Corrosion fatigue: An Anatomy of Die-Cast Die Insert Failure.

A rash of large H-13 diecasting die insert failures prompted a study that revealed a corrosion fatigue mechanism as the culprit.

Cracking of large die inserts is a constant problem in the diecasting industry. Large die-cast inserts range from $8000-$36,000 or more. A typical large insert has a replacement value of $23,000 and a leadtime of 14 weeks.

Obviously, maintaining a large inventory of inserts is a burdensome expense to most diecasters. Plus, the removal of a failed insert and the installation of a replacement detail can easily consume several shifts if the insert is large and part of a complicated sliding core arrangement. This disrupts schedules for both the casting and machining plants. Replacement and downtime penalties associated with premature failure dictate that every effort must be made to ensure long insert life.

Recently, a stream of unexplained die insert failures appeared at a large diecasting facility. This article reviews the findings of a study initiated to determine the cause of the failures.

The Problem

High-pressure aluminum diecasting has been plagued by cracking failures of H-13 steel inserts due to various causes. A premature failure can be defined as one that occurs before the cavity insert completes its expected life, which is typically 180,000 shots. Most failures occur in the 15,000-65,000 shot range. Normal operating pressures range from 9000-14,000 psi and thermal cycling can range from 250-1100F (121593C). These factors combine to produce high residual stresses and progressive softening of the material that, in turn, promote heat checking and crack propagation. EDM employed in the build of the H-13 cavity inserts also adds to the possibility of crack initiation.

Several factors or improvements have significantly reduced the incidence of premature failures due to cracking. The introduction and application of modem specifications for premium grade H-13 steel have all but eliminated most gross cracking failures. These specifications require modem high tech processing of the raw material that produces metallurgical characteristics to ensure good microstructure and high impact toughness values. High impact toughness increases the resistance to crack propagation. Good structure also enhances toughness and improves resistance to both softening and heat checking.

The heat treatment of H-13 steel has undergone a similar evolution. In general, poor heat treat ruins good steel; conversely, good heat treat yields great results if the material is of world-class quality. Most tools are hardened in vacuum furnaces and are rapidly quenched via high-pressure (6-10 atmosphere) nitrogen backfill coupled with efficient heat exchange. This guarantees heat treat characteristics close to those achievable in an ideal laboratory heat treat, thus getting the most from the steel supplied by the mill.

Most die failures start on the cavity surface. Typical failure modes include washout, heat checking, pitting, EDM-derived cracking and catastrophic cracking. Concurrent with the elimination of premature failures, a novel class of failure originating from the interior of the insert appeared to have emerged. This failure mode may have been present for a long time but overlooked because other failure modes intervened to cause havoc before its effects could be seen.

In this study, at a major diecaster, the H-13 inserts experiencing failure all had the following common characteristics:

* the material was the best premium H-13 steel available;

* heat treat was the best commercially available;

* material and heat treat met stringent specifications;

* material and heat treat quality of each insert was individually tested by an outside testing laboratory;

* all inserts were polished/sandblasted and stress relieved after EDM.

If the inserts were built to the above quality standards, why were they failing prematurely? The failure rate had dropped at the diecasting facility for several years but had hit a disturbing plateau. Could the facility be witnessing a new failure mode? The following study was initiated to yield an answer.

Method

The study's method is briefly summarized below:

Original Material & Heat Treat Testing--All H-13 inserts have an assigned serial number once the purchase order is issued to buy the material. All test data, insert build schedules and production data for H-13 inserts are tied to the serial number, so all inserts had traceable records from rough machining to failure. All of the eight failed inserts had readily retrievable raw material and heat treatment test results during their construction (Table 1).

