Aerospace Titanium Alloy Melt Process Quality Improvements

This Jet Engine Titanium Quality Commi�ee (JETQC) paper describes industry quality improvements since 1990. Quality refers to freedom from melt-related hard-alpha and high-density inclusions (HDI). JETQC, formed under the auspices of the U.S. Federal Avia on Administra on (FAA) following the Sioux City aircra� accident in 1989, is comprised of U.S., E.U. and Japanese aircra� engine manufacturers to address the quality of premium / rotor quality tanium alloy produc on. Titanium suppliers provide melt-related inclusion data. JETQC focuses on hard-alpha and HDI inclusion rates in premium quality (PQ) tanium alloy products for cri cal rota ng aircra� engine applica ons. PQ materials typically are produced via triple vacuum arc re-melt (3XVAR) or hearth melt VAR (HMVAR) processes, but more recently, the Skull plus VAR (SVAR) process has been introduced. Hard-alpha rates have con nued to decline over the last decade primarily for the HMVAR process. HDI rates declined in the early 90’s, but more recently the overall rate has stayed approximately constant with inclusions confined to the 3XVAR process. Combining the trends for both hard-alpha and HDIs, the HMVAR process has demonstrated in recent years to be higher quality compared with the 3XVAR process.

JETQC is to track and monitor the PQ (Premium Quality) tanium alloy mill product and components used in jet engines powering commercial aircra�. The PQ tanium billet and bar materials are used to manufacture cri cal rota ng parts such as disks, blisks or integrally bladed rotors, sha�s, airfoils, etc. A secondary purpose of JETQC is to undertake proac ve measures designed ul mately to lead to lower overall inclusion rates in PQ tanium alloy products.

JETQC Membership
JETQC membership has historically included (i) the PQ titanium suppliers, currently including ATI, Arconic-RTI, PCC-TIMET, and VSMPO, (ii) the jet engine manufacturers, currently including GE Aviation, Honeywell International, MTU, Pratt and Whitney (US and Canada), Rolls-Royce (UK and US), Safran (aircraft engines and helicopter engines) and Williams International, and (iii) the FAA. Several years ago, JETQC recognized the impending global expansion of both PQ Ti suppliers and engine manufacturers and instituted a qualification process for the incorporation of new PQ Ti suppliers and engine manufacturers into JETQC. At the end of 2018, IHI Corporation was added to JETQC membership and there are currently PQ Ti melters that are in active qualification.

JETQC Repor ng
Every quarter the PQ Ti suppliers provide produc on volumes by melt method, by alloy to JETQC together with any inclusion finds in mill products. In the event an inclusion is found during inspec on, a detailed characteriza on is performed including both size and composi on. The size informa on, together with the inspec on informa on has been provided since 1990 to the Rotor Integrity Steering Commi�ee (RISC) that has used the informa on to develop probabilis c lifing methods that have been adopted by the engine manufacturers [2]. The composi on informa on has been used by the PQ Ti suppliers to understand the likely root cause(s) for the inclusion and implement correc ve ac ons within their raw material supply chain and their melt shop designed to eliminate any re-occurrence. Over the years, inclusion evalua ons by the suppliers have substan ally improved, allowing root cause(s) confidence to be significantly improved. In addi on to inclusions reported by the suppliers, the engine manufacturers also report inclusion finds in components, with similar characteriza on in terms of size and composi on.
JETQC collates the annual PQ Ti volume and inclusion sta s cs from mill products (suppliers) and components (engine manufacturers) and provides an anonymous industry summary back to the suppliers. This annual summary allows the suppliers to understand how they stack-up rela ve to the overall industry, but crucially does not divulge supplier-specific process proprietary informa on. Addi onally, over the last approximately 10 years, the annual summary has included proposed inclusion root cause(s) and the assessed confidence in the root cause(s). This has enabled suppliers to be�er understand poten al inclusion risks. This repor ng approach also provides an early warning system that enables both engine manufacturers and suppliers alike to understand if there are new (or old) threats that are present and require addi onal monitoring and/or ac on.

Melt Related Inclusions
Quality as defined within JETQC relates to the incidence of melt-related inclusions in the ingot used to manufacture billet and bar, and subsequently the cri cal rota ng parts. These inclusions include (i) Hard-alpha, also known as Type I or low-density inclusions that have elevated levels of inters al elements oxygen, nitrogen and/or carbon, (ii) high-density inclusions that are enriched in dense metals such as tungsten, tantalum, molybdenum, etc., and (iii) Type II inclusions that are enriched in Al. Figure 1 illustrates each of these three types of inclusion.

Hard-alpha inclusions
These inclusions typically form because of contamina on of raw materials or exposure of hot tanium to the presence of air or water. The presence of nitrogen in an inclusion results in a significant increase in mel ng point and slow dissolu on kine cs in molten Ti. The presence of nitrogen, oxygen and/or carbon in the inclusion results in a hard, bri�le phase that may crack during hot working of the ingot and component and/or crack during service. [3]. Hard-alpha inclusions have historically been associated with the most serious consequences including the Sioux City accident, and there have been mul ple other instances of component failures over the decades ini a ng from this type of inclusions.

