Available online at www.sciencedirect.com

Scripta Materialia 62 (2010) 540–543

www.elsevier.com/locate/scriptamat

Corrosion and corrosion fatigue of Vitreloy glasses containing lowfractions of late transition metals

Aaron Wiest,a,* Gongyao Wang,b Lu Huang,b Scott Roberts,a Marios D. Demetriou,a

Peter K. Liawb and William L. Johnsona

aKeck Laboratory, California Institute of Technology, Pasadena, CA 91125, USAbDepartment of Materials Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USA

Received 24 September 2009; revised 10 December 2009; accepted 11 December 2009Available online 16 December 2009

Corrosion resistance and fatigue performance of Vitreloy glasses with low fractions of late transition metals (LTMs) in 0.6 MNaCl are investigated and compared to a traditional Vitreloy glass and other crystalline alloys. Owing to their ability to form uni-form passive films, low LTM Vitreloy glasses exhibit corrosion rates that are an order of magnitude lower than those of other alloysconsidered. The fatigue ratios in solution are substantially lower than crystalline alloys considered and similar to traditional Vitreloyglasses, which fail at low stresses due to limited repassivation in solution.� 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Bulk amorphous materials; Metallic glasses; Fatigue test; Corrosion; Titanium alloys

Owing to the lack of long-range order in theiratomic structure and the absence of microscopic defectssuch as vacancies, dislocation or grain boundaries thattypically arise in crystalline microstructures, bulk metal-lic glasses (BMGs) demonstrate a unique combinationof mechanical properties, such as high strength, highhardness and a high elastic strain limit [1–4]. Theabsence of microstructural defects is also thought topromote the formation of uniform and chemicallyhomogeneous passive layers without weak points thatcan provide enhanced protection against corrosion orchemical attack. Some metal-metalloid BMG alloyscontaining Cr indeed demonstrate superb corrosionresistance [5], though the resistance to corrosion is notuniversally high for all BMG alloys. Zr-based BMGsof the Vitreloy alloy family were found to exhibit corro-sion resistance in saline solutions that was higher thanfound for many crystalline engineering metals, thoughon a par with the most advanced corrosion-resistantmetals. For example, the corrosion resistance of ZrTi-NiCuAl glass (Vitreloy 105) in phosphate-buffered salinewas found to be on a par or slightly lower than widelyused metallic biomaterials such as stainless steels, Ti–6Al–4V and CoCrMo [6]. The corrosion resistance of

1359-6462/$ - see front matter � 2009 Acta Materialia Inc. Published by Eldoi:10.1016/j.scriptamat.2009.12.025

* Corresponding author. Tel.: +1 626 395 8454; e-mail: aaronwiest@alumni.caltech.edu

Zr41.2Ti13.8Cu12.5Ni10Be22.5 (Vit1) in NaCl solutions isthought to arise due to the formation of a protectiveZr- and Ti-based oxide film. Schroeder et al. studiedthe corrosion resistance of crystalline and amorphousZrTiNiCuBe in chloride solutions and found that theamorphous structure is only slightly more resistant topitting corrosion than the corresponding crystallinestructure [7].

Despite the generally good corrosion behavior ofVitreloy type BMGs, their behavior in cyclic stress(fatigue) corrosion environments is rather poor.Corrosion fatigue testing couples two mechanisms thatlead to ultimate fracture of the material, namely crackinitiation and crack growth. Menzel and Dauskardtobserved that crack initiation took only a few cycles ofloading in the case of the Vit1 [8]. The stresses at whichVitreloy type BMGs endure 107 cycles in saline solutionswas found to be just 10–20% of their corresponding valuesin air due to significantly higher (2–3 orders of magnitude)fatigue crack growth rates realized in saline compared toair [9,10]. A stress corrosion mechanism is proposedfor the observed behavior. The high crack extensionvelocities were attributed to incomplete repair of theprotective oxide at the crack tip, a consequence of thelow pitting potentials of these alloys [11].

