I. Introduction
Our previous investigation of the interfacial regions in the ”(p-n) thermocouple” / Cu- connector assembly (Project No. 2012-3067-0616) was aimed at providing a framework for the development of long-lasting joints between the thermoelectric materials and couple interconnects. The most salient features of the “Thermoelectric Convertor” design discovered in this study are as follows:
1). The Thermoelectric Convertor (hereafter TE convertor) utilizes PbTe for the n-type thermoelement and one of the pseudo-binary compounds (GeTe)x(AgSbTe2)100-x, commonly referred to as TAGS-x (e.g. TAGS-80, TAGS-85), for the p-type thermoelectric component of the thermocouple assembly. It is important to mention that the semiconducting materials used are not homogeneous, i.e. they are not single-phase! This thermoelectric materials (hereafter TE materials) combination is commonly used for “500-700 K applications”.
2). Pure Lead (Pb) was used for soldering of thermoelectric elements to the Cu-connector. Close inspection of the interconnect cross-sections revealed a very thin, often fragmented (PVD) Mo-(inter-) layer adjacent to the Cu-substrate. This can be appreciated just by looking through the micrographs shown in Figs. 1 and 2. The Back-scattered Electron Images (BEI’s) were taken somewhere in the central part of the samples (joints).
Fig.1: Back-scattered Electron Images (BEI’s) of the transition zone between Cu and n-type thermoelectric material (PbTe) in the MT 12-187 sample. The “irregular” Mo-layer is clear visible between Cu-connector and Lead, likewise Ni,Sb-based interlayer is present at the PbTe/Pb interface. (Note that some SiC-particles (“grinding debris”) are embedded in the soft Pb-layer)
Fig. 2: BEI’s of the central part of the transition zone between Cu and p-type thermoelectric material (TAGS-x) in the MT 12-186 sample. The often intermittent Mo-layer is clear visible at Cu-connector contact surface (Note that some SiC-particles (“grinding debris”) are embedded in the soft Pb-based layer)
3). Most likely, the Mo-interlayer at the Cu/Pb contact surface should act as a diffusion barrier to ensure that chemical interaction of Cu with the constituent of the thermoelements will not increase the thermal and electrical contact resistance. It is, however, to be noticed here that in some areas of the transition zone where Mo-interlayer was disintegrated, the formation of Cu-Sn intermetallic compounds (IMC) was detected (Fig. 2a). Perhaps, quality (soundness) of the Mo-interlayer is also critical to the structural integrity of the joints.
4). The p-thermoelement (hereafter P-Leg ) of the TE convertor is “segmented”, because although the TAGS material is “phase-stable” up to 900 K [1], its sublimation (decomposition) rate above 673 K is above the desired value and its operation is therefore limited to ~ 673 K in the (thermo)couples. Note that sublimation of the TE materials near the hot junction of the couple results in a cross-sectional reduction of the thermoelements and increase in thermoelement resistance as well as in the interfacial contact resistance of the (thermo)couples. Most likely, to suppress sublimation the P-Leg in the “Thermoelectric Convertor” design is segmented to a top (Pb)SnTe segment operating between 973 K and 673 K [2]. The (Pb)SnTe has a much lower sublimation rate that the TAGS materials in this temperature range [3]. It is also important to mention that in the “Fokker design” up to 5 μm SnTe-layer (presumable deposited by CVD method) was found all around the cylindrical TAGS-segment (i.e. the TAGS material was encapsulated into SnTe).
5). As to the Ni,Sb-based interlayer detected at the PbTe/Pb-interface in the N-Leg of the “Fokker Thermoelectric Convertor” design (Fig. 1), it is still unclear at this moment why this type of layer is introduced between the n-PbTe TE material and Pb (“solder”)! What for do we need Ni and Sb as constituents of the layer? What is the point? Is this merely a certain sort diffusion (reaction) barrier between PbTe and Pb? Or, perhaps, this layer can prevent the base semiconducting material from the excessive sublimation. If this layer must be there, then how was it produced ?
The purpose of the present investigation is fourfold.
- To investigate a possibility of soldering of the thermoelements to Cu-connector with pure Pb using radiation rather than induction heating. The overriding objective here is to find a joining technique which allows significantly increase throughput in the TE convertor manufacturing. Samples produced in the present work will be used for the assessment of the thermoelectric performance of the assembly.
