Impacts of accelerated aging on the mechanical propertiesof Cu–Nb nanolaminates

D. R. Economy • B. M. Schultz • M. S. Kennedy

Received: 27 April 2012 / Accepted: 7 June 2012 / Published online: 3 July 2012

� Springer Science+Business Media, LLC 2012

Abstract Accelerated aging (30 min at 400 �C) has been

shown to alter the mechanical properties of Cu–Nb nanola-

minate systems. The Cu–Nb nanolaminates produced were

1,000-nm thick with alternating 20 or 100-nm-thick individual

layers, which were fabricated by magnetron sputter deposi-

tion. Unaged Cu–Nb systems increased in hardness (from 4.3

to 5.5 GPa) with decreasing layer thickness. After aging, the

nanolaminates with 20 nm layers softened greatly (5.5 GPa

decreased to as little as 1.3 GPa), yet nanolaminates with

100 nm layers hardened slightly (4.3–4.8 GPa). Both nan-

olaminate structures exhibited significant residual tensile

stress, which was further increased by up to 70 % (100 nm

layers) and 120 % (20 nm layers) after accelerated aging.

X-ray diffraction showed the presence of primary textures and

high stress in niobium layers for unaged systems.

Abbreviations

MEMS Microelectromechanical systems

DC Direct current

RF Radio frequency

Introduction

Nanolaminates are composite thin film systems, which for

the purposes of this article, are alternating layers of two

differing metals (ranging from 1 to 100 nm). They have been

previously studied for possible inclusion in microelectro-

mechanical systems [1, 2], freestanding high-strength foils

[3, 4], and wear-resistant coatings [5]. Though these appli-

cations make use of the structures’ high strength and resis-

tance to dislocation flow, nanolaminates composed of two

‘‘soft’’ metals can have a yield strength higher than estimates

based on bulk yield strength values [3]. This type of film

system also exhibits mechanical strengthening as the indi-

vidual layer thickness decreases [3, 6–9].

While the mechanical properties of as-deposited nanola-

minate systems have been examined at length, only a limited

number of studies on microstructural evolution of nanola-

minate systems during aging have been conducted [4, 10–

13]. Determining the aging behavior of nanolaminate sys-

tems is important, due to the predicted temperature ranges

and environmental variation that the nanolaminates experi-

ence during use as wear-resistant coatings and high-strength

foils. A few studies exist on the mechanical response of aged

nanolaminate systems [4, 10, 14, 15] and additional research

is needed. These reports in the literature have identified two

distinct mechanisms that degrade the mechanical strength:

oxidation [10] and the stability of interfaces [4]. Misra and

Hoagland [4] suggested that the decrease in hardness fol-

lowing aging occurs more readily in systems with reduced

layer thicknesses. Softening was hypothesized as more

favorable in thinner layered systems due to an increase in

layer breakdown during three-point junction reorientation

[4, 11]. These two primary studies suggest that different

mechanisms cause loss in mechanical durability during the

aging of metallic nanolaminates. Although both show that

environment is a factor in the eventual structure of the nan-

olaminate, more work is needed to elucidate the evolution of

mechanical properties of metallic nanolaminates when

subjected to high temperature environments [4, 10].

D. R. Economy � B. M. Schultz

School of Materials Science and Engineering, Clemson

University, Clemson, SC 29634, USA

M. S. Kennedy (&)

Center for Optical Materials Science and Engineering

Technologies (COMSET), School of Materials Science and

Engineering, Clemson University, Clemson, SC 29634, USA

e-mail: mskenne@clemson.edu

123

J Mater Sci (2012) 47:6986–6991

DOI 10.1007/s10853-012-6649-y

Experimental details

Nanolaminates of Cu–Nb were deposited for this study

using magnetron sputter deposition; specifically 100 W

direct current (DC) for the Cu films and 200 W radio

frequency for the Nb films. Two sets of nanolaminates

were deposited, one with individual layer thicknesses of

20 nm and the other with 100 nm layers. The total

thickness of both nanolaminate systems was 1,000 nm.

Prior to deposition, the vacuum chamber was evacuated to

a base pressure of *7 9 10-7 Torr. Deposition was

conducted at 10 mTorr Ar, which yielded a deposition rate

of 4.9 and 2.9 nm/min for the Cu and Nb, respectively.