In this paper, we investigate the corrosion and corro-sion fatigue behavior of certain Vitreloy alloy variants ina 0.6 M saline environment (simulated sea water) and

sevier Ltd. All rights reserved.

Current Density [A/cm2]

Cyclic Anodic Polarization

Vol

tage

[V]

10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100

0

0.1

0.2

0.3

0.4

-0.1

-0.2

-0.3

-0.4

-0.5

Zr35Ti30Be35

Zr35Ti30 Be35Co6

Zr52.5Cu17.9Ni14.6 Al10Ti5

Figure 1. Cyclic anodic polarization curves of Zr52.5Cu17.9Ni14.6Al10Ti5,Zr35Ti30Be29Co6 and Zr35Ti30Be35 in 0.6 M NaCl solution at a scanrate of 0.167 mV s�1.

A. Wiest et al. / Scripta Materialia 62 (2010) 540–543 541

contrast it to traditional Vitreloy alloys and other metalsused in marine applications. We demonstrate that smallvariations in the Vitreloy alloy composition can lead tosignificant improvements in corrosion resistance. Theimprovement in corrosion resistance, however, is notaccompanied by an analogous improvement in corro-sion fatigue endurance, suggesting that the protectivemechanism that gives rise to the enhanced corrosionresistance of these glasses does not adequately protectagainst stress corrosion cracking (SCC).

A series of Vitreloy type BMG compositions with lowatomic fractions of late transition metals (LTMs) haverecently been reported [12]. Many of these alloys werefound to exhibit a combination of exceptionally largesupercooled liquid region and good glass-forming abil-ity. Notable examples include the ternary Zr35Ti30Be35

and quaternary Zr35Ti30Be29Co6, with supercooled li-quid regions of 120 and 150 �C and critical casting thick-nesses of 6 and 15 mm, respectively. Owing to the lowLTM atomic fractions, these compositions were thoughtto also exhibit good corrosion characteristics. The cor-rosion and corrosion fatigue behavior of Zr35Ti30Be35

and Zr35Ti30Be29Co6 is investigated here, and is con-trasted to Zr52.5Cu17.9Ni14.6Al10Ti5, a traditional Vitre-loy alloy (Vit 105) with a much higher LTM atomicfraction. The measured values are also contrasted tothree traditional crystalline alloys used in sea water envi-ronments: 18/8 stainless steel, Monel (Cu–Ni based) andAlclad (Al based) [13].

Alloys of Zr35Ti30Be35 and Zr35Ti30Be29Co6 were pre-pared by arc melting elements of >99.9% purity on awater-cooled Cu plate in Ti-gettered argon atmosphere.Three millimeter diameter amorphous rods of Zr35Ti30-

Be29Co6 and 2 mm diameter amorphous rods ofZr35Ti30Be35 were cast using an Edmund Buhler miniarc melter suction casting setup. The amorphous natureof the rods was verified using X-ray diffraction and dif-ferential scanning calorimetry (DSC).

Cyclic anodic polarization experiments were con-ducted on unloaded samples of Zr35Ti30Be35 andZr35Ti30Be29Co6 in 0.6 M NaCl solution at a scan rateof 0.167 mV s�1. Data for Vit 105 were gathered fromthe study of Morrison et al. [9]. Epit, the pitting poten-tial, and Ecorr, the steady-state corrosion potential, weremeasured multiple times for each sample. In betweenmeasurements, samples were polished with 1200 gritsandpaper in order to remove the reaction layer. Corro-sion rates were calculated from corrosion current den-sity measurements. A more detailed description of theexperimental method can be found in [6,9]. Theaveraged cyclic anodic polarization curves for the threealloys are presented in Figure 1. Average Epit and Ecorr

values for each alloy along with corrosion rates arefound in Table 1.

