Hydrogen Embrittlement – Part 3
High Strength Steels Achilles Heel

Susceptible Materials

While some stainless steel grades are susceptible, high strength steels with tensile strengths and hardness above 130,000 psi and Rockwell C35, respectively, are the most prone to hydrogen embrittlement. Steels below these tensile and hardness levels are generally immune. Why?

Increasing hardness, most commonly by heat treating, is accompanied by a corresponding decrease in ductility. In simple terms, ductility is the ability to deform under stress rather than crack or fracture. When hydrogen atoms combine into molecules in a steel that exceeds the tensile strength and hardness threshold, the steel cracks under the pressure increase. But if the tensile and hardness levels are below the critical threshold, the higher degree of ductility allows the steel to deform, absorbing and redistributing the pressure increase, rather than cracking.

Susceptibility to hydrogen embrittlement increases in alloy steels with heat treatment to higher strength. The strength/susceptibility relationship, in fact, approaches exponential levels. In other words, doubling the heat treated strength, quadruples the steel’s susceptibility to hydrogen embrittlement.

Identifying and sorting embrittled parts from good components before they fail is virtually impossible. The detection limit of chemical analysis for hydrogen is generally well above the 5 to 10 ppm level at which embrittlement has been shown to occur. Even if such detection capability was available, hydrogen tends to concentrate at isolated locations within the part. This leaves the majority of the part at “low” or undetectable hydrogen levels. Chemical analysis of parts after failure, to determine if hydrogen embrittlement is the cause, is also not viable since hydrogen diffuses from the part after fracture.

Since most hydrogen embrittlement results from hydrogen absorbed during the manufacturing process, parts which are “batch” processed are usually either all embrittled or all “good”. The failure of one part from a “batch”, therefore, is usually a good indication that others from the same batch will fail over a similar time period.

Prevention

The two keys to avoiding hydrogen embrittlement are at the design stage and during the manufacturing process. Designers with little metallurgical training may not realize the implications of the materials and manufacturing processes they call for in the drawing specifications. As with other failure modalities, hydrogen absorption can inadvertently be “designed into” a part.

On the manufacturing side, avoiding reducing acids where possible removes an abundant source of hydrogen from potential exposure to the part. Where electro- plating is required, minimizing plating time and maximizing current density reduce the volume of absorbed hydrogen. Consideration of electroless plating or vapor deposition as alternatives eliminates the possibility for hydrogen absorption from the coating process altogether.

Protecting furnace charges or components that will be welded from corrosion, or cleaning them prior to use will avoid hydrogen introduction by these routes. This applies equally to parts that will be heat treated. Any processing that will elevate the temperature of the part, and most do, will also raise hydrogen to a higher level of mobility, increasing the potential for absorption.

Hydrogen Management

Despite the most stringent precautions, processing requirements will sometimes introduce hydrogen to parts that exceed the hydrogen embrittlement tensile and hardness threshold. Fortunately, there is a procedure that will effectively remove absorbed hydrogen. This oven heating process, referred to as “baking”, is performed as follows:

1. Parts must be baked within 4 hours of hydrogen exposure. Less is better.
2. Parts must be baked at 400º F.
3. Parts must be held at 400º F for a minimum of 4 hours. Longer may be required depending on part size, processing, etc.

To be effective, the time and temperature parameters must be strictly followed. Short-cuts or delays on any step will dramatically reduce the effectiveness of the entire process. For example, twice as much hydrogen will be baked out at 400º F versus 350ºF, and doubling the bake time doubles the amount of hydrogen that is baked out.

The sooner baking begins after exposure to hydrogen, the better. The 4 hour “window” is a maximum. Note that baking must be performed after each hydrogen exposure if 4 hours will elapse between multiple exposures. No amount of baking will salvage embrittled parts if these time and temperature parameters have not been met.

A final word of caution. A 30 year analysis of hydrogen embrittlement failures in the aircraft industry found that over 70% resulted from improper baking procedures.

Closing

There are competing theories on the mechanism by which hydrogen embrittlement occurs. That presented here appears to be the most widely accepted by the scientific community.

The subject of this article has been limited to hydrogen embrittlement in low alloy steels. Other alloys are subject to hydrogen embrittlement, thought the mechanisms discussed are the same. However, other hydrogen driven failure mechanisms have been observed, though they appear to be less common. These include hydrogen induced blistering, internal hydrogen precipitation and hydride formation in nonferrous metals.

