Case Study: Metallurgical Failure Analysis of a Torsion Spring

Click here to view a .pdf of this Case Study including additional tables, images & charts.

Springs are one of the few engineered components that are intentionally designed to deform. As they absorb, store and release energy in service they must be able to deform and then return to their original shape over thousands, millions, or even tens of millions of cycles. These service conditions make springs extremely susceptible to fatigue failures, often initiating from relatively small nicks, scratches or seams that act as stress concentrations.

As this case study shows, however, there are many other more subtle conditions that can result in spring failures by other fracture modes.


This case study of a failed torsion spring is presented in the format used in MAI Failure Analysis reports, though in a condensed form for reasons of space. Images and data tables are located after the text. MAI reports include exact identification of “how” and “why” the failure occurred, and specific recommendations to prevent future failures, along with the data to support those conclusions. The answers to these questions reduce our clients’ costs and liability exposure, and enhance the quality of their products.

Comments offered as context and explanation in this case study are in italics.


A fractured torsion spring was submitted for evaluation of the material, heat treatment, and root cause of the failure. It was reported that no failures of this component had occurred after several years of production before three recent failures were reported in service. Four of these springs are used to provide a counter force which allows one person to lift an 1100 pound cover on an aircraft ground support (fuel storage) vault.

The springs are specified to be made of SAE 5160H medium carbon alloy steel, hot-formed and are then oil quenched. They are then specified to be tempered at 750 to 800 ºF to a hardness of approximately 51 to 52 Rockwell C. After heat treatment the springs are shot peened for increased fatigue resistance.


Visual examination is often taken for granted. That’s a mistake. A thorough and thoughtful visual examination lays the foundation for an effective failure analysis, identifying critical features and aspects of the failed component.

The spring fractured approximately seven coils from one end, and approximately 2-1/2 coils from the opposite end as shown in Figure 1. No localized “necking” or other macroscopic indications of ductility are present adjacent to the fracture. No substantial damage or corrosion is present on the wire surface adjacent to the fracture that could have acted as a stress concentration and contributed to the failure. Some corrosion is present on the fracture surface that occurred after the failure as shown in Figure 2. The fracture exhibits granular macroscopic features that are consistent with a brittle single cycle overload. Examination of the fracture surfaces by optical stereomicroscopy at magnifications up to approximately 40X did not reveal any discontinuities that could have contributed to the failure.


Chemical and mechanical properties are typically presented under separate headings but are combined in this case study for the sake of brevity. Determining that the appropriate material was used is an essential baseline to the analysis.

Metallurgical-Spring-Failure-Cross-SectionThe chemical analysis of this spring conforms to SAE 5160H medium carbon alloy steel. The sulfur content is elevated (0.042%), but is within the standard product analysis tolerance. This elevated sulfur would decrease the ductility and toughness in the radial direction, and would increase the likelihood of longitudinal or angular cracking, however, it is not a contributory factor to the failure. The spring material is deoxidized with silicon and vanadium, indicating that it was produced to a fine-grained melting practice, probably by a continuous (strand) casting process. This spring contains an elevated amount of arsenic (0.014%), which can contribute to temper embrittlement and is a significant contributory factor to the failure.

The surface hardness of this spring (46 Rockwell C) in slightly lower than the core hardness (48 Rockwell C), which is indicative of some carbon depletion. The hardness is lower than the 51 to 52 Rockwell C specified on the part drawing. This contributed to the failure by decreasing the tensile properties and increasing the likelihood of overload failures in service. The chemical analyses and hardness indicate that the spring was tempered at approximately 800 ºF, which is towards the high end of the range specified on the part drawing. This tempering temperature is also towards the high end of the range at which tempered martensite (blue) brittleness can occur, and is at the low end of the range at which temper embrittlement can occur.

A standard round tensile test bar was machined from one of the straight tang ends of the spring. The tensile strength is lower than the approximately 232,000 psi expected from a steel having a hardness of 48 Rockwell C. The yield to tensile strength ratio is very high and indicates that this spring was properly quenched. The tensile test bar fractured outside of the gage length, precluding accurate measurement of elongation. The reduction in area, however, is consistent with the measured tensile properties. The tensile test fracture exhibits substantial ductile shear indicating that the spring material has inherently good ductility.

Standard Charpy V-notch impact test samples were machined from the tang at the opposite end of the spring. The Charpy V-notch impact properties are low, which is consistent with temper embrittlement. The impact test bar fractures exhibit primarily granular features that are consistent with brittle fracture. Only very small amounts of ductile fracture are present at the edges of the test bar. This indicates that the spring has inherently low toughness at room temperature.


The corrosion products were removed from the fracture by ultrasonic cleaning in an inhibited hydrochloric acid solution and the fracture was examined using a scanning electron microscope. Some mechanical damage is present as shown at higher magnification in Figure 3. The remaining fracture features in this area indicate that the failure is the result of intergranular rupture that is indicative of embrittlement of the spring material as shown in Figures 4 and 5. No evidence of hydrogen embrittlement are present. Similar intergranular rupture is also present at the mid-radius and center of the fracture surface along with small areas of transgranular cleavage, which is the normal brittle single cycle failure mode in this material.