A Perkin Elmer Phi X-ray Photoelectron Spectrome-ter (XPS) was used to investigate the surface chemistryof low LTM compositions with Al Ka X-rays of1486 eV. The instrument was used in “survey scanmode,” with a step size of 1 eV, a dwell time of200 ms step�1, scanning from 160 to 1000 eV four times.

Fatigue and corrosion fatigue measurements wereconducted on 3 mm diameter � 6 mm tall rods ofZr35Ti30Be29Co6 and 2 mm diameter � 4 mm tall rods

of Zr35Ti30Be35 in a compression–compression loadinggeometry at 10 Hz using a stress ratio R = rminrmax = 0.1.For Vit 105, four-point bending fatigue and corrosionfatigue data from the study of Morrison et al. [9] wereutilized. Even though compression–compression andfour-point bending fatigue experiments often result insomewhat different endurance limits, the relativedrop in the endurance limits between air and salineenvironments, which is of interest here, is not expectedto be dramatically different in the two tests. The S/Ncurves for the three alloys in air and saline environmentsare presented in Figure 2. The ratio of fatigue limit toyield stress in air and in saline solution for the threeBMGs are listed in Table 1. Data for 18/8 stainless steel,Alclad 24S-T and Monel taken from Atlas of FatigueCurves and other sources [15–19] are also displayed inTable 1.

Analysis of the cyclic anodic polarization data, pre-sented in Figure 1, reveals that the pitting potential ofVit 105 is the greatest, followed by Zr35Ti30Be29Co6 andZr35Ti30Be35. This suggests that the Zr35Ti30Be35 is themost susceptible and Zr52.5Cu17.9Ni14.6Al10Ti5 the leastsusceptible to attack by pitting in 0.6 M NaCl. However,the corrosion current densities of the alloys with lowLTM fractions are approximately two orders of magni-tude lower than those of Zr52.5Cu17.9Ni14.6Al10Ti5 in therange between Ecorr and Epit. A higher corrosion currentdensity naturally leads to a higher corrosion rate. The cor-rosion rate of Vit 105 in 0.6 M NaCl is reported to be29 ± 6 lm year�1 [9]. By contrast, the corrosion rates ofZr35Ti30Be29Co6 and Zr35Ti30Be35 in the same solutionand under the same conditions were measured in thisstudy to be 0.5 ± 0.2 and 0.9 ± 0.3 lm year�1, respec-tively. These rates are lower than that of Zr52.5Cu17.9-

Ni14.6Al10Ti5 by factors of �150 and �30, respectively.Corrosion rates for 18/8 stainless steel, Alclad 24S-Tand annealed Monel in sea water are reported to be 13,11.2 and 15.9 lm year�1, respectively [13]. These ratesare slightly lower than that of Vit 105, but more than anorder of magnitude higher than the rates of Zr35Ti30Be35

and Zr35Ti30Be29Co6.In Table 1, fatigue ratios in air and saline are pre-

sented. Fatigue ratio is defined as the stress amplitude,ra, at which the material can survive 107 cycles normal-

Table 1. Data for corrosion and corrosion fatigue in 0.6 M NaCl. Fatigue values are the stress amplitudes at which the samples endured 107 loadingcycles normalized by the material yield strength. The yield strengths [14] of Zr35Ti30Be35, Zr35Ti30Be29Co6 and Zr52.5Cu17.9Ni14.6Al10Ti5 are 1850,1800 and 1700 MPa [9], respectively. Corrosion data for 18/8 stainless steel, annealed Monel, and Alclad 24S-T are taken from [13], while data forfatigue in air and 0.6 M NaCl are taken from [15–19].