Part 4 will briefly discuss the analysis of hydrogen embrittlement failures and illustrate several “infamous” and unusual failures.

The Metallurgical Phenomenon

Hydrogen atoms are the smallest of any element. So small, that they easily travel between iron atoms. The boundaries between crystals, or grains, which are the structure of metals, are gapping canyons in relative size to hydrogen atoms. Once absorbed, hydrogen atoms are attracted to microscopic crystal defects, or misalignments, where there is slightly more space between grains. They are also attracted to areas under tensile stress that cause a very slight increase in the space between grains from the opposing “pull” of the stress.

As more hydrogen atoms accumulate at these areas, they combine to form relatively very large hydrogen molecules (H₂) which raises internal pressure, expands the size of the defect or grain boundary interface and attracts still more hydrogen atoms, accelerating the process. This cycle produces a raising tensile stress inside the component which eventually results in a micro-crack. These micro-cracks grow rapidly and simultaneously at numerous locations within the part, reducing the actual intact load bearing cross section by as much as 10-20%.

In order for hydrogen embrittlement to occur, three conditions must coincide:

1. The part must have a tensile strength in excess off approximately 130,000 psi. This generally corresponds to a hardness of Rockwell C 35.
2. The part must be in contact with a source of hydrogen. This may occur during manufacture, in service, or both.
3. The part must be subjected to a tensile stress.

This last condition can be deceptive because parts do not need to be assembled or in service to be under tensile stress. Residual internal stresses from casting, forging, welding and other manufacturing processes are significant and, in fact, are probably the root cause of most hydrogen embrittlement failures. Heat treating to raise strength levels above 130,000 psi induces substantial levels of residual stress. The disturbing phenomenon of “shelf popping”, unassembled parts cracking in storage or inventory with an audible “pop”, results from hydrogen embrittlement associated with residual stress.

Since the majority of the hydrogen is absorbed through and accumulates at the grain boundaries, hydrogen embrittlement cracking is primarily intergrannular (fracture at the grain boundary) rather than transgrannular (fracture through the grains) as in some other forms of brittle cracking.

Hydrogen Sources

One of the challenges in predicting and preventing hydrogen embrittlement is the wide range of available sources of hydrogen, both in the manufacturing and service environment.

Sources from manufacturing include the original steel making process, subsequent casting or forging, grinding operations, soldering and brazing fluxes, blasting and tumbling media, welding electrodes, acid cleaning or pickling and electro-plating, etc.

Service related sources of hydrogen may include incidental contact with acids or hydrogen containing cleaning solutions, or absorption from hydrogen containing product by equipment used in its processing. The most common source in service by far, however, is corrosion. Corrosion can also act as a source of hydrogen in the manufacturing process. Rusted ingots and scrap used in casting melts, welding on parts that have corroded, and heat treating corroded parts, are potential sources of absorbed hydrogen, particularly when exposed to elevated temperatures which increase the mobility of hydrogen atoms.

In Part 3, we will discuss the conditions that render steels susceptible to hydrogen embrittlement, and how to prevent its occurrence in both the manufacturing and service environments.

Part 4 will briefly discuss the analysis of hydrogen embrittlement failures and illustrate several “infamous” and unusual failures.

The Phenomenon

Sudden brittle fracture in high strength steels resulting from hydrogen embrittlement represents an extremely dangerous phenomenon to industry, particularly since it is usually the result of factors that occur during the manufacturing process.

Hydrogen embrittlement reduces ductility, often to the point where metals behave like ceramics. Consequently, fatigue strength and fracture toughness are also dramatically reduced. Brittle fracture occurs without warning and can be immediate, within hours of manufacture, or after years in service. Hydrogen embrittlement failures have even been observed in unassembled parts in inventory, a phenomenon known as “shelf popping”.

Generally, the higher the strength of the steel, the more at risk it is to hydrogen embrittlement and the more vulnerable it is to lower levels of hydrogen. Embrittlement at levels of 10 parts per million and less are not uncommon. Some research suggests this relationship is exponential. In other words, doubling the strength, quadruples the susceptibility to hydrogen embrittlement.

Although hydrogen embrittlement occurs in many different metal alloys, high strength steel appears to be the most sensitive, is the most widely used and accounts for the largest number of hydrogen embrittlement failures. This article offers an overview on hydrogen embrittlement as it relates to high strength steels only, though details of the phenomenon generally apply to other susceptible metals.

