Brittleness in Copper and Copper Alloys With Particular Reference to Hydrogen Embrittlement

Prof. Dr.-Ing.habil. A. Matting and Ruth Ziegler, Institut A für Werkstoffkunde der Technischen Hochschule Hannover


From: Failure Analysis: The British Engine Technical Reports, F.R. Hutchings and P.M. Unterweiser, Ed., American Society for Metals, 1981

Abstract: Hydrogen embrittlement is the brittleness affecting copper and copper alloys containing oxygen which develops during heat treatment at temperatures of about 400 deg C (752 deg F) and above in an atmosphere containing hydrogen. The phenomenon of hydrogen embrittlement of copper and its alloys is illustrated by examples from practice and reference is made to data from recent publications on the subject. Embrittlement due to this cause can only be identified by microscopic examination because other modes of failure in copper, e.g. from heat cracking, mechanical overload, the formation of low melting point eutectics or corrosion, show a similar appearance when investigated on a macroscopic scale.

Keywords: Conductors (devices); Plate metal; Turbogenerators

Material: Copper (Copper, general)

Failure type: Hydrogen damage and embrittlement


This article originally appeared in the issue of “Der Maschinenschaden” for December 1965, published by the Allianz Versicherungs-Aktiengesellschaft, Munich, and is reproduced by kind permission of that Company.
Embrittlement due to hydrogen is known in Germany as “Wasserstoffkrankheit,” the literal translation of which is “hydrogen disease.” This term is not used in this country, the phenomenon being referred to as “hydrogen embrittlement” or “gassing.”
Much research has been carried out on the hydrogen embrittlement of copper and copper alloys. The onset of the trouble cannot be recognised macroscopically and evidence regarding the cause of failure cannot be judged from the appearance of components which have failed. It is for this reason that hydrogen embrittlement can be such a source of danger.

Hydrogen Embrittlement

The term “hydrogen embrittlement” describes the brittleness affecting copper and copper alloys containing oxygen which develops during heat treatment at temperatures of about 400°C. and above in an atmosphere containing hydrogen. It arises from the hydrogen which diffuses through the metal to combine with either dissolved oxygen, or oxygen present at the grain boundaries as cuprous oxide to form water vapour. The latter cannot diffuse through the metal and therefore high vapour pressures are generated at the sites where the reaction takes place (such pressures may attain a value of 5120 atmospheres = 75,000 psi). 1 Intercrystalline cracks appear subsequently and their development is influenced by such factors as the temperature and duration of the heat treatment, the hydrogen content of the atmosphere and the oxygen content and constitution of the copper or alloy.
In Figure 1 are shown specimens of copper strip which were bent after being heated in a stream of hydrogen for a half, two, and six hours respectively to result in varying degrees of hydrogen embrittlement. 2 It can be seen that a relatively short reaction time results in only a slight reduction in the elongation and that a higher hydrogen content in the copper is needed in order to cause serious deterioration of the properties of the material. In Figure 2 the structure of electrolytic copper showing pronounced deposits of cuprous oxide (Cu2O) at the grain boundaries is depicted. A similar sample, after annealing for five hours in hydrogen at a temperature of 850°C., is shown in Figure 3. It will be apparent that the treatment has reduced the amount of cuprous oxide at the grain boundaries and it has been replaced by voids and cracks.
Figure 1

Fig. 1  Test pieces bent after heating for $\genfrac{}{}{0.1ex}{}{1}{2}$, 2 and 6 hours in hydrogen.

Figure 2

Fig. 2  Copper showing cuprous oxide at grain boundaries. × 320.

Figure 3

Fig. 3  As figure 2. after annealing in hydrogen. × 130.

In order to provide some further clarification of the problems of hydrogen embrittlement, the influence of the annealing temperature has been investigated on plates and rods made of oxygen-containing copper. 3 Annealing for one hour in an hydrogen stream at 500°C. did not produce any cracks in the material but heat treatment for the same time at 800°C. caused severe damage. A network of cracks developed at the outer surface of the copper plate as illustrated inFigure 4. One of the copper rods was severely attacked, separation of the individual grains taking place as is shown in Figure 5.
Figure 4

Fig. 4  Surface attack of copper plate after heating for 1 hour at 800°C in hydrogen. × 65.

Figure 5

Fig. 5  Surface attack of copper rod after heating for 1 hour at 800°C in hydrogen. × 65.