Ecorr [mV] Epit [mV] Corrosion rate [lm year�1] Fatigue (air) [% strength] Fatigue (0.6 M NaCl) [% strength]

Zr35Ti30Be35 �445 ± 42 84.5 ± 23.5 0.871 ± 0.266 27 8Zr35Ti30Be29Co6 �424 ± 8 257 ± 64 0.544 ± 0.215 28 6Zr52.5Cu17.9Ni14.6Al10Ti5 �264 324 29 ± 6 25 618/8 stainless steel 13 25 15Monel annealed 15.2 40 35Alclad 24S-T 11.2 20 10

0

100

200

300

400

500

600

700

800

Cycles to Failure (NF)104 105 106 107

Str

ess

Am

plitu

de (M

Pa)

0.6M NaCl

Air

Figure 2. Fatigue performance in air (N Zr52.5Cu17.9Ni14.6Al10Ti5,Zr35Ti30Be29Co6, Zr35Ti30Be35) and in 0.6 M NaCl (D Zr52.5Cu17.9

Ni14.6Al10Ti5, Zr35Ti30Be29Co6, Zr35Ti30Be35) at a frequency of10 Hz and R = 0.1. Zr52.5Cu17.9Ni14.6Al10Ti5 is tested in four-pointbending geometry (data taken from [9]). Zr35Ti30Be29Co6 andZr35Ti30Be35 are tested in compression–compression geometry (presentstudy).

103

104

105

160360560760Binding Energy (eV)

Coun

ts (A

rb. R

ef.) Zr

O2 3

dO 1

s

C 1

sZr

O2 3

pC

a 2p

TiO

2 2p

Ti 2

pZr

3s

Cu

/ CuO

2p

Zr35Ti30Be35

Zr35Ti30 Cu7.5Be27.5

Figure 3. XPS survey scans of Zr35Ti30Be35 and Zr35Ti30Cu7.5Be27.5

showing Zr and Ti oxide peaks. The Ca and C peaks evident on thescans are due to contamination. The peak at approximately 933 eV isunique to the quaternary alloy and indicates the presence of Cu orCuO in the surface layer of the alloy.

542 A. Wiest et al. / Scripta Materialia 62 (2010) 540–543

ized by the yield strength [14]; ra = (rmax � rmin)/2. Inair, the ratios of the glassy alloys range between 0.25and 0.28. In saline, however, the fatigue ratios of theglasses drop to 0.06–0.08, despite the vast differencesin corrosion resistance. Alclad also experiences a largedrop in fatigue ratio between air and saline, going from0.2 to 0.1. Steel exhibits a fatigue ratio of 0.25 in air,which is similar to metallic glasses, but 0.15 in saline,which is twice that of metallic glasses. Annealed Monelexhibits the highest fatigue ratios: 0.4 in air and 0.35 insaline.

Owing to the dramatically improved corrosion resis-tance demonstrated by Zr35Ti30Be29Co6 and Zr35Ti30Be35

over the traditional Vit 105, one might expect to observean analogous improvement in corrosion fatigue endur-ance as well. As seen in Table 1, however, no statisticallysignificant improvement in corrosion fatigue endurance isrealized. To understand this behavior, one must investi-gate the key mechanism controlling the corrosion resis-tance of these glasses, i.e. the passive oxide film [20–23].

The passive layers of Zr35Ti30Be35�xLTMx wereinvestigated using XPS. The XPS scans revealed thatthe passive layer is comprised of oxides of the constitu-ent elements in concentrations similar to the bulk sam-ple composition. As examples, XPS survey scans of aCu-containing Vitreloy-type alloy, Zr35Ti30Cu7.5Be27.5,

and the LTM-free ternary composition, Zr35Ti30Be35

are presented in Figure 3. The scan of the Cu-containingalloy reveals a Cu (or possibly Cu oxide) peak alongwith the Zr and Ti peaks, while the scan of the Cu-freealloy reveals only Zr and Ti oxide peaks. Given the vul-nerability of Cu (or Cu oxide) to chloride attack, thesurface Cu atoms can be thought as “weak spots” thatdegrade the protective capabilities of the ZrTi oxidefilm. It is likely that a higher fraction of LTM in a Vit-reloy-type alloy composition will result in a concentra-tion of “weak spots” in the passive film and thus lowerthe corrosion resistance. The study of the oxide filmtherefore provides insight into the origin of the en-hanced corrosion resistance demonstrated by the lowLTM glasses.