First Appearance

In the late 1940’s a revolution was underway in aviation. Jet propulsion was rapidly replacing the old piston engine driven propeller technology and aircraft performance began to exceed levels that had been considered physically impossible just ten years earlier. Weight reduction and more power propelling airframes that could withstand higher loading were critical to these improvements. This resulted in demands for higher strength alloys from which smaller, lighter and stronger components could be made.

Low alloy steels such as 4130 had been used in aviation in the past. However, these materials were typically used in the normalized heat treated condition, at tensile strengths in the 90,000 to 120,000 psi range – well below levels susceptible to hydrogen embrittlement. In response to demands for more strength, “radical” heat treatments to tensile strengths approaching 200,000 psi were applied to 4130 and other “anemic” low alloy steels. Some of the first hydrogen embrittlement failures appeared, though they weren’t initially recognized as such.

Enhanced low alloy steels, such as 4140 and 4340 were used in response to these failures, and the cycle was repeated, with the demand for more performance from smaller components resulting in processing to ever higher strength levels.

One of the unfortunate consequences of increasing the strength of low alloy steels is a corresponding reduction in corrosion resistance. To combat increased corrosion in service, a variety of electroplated coatings, such as chromium, nickel and cadmium, were applied. With a new potent source of hydrogen now available from the plating baths used in these processes, a dramatic increase in hydrogen embrittlement failures occurred in both the aerospace industry and in other industries to which the new materials technology had filtered down.

Part 2 in this series will discuss this phenomenon from a metallurgical perspective – what actually occurs, on a microscopic scale, that causes hydrogen embrittlement?

In Part 3, we will discuss the conditions that render steels susceptible to hydrogen embrittlement, and how to prevent its occurrence in both the manufacturing and service environments.

Part 4 will briefly discuss the analysis of hydrogen embrittlement failures and illustrate several “infamous” and unusual failures.

There are three stages in the life of a fatigue failure; 1. Initiation, 2. Crack Growth (propagation), and 3. Final Fracture. These stages are illustrated in the SEM image of a fractured rectangular wire above. The Initiation is indicated by the large red arrow at the lower left. The area of progressive Crack Growth extends from this arrow to the line indicated by the three smaller red arrows. Final Fracture, the point at which the remaining intact cross section of the wire could not sustain the next cyclic stress load – “the straw that breaks the camels back” – and complete fracture occurred, is the light area above the three arrows. This fracture is an example of bending fatigue (in one direction) initiating from a single point of origin.

The fractured crane lifting hook above is an example of reverse bending fatigue (back and forth in two directions). In this case, the major bending stress was applied from the bottom of the fracture as oriented in this photo, and a minor stress from the top. The darker gray area indicates final fracture in a single stress cycle. The thin horizontal band at mid fracture indicates a significant “jump” in the fracture progression that occurred in the cycle proceeding final fracture which almost, but not quite, resulted in complete fracture. This fatigue fracture initiated from multiple origins. Multiple origins are indicated by the steps, or “ratchet marks”, at the outer diameter of the fracture indicated by the arrows. Ratchet marks occur when multiple fatigue cracks initiate at slightly different planes on a component’s surface. As these multiple cracks progress into the component, they eventually join into a single fracture plane as show above.

Rachet marks resulting from multiple fatigue origin locations are shown at high magnification in these images taken on our Scanning Electron Microscope (SEM). Fatigue cracking penetrated only a short distance into this NASCAR Racing suspension component before it failed completely in a single load cycle. As a result, the multiple origin fatigue cracks never progressed far enough to coalesce into a single fracture plane. Several of the individual origin sites are indicated by arrows.

The diagonal bands exhibited by the fatigue fracture of this compressor connecting rod are commonly called “arrest lines”. These indicate a change in the frequency of cyclic stresses, such as “stop-start” sequences, changes in RPM, or variations in load. The initiation site at the crankshaft journal bore (arrow) is heavily damaged. This is not uncommon in fatigue failures. As the first location to crack, the initiation site is exposed to potential relative movement of the two sides of the crack during propagation. This presents a significant challenge to the analyst in determining the root cause of fatigue cracking.

Fatigue fractures exhibit distinct features, called striations, when viewed at high magnification using Scanning Electron Microscopy. Striations appear as relatively evenly spaced parallel lines. Each striation is actually a shallow crack that results from a single load, or stress, cycle. Repetition of these cycles produces an advancing repetition of shallow cracks as shown above in this fatigue fracture in a hydraulic valve body. This process is characterized by the term, “fatigue crack propagation”.