The tensile test bar fracture and a typical Charpy V-notch impact test bar fracture were also examined by scanning electron microscopy. The flat portion of the tensile test bar fracture occurred due to brittle transgranular cleavage as shown in Figure 6. No intergranular rupture features are present in this area of the fracture. The angled areas of the fracture occurred due to microvoid coalescence that is consistent with a ductile shear overload. This indicates that this spring did not fail due to a slowly applied single cycle overload.

The Charpy V-notch impact test bar fracture exhibits substantial amounts of intergranular rupture as shown in Figure 7. This indicates that this spring fractured due to temper embrittlement from a relatively low impact load in service. These results indicate that the elevated arsenic in the spring material is a primary contributing factor by increasing the likelihood of temper embrittlement.


Metallography is the examination of the microstructure of metal alloys, typically using an optical microscope. Because of the limited depth of focus available with an optical microscope, metallography samples are polished to a flat plane. They are then etched with acid to expose their grain structure. Variations in grain structure reveal the processing “history” of the metal. Some service applications, such as high temperature exposure, also introduce microstructural indications.

A longitudinal metallographic section was prepared through the fracture shown in Figure 4. Some oxides are present at the coil OD, indicating that this spring was not chemically cleaned prior to painting. No excessive nonmetallic inclusions are present that could have contributed to the failure, despite the elevated sulfur content of the spring material. The microstructure adjacent to the coil OD consists of uniform tempered martensite that is consistent with the specified quench and temper heat treatment as shown in Figure 8. Only slight amounts of grain boundary ferrite extend from the wire surface, which is indicative of slight amounts of carbon depletion. This is not a contributory factor to this failure.

A transverse metallographic section was also prepared adjacent to the fracture. No seams or other material discontinuities extend from the wire surface that could have contributed to the failure. No plastic deformation or microstructural alterations that could be indicative of the specified shot peening is evident adjacent to the wire surface. No abnormalities are present in the microstructure of this spring that could have contributed to the failure.


This investigation indicates that the fractured spring submitted for this investigation is made of the specified SAE 5160H medium carbon alloy steel. The spring is deoxidized with silicon and vanadium indicating that it was produced to fine-grained melting practices, which normally results in optimum mechanical properties after heat treatment. The use of vanadium, rather than the more common aluminum deoxidizing agent indicates that it was produced by a continuous (strand) casting process, rather than an ingot casting process. Depending upon the starting billet size, this could result in decreased hot working to the final wire size, which can result in increased chemical segregation and decreased mechanical properties. However, there are no indications in the microstructures of these two springs, however, that chemical segregation is a contributory factor to the failures.

The spring contains elevated amounts of arsenic that is not specified in SAE 5160H and can contribute to temper embrittlement. The elevated arsenic in this spring is a significant contributory factor to the failure.

The spring was through hardened by quenching and tempering to a hardness that is below the value specified on the part drawing. This lower hardness is indicative of decreased strength, which could increase the likelihood of failures due to overloading in service. The tensile strength of this spring is somewhat lower than expected from the measured core hardness, however, the tensile test bar exhibits good ductility and no intergranular rupture is present on the tensile test bar fracture. This indicates that this spring did not fail due to a slowly applied single cycle overload. The high yield to tensile strength ratio indicates that this spring was properly quenched.

The room temperature Charpy V-notch impact energy is relatively low, and a typical impact test bar fracture exhibits substantial amounts of intergranular rupture. This intergranular rupture is indicative of temper embrittlement due to the elevated arsenic content. Temper embrittlement occurs when asusceptible steel is heated to approximately 700 to 1070 ºF or is slowly cooled through this temperature range. The hardness is indicative of tempering within this embrittling range. Therefore, the combination of the arsenic in the steel and the tempering temperature required to obtain the measured hardness is the cause of the temper embrittlement that resulted in the failure of this spring.

No features that could be indicative of the specified shot peening is evident on this spring. However, shot peening only improves the low stress, high cycle fatigue resistance and does not increase the resistance to overload failures.

The spring was subjected to a relatively low impact load in service, which could be related to the opening or closing of the vault cover. Due to the temper embrittlement, however, a high impact load was not required to cause the failure.

The prevent fractures from these causes in future spring production:

• Specify the use of steels with low levels (below approximately 0.010% and preferably below 0.005%) of phosphorus, tin, arsenic and antimony.

• Use as low a tempering temperature as possible while maintaining the specified hardness. Preferably, this tempering temperature should be close to 700 ºF, which is the approximate lower limit where temper embrittlement can occur, and is the approximate upper limit where tempered martensite (blue) brittleness can occur.

• The two fractured springs examined to date have hardnesses that are lower than desired, which is indicative of higher tempering temperatures that increase the likelihood of temper embrittlement. This may require the development of a proprietary material and heat treatment specification.

• Consideration could also be given to changing the spring material to one that is less susceptible to temper embrittlement. Steels containing nickel and molybdenum are less susceptible to this failure mechanism, so consideration could be given to changing the spring material to SAE 4161, SAE 4340, SAE 8655, or SAE 8740. These materials, however, would be more expensive than the currently specified SAE 5160H and would not be as readily available.

• The design and operation of the vault covers should also be reviewed to determine if changes can be made to decrease the likelihood of impact loading.