Under stress (or cyclic stress) corrosion conditions,however, a highly protective passive layer that promotesgood corrosion resistance would not necessarily pro-mote good resistance to SCC. For an alloy to resistSCC and exhibit good corrosion fatigue performance,an ability to repassivate spontaneously in solution isalso necessary. Schroeder et al. [11] attributed the poorcorrosion fatigue performance of ZrTiCuNiBe primarilyto the electrochemistry of this alloy, which yields passivelayers that lack the ability to reform quickly enough insolution to blunt the extending crack tip. Therefore eventhough reduction of the LTM fractions in Vitreloy com-positions improves passive film uniformity and enhancescorrosion resistance, it appears not to promote SCC

A. Wiest et al. / Scripta Materialia 62 (2010) 540–543 543

resistance. Other metallic glass families with differentelectrochemical character are known to perform wellin corrosion and also resist SCC. One example is thefamily of Cr-bearing Fe metalloid glasses (e.g. FeCrPC).These alloys are found to resist pitting at high anodicpolarization (�1 V), and exhibit corrosion rates in acid-ified chloride solutions that are many orders of magni-tude lower than conventional (crystalline) stainlesssteel alloys [5,24–27]. Their high corrosion resistance isattributed to their ability to passivate spontaneouslyupon immersion, and specifically to the inclusion of hy-drated chromium oxyhydroxide in the oxide layer [5]. Inexperiments where the passive layer is broken down insolution (scratched off using sand paper), the reforma-tion or repair process was found to be more rapid thanconventional stainless steel alloys [28]. As a consequenceof good repassivation capability, these alloys exhibitvery limited SCC and embrittlement in tensile tests per-formed in neutral saline solutions or in acidic chloridecontaining solutions [27].

In conclusion, the corrosion resistance and corrosionfatigue performance of low LTM Vitreloy glassesZr35Ti30Be29Co6 and Zr35Ti30Be35 in 0.6 M NaCl wereinvestigated and compared to traditional Vitreloy glassZr52.5Cu17.9Ni14.6Al10Ti5 and to other crystalline engi-neering alloys used widely in saline environments suchas 18/8 stainless steel, Alclad 24S-T and annealed Monel.The low-LTM Vitreloy glasses were found to exhibit cor-rosion rates of less than 1 lm year�1 which are lower bymore than one order of magnitude compared to the tradi-tional Vitreloy glass and the conventional engineeringmetals. The high corrosion resistance of the present alloysis attributed to the low fraction or complete absence ofLTM elements, facilitating the formation of a chemicallyhomogeneous passive layer without “weak spots.” De-spite their superb corrosion resistance, the corrosion fati-gue performance of Zr35Ti30Be29Co6 and Zr35Ti30Be35 isfound to be rather poor, as less than 10% of their yieldstrength is retained at 107 cycles, a value comparable totraditional Vitreloy glasses but significantly lower thanconventional crystalline alloys. The poor corrosion per-formance of the present alloys is probably due to a re-tarded reformation of the passive layer at the extendingcrack tip, possibly caused by their low pitting potential.

Valuable discussions with Dr. Vilupanur A.Ravi are gratefully acknowledged. This work is sup-ported by the NSF International Materials InstitutesProgram under DMR-0231320, with Drs. C. Huber,U. Venkateswaran, and D. Finotello as contract moni-tors, and by the Office of Naval Research underONR06-251 0566-22.

[1] A.L. Greer, E. Ma, MRS Bull. 32 (2007) 611.[2] W.L. Johnson, MRS Bull. 24 (1999) 42.

