Engineering Failure Analysis 115 (2020) 104618

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Failure analysis of a titanium Coriolis mass flow meter: A case of

T
hydrogen embrittlement
I.G.R. Santosa, , G.S. Vacchia, R. Silvaa, C.L. Kugelmeiera, D.C.C. Magalhãesa,
⁎

G.R. Campesanb, C.A.D. Roverea,

⁎

a
Federal University of Sao Carlos, Graduate Program in Materials Science and Engineering, Rodovia Washington Luis, Km 235, 13565-905 São
Carlos, SP, Brazil
b
Materials Characterization and Development Center, Department of Materials Engineering, Federal University of São Carlos, Rodovia Washington
Luis, Km 235, 13565-905 São Carlos, SP, Brazil

ARTICLE INFO ABSTRACT

Keywords: Titanium alloys are suitable for construction of the Coriolis mass flow meter due to their high
Coriolis Mass Flow Meter strength to weight ratio, high thermal stability and excellent corrosion resistance. However,
Grade 9 titanium environments with high hydrogen content can be very harmful to titanium and its alloys, since
Raney nickel Interaction this element may cause hydride formation and leads to hydrogen embrittlement. In this context,
Hydrogen Embrittlement
this paper reports on the analysis of a grade 9 titanium tube of a mass flow meter, which was
Premature Brittle Fracture
performed using cyclohexanol (68%), water (2%) and Raney nickel catalyst (30%). After the
active period, the flow meter was deactivated and left on standby for one year, and then re-
assembled on a production line, whereupon it failed prematurely. The failure was analyzed by
OM, SEM, EDS, Vickers microhardness and XRD, whose results indicated that the cause was
hydride embrittlement due to the chemical interaction between Raney nickel and the titanium
tube during the deactivated period.

1. Introduction

One of the most widely measured physical quantities in industrial processes is flow. It is measured based on the volumetric or
mass quantity of a fluid that flows through a section of a channel per unit of time, using the most varied physical principles. The
Coriolis mass flow meter, an instrument for taking direct measurements of the mass of liquid or gaseous fluid flows, is widely used for
batching and precision filling in the food, chemical, petrochemical and pharmaceutical industries, due to its good accuracy, low
maintenance requirements and compact size [1,2].
The device consists of a pair of parallel tubes (or a single tube with two parallel sections), which vibrates at low amplitude and
resonant frequency through the interference of an electromechanical actuator. When the flow starts, the Coriolis force acts in pro-
portion to the fluid mass, modifying the tube’s vibration frequency in response to the principle of inertia. This change in vibration is
then captured by an outlet sensor, which records it and compare with an inlet sensor signal, thereby enabling the density and mass
flow rate that passes through the measuring tube to be determined [1,3].
In order to avoid corrosion, wear and fatigue problems, mass flow meters are usually made of stainless steels, although these
materials have limitations in terms of geometry and temperature. Thus, zirconium, tantalum, titanium, ethylene tetrafluoroethylene
and glass are often suitable alternatives for some applications [2].

⁎
Corresponding Authors.
E-mail addresses: icarogrsantos@gmail.com (I.G.R. Santos), carlosdrovere@gmail.com, rovere@ufscar.br (C.A.D. Rovere).

https://doi.org/10.1016/j.engfailanal.2020.104618
Received 6 July 2019; Received in revised form 6 May 2020; Accepted 25 May 2020
Available online 28 May 2020
1350-6307/ © 2020 Elsevier Ltd. All rights reserved.
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 1. Coriolis mass flow meter: (a) Overall view, and (b) disassembled device.

Among these materials, titanium alloys are cost efficient because of their high strength to weight ratio, high thermal stability and
excellent corrosion resistance, which increases the product lifetime. In addition, the thermal expansion coefficient of titanium is
almost half that of stainless steels, enabling fluid applications for temperatures up to 150 °C without damaging the device through
thermal stress. However, environments with high hydrogen content can be very harmful for titanium and its alloys, as the high
affinity of titanium for hydrogen can cause the hydrides formation and lead to premature failures [3–5].
The next sections describe the failure analysis of a Coriolis mass flow meter made out of grade 9 titanium. The device was used to
measure the mass of a flow of cyclohexanol (68%), water (2%) and Raney nickel catalyst (30%) on a nylon 6.6 fiber production line.
After being in active use for ten months, the device was put on standby for a period of one year and then reassembled on the
production line. During the reactivation process, the electronic components in the flow meter malfunctioned, rendering the device
inoperative. A visual inspection revealed a crack in the measuring tube, which caused leakage of the solution. The failure was
examined by macroscopic techniques, optical microscopy (OM) and EDS composition analysis. Vickers microhardness, X-Ray dif-
fraction (XRD) and scanning electron microscopy (SEM) techniques were also employed to gain a better understanding of the failure
process.