Type II inclusions
These inclusions typically are associated with ingot shrinkage cavi es near the ingot top with local aluminum evapora on into the gas-free cavi es and re-deposi on onto the cavity wall during high temperature exposure related to ingot cas ng and subsequent high temperature working opera ons. One component failure was a�ributed to the presence of a Type II inclusion in 1970 [1], but since then, the PQ tanium alloy industry has effec vely dealt with this threat.

HDI inclusions
These inclusions typically are a result of contaminants in raw materials and/or process equipment and survive during mel ng due to (i) limited residence me in the molten pool due to their high density resul ng in their falling to the mushy zone during cas ng, and (ii) high mel ng point resul ng in rela vely slow dissolu on in the molten pool. There are no known component failures associated with this type of inclusion, however, they can cause significant produc on disrup on.

Premium Quality Process Routes
At the me of the Sioux City accident, the typical melt route for PQ Ti material was the triple vacuum arc re-melt (3XVAR) process. This process starts with an electrode that may be formed by (i) pressing raw materials into compacts that are welded together in a chamber, (ii) welding bulk re-cycled materials together, or (iii) consolida on melted. This electrode is then consumed during vacuum arc mel ng in a cooled copper crucible. The melted electrode is then vacuum arc re-melted a second and a third me to produce an ingot that is free from significant chemical segrega on with desirably no melt inclusions. The primary purpose of the first melt is to produce a solid ingot without any vola le residuals, such as magnesium chloride, that are associated with the Ti sponge. The second and third melts are designed to provide addi onal residence me for dissolu on of any inclusions and to provide a final ingot with a minimum of chemical segrega on from top-to-bo�om and from center-tooutside diameter.
It was recognized that the 3XVAR process was not fully capable of removing hard-alpha and high-density inclusions due to the limited residence me in the molten tanium, so the industry inves gated alternate methods to produce PQ Ti materials. This led to the development of cold hearth mel ng routes using either electron beam or plasma arc mel ng with the hearth being water-cooled copper. The electron beam mel ng method is conducted under vacuum and tends to evaporate vola le alloy elements, such as aluminum. The plasma arc mel ng method is performed under inert gas and tends to absorb some of the inert gas. Both routes use a final VAR melt to provide uniform chemistry in the case of electron beam and remove any absorbed inert gas in the case of plasma arc. The primary quality benefits of the cold hearth process are (i) molten metal flows horizontally across the hearth and so any high-density inclusions will fall to the bo�om of the hearth, whereas in the inline 3XVAR process, any high-density inclusions will fall rapidly to the bo�om and poten ally survive into the final ingot, (ii) low density inclusions will float on the surface and be subject to direct impingement of the electron beam or plasma arc, resul ng increased probability that inclusion dissolu on will occur, and (iii) increased overall residence me of up to an order of magnitude over 3XVAR allowing for a higher probability that inclusion dissolu on will occur. Addi onally, the cold hearth melt processes poten ally have a greater flexibility of raw material forms that can be used, enabling improved economics. The hearth melt plus vacuum arc re-melt (HMVAR) process was introduced around the me of the Sioux City accident [4], and over the ensuing 30 years has become the predominant PQ melt method as discussed below. One of the key reasons for the success of the PQ HMVAR process is that seeded heat trials are conducted during qualifica on and validate that the hearth melt process is extremely robust and eliminates all inclusions of all types that are inten onally seeded into the input materials. These process parameters that result in complete inclusion elimina on are then used for produc on. There is no equivalent process demonstra on for the 3XVAR process.
More recently, an alternate melt method has been introduced by one supplier that uses a water-cooled skull and arc mel ng to produce a large quan ty of molten Ti alloy and then a�er a pre-determined me, the furnace is lted and the molten Ti alloy is poured into a mold. Due to the solidifica on condi ons, a large center-line shrinkage pipe is formed along a significant propor on of the ingot requiring a subsequent VAR melt to reduce [5]. Figure 2 shows schema cally the three PQ Ti melt routes described above and includes the subsequent high-level manufacturing steps to convert an ingot into an engine component. Mul ple inspec ons are performed at different points in the process to validate that the finished component is inclusion-free. The mill product is ultrasonically inspected, the forging is ultrasonically inspected and macroetched. A key improvement in ultrasonic inspec on for both mill product and forging occurred following the Sioux City accident with the introduc on of zoned ultrasonic inspec ons that enabled a much higher sensi vity inspec on to be performed. This ini ally resulted in an increase in mill product inclusion finds, but the industry quickly iden fied the root causes of the inclusions and implemented the appropriate correc ve ac ons. More recently, phased array ultrasonic inspec ons have also been introduced.
The FAA has issued AC33-15.1 that captures the high-level lessons learned regarding mel ng, processing and inspec ons related to the produc on of PQ Ti parts [6].