[3] A. Inoue, Acta Mater. 48 (2000) 279.[4] W.L. Johnson, JOM 54 (2002) 40.[5] K. Hashimoto, Corrosion 58 (2002) 715.[6] M.L. Morrison, R.A. Buchanan, R.V. Leon, C.T. Liu,

B.A. Green, P.K. Liaw, J.A. Horton, J. Biomed. Mater.Res. A 74A (2005) 430.

[7] V. Schroeder, C.J. Gilbert, R.O. Ritchie, Scripta Mater.38 (1998) 1481.

[8] B.C. Menzel, R.H. Dauskardt, Acta Mater. 54 (2006) 935.[9] M.L. Morrison, R.A. Buchanan, P.K. Liaw, B.A. Green,

G.Y. Wang, C.T. Liu, J.A. Horton, Mater. Sci. Eng. A467 (2007) 198, Correspondence with the Liaw groupverified that the corrosion rate value is 29 ± 6 lm year�1,not 29 ± 60 lm year�1 as reported.

[10] V. Schroeder, C.J. Gilbert, R.O. Ritchie, Mater. Sci. Eng.A 371 (2001) 145.

[11] V. Schroeder, R.O. Ritchie, Acta Mater. 54 (2006) 1785.[12] A. Wiest, G. Duan, M.D. Demetriou, L.A. Wiest, A.

Peck, G. Kaltenboeck, B. Wiest, W.L. Johnson, ActaMater. 56 (2008) 2625.

[13] C.V. Brouillette, Corrosion Rates in Sea Water at PortHueneme, California, for Sixteen Metals, AD81212Armed Services Technical Information Agency, 1954.

[14] The reported strength values correspond to the materialplastic yield strength rather than the fracture strength,because they represent the stress at which plastic flowbands (shear bands) initiate, as detected by scanningelectron microscopy.

[15] H.E. Boyer, Atlas of Fatigue Curves, American Societyfor Metals, Materials Park, OH, 1986, pp. 37–38, 66, 177–180, 319, 321, 327, 391–392.

[16] H.W. Russell, L.R. Jackson, H.J. Grover, W.W. Beaver,Fatigue Strength and Related Characteristics of AircraftJoints, National Advisory Committee for AeronauticsTechnical Note No. 1485, Battelle Memorial Institute,Washington, 1948.

[17] A.C. Bond, Fatigue Studies of 24S-T and 24S-T AlcladSheet with Various Surface Conditions, Master’s Thesis,Georgia Institute of Technology, 1948.

[18] T.W. Crooker, R.E. Morey, E.A. Lange, Low CycleFatigue Crack Propagation Characteristics of Monel 400and Monel K-500 Alloys, NRL Report 6218, 1965.

[19] Monel Alloy R-405, Technical Brochure, UNS N04405.Available from: <www.specialmetals.com>.

[20] K. Hashimoto, K. Asami, M. Naka, T. Masumoto, Res.Inst. Iron Steel Met. 1694 (1979) 237.

[21] K. Hashimoto, K. Asami, M. Naka, T. Masumoto, Res.Inst. Iron Steel Met. 1695 (1979) 246.

[22] S. Hiromoto, A.P. Tsai, M. Sumita, T. Hanawa, Corros.Sci. 42 (2000) 1651.

[23] S. Hiromoto, A.P. Tsai, M. Sumita, T. Hanawa, Corros.Sci. 42 (2000) 2193.

[24] M.D. Archer, R.J. McKim, J. Mater. Sci. 18 (1983) 1125.[25] J.R. Scully, A. Gebert, J.H. Payer, J. Mater. Res. 22

(2007) 302.[26] R.H. Jones, R. Wang, Corrosion 42 (1986) 504.[27] A. Kawashima, K. Hashimoto, T. Masumoto, Corros.

Sci. 16 (1976) 935.[28] R.B. Diegle, Corrosion 35 (1979) 250.