Figure 3 depicts the adaptor apparatus and the sample setup in the MTS testing machine.

Table 2 summarizes the results of the different sample groups and the improvement (by percentage) for the different categories. The mean pull-out force was 9.07 kN (o′ = 3.8 kN) for the reference samples, 10.7 kN (o′ = 1.5 kN) for the samples with the 5 μm Ni interlayer, and 12.1 kN (o′ = 5.4 kN) for the samples with WNS. The mean pull-out force on these rods can be used to estimate the stress in a tubular cross section.

Figure 4 shows the representative, post-testing pictures of the rod surfaces on which DB occurred: a reference sample (left), a WNS sample (center), and 5 μm sample (right) are depicted. The improved DB on the WNS sample resulted in visible areas where base material is peeled off from the rod. Areas with such sound DB are smaller on the 5 μm sample and much smaller on the reference sample. The light oxidation of the surfaces, as shown in Fig. 4, suggests insufficient surface contact, since with ideal surface-matching, the touching areas should not have been exposed to the oxygen environment.

In addition to quantitative pull-out testing, which was the success criterion of the study, qualitative analysis using OM and SEM was also conducted. Figure 5 shows representative SEM pictures of a reference sample (left) and a sample with 5 μm interlayer (right). The pictures show that the low pressure applied prior to DB was not sufficient to overcome the surface asperities induced by the cutting tool. Though the pressure was not sufficient to force the induced asperities to yield, partial DB could be observed with the OM and SEM.

When the asperities yielded and the stainless steel's protective oxidation layer broke, sound DB was observed. In these areas, the material grains grow through the interface matching line (tube-to-tubesheet interface). Figures 6 and 7 (right) are micrographs of these areas, taken with OM and SEM/electron backscatter diffraction (EBSD), respectively. Figure 6 depicts the material grain growth through the tube-to-tubesheet interface line to create a continuous metallographic structure. Figure 7 compares the partial DB area (left picture) to the sound DB area (right picture); the sound DB is characterized by a continuous metallographic structure. The figure reveals that the tubesheet grains (bottom portion of pictures of Fig. 7) are twice the size of the rod (tube) grains (top portion), which is a consequence of the different manufacturing processes for the two components.

This study confirms that low-pressure DB of press-fit, 316 L stainless steel tube-to-tubesheet joints (with the tube simulated by a rod) meets the overall success criteria, namely, a pull-out force exceeding the force required for tube rupture. This study also suggests that the DB joint can be improved by using a Ni interlayer (Table 2; Fig. 4). The pull-out force increased by 18% when 5 μm, electroless-plated Ni interlayer was used and by 33% when an electroplated WNS interlayer (0.25 μm thick) was used. The SEM analysis and the visual observation of slight oxidation of the surfaces confirm the challenge of manufacturing perfectly matching surfaces. Imperfect matching of surfaces resulted in areas where the rod's surface could not be pressed against the tubesheet's surface with the necessary pressure, resulting in only partial DB of the two parts, as shown in Figs. 5 and 7 (left). The sound DB areas were characterized by the growth of material grains through the tube-to-tubesheet interface line to create a continuous metallographic structure (Figs. 6 and 7 right). The imperfection in surface matching may be resolved by using additional manufacturing techniques, such as grinding, to create smoother surfaces and thus better surface-matching.

The authors want to express their gratitude to the operators of the Mechanical Engineering Machine Shop at UCB, without whose guidance and assistance this work would have not been possible. This research was performed using funding received from the Nuclear Energy University Programs of the Office of Nuclear Energy in the U.S. Department of Energy (DOE).

CTGH =

coiled tube gas heater

D_si =

tubesheet hole diameter, mm

D_so =

tubesheet disk diameter, mm

D_ti =

tube inner diameter, mm

D_to =

tube outer diameter, mm

DB =

diffusion bonding

E =

Young's modulus, Pa

E_s =

Young's modulus of tubesheet material, Pa

E_t =

Young's modulus of tube material, Pa

EBSD =

electron backscatter diffraction

EDS =

energy dispersive X-ray spectroscopy

F_p =

applied force, N

HFDB =

hybrid friction diffusion bonding

JSFR =

Japan sodium-cooled fast reactor

L_c =

contact length, mm

OM =

optic microscopy

P =

interference pressure as a result of diametrical interference, Pa

r =

collar bottom inner radius, mm

R =

collar outer radius, mm

SEM =

scanning electron microscopy

S-CO₂ =

supercritical carbon dioxide

UCB =

University of California, Berkeley

ν_s =

Poisson's ratio of the tubesheet disk material

ν_t =

Poisson's ratio of the tube material

WNS =

wood's nickel strike

δ =

interference depth, mm

Δ_s−tol =

lower tolerance in the hole diameter, mm

Δ_t+tol =

upper tolerance in the tube outer diameter, mm

θ =

taper angle

μ =

static friction coefficient