Hydrogen embrittlement only develops after a certain period of time, the so-called “incubation period.” It is characterised by surface deoxidation of the material which, according to Ransley, 4 results from the diffusion of oxygen to the metal surface and its subsequent reaction with hydrogen and this provides an explanation of the pronounced deoxidised rim 5 which is characteristic of hydrogen embrittled components. 6 This effect is depicted in Figure 6. If the depth of the deoxidised zone exceeds a critical value, the rate at which the oxygen reaches the surface will be less than that of the hydrogen adsorption. Hydrogen will penetrate into the metal and the reaction between it and cuprous oxide will commence.
Figure 6

Fig. 6  Intercrystalline cracking and deoxidised rim. × 125.

The influence of the hydrogen content of the atmosphere and the annealing temperature upon the incubation for an annealing time of $1 \genfrac{}{}{0.1ex}{}{1}{2}$ hours is shown in Figure 7. 2 The upper curve corresponds to the values determined experimentally but curve AA is recommended for practical purposes since the onset of embrittlement is also related to the structure of the material. The lower the annealing temperature and the content of hydrogen in the atmosphere, the longer will be the incubation time and the slower the formation of cracks. An increased oxygen content in the copper will have the same delaying effect, but once penetration of hydrogen has commenced the amount of water vapour formed and the extent of the cracking will be greater the higher the oxygen content of the metal. 5 6 The embrittlement which occurs when copper is annealed in an atmosphere containing carbon monoxide was, for a long time, thought to result from diffusion of this gas into the metal, 7 but today it is fairly evident that the embrittlement is, in all cases, due to the diffusion of hydrogen present as an impurity in the carbon monoxide. 8 The diffusion rate of hydrogen is more than 105 times greater than that of carbon monoxide at the same temperatures. 8
Figure 7

Fig. 7  Influence of hydrogen and temperature on incubation for heating periods of $1 \genfrac{}{}{0.1ex}{}{1}{2}$ hours.

In some instances, the hydrogen driven off from a residue of drawing or lubricating grease on a metal surface by heating subsequently is sufficient to induce hydrogen embrittlement. 8 Insulation breakdown in the stator windings of a turbo-generator was found, on dismantling, to be associated with damage to the copper conductors, partial fracture having taken place. 9 The damaged parts were bent and showed gaping brittle cracks as seen at “R” in Figure 8.
Figure 8

Fig. 8  Brittle cracks in copper conductors.

Temper colours and traces of burning on the surfaces indicated that excessive heating of the conductor had taken place. A photo-micrograph of a section through the fractured conductor indicated the presence of intergranular cracks as depicted in Figure 9, where A, represents the surface of the fracture, B, fine intergranular cracks, and C, some of the larger inclusions that were present. Examination at higher magnification showed the presence of a system of interconnected cracks and oxides mostly adjacent to the surface as is evident in Figure 10. The cuprous oxide, one of the prerequisites for hydrogen embrittlement, was present in the material and had been reduced by hydrogen. The temper colours and the traces of burning on the surface of the conductor led to the conclusion that failure of the insulation had occurred. The material used for the insulation, as a result of the temperature attained, gave off hydro-carbons, hydrogen and tar substances to result in hydrogen embrittlement of the bars. 7
Figure 9

Fig. 9  Section through one of the fractured bars shown in figure 8. (× 35).

Figure 10

Fig. 10  Intergranular cracks and oxides adjacent to surface. (× 400).

The possibility that hydrogen embrittlement may occur requires to be considered during welding operations. The presence of the usual de-oxidation additives to copper reduces the electrical conductivity and a small amount of oxygen is unavoidable in materials for conductors in electrical engineering. In gas welding even a neutral oxygen-acetylene flame will cause “gassing” as seen in Figure 11. Although the use of a shielding gas will serve to prevent this to a large extent, a phenomenon similar to hydrogen embrittlement may be observed in oxygen-bearing copper (more than 0.2% O2). The cuprous oxide which is present at the grain boundaries originates from the melting process and is soluble in liquid copper to form, with copper, a strongly embrittling eutectic at 3.45% Cu2O (approximately 0.4% O2).
Figure 11

Fig. 11  Effect of welding oxygen-containing copper in neutral oxy-acetylene flame. (× 160).

The two test pieces shown in Figure 12 were welded under a protective gas, (a) being an oxygen-containing copper and (b) an oxygen-free copper. While test specimen (b) could be bent through 180°, specimen (a) fractured after slight deformation, failure occurring in the transition zone between the weld and the parent material. The respective structures are illustrated in Figures 13 and 14, that from specimen (a) showing the presence of the Cu-Cu2O eutectic at the grain boundaries, this not being present in specimen (b).
Figure 12

Fig. 12  Effect of welding under protective atmosphere. (a) oxygen-containing copper. (b) oxygen-free copper.