2. Case study

The Coriolis mass flow meter (see Fig. 1) was in active use for 10 months in a chemical plant that synthesizes nylon 6.6 fiber.
During that period, the fluid that flowed through the measuring tube was a solution of cyclohexanol (68%), water (2%) and Raney
nickel catalyst (30%). After being deactivated for one year, the device was reassembled on the production line, but the electronic
components of the device stopped working during the reactivation process. An emergency shutdown was called and the device was
disassembled for a careful inspection. The first investigation concluded that a solution leakage occurred from a crack in the measuring
tube. This required a more in-depth analysis to determine what caused the crack, since the tube had a history of normal operation.

2
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 2. (a) Longitudinal crack in grade 9 titanium measuring tube, (b) and (c) crack macrograph with failure starting point (red arrows) and
undamaged brazing filler (blue circle) highlighted.

3. Experimental procedure

The failure analysis began by disassembling the flow meter and visually inspecting the integrity of its components, locating the
failure and examining the environment. Then, microfractographic analysis using a Philips XL-30 FEG scanning electron microscope
and chemical analysis by EDS of the outer surface of the measuring tube were performed. In addition, chemical analysis of low atomic
mass elements was carried out in a LECO gas analyzer.
A cross section was cut out 10 mm from where the crack started and the retrieved specimen was prepared following metallo-
graphic techniques, etched with Kroll solution (10 mL HF + 5 mL HNO3 + 85 mL H2O) and analyzed under an Olympus BX41M-LED
optical microscope. Philips XL-30 FEG scanning electron microscopy was performed to surface analysis of the cross section without
etching. XRD analysis of the inner and outer surfaces of the tube was conducted with a Bruker D8 Advance ECO diffractometer, using
Cu-Kα radiation, 2θ range of 30° to 90° and a 0.02° step size. Lastly, hardness was measured with a Shimadzu FM-800 microhardness
tester, under a load of 300 gf maintained during 15 s, throughout 128 measure points in 0.15 mm steps on the entire cross section of
the tube wall thickness.

4. Results

4.1. Visual inspection and chemical analysis

Fig. 2 shows the crack of measuring tube, where can be noted that it started at two different points next to the brazed ring (red
arrows) of the measuring tube and converged into a single front at 10 mm. The total length of the crack was about 65 mm and the
propagation occurred in the longitudinal direction. In addition, the tube has been presented brittle regions next to the crack with
minimal mechanical stress.
EDS chemical analysis of two sections of the tube’s outer surface, as shown in Fig. 3, revealed the presence of a higher nickel
content in a dirty section with leaked solution than in a cleaned section, which may be ascribed to the flow solution. Some K, Na, Br
and Cl was also present in this dirty section, possibly originated from the solution (of the reactive medium) or from handling during
the disassembling process. Furthermore, the EDS analysis indicated that the tube is a standard ASTM tube (see Table 1), as specified
in the manufacturers manual [6].
An additional EDS chemical analysis conducted at the brazed region (not shown) revealed it to be Ag-7.5Cu filler material. It was
well-established that the brazing processes in grade 9 titanium alloy were conducted at nearly 870 °C, and a heat treatment is
required after this joining process to minimize microstructural changes [7]. This caught our attention, given that it could be the root
cause of failure. As cracking began near this region, incorrect parameters of thermal treatments, such as the brazing atmosphere, may

3
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 3. (a) SEM analysis of external surface, (b) and (c) EDS chemical compositions of (1) cleaned and (2) dirty surface of the tube, respectively.

Table 1
– Composition of the analyzed tube and grade 9 titanium alloy (wt.%).
a a a b b b b c
Ti Al V Fe C O N H

Tube 94.7 2.8 2.5 n/d n/d n/d n/d 0.28

Grade 9 Ti Bal. 2.5–3.5 2.0–3.0 0.25 0.10 ≤0.012 ≤0.02 ≤0.013

a
EDS chemical analysis; bnot determined; cLECO gas analysis.

have been one of the causes of the premature failure of the titanium tube; however, the brazing aspect appears to be adequate
(supplementary material) and a more detailed microstructural characterization would be necessary to confirm this assumption.