Following deposition, each sample was subjected to heat

treatment to simulate accelerated aging effects. Based

upon the different results in atmospheres of both ambient

air [10] and ultra high vacuum [4], three additional heat

treatment atmospheres were chosen: a high purity argon

(99.9995 % Ar, \0.5 PPM O2), an argon/oxygen blend

(97.9 % Ar, 2.1 % O2), and ambient air (assumed &78 %

N2, &21 % O2, &1 % Ar). The argon/oxygen blend was

chosen as an order-of-magnitude reduction in oxygen

concentration relative to ambient air. The heating was

conducted in a glove box on a temperature-controlled hot

plate. Each sample was heated to 400 �C for 30 min, at

which point they were removed from the hot plate and

allowed to cool.

The structure was observed using cross-sectional SEM,

which was performed with a Hitachi SU-6600 variable

pressure field-emission-gun SEM. Cross-section views

were obtained by cleaving the silicon substrate immedi-

ately prior to insertion into the microscope and viewing the

fracture surface at 10 kV under high vacuum. The grain

size of the top niobium layer was calculated by the line

intercept method using one-micrometer square images

taken by a Digital Instruments Nanoscope IIIa AFM in

contact mode. Five AFM images were collected for each

sample with seven lines used for calculating the grain size

of each image, which were then used to calculate the

average and standard deviation of grain size. Residual

stress was estimated by measuring changes in substrate

curvature and using Stoney’s equation [16]. Curvature

measurements were performed on a Veeco Dektak3 contact

profilometer, and XRD was performed on a Bruker D8

Discover system with Cu Ka X-rays. The hardness of both

the unaged and accelerated-aged nanolaminate films was

measured using a Hysitron Triboscope nanoindentation

system with a diamond Berkovich tip. The tip was cali-

brated using fused quartz prior to nanoindentation testing

of nanolaminates. For each sample, 22 indentations were

performed, two at each of 11 maximum depths. Indentation

spacing was a minimum of 10 lm between each test

location to ensure a negligible influence of pre-existing

deformation zones and was shallower than 100 nm to

minimize effects of the underlying substrate.

Results and discussion

Initial SEM images of the unaged nanolaminate films

showed the desired laminar geometry (Fig. 1 a, c). Fol-

lowing accelerated aging, samples were freshly cleaved

and observed under SEM, where a significant loss of

overall laminar geometry was observed for the 20 nm Cu–

Nb system (Fig. 1b). No effect was evident in the 100 nm

system, however (Fig. 1d).

Though cross-section views of the samples aged in high

purity argon are shown in Fig. 1, similar results were

observed for all aging atmospheres. The hardness of

as-deposited Cu–Nb nanolaminates was measured at

5.5 ± 0.3 GPa for the 20 nm layer thickness system and

4.3 ± 0.3 GPa for the 100 nm system, comparable to the

hardnesses observed elsewhere for similar Cu–Nb nanola-

minates [4, 7]. The heat treatment of the 20 nm Cu–Nb

systems caused significant losses in hardness regardless of

the heat treatment atmosphere (Fig. 2). This decrease did

not occur in the 100 nm Cu–Nb systems, however, which

remain consistent in hardness regardless of atmosphere.

The softening in thinner layered systems has been

observed previously [4, 10], which is consistent with the

changes in hardness, as shown in Fig. 2. However, Bellou

et al. [10] determined that the aging of the Pt–Mo systems

caused significant oxidation, which in turn drastically

decreased the hardness. Significant oxidation was not

observed in this study (to the extent of that observed in

Bellou et al.), which can be seen in the cross-sectional

SEM images (Fig. 1). This indicates that oxidation is not

the only major factor to consider in determining the

mechanical stability of nanolaminate systems. As previ-

ously discussed, Misra and Hoagland [4] did not observe

changes in hardness until higher temperatures and longer

aging times for similar Cu–Nb systems. As seen in Fig. 3,

the softening was observed in this study at significantly

lower temperatures than seen elsewhere in the literature

[4]. It is hypothesized that the cause of the change in

temperature needed for softening is due to the stresses

imposed by the silicon substrate, which was absent in the

other study [4].