Figure 13

Fig. 13  Photomicrograph of specimen (a) in figure 12. (× 80).

Figure 14

Fig. 14  Photomicrograph of specimen (b) in figure 12. (× 160).

A further example is provided by a drainage cock made of (α + β) brass which tore at the threads to result in considerable damage. Embrittlement of the material was due to the presence of cuprous oxide the crack running along a chain of Cu2O particles. The critical position in which the eutectic formed is shown diagramatically in Figure 15. The lower photo-micrograph shows that Cu2O was present in the material and the formation of the resulting grain boundary eutectic is apparent from the upper illustration. 10 For welding by shielded-arc processes, oxygen-free materials are therefore to be recommended.
Figure 15

Fig. 15  Showing influence of welding on formation of eutectic.

Embrittlement can develop in copper or copper alloys containing oxygen if they are brazed in a reducing flame containing an excess of hydrogen. The bars of an 80 kW squirrel cage rotor broke off in serve at the junction with the end rings, as seen in Figure 16. 9 The oxidised fracture surfaces marked A, had a hammered appearance and the exposed crack surfaces B were oxidised. The uncracked bar marked C was also broken off. Other bars in the same position showed incipient cracks and microscopical examination of a specimen taken tangentially from an end ring in the region of a cracked bar showed that the cracks originated at the end of the brazed area and developed in an arc-like manner through the ring in the manner depicted in Figure 17. The material of the end ring showed the presence of gross oxide impurities and, in the brazed area, numerous fine cracks were present as seen in Figures 18 and 19, the former being unetched and the latter etched. The type and location of these cracks were related to the numerous inclusions of cuprous oxide and indicated that failure resulted from hydrogen embrittlement. Brazing of the rings, which contained cuprous oxide, in a reducing flame containing an excess of hydrogen caused intercrystalline cracks to develop which subsequently extended in service leading to fracture at the junction with the rotor bars.
Figure 16

Fig. 16  Failure of bars at junction to end ring.

Figure 17

Fig. 17  Showing crack origin associated with brazing metal.

Figure 18

Fig. 18  Unetched. Intercrystalline cracking adjacent to brazing (× 200).

Figure 19

Fig. 19  Intercrystalline cracking adjacent to brazing. (× 200)

Failure of Copper Parts due to causes which resemble Hydrogen Embrittlement

At elevated temperatures, certain temperature ranges and deformation rates exist where the ductility of copper and its alloys is diminished and cracks and fractures readily occur at slight degrees of deformation. At a superficial glance, these may resemble those resulting from hydrogen embrittlement. The curve showing the elongation of electrolytic copper passes through a minimum in the temperature range 550–600°C., as shown in Figure 20. 12 Microscopical examination of specimens heated in this range showed the presence of numerous pores and separation had taken place at those grain boundaries which lay perpendicular to the direction of the stresses as indicated in Figures 21 and 22. In Figure 22 specimen A was broken at 400°C, B at 600°C and C at 800°C. The pores that form initially extend and unite as the result of deformation to result in cracks. The degree of deformation necessary for this process to occur is lower the lower the deformation rates. The ductility increases again at higher temperatures as soon as re-crystallisation prevents the formation of pores and cracks and transfers separation to within the grains.
Figure 20

Fig. 20  Showing minimum elongation of copper in region of 550°–600°C.

Figure 21

Fig. 21  Intergranular cracks in specimen heated to 550° 600°C. × 500

Figure 22

Fig. 22  Microstructure of × 200 specimens broken at a. 400°C, b. 600°C and c. 800°C.

Several theories have been proposed regarding the causes of the nucleation of pores. 13 14 One such theory assumes inclusions to be responsible. 15 Another concludes that grain boundary slip is the cause. Precipitations, especially those at the grain boundary, can often be invoked to explain the development of cracks. Bismuth for instance even in the smallest amounts (less than 0.003%) occupies the grain boundaries and results in considerable embrittlement of copper. 12 The effect of antimony is controversial at the moment. According to American investigations, specimens of tough-pitch copper containing 0.02% arsenic were not subject to hot cracking, but at a content of 0.06% small cracks appeared. 16 Tellurium has been found to favour hot cracking. Other failures which in a macroscopical manner resembles hydrogen embrittlement, may be due to brittle fracture. Under the microscope, however, it is sometimes possible in such cases to observe definite deformation of the structure as seen in Figures 23 and 24. The former illustrates an unetched microsection, whereas the latter is of a further portion of the same specimen in the etched condition.
Figure 23

Fig. 23  Section through fracture-unetched. (× 40).