4.2. Microstructural and fracture surface analysis

Analysis of OM images indicated that the material contains equiaxial α grains with β structure around the grain boundaries (see
Fig. 4b), which is characteristic of annealed grade 9 titanium. Moreover, as shown in Fig. 4c, 4d and 4e the tube exhibits two different
microstructures next to the crack, showing equiaxial grains at the outer surface and a needle shaped structure with high density of
microcracks near the tube inner surface.
To further evaluate these microstructural features, SEM analysis was made in the cross section next of the crack, revealing cavities
and microcracks on the tube’s inner surface. In addition, the elemental mapping of the cross section (see Fig. 5) indicated a slight
titanium-depletion in this region and some aluminum-rich particles, which were probably residues from the sample’s preparation.
As the SEM analysis was inconclusive about these differences between the structures of tube’s inner and outer surfaces, XRD
analysis was made on both surfaces near the crack and the peaks were correlated with data in the literature, as presented in Fig. 6
[8–11]. At the outer surface, the XRD patterns indicated α, β and Al2V3 phases, typical of grade 9 titanium. On the other hand, at the
inner surface, the analysis revealed mainly titanium hydrides peaks, demonstrating that hydride precipitation occurred, probably, in
response to interactions between the material and the chemical environment.
Finally, the fracture surface analysis performed under high-magnification SEM, shown in Fig. 7, revealed brittle aspect of the
failure. Furthermore, the crack propagation occurred alongside the grain boundaries, which characterizes an intergranular fracture,
implying that the microstructural change probably caused embrittlement in the surroundings of the crack.

4.3. Hardness 2D mapping

Hardness measurements revealed some variations among the tube wall thickness and as function of distance from the crack. In
Fig. 8 can be seen that at the bottom half of the tube (near the crack), the average hardness of the inner surface was 373 HV, while
that of the outer surface was 221 HV, representing a 68.8% increase in a wall thickness, from the outer to the inner surface. On the
other hand, the hardness in the upper half of the tube shown an increase in hardness of only 16.9% from the outer to the inner
surface.
Along these lines, the hardness mapping revealed that the main region affected by the microstructural change was the bottom part
of the tube, and a phase transformation led to great hardness increase. The hardness differences indicated that this transformation
occurred from the inner towards the outer surface, thus refuting the initial hypothesis of unsuitable heat treatments after brazing.
However, as this change occurred mainly in the bottom half of the tube, may have been a hydrogen attack caused by some residual
solution during the time where the tube was left on standby.

5. Discussion

The Coriolis mass flow meters are used in the chemical industry, where hydrogen-rich environments are very common during the

4
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 4. Microscopic analysis of cross section after Kroll etching: (a) OM image of a cross section of the tube; (b) opposite end of the crack; (c) overall
microstructure of crack nearest region, (d) higher magnification of dual structure at the crack nearest region and (e) needle shaped microstructure at
inner surface of the tube.

synthesis of polymeric and organic products. To exemplify, some phenols and alkenes are usually hydrogenated using Raney nickel
catalyst, which is very efficient because of its high activity and adsorption of hydrogen content [12]. On polyamide manufacturing
lines one traditional chemical route is shown at Fig. 9, where Raney nickel is employed to hydrogenate phenol to cyclohexanol, which
is then further oxidized with HNO3 to form adipic acid and Nylon 6.6 or dehydrogenized to form cyclohexanone and Nylon 6 [13].

5
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 5. Elemental mapping near the crack and on the inner surface of the tube.

Fig. 6. XRD patterns of titanium measuring tube: (a) Outer surface, and (b) inner surface.

Fig. 7. SEM images of the fracture: (a) crack surface, (b) detailed region of crack surface.

6
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

Fig. 8. Hardness mapping of the tube wall cross section.

Fig. 9. Nylon synthesis routes using Raney nickel [13].