- To demonstrate that high-temperature soldering with lead as a filler metal can be performed under flowing mixture of Nitrogen with 5 vol. % of Hydrogen (“Forming gas”) instead of using pure Hydrogen. This will made soldering process less costly and much more safely.
- To take a look at the feasibility of electrolytic Nickel as an alternative to the Mo- reaction barrier interlayer at the Cu/Pb contact surface. The use of the electrochemical deposition of Ni onto Cu-substrate instead of using PVD process for the fabrication of Mo-layer may reduce total production cost significantly.
- To explore solderability of “bare” Copper connectors directly to the TE materials, PbTe and TAGS, without any reaction barrier (Mo or Ni) or encapsulating layer (SnTe for the TAGS and Ni,Sb intermetallic for the PbTe).
II. Experimental procedure
Three different Cu-based connectors (hereafter “follower”) were received from Fokker. These are made of:
“Bare” (without any surface layer) Copper
Copper with the PVD surface (few microns) layer of Molybdenum, and
Copper electrolytically plated (few microns) with Nickel
Two types of followers were supplied:
A so-called “mid-follower” to which both N- and P-Legs are to be soldered (Fig. 3a
“End-followers” where only one (N- or P-Leg) is to be joined (Fig. 3b)
Fig. 3: Two types of samples were made in the present study: a) mid-follower and b) end-follower
Pre-forms of approximately 3 mm in diameter made of 0.1 mm foil of pure Pb were used for fabrication of the soldered interconnects. The assembly to be joined was set in a specially designed clamp (Fig. 4a) and placed in the reactor (Fig. 4b) through which mixture of Nitrogen
with 5 vol. % of Hydrogen can flow. Eventually, the “gas-flow” reactor with the samples inside was heated in the force-air furnace.
Fig. 4: a) General view of the (“clamped”) assembly to be soldered and b) “gas-flow” reactor used in the present study
In addition, for sake of comparison, a number of Cu/Cu joints were made using the same Pb pre-forms.
III. Results and Discussion
In this MEMO we will concentrate only on discussion of some outstanding observation of the soldered interconnect failures. Microstructural aspects of the connector/solder interface and, perhaps, fracture analysis of the solder joints will be main topics at the later stage of the project.
Let us start with the experiments on soldering Cu with Pb. There is (virtually) no terminal solubility of the components in the binary Cu-Pb system (Fig. 5). It is, however, found
Fig. 5: Binary Phase diagram Cu-Pb [4]
that although no reactive phase formation occurs at the Cu / liquid Pb interface, rather sound Cu /Pb/Cu solder joints were produced at 370 as well as at 550 °C. The representative micrographs of the joint cross-sections are given in Fig. 6.
Fig. 6: Optical (bright-field) images of the transition zone between Cu and Pb in the Cu /Pb/Cu interconnections after soldering under flowing mixture of Nitrogen with 5 vol. % of Hydrogen (“Forming gas”) at: a) 550 °C and b) 370 °C. (Note that some SiC-particles (“grinding debris”) are embedded in the soft Pb-based layer)
Even though the experimental results on soldering Copper with pure Pb are certainly promising, there are still some “conceptual” problems in understanding of the operational mechanism that may facilitate wetting of Cu-substrates by liquid Lead. The main question here (perhaps, of academic nature) is: “What type of interaction can provide a thermodynamic driving force for the wetting in the Cu-liquid Pb “non-reactive” system?
In this context, it is relevant to note that in the Thermoelectric Convertor design the P- and N-Legs are soldered to the Cu-based connector through the thin PVD layer of Molybdenum. No compound formation as well as no mutual solubility of the components occurs in the Mo-Pb system [5]. Nevertheless, satisfactory joining of the TE segments to the Cu-based connector was achieved with pure Pb as a solder using induction heating.
Perhaps, the most remarkable failure was observed during soldering of TE segments to Cu-based connector at 550 °C for 40 min in the flowing N2+ 5 vol. % H2 gas mixture. Under these circumstances the N-Leg of the couple was always joined to the Cu-based connector, whereas the P-Leg was detached (Fig. 7).