Results
JETQC has published the hard-alpha inclusion sta s cs at two prior World Titanium Conferences [7,8]. The ini al hardalpha inclusion heat rate was more than 1 per million pounds of PQ Ti material produced in the a�ermath of the Sioux City accident when the industry data was first collected. JETQC uses inclusion heat rates to measure overall inclusion rates since it is possible that an inclusion that is present during mel ng may (i) remain as a single en ty, or (ii) disintegrate into mul ple fragments either during mel ng or billet conversion. The use of inclusion heat rate eliminates this issue, providing that in the case of mul ple inclusions in a heat they are confirmed to be of the same type and chemistry and therefore likely origina ng from the same root cause. Reference [8] showed that the hard-alpha inclusion rate in the 2002-2005 meframe had decreased by an order of magnitude to approximately 1 in 10 million pounds of PQ Ti material. This reference also described the many raw material and process improvements that were responsible for the more than an order of magnitude improvement in hardalpha inclusion heat rate. At a very high level, these improvements included (i) use of vacuum dis lled sponge, (ii) increased controls surrounding all raw materials, (iii) more robust melt processing methods related to electrode prepara on, mel ng, etc., and (iv) a�en on to melt shop house-keeping to avoid contaminant pick-up onto in-process electrodes and contaminant drop-in into melt furnaces.
This paper now updates the hard-alpha inclusion sta s cs up to 2016 and in addi on provides a summary for HDI inclusion sta s cs. This paper also addresses at a high level for the first me the difference between 3XVAR and HMVAR inclusion sta s cs since it has become clear that there is a significant difference in quality between the two processes. Figure 3 shows the updated hard-alpha inclusion heat sta s cs in 3-year increments and it shows that since the last report which covered through the end of 2005, there has been a further reduc on to approaching 1 hard-alpha heat per 50 million pounds of PQ Ti material over the last three reported years (2014-2016). This represents a con nued outstanding performance by the Ti suppliers, driven by incorpora ng prior lessons learned and con nuing to pay great a�en on to detail in the manufacture of PQ Ti used in cri cal rota ng aircra� engine applica ons. HDI heat sta s cs Figure 3 also shows for the first me the high-density inclusion heat sta s cs in 3-year increments. The data show the ini al rate was on the order of about 0.4 HDI heat per million pounds of PQ Ti through the mid-nine es and then declined to less than 0.1 HDI heat per million pounds of PQ Ti through 2016. The primary reason for this reduc on is a�ributed to the significant increase in controls over raw materials and melt process equipment repair. Unlike the hard-alpha inclusion heat rate data shown in Figure 3, the HDI inclusion heat rate data do not show a con nued decline over the last approximately 10 years. This is largely been a�ributed to the presence of small, molybdenum-rich inclusions that have been detected in molybdenumbearing alloys such as Ti-6246, Ti-6242 and Ti-17 as zoned ultrasonic billet inspec on was adopted across the industry. Figure 4 shows for the first me the split between hard-alpha and HDI inclusion heat rates in billet produced by 3XVAR and HMVAR. The absolute inclusion rates are not shown in Figure 4 to protect supplier proprietary informa on; however, the figure is informa ve as it shows an approximately similar overall inclusion rate in the early years between produc on of HMVAR and 3XVAR material. In the early years, the hard-alpha inclusion heat rate is clearly higher for HMVAR material, while the HDI inclusion rate is clearly higher for 3XVAR material. As described earlier, HMVAR was introduced for several reasons, including density separa on that would allow HDIs to sink to the bo�om of the hearth. Figure 4 in the early years clearly bears this an cipated benefit out. Based on the longer residence me in the hearth, it was also expected that the hard-alpha inclusion heat rate should have been lower in HMVAR compared to 3XVAR but Figure 4 does not bear this out. It is believed that this is due to a combina on of (i) methods used to condi on the HM electrode surface prior to the VAR melt step (i.e. post-hearth melt), and (ii) adop on of zoned billet inspec ons that were adopted predominantly in HMVAR materials in the early produc on years. Figure 4 shows in recent years that there has been a drama c shi� in billet inclusion rates, par cularly for HMVAR, with a total elimina on of HDI heats, and close to zero hard-alpha inclusion heats represen ng about a 50X rate reduc on. For 3XVAR, there has been a smaller, approximately 3X reduc on in inclusion heats with both hard-alpha and HDI inclusion heats remaining. The total billet volumes represented in recent years in Figure 4 is approximately 100 million pounds for both HMVAR and 3XVAR.

Conclusions
Several conclusions can be drawn from this work:

1.
JETQC membership has expanded and is expected to con nue to expand.

2.
The en re PQ Ti industry has focused on the threat of melt-related inclusions to improve air travel safety and through co-opera ve efforts with the FAA, the suppliers and the OEMs has resulted in a significant reduc on in risk.

3.
HMVAR melt process has been demonstrated to be extremely robust and has consistently delivered the highest level of PQ Ti quality to the industry. The sta s cs support that the final VAR melt step has been well-controlled.