Due to high activity of Raney nickel catalyst, the direct contact with titanium and its alloys can be very risky, since this materials
can suffer from hydrogen attack. Ivey and Northwood [14] showed that there is a correlation between hydride formation and the
electronegativity of hydrogen-related species. Moreover, when Ni or Pd are in contact with some more electropositive metals (hy-
drogen-related), they may cause an instant attack due to dissociated hydrogen. Thus, the active Raney nickel can be a powerful source
of hydrogen to the titanium hydride formation.
As previously reported in [15], hydrogen-rich environments are dangerous to systems which form stable hydrides, because it can
cause microstructural changes and negatively affect the mechanical properties. In α + β titanium alloys composed mainly of α phase,
hydrides can form at the metal/environment interface, where the alloy becomes degraded by the interaction with hydrogen. This
leads to microstructural changes due to hydride precipitation and successive breaks in the hydride film. However, hydrogen diffuses
more readily in β phase, thus enabling some atoms to diffuse throughout the material’s structure. Hence, if the concentration reaches
the hydrogen solubility limit, the result is hydride precipitation at the α/β interface [16].
The solubility limit, also reported in the literature [17] as the critical hydrogen content for hydride precipitation and hydrogen
damage in α and α + β titanium alloys, is influenced by the alloying content, even reaching levels below 0.015 wt% in the case of
commercially pure titanium alloys. In recent work, Tao et al. [18] showed that a Ti-6Al-4 V (α + β alloy), hydrogen charged with
0.035, 0.070 and 0.140 wt% after welding process, suffered a decrease of fatigue life as the increase in the hydrogen content, despite
the nonexistence of hydrides. Furthermore, the authors showed that with more hydrogen addition, the fatigue source region changed
from ductile to brittle fracture, where a quasi-cleavage feature was present in the fracture of the 0.140 wt% hydrogen charged alloy.
In the present study, the LECO gas analysis, shown at Table 1, indicates that the hydrogen content in the analyzed tube was
twenty-fold above the upper limit of the alloy, as result of the hydrogen attack caused by the solution/titanium interaction.
Moreover, in Fig. 4, the cross-sectional SEM images shows a dual microstructure in the tube wall. Equiaxial grains are present at
the outer surface of the tube and at the inner surface a needle shaped structure appeared, due to a phase transformation. The exposure
to a hydrogen-rich environment caused strong hydride precipitation at the inner surface of the measuring tube, as observed by the
inner XRD pattern (see Fig. 6b). In addition, hydride evidently precipitated not only at the surface but also throughout the grain

7
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

boundaries, proving that the hydrogen attack followed the two above-cited mechanisms. SEM and OM analyses also indicated that
this hydride precipitation was detrimental to the device, in view of the high density of microcracks, due to TiH2 phase formation.
As for hardness, hydride precipitation in the microstructure of the titanium tube evidently leads to localized hardness en-
hancement and decreases the mechanical strength, given the tube’s fragility. Moreover, the localized hydrogen attack created a
favorable environment for crack initiation and as the titanium hydrides also precipitated at the α/β interface, a convenient path for
crack propagation was formed due to the lower stress–strain response and higher crystal lattice parameter of titanium hydride in
relation to α and β phases. For these reasons, when stress was applied to the measuring tube during reactivation of the Coriolis mass
flow meter, the internal stress to microcrack propagation was probably reached, causing the separation of grain boundaries and
leading the material to fail prematurely.
Finally, the scarcity of data about hydrogen embrittlement failures in Coriolis titanium tubes shows the importance of our paper,
since it describes a distinctive case of engineering failure of this component caused by management mistakes and incorrect materials
selection. The technical paper “The Nature of Hydrogen Gas and the Benefits of Coriolis Fluid Measurement Technology” [19] shows
that for hydrogen-rich environments, the best materials for tubing construction are 304L and 316L Austenitic Stainless Steels or C-22
Hastelloy, due to its improved hydrogen embrittlement resistance. Thus, the titanium is not even considered to construct these
components for hydrogen-rich environments and the use of this material for tubing in Coriolis mass flow meters should be warned to
avoid future failures in Nylon manufacturing lines and related situations.

6. Conclusion

Based on the findings of this investigation, the titanium tube of the Coriolis mass flow meter failed due to hydrogen embrittle-
ment. The presence of Raney nickel and high hydrogen content in the residual solution led to an aggressive attack at the inner surface
of the tube, triggering hydride formation. Since the tube was manufactured with grade 9 titanium, the β phase contained in this alloy
favored the diffusion of hydrogen and the precipitation of hydride at the grain boundaries. Furthermore, these microstructural
changes in the alloy created a favorable environment for crack nucleation and propagation, leading the material to premature failure
in response to the external stress solicitation.
The recommended solutions to this problem are: (i) avoid the stagnation of liquids with Raney nickel inside the Coriolis mass flow
meter measuring tubes; and (ii) perform annealing in a vacuum atmosphere before reassembling Coriolis mass flow meters at
manufacture lines, despite the high cost of this process. The use of a Coriolis mass flow meter whose measuring tube is made of alloys
resistant to hydrogen embrittlement or surface treated titanium alloys is also encouraged on nylon 6.6 production lines.

Acknowledgements

The authors gratefully acknowledge CNPq (National Council for Scientific and Technological Development − grant no. 311163/
2017-3) for their financial support of this work. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal
de Nível Superior - Brasil (CAPES) - Finance Code 001. We also acknowledge the Laboratory of Structural Characterization (LCE) of
the Department of Materials Engineering at the Federal University of São Carlos for the use of its facilities.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.engfailanal.2020.104618.

