Effects of substrate rotation on the microstructure of metal sheet fabricatedby electron beam physical vapor deposition

Yue Sun a, Xiu Lin a, Xiaodong He a,b,*, Jiazhen Zhang a, Mingwei Li a, Guangping Song a,Xinyan Li a, Yijie Zhao a

a Center for Composite Materials, Harbin Institute of Technology, Chinab School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore

Applied Surface Science 255 (2009) 5831–5836

A R T I C L E I N F O

Article history:

Received 20 November 2008

Accepted 6 January 2009

Available online 13 January 2009

PACS:

68.55.Jk

81.10.Bk

81.15.Jj

Keywords:

Physical vapor deposition (PVD)

Helical microstructure

Film growth

A B S T R A C T

The effects of substrate rotation speed and rotation mode on the microstructure of large-sized metal

sheet fabricated by electron beam physical vapor deposition technique were investigated. Helical and

columnar microstructures were found in the deposited sheet. Both types of microstructures exhibit no

preferential crystallographic orientation. The column inclination under asymmetric vapor incidence

pattern was discussed. Integrated vapor incidence angle was found to be effective in evaluating the

column inclination.

� 2009 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Applied Surface Science

journa l homepage: www.e lsev ier .com/ locate /apsusc

1. Introduction

Electron beam physical vapor deposition (EBPVD) is wide-spread in fabricating coatings and metallic sheets such as yttriastabilized zirconia thermal barrier coating, SixCy high emissivitycoating, high temperature alloy sheets, and so on. In most of theapplication, the substrate was rotational. However, the effect ofsubstrate rotation on microstructure has not been clarified [1–4].

The microstructure of inclined columns is generally found inEBPVD prepared thermal barrier coatings and obliquely depositedor sputtered thin films. It causes anisotropy and strongly influencescrystallography and properties of the deposits. And the orientationof intercolumnar gaps is closely related with column inclination,which is of great concern for fabricating porous materials orthermal barrier coatings [5,6].

As for deposits with stationary substrate, the tangent rule isespecially effective in predicting the relationship between columninclination angle and vapor incidence angle (VIA) in the VIA rangeof less than 608 [7–12]. However, as for the column inclination indeposits with rotational substrates, especially with asymmetricvapor incidence pattern, little work has been done.

* Corresponding author. Postal address: P.B. 3011, No. 2 Yikuang Street, Nangang

District, Harbin 150080, China. Tel.: +86 451 86402323; fax: +86 451 86402323.

E-mail addresses: hexd@hit.edu.cn, lx.hit@tom.com (X. He).

0169-4332/$ – see front matter � 2009 Elsevier B.V. All rights reserved.

doi:10.1016/j.apsusc.2009.01.013

In this paper, EBPVD technique is employed to fabricate large-sized metal sheet. This paper mainly discusses the effect of substraterotation on the morphology and the column inclination in the sheetunder asymmetrically changing vapor incident pattern.

2. Experimental details

A carbon steel sheet was fabricated by a GEKONT L5 electronbeam facility in a vacuum chamber evacuated to 1 � 10�3 Pa. Thearrangement of the facility was illustrated in Fig. 1. The diameter ofthe stainless steel substrate was 1000 mm and the vertical distancefrom the vapor source to the substrate was 550 mm. A carbon steelvapor source was placed 250 mm far from the central axis of thesubstrate.

Three substrate rotation speeds of 1.6, 10.7, and 25.4 rpm wereapplied in the deposition and each speed lasted about 15 min.The substrate temperature was 635 8C with a 10 8C error range. Theelectron beam current was controlled to be 1.85 A with a 0.05 Aerror range.

Samples were fractured respectively along the tangent direc-tion (TD) and radius direction (RD) at different positions of thesheet in liquid nitrogen. In this paper, the plane determined by RDand the substrate normal is denoted as RP. The plane determinedby TD and the substrate normal is denoted as TP. The cross-sectionswere observed on a HITACHI S-570 Scanning Electron Microscope.

Fig. 1. Arrangement of the substrate and the vapor source.

Y. Sun et al. / Applied Surface Science 255 (2009) 5831–58365832

X-ray diffraction was conducted for both sides of the sample at250 mm from the center of the sheet on a D/max-rB X-raydiffractometer with Cu Ka radiation.