Testing of H-13 Inserts After Failure--The inserts were tested again when received by the laboratory for failure analysis. A chemical analysis was performed using OES and compared to the specification. All eight pieces were confirmed as premium H-13 material. Charpy v-notch impact toughness testing was then performed in two stages. The first stage assessed the as-received quality of the commercially hardened insert. The second stage took the same sampling plane coupons machined for stage 1 and subjected them to a laboratory heat treat. Variations in hardness accounted for some differences in charpy impact values when compared to original test results (Table 1).

Visual Examination--An overall visual examination of each insert was made with a stereomicroscope at magnifications up to 40X to establish the condition of the insert and the location of the cracks. Dye-penetrant inspection was also used to highlight cracking and heat checking.

SEM Examination--A section of the fractured surface containing the origin area from each failed insert was placed in the vacuum chamber of a scanning electron microscope (SEM) to help determine the failure origin and crack propagation characteristics of the fracture. Beach marks and chevron patterns indicated the fracture origins and the cyclical nature of each primary fracture. Intergranular cracking was observed along the primary fractures with cleavage or quasi-cleavage observed along secondary fracture paths. Images from the SEM were digitally recorded and stored for the failure analysis reports and future reference.

Microstructure Examination--A section through the apparent crack origin area was removed from each failed insert, mounted in a Bakelite mold and prepared according to standard metallographic procedures. The prepared specimens were examined in both the as-polished and etched conditions using a metallurgical microscope capable of magnifications up to 1000X. A microcleanliness exam was performed and the specimens were then etched in a 5% nital solution and re-examined. Core microstructure, banding, carbide precipitation, primary carbides as well as other characteristics were determined and each structure was rated. Digital images were recorded, made part of each failure analysis report and archived for future reference.

Microhardness Testing--Microhardness traverses were performed on all relevant metallographic specimens using a digital microhardness tester equipped with a diamond indenter and a 500-gram load. Microhardness was measured at various depths below the surface adjacent to the transverse fracture and the cross-section of the longitudinal fracture.

Energy Dispersive X-Ray Spectroscopy--Specimens from the crack surface were repolished, rinsed and dried. They then were inserted in the vacuum chamber of the SEM and an energy dispersive X-ray spectroscopic unit was used to quantitatively analyze deposits observed within the cracks.

Deposit and Water Analysis--Deposits were scraped from several of the cooling lines in a failed spline core insert and were subject to X-ray diffraction analysis for compound identification. A water sample was collected from the cooling system and was subjected to chemical analysis using ICP and OES techniques. The cooling water chemistry showed: pH-8.07 units; sulfur-73 ppm; chloride-2.96 ppm; and phosphorous-2.54 ppm.

Results

The results of this testing revealed that all eight cavity inserts (which met all the chemical, metallographic and impact toughness requirements for premium quality H-13 steel) failed prematurely by a corrosion fatigue fracture mechanism (see sidebar on p. 47). The inserts failed in relatively similar manners and shared the following attributes:

* all eight inserts had been commercially hardened using high-pressure vacuum quenching and met the microstructure, hardness and impact toughness requirements;

* although some cracking and heat checking was detected along outside impression surfaces, visual and macroscopic examinations revealed that the failures actually initiated internally along the bore diameter of cooling lines;

* SEM inspection revealed an inter-granular fracture mode at most crack origins, often accompanied by shallow pitting corrosion;

* multiple parallel crack origins were detected, many filled with corrosion byproducts;

* crack propagation appeared to occur in a repetitive, cyclical manner rather than by a single catastrophic event;

* scale and deposits inside the cracks and along the diameter of cooling lines were composed of iron oxides and calcium carbonates, along with compounds of sulfur and chlorine. Cooling water chemistry revealed a low chloride content with a moderate sulfur concentration in a slightly alkaline pH.