Fig. 7 Optical image of TE assembly after soldering under flowing N2+ 5 vol. % H2 gas mixture at 550 °C for 40 min (a) and the P-Leg (TAGS-based) detached from the Cu-based connector. (Note very severe erosion of the Cu-based connector occurred in the vicinity of the contact surface with the P-Leg of the initial assembly, and shape of the P-Leg after soldering cycle is not cylindrical)
In order to understand this peculiar phenomenon let us turn attention to the pertinent phase diagrams available in the literature. As mentioned earlier, the p-type thermoelectric component of the thermocouple assembly in the “Fokker Thermoelectric Convertor” design is one of the pseudo-binary compounds (GeTe)x(AgSbTe2)100-x, commonly referred to as TAGS-x (e.g. TAGS-80, TAGS-85). This TE material combination is commonly used as base constituent of the P-Leg for the “500-700 K applications”. Binary phase diagram of the Ge-Te system is shown in Fig. 8. It follows from this diagram that at soldering temperature 550 °C GeTe compound is thermodynamically stable. However, if for what ever reasons elemental (non-reacted) Tellurium (with melting point of ~450 °C) is still present in this p-type TE material, the situation will be completely different. This is exactly what was observed earlier in the course of our Reverse Engineering Study (Project No. 2012-3067-0616).
Fig. 8: Binary Phase diagram Ge-Te [4]
Very telling example of the inhomogeneity of the “basis” semiconducting material in the N-Leg is afforded by the micrographs taken in the sample cross-sections (Fig. 9). It remarkable that although “black” inclusions in Fig. 9a are rich in Germanium (Fig. 10a), and “dark grey” areas surrounded precipitates contain mainly Copper (“smearing” from the Cu-containing plastic, or maybe low-temperature interaction ??) and Tellurium (Fig. 10b).
Fig. 9: a) BEI and b) optical (bright-field) image taken in the vicinity of the sample cross-section showing inhomogeneity of the “basis” semiconducting material of the N-Leg in the TE assembly of the Thermoelectric Convertor design
It is not at all inconceivable that some excess of Tellurium (beyond stoichiometric ratio) may still present in the material after synthesis (by induction melting) of the TAGS pseudo-binary compound. It is also possible that some excess of Tellurium in the TAGS-base component was kept deliberately by the manufacturer in order to ensure n-type conductivity of the semiconducting material.
Fig. 10: X-ray spectra taken from a) inclusions exhibiting “dark” contrast in Fig. 9a and b) from the “dark grey” area surrounded “black” inclusions. (Note that all intensities in the X-ray spectrum are presented in a “logarithmic scale”)
Obviously, presence of non-reacted Tellurium within the base material of the N-Leg will have detrimental effect on structural integrity of the TE assembly during soldering at 550 °C, especially in the case of using radiation heating. Under these conditions, the low-melting Tellurium-rich liquid phase will leach out of N-Leg body towards the Cu-base substrate. This can be appreciated just by looking at the “distorted shape” of the N-Leg after soldering (Fig. 7b). Furthermore, at elevated temperature the Tellurium-rich melt can react heavily with Copper resulting in the formation of (rather stable) intermetallic compounds (Fig. 11). The latter can explain very severe erosion of the Cu-substrate observed in our soldering experiments at 550 °C.
Fig. 11: Binary Phase diagram Cu-Te [4]
As to the “homogeneity of the “basis” semiconducting material in N-Leg of the couple, the electron micrograph (BEI) shown in Fig. 12 gives clear evidence that the PbTe-matrix contains some inclusions of virtually pure Lead. It seems that these precipitates are situated along boundaries of crystallites in the “sintered PbTe”.
Fig. 12: BEI image taken from the “basis” n-type semiconducting material of the N-Leg in the TE assembly of the Thermoelectric Convertor design revealing Pb-inclusions in the PbTe-matrix
Fig. 12: BEI image taken from the “basis” n-type semiconducting material of the N-Leg in the TE assembly of the Thermoelectric Convertor design revealing Pb-inclusions in the PbTe-matrix
Fig. 13: Binary Phase diagram Pb-Te [4]
There exists, of course, another possibility. It is plausible that some excess of Lead in the PbTe-base component was kept deliberately by the manufacturer in order to ensure n-type conductivity of the semiconducting material. No matter for what reason a (virtually) pure Lead (with melting point ~327 °C) remains inside the TE material, soldering at 550 as well as at 370 °C will result in the occurrence of the liquid phase in the N-Leg.
In systems exhibiting isotropy of solid-liquid interfacial energy, any region of liquid phase will aggregate at grain boundary junctions (or in the case of sintering, at the contact surfaces of particles within the powder compact) and take up a uniform shape controlled by a unique dihedral angle. One sees from the microstructure shown in Fig. 12 that in our case of
Lead Telluride sintering and subsequent soldering to Cu-based connector, dihedral angle has a value between 0° and 60°. This implies that most (but not all) of the grain (particles) edges and corner were replaced by continuous channels of liquid Lead. Perhaps, fraction of the grain boundaries completely wetted by molten Lead is not high, and therefore, during exposure to elevated temperature (above melting point of Pb) distortion of the N-Leg (cylindrical) body is not significant (Fig. 7a).