3. Results

3.1. Helical microstructures and columnar microstructures

It is observed from Fig. 2(a) that the cross-section is mainlycomposed of a helical microstructure zone and a columnarmicrostructure zone. The thickness of the helical microstructurezone is about 61.8 mm, one third of the total thickness of the sheet.It is thus taken that the helical microstructures were formed underconditions when the substrate rotation speed was 1.6 rpm, asshown in Fig. 2(b). And columnar microstructures were formedunder conditions when the substrate rotation speed was 10.7 rpmand 25.4 rpm, as shown in Fig. 2(c).

The deposition time was 15 min for the substrate rotationspeed of 1.579 rpm, which means the substrate finished about 24cycles of rotation during low rotation speed deposition. Thethickness of one helix is about 2.4 um, and there are about 25helixes in the low rotation speed zone. Which indicates that onehelix may correspond with one cycle of substrate rotation.

3.2. Inclination of columns

Fig. 3(a) shows cross-sectional fractures along the radius of thesheet at positions 100, 200, and 300 mm far from the center of thesheet. Samples were firstly sliced at the appointed positions, andthen fractured respectively along TD and RD in the liquid nitrogen.Three samples were examined for each distance. It is indicated thatthe position right above the vapor source exhibit inclined columnswhile the position that is the rotation center with inclined incidentvapor exhibit vertical columns. Fig. 3(b) shows cross-sectionalfractures vertical to the radius of the sheet at same positions withFig. 3(a). It is indicated that the columns are parallel to the surfacenormal in the plane defined by the surface normal and the tangentat any position of the sheet.

The inclination of the columns does not change with thesubstrate rotation speed obviously.

3.3. Crystallographic orientations

Fig. 4 shows XRD patterns examined both from the near-substrate side and from the deposit surface of the sample at

250 mm from the center of the sheet. The patterns indicate noobvious preferred orientation for the two sides.

4. Discussion

4.1. Effect of substrate rotation on column inclination and

morphology (growth mode)

Fig. 5 illustrates the deposit growth at three typical positionson the substrate, which are 0, 250, and 425 mm from the center ofthe substrate, denoted as A, B, and C, respectively. We are going tointerpret the column inclination by tracking the displacement ofthe column tips, as inspired by idea of K. Wada in terms of theformation of crescent-shaped grains [13]. Assuming that thecolumns grow in the direction of incident vapor flux and thusshadowing from neighboring grains has a negligible effect on thegrowth of column tips, which means that the growing directionfollows the variation of VIA. In the actual situation the vaporsource is stationary and the substrate is rotated around a verticalaxis. To better understand the variation of their relative positionsit is supposed that the substrate is stationary while the vaporsource moves around the vertical axis. So the incident vaporfluxes for position A move along the generatrix of a symmetriccone. The time of each substrate revolution is divided into manyminute durations, each arrow-headed line section indicates thecolumn growth for certain duration. So as the substrate rotatesthrough 2p radian, the column tips finish a complete helix. As forposition B and C, the situation is similar except that the incidentvapor fluxes move along an asymmetric cone thus the tips finishan inclined helix.

In this assumption, the variation of deposition rate at differentVIA and with different distance from the vapor source was takeninto account. The column inclination was caused by the differentdisplacement of column tips along different direction and can beestimated by the inclination of helix axis.

As for any position on the substrate whose distance from thecenter is d, the instantaneous VIA during one substrate revolutioncan be expressed by the following equation:

VIA ¼ arctan

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2502 þ d2 � 500d cos u

p550

(1)

where d is the distance from the point on the substrate to thecenter of the substrate and is in millimeters; u representsthe instantaneous rotation angle of the substrate and is indegrees. VIA is the instantaneous vapor incidence angle and isin degrees too.

Because the column inclination is observed from RP and TP, VIAis also calculated in terms of its projections on the two planes. Theinstantaneous VIA projected on RP can be expressed as:

VIAr ¼ arctand� 250 cos u

550(2)

where VIAr is the instantaneous VIA projected on RP and is indegrees. When the value is positive, it means the incident vaporflux inclines towards the center of the substrate. While the value isnegative, it means the incident vapor flux inclines towards theedge of the substrate.

The instantaneous VIA projected on TP can be expressed as:

VIAt ¼ arctan250 sin u

550(3)

where VIAt is the instantaneous VIA projected on TP and is indegrees.

The instantaneous VIA projected on RP and TP during onesubstrate revolution are separately illustrated in Fig. 6(a) and (b).As for position A, the value of VIA keeps constant to be 24.48, while

Fig. 2. Cross-sectional morphology corresponding to substrate rotation speed of (a) 1.6 and 10.7 rpm, (b) 1.6 rpm and (c) 25.4 rpm, observed in TP at 100 mm from the center

of the sheet.