Overall, the different atmospheric oxygen concentra-

tions during heating did not cause a distinct effect on

hardness (Fig. 2). The 20 nm sample heated in the Ar/O

blend shows the highest hardness of the aged 20 nm sys-

tems. The hardness of the Ar/O 20 nm sample still shows a

significant decrease when compared with the unaged sys-

tem, however. As reducing oxygen concentration to below

1 ppm cannot control the major factors affecting

J Mater Sci (2012) 47:6986–6991 6987

123

Fig. 1 Cross-sections of Cu–Nb nanolaminates: 20 nm layer thick-

ness samples before (a) and after aging in Ar (b). Also, 100 nm layer

thickness systems are shown before (c) and after aging in Ar (d).

These systems were composed of Cu as the base layer and Nb as the

capping layer. Images include silicon substrate

Fig. 2 The observed changes in hardness upon accelerated aging as

measured by nanoindentation

Fig. 3 Observed changes in hardness of this study compared with

those found in the literature [4]. The previous study peeled the

nanolaminate films from their substrates prior to aging [4], which was

not done in this study

6988 J Mater Sci (2012) 47:6986–6991

123

accelerated aging in nanolaminate Cu–Nb systems, further

study on the effect of oxygen in these systems is desired.

As an accurate measure of film hardness requires con-

sidering and avoiding the potential substrate effects

[17, 18], the indents should not penetrate deeper than 10 %

of the film’s total thickness [18, 19]. This is due to the

volume of material undergoing plastic deformation below

the nanoindentation probe [19]. The size of the plastic

deformation zone can be estimated by a relation consid-

ering the contact depth of the indentation probe [20]. For

fully plastic indents, the ratio of the deformation zone size

to the indentation size (c/a) is equal to 2.5; for indentations

undergoing elastic deformation as well the ratio increases

(c/a & 4) [20]. For the Cu–Nb systems, a c/a & 4 was

assumed and indentations were kept at a 75 nm contact.

This resulted in an approximate 300 nm deep deformation

zone beneath the indenter tip, which is consistent for both

the as-fabricated and aged systems. This assumption is

valid due to no observable change in thickness upon

observing the films following aging (Fig. 1).

The grain size of the top Nb layers was similar for the

unaged 20 and 100 nm nanolaminates (Fig. 4). For the

unaged films, the 20 nm system had a grain size of

51 ± 8 nm, compared with 42 ± 5 nm for the 100 nm

sample. After aging in pure argon, the grain size of the top Nb

layer was calculated to be 119 ± 16 nm, significantly

greater than the 100 nm sample for the same aging condi-

tions (50 ± 3 nm). When the nanolaminate films were aged

under the other conditions, similar results were observed. It

has been shown in multiple studies that thicker nanolaminate

systems exhibit greater microstructural stability [4, 11],

which is consistent with grain sizes observed in the top Nb

layer for 100 nm systems following aging. However, it was

noted that the grain size of the 20 nm samples behaved non-

uniformly for different heating atmospheres (Fig. 4). Pin-

ning of grains by oxygen atoms has been observed for Nb

films [21] as well as other film systems [22]. In previous

work, the pinning force of impurities, or resistance to grain

growth, in Nb films was shown to follow a negative parabolic

trend with respect to oxygen impurity content on a loga-

rithmic basis [21]. It is thus supposed that for the three

conditions shown: Ar, Ar/O, and ambient, the sample heated

under Ar/O had the highest pinning force followed by

ambient and Ar. This ultimately led to the least grain growth

in the Ar/O atmosphere and greatest grain growth under Ar

atmosphere due to impurities from the heating gas.

The residual stress estimations based on changes in

substrate curvature were calculated using Stoney’s equa-

tion with changes in curvature quantified using a Veeco

Dektak3 diamond stylus profilometer. Initial stress values

are consistent (on the order of 200 MPa) with monolithic

Cu and Nb film stresses presented elsewhere [23, 24].

Calculated stresses are consistent with the changes in

hardness and microstructure (Table 1). After fabrication,

all Cu–Nb systems contained significant tensile residual

stresses. After aging, all nanocomposites were under

greater tensile stress than the as-deposited state, with the

20 nm systems experiencing greater gains relative to the

100 nm systems. The overall trend in stress development

following a thermal cycle is consistent with Vinci et al.