It seems not profitable in this MEMO to go further with the analysis of all TE segments soldered up to this moment. However, few things about wetting behaviour of liquid Lead on “bare” Copper and on Nickel- (electrolytically) plated Cu are to be mentioned here.
Very interesting example is provided by Fig. 14 which is optical images of various views of the TE assembly after soldering of “uncoated” semiconducting segments to “bare” Copper connector at 370 °C for 40 min under flowing N2+ 5 vol. % H2 gas mixture. The appearance (shape) of the Pb-layer within the fractured sample, provide clear evidence of very poor wetting of the Cu-substrate.
Fig. 14: Optical images of the TE assembly based on “uncoated” semiconducting segments and “bare” Copper after soldering under flowing N2+ 5 vol. % H2 gas mixture at 370 °C for 40 min: a) “side” and b) “top” view. Note that the P-Leg (TAGS-based) detached from the “bare” Cu-based connector, and appearance of Pb-layer at the fractured surface is indicative for poor wetting of the Cu-substrate
When Nickel- (electrolytically) plated Copper connectors were used in the TE assembly, wetting by liquid Pb under the same experimental conditions was also very poor. This can be concluded from the analysis of the optical images shown in Fig. 15.
Fig. 15: Optical images of the TE assembly based on “uncoated” semiconducting segments and Ni- (electrolytically) plated after soldering under flowing N2+ 5 vol. % H2 gas mixture at 370 °C for 40 min: a) “side” and b) “top” view. Note that the P-Leg (TAGS-based) detached from the Ni- (electrolytically) plated Cu-substrate, and appearance of Pb-layer at the fractured surface indicates very poor wetting of the substrate
IV. Final Thoughts
Looking at the overall picture a number of things are coming in mind:
- I have a strong feeling that we need to understand much more about “technological steps” involved in the preparation of the “thermoelectric parts” of the device. The questions like “Why do we need Ni,Sb layer on the PbTe?” and “Why SnTe should be present at the top of the TAGS-x material?” should be clarified. Obviously, it would be very cost saving to use “uncoated” materials.
- The next important question is: “What would be the effect of inhomogeneities on the performance of thermoelectric materials?” It seems that presence of inhomogeneities may have a detrimental effect on structural integrity of the interconnections not only upon soldering, but also during operation of the TE devices.
- As to the soldering of the N- and P-Legs to the Cu-based connector by pure Lead, it looks possible to produce joints using radiation instead of induction heating. This allows significantly increase throughput of the TE convertor manufacturing. However, the soldering temperature 550 °C appeared to be too high, given the presence of (chemical) inhomogeneities in the base TE materials.
- High-temperature soldering with lead as a filler metal can be performed under flowing mixture of Nitrogen with 5 vol. % of Hydrogen (“Forming gas”) instead of using pure Hydrogen. (However, quality (mainly water content) of the “Forming gas” has to be controlled). This will made soldering process less costly and much more safely.
- Wetting of the various connector materials used, namely “bare” Cu, Ni-plated Cu and Mo-coated Cu in flowing mixture of Nitrogen with 5 vol. % of Hydrogen is rather poor.
- Quality of the Mo diffusion (reaction) barriers is very critical to the structural (and functional) integrity of the TE segments. Whether or not layer of electrolytic Ni can be used instead of PVD-layer of Mo we can understand after some additional experiments including electrical measurements, analysis of the sample cross-sections and fracture surfaces.
References
[1] J. Yang and T. Caillat, MRS Bulletin, 31 (2006) 224-229
[2] Y. Chen, Y.J. Zhu, S.H. Yang, S.N. Zhang, W. Miao and X.B. Zhao, J. Electron. Mater., 39
(2010) 1719-1723
[3] A.V. Novoselova, Metodu Issledovaniya Geterogennych Ravnovesiy, Moscow, Vishaya
Shkola, 1980, pp. 26-55 (in Russian)
[4] Th.B. Massalski, Ed. Binary Phase Diagrams, American Society for Metals, 1986
[5] F.R. de Boer, R. Boom, W.C.M. Mattens, A.R. Miedema and A.K. Niessen, Cohesion in
Metals, North-Holland, 1988