Figure 24

Fig. 24  Section through fracture showing deformation near crack-etched. (× 80).

In one example, the bars of an AC motor fractured where they where brazed to the short circuit ring as shown in Figure 25. 9 The fracture surface B was grainy and banded on those regions which had not been damaged. Microscopical examination showed that incipient cracks were present in some bars, these being situated in the same region as other bars in which fracture had occurred, as depicted in Figure 26. At higher magnification this cracking was shown to be of an inter-crystalline nature as shown in Figure 27. No defects resulting from oxides were present and the appearance and course of the cracks suggested that they most probably originated as a result of the simultaneous effect of mechanical stress and high temperature possibly associated with overloading of the cage.
Figure 25

Fig. 25  Failure of bars of a.c. Motor at brazed ends.

Figure 26

Fig. 26  Section through incipient cracks in unbroken bars. (× 3.5).

Figure 27

Fig. 27  Intergranular nature of cracking. (× 80).

In Figure 28, the cracking on the bore of a copper tube resulting from a combination of vibration and corrosion is shown. 17 A longitudinal specimen showed the cracks to be of the form illustrated in Figure 29. Such fine, parallel cracks usually result from alternating stresses, possibly caused by pressure fluctuations or the effects of water hammer, as well as simple vibration. Repeated expansion and contraction as a result of considerable temperature fluctuations may show similar effects. Corrosion fatigue cracks, which result from this cause, are often initiated at small corrosion spots and develop in a transcrystalline manner in contrast to cracks resulting from stress corrosion which are mainly of the intercrystalline type.
Figure 28

Fig. 28  Corrosion fatigue cracking on bore of copper tube. (× 7).

Figure 29

Fig. 29  Section through cracking shown in figure 28. (× 50).

Reference

  1. Baukloh, W., und W. Stromburg: Uber die Wasserstoffkrankheit einiger Metalle. Z. Metallkunde 29 (1937), S. 427/30.
  2. Schückher, F., und E. Mattson: Wasserstoffkrankheit in Kupfer. Pro Metall 1960, S. 281/87.
  3. Unpublished Report.
  4. Ransley, C. E.: J. Inst. Metals 65 (1939), p 147 etc.
  5. Mattson, E., and F. Schückher: An investigation of hydrogen embrittlement in copper. J. Inst. Metals 87 (1959), p 241/47.
  6. Houlden, B. T. and W. A. Baker: Metallurgia 47 (1953), P. 223 .
  7. Werkstoffhandbuch NE-Metalle, Blatt D 9. Berlin: VDI-Verlag 1938.
  8. Erdmann-Jesnitzer, F., F. Wünser und A.Schlegel: Uber das Blankglühverhalten von Kupfer und Kupfer-Edelmetall-Legierungen. Metall 13 (1939), S. 87/92.
  9. Unpublished Report.
  10. Köcher, R.: Das Schweissen von Kupfer. Metall 13 (1959), S. 107/14.
  11. Matting, A.: Werkstoffwahl und Werkstoffeigenschaften als Ursache von Schäden. Maschinenschaden 37 (1964), S. 144/52.
  12. Bauser, M.: Verformbarkeit und Bruchverhalten bei der Warmformebung von Metallen. Metall 17 (1963), S. 420/29.
  13. Chen, C. W., and E. S. Mecklin: The effect of Grain Boundary Migration on the Formation of Intercrystalline Voids during Creep. Trans A.I.M.E. 218 (1960), p 177 etc.
  14. Hull, D., and D. E. Rimmer: The Growth of Grain-boundary Voids under Stress. Phil. Mag. 4 (1959), p 673 etc.
  15. Cottrell, A. H.: Structural Processes in Creep. Iron Steel Inst. Spec. Rep. No. 70 (1961), p 1 etc.
  16. Smart, J. S., and A. A. Smith: Effect of Certain Fifth-period Elements on some Properties of High-purity Copper. American Institute of Mining and Metallurgical Engineers A.I.M.E. (1943), p 1 to 14.
  17. Camenisch, P.: Kupferrohrschäden und deren Vermeidung. Pro Metall 71 (1959), S. 222/24.

Related Information

Hydrogen Damage and Embrittlement, Failure Analysis and Prevention, Vol 11, ASM Handbook, ASM International, 2002, p 809–822