Analysis

In diecasting tooling, thermal stresses act on the cavity insert along with mechanical stresses and residual stress. The thermal stresses result from the rapid temperature excursions and temperature differences experienced by the detail of every shot. Conventional wisdom assumes that the impression surfaces are exposed to the greatest thermal stresses as these surfaces experience the greatest temperature fluctuations during diecasting. Temperatures fluctuate from 1200F (650C) as the molten aluminum contacts the die surface to about 400F (200C) at the end of the die lube spray cycle. Work recently published by J.F. Wallace and D. Schwam, however, shows that the surface stresses resulting from this temperature excursion may be mostly compressive in nature, reaching a slightly positive or tensile stress at the lowest temperature portion of the surface temperature profile. Surprisingly, although the bore diameter surface of a cooling line experiences a smaller temperature gradient during the diecasting cycle, th e resultant stresses may be entirely tensile in nature and far greater in magnitude than the tensile stresses acting on the surface of the insert.

In addition to cyclical tensile stress, corrosion fatigue requires the simultaneous action of a corrodent. While it is obvious that water will attack H-13 steel, as stated above, once a thin passive film of iron oxide is established, the rate of corrosion is usually insignificant to the overall life of a die, The results of this investigation, however, found severe corrosive action inside the water-cooling lines. An important characteristic was the fact that the sulfur and chloride contents found in the corrosion deposits were considerably greater than the level measured in the cooling water.

This concentration of aggressive constituents suggests that repetitive vapor blanketing occurred within the cooling line. Coolant chemicals and mineral deposits precipitate from the water when vapor is formed due to overheating. In fact, studies have shown that impurities can concentrate 10-100 times in a cyclical water-vapor-water cycle. Further, a near-neutral or slightly alkaline pH of the cooling water can become extremely aggressive, either acidic or caustic, as constituents in the water concentrate from the alternating liquid-vapor-liquid cycle.

All eight die cast die inserts failed prematurely by corrosion fatigue that initiated at the internal cooling line surfaces. This failure mechanism requires a combination of a corrosion accelerant and cyclic stress. The corrosion accelerant, determined by EDS, XRD and water analysis, was the concentration of sulfur and chlorine due to formation of a vapor phase inside the cooling passage. The cyclic stress was most likely due to thermal shock. Both the corrosive environment and the thermal shock were apparently due to insufficient or stagnant flow of coolant through the cooling line during diecasting. This allowed the cooling water to overheat, resulting in the formation of a vapor blanket between the surface of the cooling line and the liquid coolant. While the vapor blanket was present, the rate of heat extraction from the cooling line surface is drastically diminished. The repeated formation and collapse of the vapor blanket would result in deposition and concentration of coolant chemical and mineral deposits. This cyclic formation and collapse of a vapor blanket would also result in thermal shock due to rapid cooling upon colla pse of the vapor blanket.

In addition to creating thermal shock at the cooling line surface, this condition also may result in greater thermal shock at the working surface. The lack of heat extraction due to the vapor blanket would result in higher temperatures at the working surface. During extraction of the spline core from the casting, the working surface would become exposed to air and die lubricant, resulting in thermal shock.

In summary, the failures studied initiated internally and not from surfaces that were in contact with the aluminum shot. The raw material quality and heat treatment quality of the details were found to be satisfactory.

The corrosion fatigue condition appears to be the result of insufficient water flow through the cooling lines. A cyclical vapor blanket appears to have formed with each shot, causing the concentration and precipitation of aggressive constituents on the cooling line diameter surfaces. Acting simultaneously with corrosion, the collapse of the vapor blanket caused a repetitive thermal shock due to the increased cooling as the liquid phase was reformed, Thus, due to insufficient or stagnant flow through the cooling channels, both a corrosive environment and repetitive thermal stresses were established to initiate and propagate a series of corrosion fatigue failures. The best preventive method is to ensure that sufficient water flows in all water fountains during downtime and startup periods of die operation.