The appearance, or morphology, of fatigue fracture striations varies depending on the magnitude and frequency of the applied load and the physical characteristics of the affected component such as hardness, microstructure and chemical composition of the alloy. These SEM images illustrate fatigue striations in an aluminum valve body (left) and an alloy steel high pressure hydraulic cylinder (right).

In some cases, the root cause of a fatigue failure can only be discovered by an analysis of internal characteristics of a component at the crack location. In this example, a metallographic cross section revealed decarburization (dark phase at arrow) of the surface of a steering arm due to faulty heat treating. This carbon depleted layer has significantly reduced hardness and strength, as well as residual tensile stress, conditions highly conducive to fatigue crack initiation.

Other types of internal defects which act as initiation sites for fatigue are apparent on the fracture surface. Examination of this brake return spring by SEM revealed fracture features which radiate from a single initiation point. Viewed at higher magnification, this initiation point exhibits a void containing a non-metallic inclusion which acted as a stress concentration.

Identifying the root cause of service environment initiated fatigue failures can be challenging. Some years ago, we provided analytical support on a lawsuit which was filed after an individual sustained a back injury when the metal leg of a “stacking chair” fractured. Stacking chairs are the type of institutional chairs you often see in school auditoriums, government office waiting rooms, etc. and are designed to be stacked, one upon the other, for more compact storage when not in use. This particular chair came from a college in Ohio. Our analysis proved low stress, high cycle fatigue as the failure mode. In other words, low magnitude stresses applied at high frequencies, in this case over a million cycles.

The chair had been in use for a relatively brief time, and even if it had seen longer service, it seemed unlikely that it had been sat in on a million occasions. This presented something of a mystery, as the failure mode was indisputable. Investigation of the service environment revealed that the chairs were used sporadically and when not in use, were stacked in a storeroom. The college staff was methodical in setting up the chairs in orderly rows in an adjacent auditorium, then stacking them from the same row end when they were no longer required, with the same chair ending up on the bottom of the stack before going back into storage. The stack was higher than the maximum specified by the manufacturer, providing a load in excess of the design limit. A survey of the area revealed that the storeroom was immediately above the main HVAC installation, the key and final piece of the puzzle. Vibration from the HVAC system, transmitted through the storeroom floor, and loads from the weight of chairs stacked in excess of the design limit provided the stresses required to initiate the fatigue crack. Once the crack grew to the point at which the remaining intact tubular leg could no longer sustain the load of a sitting person, final fracture occurred.

As with all failure analyses, the analyst must provide specific answers to three critical questions when evaluating a fatigue failure.

They are:

1. How did it fail?
2. Why did it fail?  and
3. What will prevent future failures?

If you have commissioned a failure analysis, and all three of these questions are not answered, all you have paid for is some interesting pictures and a possible lawsuit when your product fails again.

Read part 4 – Analyzing Material Fatigue

The exception to “usually”, the cases where fatigue fractures initiate from component surfaces that are free of stress concentrations, typically result from one of two causes; under-design of the component, or abusive service conditions. Just as all materials have an ultimate tensile strength, they also have a fatigue strength, sometimes called the fatigue limit or endurance limit. Once a component is subjected to cyclic stresses that exceed this limit, fatigue fracture occurs.

Fatigue failures of this type are less common than fatigue failures initiating from stress concentrations. Usually components are intentionally over-designed to deal with stresses several times greater than what they would be subjected to in service as a safety margin.

Fatigue Crack Initiation – The Critical Event

If the initiation stage can be prevented, fatigue fracture will not occur.  It sounds so obvious and simple. It’s not. As noted above, initiation is the most complex stage of fatigue fracture. A low magnitude load, which would have no effect whatsoever on a component in a single application, can be devastating when repeatedly applied as thousands or millions of cycles. The cumulative effect of these cyclic loads are microscopic “shifts” in the material’s structure which ultimately produce a “dislocation” – at this scale it is too small to be called a crack – and the focal point of stress concentration is born. Corners, holes, rough surface finish, welds and other features only accelerate the process.  To further complicate the issue, vibration harmonics, dampening of the system and the environment in which the component functions add a large unknown factor. Collectively, these affects become difficult to predict in the design stage.

Confronting Fatigue – Attack and Defense

From a practical standpoint, fatigue failures present a danger to you, the manufacturer, at three points in a components life. These are the design stage, the manufacturing process, and the service environment.

Design

The design engineer is the first line of defense against fatigue fracture.  He or she can’t prevent failures originating in the manufacturing process or service environment, but the designer lays the foundation of prevention.