[24]. This estimation of stress based on substrate curvature

is inadequate for analyzing the stress state during heating,

however. Residual stress is a possible significant factor on

mechanical behavior and diffusivity during heat treatment.

This is significant due to the comparisons made by Misra

and Hoagland [4] who utilized stress-free deposition

parameters for film fabrication. Following the results

shown in Vinci et al. [24], though a hysteresis loop for the

residual stress should be expected upon thermal cycling for

the film systems, the effects of initial stress cannot be

predicted.

Fig. 4 Observed changes in grain size as calculated by the lineintercept method using AFM images of the upper Nb film surface.

Grain growth and possible pinning were observed in the 20 nm

systems

Table 1 Observed changes in residual stress estimations calculated

from change in substrate curvature

Aging condition Cu–Nb 20 nm Cu–Nb 100 nm

Unaged 321 ± 2 MPa 170 ± 3 MPa

Ar 599 ± 2 MPa 202 ± 1 MPa

Ar/O 707 ± 2 MPa 235 ± 1 MPa

Ambient 646 ± 2 MPa 291 ± 16 MPa

Positive values denote tensile stress. Upon accelerated aging treat-

ment, all films are left with higher tensile stress than prior to aging.

These calculations represent an average stress over a long (20 mm)

distance

J Mater Sci (2012) 47:6986–6991 6989

123

X-ray diffraction of the unaged Cu–Nb systems shows

clear peaks for both Cu and Nb (Fig. 5). Peak broadening

and shifting to lower 2h values for the Nb peaks indicate

that the niobium layers are under significant tensile stres-

ses, which is consistent with the residual stress estimated

by wafer curvature. This is most evident in the Nb (110)

peak observed in the 100 nm sample, which has shifted

&1� due to tensile stress (Fig. 5). The XRD patterns also

show grain orientations other than the primary crystalline

planes, indicating that the as-fabricated films produced in

this study are not highly textured.

Summary

Nanolaminate films exhibited hardnesses consistent with

film systems produced elsewhere [4, 7]. The stability of

nanolaminate systems is limited due to the increased dif-

fusion in the high degree of interfacial area (grain bound-

aries and layer interfaces). As other studies have

emphasized, the increased hardness in laminate geometries

is due to the constrictions on dislocation motion imposed

by interfaces and grain boundaries [7, 25]. These features

contribute to systems that do not retain their properties

even after short exposures to elevated temperatures, which

were shown through decreases in the hardness of 20 nm

individual layer thickness Cu–Nb samples measured by

nanoindentation. Consequently, applying metallic nanola-

minates in elevated temperature environments may require

a compromise; decreased layer thickness yields greater

hardness initially, but the increased hardness is at the

expense of decreased stability. Through this work, the

effect of oxygen concentration in heat treatment atmo-

sphere on Cu–Nb nanolaminates is not immediately clear,

but could be due to the pinning force of oxygen impurities

[21]. However, the observed softening occurred at lower

temperatures and shorter times than previously reported in

Cu–Nb systems [4]. It is hypothesized that this softening is

caused by the constraint imposed on the systems by the

silicon substrate. As previously noted, in Misra and Hoa-

gland [4], films were peeled from their substrates prior to

heat treatment, allowing relaxation in the films. Research

has shown that the constraint imposed by the substrate in

copper thin films causes significant changes on the residual

stress due to thermal cycling [24]. Further examination of

the effects of residual stress on thermal stability of nan-

olaminate film systems is desired.

Acknowledgements The authors would like to thank Prof. J. Hud-

son (Clemson University Electron Microscope Facility), Dr. D. Van-

Derveer (Clemson University Molecular Structure Center), Prof.

M. Dayananda (Purdue University), and Prof. F. Akasheh (Tuskeegee

University) for helpful discussions. In addition, the authors thank Mr.

Godfrey Kimball at Clemson University for editorial assistance.