          Original Material/Heat Treat vs. Testing After Failure
                          Original
                        Material Test        Heat Treat Results
S/N      Description      Structure   Charpy     Structure      Charpy
9625     Bottom Core         B3        8.0           6           8.0
11913   Operator Core        B3        13.0          20          11.7
11756 Upper Helper Core      B2        13.0          5           8.0
10388    Spline Core         A2        12.3          3           9.3
10171    Spline Core         A2        12.3          3           9.7
32029    Spline Core         B1        13.0         4/20         11.7
32028    Spline Core         A2        14.7         4/20         12.7
32030    Spline Core         A2        12.7          4           10.7
      Post-Failure
      As-Received         Ideal Heat Treat
S/N    Structure   Charpy    Structure     Charpy
9625       --       11.0         --         16.0
11913      --        8.7         --         15.0
11756      --        7.7         --         13.0
10388      --       12.0         --         10.3
10171      --       12.0         --         13.3
32029      --        7.3         --         12.7
32028      --        7.7         --         14.3
32030      --        9.0         --         12.3
              Visual/Macroscopic Appearance & SEM Examination
S/N   Description   Crack Location (Visual/Macroscopic)
9625  Bottom Core   Through the shot block recess and
                    cooling line hole. Parallel to cooling
                    line centerline
11913 Operator Core Through core pin and bell area
11756 Upper Helper  Through bell face of
      Core          core toward side
10388 Spline Core   No apparent surface cracks. Cutups
                    revealed internal WF cracks
10171 Spline Core   5 in. from tip... through diameter
                    approx. 240 deg. radially
32029 Spline Core   Longitudinal crack along small
                    exterior radius
32028 Spline Core   Shaft broke completely off.
                    Radial crack clear-through
32030 Spline Core   7 in. from tip. Through diameter.
                    Complete separation
S/N   SEM
9625  Origin area. No gross non-metallic
      inclusions. Heavily oxidized and
      impossible to get good look at
      underlying fracture surface.
11913 Intergranular fracture. Grain boundaries
      chemically attacked.
11756 Fatigue fracture and intergranular
      fracture modes observed
10388 Intergranular fracture path.
      Corrosion present
10171 Intergranular fracture mode. Severe
      obscuring of the area just beyond
      initiation.
32029 Some fatigue fracture. Quasi-cleavage
      fracture. Intergranular fracture mode
      at a second site
32028 Intergranular fracture mode. Corrosive
      intergranular attack. Quasi cleavage
      fracture
32030 Intergranular fracture mode. Corrosive
      intergranular attack
                         X-Ray Diffraction Results
Appearance of Sample      X-ray Diffraction Deposit Composition
White deposit, chunks     Calcium carbonate
Light brown deposit       Calcium carbonate & Calcium carbonate
                          hydrate
Light & dark brown substr Calcium carbonate & Iron oxide
Dark brown substrate      Calcium carbonate; Iron sulfide; Iron ox

What is Corrosion fatigue?

Corrosion fatigue is defined as the combined action of an aggressive environment and a cyclic stress to result in premature failure by cracking. First discussed in the literature as a specific failure mode in 1917, corrosion fatigue is the synergistic action on a metal by tensile stresses simultaneously with corrosion. An insidious mechanism, the tensile stress sufficient to cause failure is usually low-to-moderate, well below the yield strength of the metal in question and insufficient to cause any damage when acting independently. Similarly, the hostility of the corrosion is relatively low, with the usual extent of the corrosion damage being shallow pitting. But due to the synergistic effect of the two independent conditions acting simultaneously, the total damage is far greater than the sum of the damage arising from the two actions alone.

Several proposed theories explain the corrosion fatigue mechanism. Most metals, including H-13 die steel, owe much of their corrosion resistance to the creation of a thin, tightly adherent oxide film. After a minor amount of corrosion to establish this passive surface deposit, the corrosion resistance of the metal is greatly enhanced. In corrosion fatigue, however, the repetitive cyclic stresses cause hairline ruptures of the passive protective surface oxide. Fine cracks initiate and then propagate as the surface oxide is successively reformed and ruptured at a specific location.