In an ideal world, each design would be subjected to extensive stress calculations and fatigue testing.  In the real world this is rarely cost effective for non-critical components. Instead, accepted and “proven” parameters are applied.  These typically include safety margins which are more than adequate.  Typically, but not always.

Computer Aided Design (CAD), Finite Element Analysis (FEA) and a variety of other computer driven design and predictive technologies can greatly enhance the fatigue resistance of a component at the design stage.  But they can not prevent fatigue failures.  That’s because the next two threats of fatigue failure are beyond the designer’s control.

The Manufacturing Process

Manufacturing processes are a rich, though unintended, source of stress concentrations from which fatigue cracks can initiate.  The list is almost endless, and includes rough machined surfaces from dull tooling or excessive feeds and speeds, burrs from cutting or drilling operations, and insufficient chamfers or corner radiuses. Welds, even when technically faultless, provide geometric stress concentrations.  Defective welds and welding procedures may result in porosity and high hardness heat affected zones from which fatigue can initiate. Mechanical fasteners – bolts, screws, studs, and rivets- are highly prone to fatigue failure, either due to defects in the fastener itself, or to insufficient tightening torque during the assembly stage of the manufacturing process.

Care in manufacturing and a good quality control program will avert many of these potential sources of fatigue initiation.  However, despite the best quality control program, the manufacturer is often at the mercy of their raw material supplier. These suppliers may open the door to fatigue failure through castings which contain excessive porosity or  microstructural defects, mill products which are work hardened, forgings with undetected laps or seams, or gross non-metallic inclusions in any of these products. Appropriate specifications on outsourced stock and components are vital in guaranteeing their quality, but as with so many aspects of production, they are a compromise. Loose specs solicit low cost bids, but a potentially high percentage of defective products, while tight specs limit the number of vendors capable of meeting them and drive costs higher, cutting into profits.

Read part 3 – The Service Environment

Fatigue is the most common type of fracture in engineered components. Fatigue fractures are also particularly dangerous because they can occur under normal service conditions, with no warning that a progressively growing crack is developing until the final catastrophic failure. The component, whether it’s the outer aluminum skin of a commercial jet or a simple tubular chair leg, often appears to be perfectly sound with no visible distortion to warn of impending failure.

A technical understanding of fatigue requires a comprehensive knowledge of metallurgy, physics, and phenomena like plastic deformation, slip planes and dislocation theory. In fact, there are several competing theories on exactly what happens at a microscopic level when a fatigue crack initiates. But a practical understanding of the process is extremely beneficial and has direct application to its prevention, and the manufacturing environment, as discussed below.

To the non-technically inclined, the term “fatigue” suggests this type of failure is related to the age of a component, that the material is “tired”. In fact, fatigue fracture can occur within hours of a component going into service. Conversely, even large, highly stressed components can operate for decades with no fatigue cracking or failure.

Fatigue fractures result from repeated, or cyclic, stresses. These stresses can take a variety of forms, such as bending (in one direction), reverse bending (back and forth in two directions), torsion (twisting in one or more axis) and rotation.  Regardless of the variation in direction, the stress on the component at the point of fatigue fracture is always tensile stress, in which the fracture initiation site is being “stretched”, or pulled in opposite directions. To illustrate this, visualize a tube which is being repeatedly bent in one direction. The side of the tube that is concave when it is bent is being compressed. The side of the tube which is convex is being “stretched”, or subjected to a tensile stress. This is the side on which a fatigue crack will initiate.

Fatigue cracks initiate at stresses below the tensile strength of the material. Tensile strength is the stress, or load, at which a material breaks when pulled in two opposing directions. This load is a specific value for each metal alloy, varying somewhat depending on heat treating and other processing operations. These values are widely available in engineering reference manuals, typically expressed as pounds per square inch in American references. The fact that fatigue cracks can occur at stress levels below the tensile strength of a material is difficult to explain. Theories on this focus on physical and structural changes at the microscopic (0.0001” or less) area of crack initiation.

Fatigue is a progressive fracture mechanism. Once a fatigue crack initiates, it is driven further into the component with each stress cycle. This crack growth process continues as long as the component is subjected to cyclic stress. Depending on the magnitude and frequency of these stresses, the crack may grow over time ranging from hours to years.  Eventually, the crack advances to a point where the remaining intact cross section of the component can not sustain the next cyclic stress load – “the straw that breaks the camels back” – and complete fracture of the component occurs.

Read part 2 – Fatigue in the “Real World”