References

1. Chih-Tang P, Kuo-Ning C (2002) In: ITherm 2002. 8th Interso-

ciety conference on thermal and thermomechanical phenomena in

electronic systems (Cat. No. 02CH37258), 30 May–1 June

2002. IEEE, Piscataway, p 1058. doi:10.1109/itherm.2002.10

12575

2. Pustan M, Rymuza Z, Schneider A, Serra SG, Huq SE (2007) In:

Microprocess and nanotech 2007. 20th International micropro-

cess and nanotechnology conference, 5–8 Nov 2007, Piscataway,

p 360

3. Misra A (2006) In: Hannick RHJ, Hill AJ (eds) Nanostructure

control of materials. Woodhead Publishing, Cambridge, p 146

4. Misra A, Hoagland RG (2005) J Mater Res 20 (Copyright 2006,

IEE):2046

5. Farhat ZN, Ding Y, Northwood DO, Alpas AT (1997) Surf Coat

Technol 89(1–2):24. doi:10.1016/s0257-8972(96)02939-8

6. Misra A, Verdier M, Kung H, Embury JD, Hirth JP (1999)

Scripta Mater 41(9):973

7. Misra A, Hirth JP, Hoagland RG (2005) Acta Mater 53(18):4817

8. Kramer DE, Savage MF, Lin A, Foecke T (2004) Scripta Mater

50(6):745

9. Was GS, Foecke T (1996) Thin Solid Films 286(1–2):1

10. Bellou A, Scudiero L, Bahr DF (2010) J Mater Sci 45 (Com-

pendex):354. doi:10.1007/s10853-009-3943-4

11. Josell D, Carter WC, Bonevich JE (1999) Nanostruct Mater

12(1–4):387

12. Josell D, Spaepen F (1999) Mater Res Soc Bull 24 (Copyright

1999, IEE):39

13. Lewis AC, Josell D, Weihs TP (2003) Scripta Mater 48

(Compendex):1079

14. Mara N, Sergueeva A, Misra A, Mukherjee AK (2004) Scripta

Mater 50(6):803

15. Mara NA, Misra A, Hoagland RG, Sergueeva AV, Tamayo T,

Dickerson P, Mukherjee AK (2008) Mater Sci Eng A

493(1–2):274. doi:10.1016/j.msea.2007.08.089

16. Stoney GG (1909) Proc R Soc Lond Ser A 82(553):172

17. Han SM, Saha R, Nix WD (2006) Acta Mater 54(6):1571

18. Saha R, Nix WD (2002) Acta Mater 50(1):23

Fig. 5 XRD patterns of unaged 20 and 100 nm Cu–Nb nanolaminate

films showing both Cu and Nb peaks, Nb peak shift denotes tensile

strain

6990 J Mater Sci (2012) 47:6986–6991

123

19. Fischer-Cripps AC (2002) Nanoindentation/Anthony C. Fischer-

Cripps. Mechanical engineering series (Berlin, Germany). Springer,

New York

20. Woodcock CL, Bahr DF, Moody NR (2001) In: Fundamentals of

nanoindentation and nanotribology II, Materials Research Society

Symposium Proceedings, 28–30 Nov 2000, Boston, p Q7.14.11

21. Pan VM, Prokhorov VG, Kaminsky GG, Tretiatchenko KG

(1984) In: Proceedings of the 17th international conference on

low temperature physics, LT-17, 15–22 Aug 1984, North-Hol-

land, Amsterdam, p 565

22. Tang F, Gianola DS, Moody MP, Hemker KJ, Cairney JM (2012)

Acta Mater 60(3):1038. doi:10.1016/j.actamat.2011.10.061

23. Booi PAA, Livingston CA, Benz SP (1993) IEEE Trans Appl

Supercond 3(2):3029. doi:10.1109/77.257236

24. Vinci RP, Zielinski EM, Bravman JC (1995) Thin Solid Films

262 (Copyright 1995, FIZ Karlsruhe):142

25. Misra A, Hirth JP, Kung H, Hoagland RG, Embury JD (2001)

Dislocation models for strengthening in nanostructured metallic

multilayers. In: Structure and mechanical properties of nano-

phasematerials—Theory and computer simulation vs. experiment

symposium, November 28–30, 2000, Warrendale, PA. Materials

Research Society Symposium Proceedings, vol 634, p 4-2

J Mater Sci (2012) 47:6986–6991 6991

123