References

[1] L. van de Ridder, W.B.J. Hakvoort, J. van Dijk, J.C. Lötters, A. de Boer, Quantification of the influence of external vibrations on the measurement error of a
coriolis mass-flow meter, Flow Meas. Instrum. 40 (2014) 39–49, https://doi.org/10.1016/j.flowmeasinst.2014.08.005.
[2] M. Anklin, W. Drahm, A. Rieder, Coriolis mass flowmeters: Overview of the current state of the art and latest research, Flow Meas. Instrum. 17 (2006) 317–323,
https://doi.org/10.1016/j.flowmeasinst.2006.07.004.
[3] A.S. Morris, R. Langari, Measurement and Instrumentation: Theory and Application, 2nd ed., Academic Press, Boston, 2016. doi:10.1016/C2013-0-15387-1.
[4] W.D. Callister, D.G. Rethwisch, Materials Science and Engineering: An Introduction, 9th ed., John Wiley and Sons, 2013.
[5] E. Tal-Gutelmacher, D. Eliezer, Hydrogen-assisted degradation of titanium based alloys, Mater. Trans. 45 (2004) 1594–1600, https://doi.org/10.2320/
matertrans.45.1594.
[6] American Society for Testing and Materials, B861–14: Standard Specification for Titanium and Titanium Alloy Welded Pipe, ASTM International, West
Conshohocken (2014), https://doi.org/10.1520/B0861-14.1.1.14.
[7] ASM International, Welding, Brazing and Soldering, ASM International, Russell Township (1993), https://doi.org/10.31399/asm.hb.v06.9781627081733.
[8] A.E. Morcelli, Study of deformation and fracture micromechanisms of titanium alloy Ti-6Al-4V using electron microscopy and X-Ray diffraction techniques,
University of São Paulo, 2009.
[9] L. Miaoquan, Z. Weifu, Z. Tangkui, H. Hongliang, L. Zhiqiang, Effect of hydrogen on microstructure of Ti-6Al-4V alloys, Rare Met. Mater. Eng. 39 (2010) 1–5,
https://doi.org/10.1016/S1875-5372(10)60071-9.
[10] L. Bolzoni, P.G. Esteban, E.M. Ruiz-Navas, E. Gordo, Mechanical behaviour of pressed and sintered titanium alloys obtained from prealloyed and blended
elemental powders, J. Mech. Behav. Biomed. Mater. 14 (2012) 29–38, https://doi.org/10.1016/j.jmbbm.2012.05.013.
[11] H. Galarraga, R.J. Warren, D.A. Lados, R.R. Dehoff, M.M. Kirka, P. Nandwana, Effects of heat treatments on microstructure and properties of Ti-6Al-4V ELI alloy
fabricated by electron beam melting (EBM), Mater. Sci. Eng. A. 685 (2017) 417–428, https://doi.org/10.1016/j.msea.2017.01.019.
[12] P. Fouilloux, The nature of raney nickel, its adsorbed hydrogen and its catalytic activity for hydrogenation reactions (review), Appl. Catal. 8 (1983) 1–42,
https://doi.org/10.1016/0166-9834(83)80051-7.
[13] R.S. Suppino, Partial hydrogenation of benzene on Ru/Al2O3 and Ru/CeO2 catalysts: effects of the impregnation method and addition of solvents to the reaction
medium in liquid phase, The University of Campinas (2010).

8
I.G.R. Santos, et al. Engineering Failure Analysis 115 (2020) 104618

[14] D.G. Ivey, D.O. Northwood, Storing energy in metal hydrides: a review of the physical metallurgy, J. Mater. Sci. 18 (1983) 321–347, https://doi.org/10.1007/
bf00560621.
[15] D. Eliezer, E. Tal-Gutelmacher, T. Boellinghaus, Hydrogen embrittlement in hydride-and non hydride-forming systems–microstructural/phase changes and
cracking mechanisms, Proc. 11th Int. Congr. Fract., Turin (2005).
[16] E. Tal-Gutelmacher, D. Eliezer, Hydrogen cracking in titanium-based alloys, J. Alloys Compd. 404–406 (2005) 621–625, https://doi.org/10.1016/j.jallcom.
2005.02.098.
[17] M.J. Donachie, Titanium - A Techinical Guide, 2nd ed., ASM International, Russell Township, 2000.
[18] J. Tao, S. Hu, L. Ji, Effect of trace solute hydrogen on the fatigue life of electron beam welded Ti-6Al-4V alloy joints, Mater. Sci. Eng. A. 684 (2017) 542–551,
https://doi.org/10.1016/j.msea.2016.12.102.
[19] Micro Motion Inc, The Nature of Hydrogen Gas and the Benefits of Coriolis Fluid Measurement, Technology, Boulder, 2003.