Y. Sun et al. / Applied Surface Science 255 (2009) 5831–5836 5833

the direction changes with substrate rotation, giving rise to asymmetric conical pattern. The instantaneous VIA projected on RPvaries from �24.48 to +24.48. The average integrated VIA duringone substrate revolution is 08. As for position B, the instantaneousVIA projected on RP varies from 08 to 42.38 with an average of 22.68.And the instantaneous VIA projected on RP for position C changesfrom 17.68 to 50.88 with an average of 35.98.

As indicated by the above expressions and illustrations, theVIA for certain position on the substrate varies every second asthe substrate rotates. To acquire a general knowledge about therelation between VIA and position, the integrated average of VIAduring one substrate revolution in terms of different positionson the substrate is calculated and displayed in Fig. 4(c) [14]. Andit is noticed that the integrated average VIA projected inRP increases with distance from the center of the sheet in sucha deposition mode that the substrate was rotated around avertical axis and the vapor source was located in a distance ofhalf-radius from the vertical axis. Moreover, the VIA projectedin TP for any position on the sheet varies the same way during

substrate rotation and the integrated average during one subs-trate revolution is zero.

As for any position P on the substrate with a distance of d mmfrom the center of the substrate, the growth rate was y and alongthe vapor incidence direction, which can be decomposed into threecomponents of yz along the substrate normal direction (ND), yr

along the RD and yt in the TD.

yz ¼ n cos a (4)

yr ¼ y sin ad� 250 cos u

550 tan a¼ n cos a

d� 250 cos u550

(5)

yt ¼ y sin a250 sin u550 tan a

¼ n cosa250 sin u

550(6)

where u is the rotation angle of the substrate and is in degrees. a isVIA and is in degrees too, which is expressed by Eq. (1).

Fig. 3. Cross-sectional fractures in (a) RP and (b) TP at positions 100, 200, and 300 mm from the center of the sheet. ND represents the direction of the substrate normal.

Y. Sun et al. / Applied Surface Science 255 (2009) 5831–58365834

The growth rate in the vapor flux direction, y, can be expressedby the following equation.

y ¼ y0 ðcosaÞn H

R

� �2

¼ y0 ðcos aÞnþ2 (7)

where y0 is the growth rate in the vapor flux direction at theposition directly above the vapor source at a distance of H. R is thedistance between point P and the source. And n is an empiricalexponent that depends on the evaporation rate [14,15].

Therefore, the displacement of column tips in ND during timeincrement dt can be expressed as follows.

dz ¼ yz dt ¼ yzduv

(8)

where v represents the rotation speed of the substrate, which is12 rpm in this work. The displacement of column tips in ND duringone substrate revolution can be roughly calculated by integrationof Eq. (8) from u = 0 to u = 2p. So do the displacement along othertwo directions.

z0 ¼Z 2p

0nz

duv¼ n0

v

Z 2p

0ðcos aÞnþ3du (9)

r0 ¼Z 2p

0nr

duv¼ n0

v

Z 2p

0ðcos aÞnþ3 d� 250 cos u

550du (10)

t0 ¼Z 2p

0nt

duv¼ n0

v

Z 2p

0ðcosaÞnþ3 250 sin u

550du (11)

Fig. 4. XRD pattern from both sides of the sheet at 250 mm from the center of the

sheet.

Fig. 5. Variation of deposit growth direction during one substrate revolution for

position A, B, and C.

Y. Sun et al. / Applied Surface Science 255 (2009) 5831–5836 5835

The column inclination observed in RP and TP can be expressedas:

tan br ¼r0

z0(12)

tan bt ¼t0

z0(13)

where br is the column inclination in RP and bt is the columninclination in TP.

Column inclinations in RP at different positions on thesubstrate are estimated by the above equations and plotted inFig. 7 for n = 1, 2, and 5. The calculated values qualitatively agreewell with the experimental data. Part of the experimental data is

Fig. 6. (a) Variation of VIA projected on RP during one revolution of substrate. (b) Variat

position A, B and C. (c) Integrated average VIA at different positions on the substrate.

from reference [16]. Column inclination in TP at any position isestimated to be zero by this model.

According to the model, the morphology of the deposits issupposed to be helical, as observed in Fig. 2(b) corresponding tothe substrate rotation speed of 1.6 rpm. However, columnarmicrostructures were observed when the substrate rotation speedwas 10.7 and 25.4 rpm. Some other researchers reported helicaland columnar microstructure due to different substrate rotationspeed yet without detailed interpretation. It was proposed that the

ion of VIA projected on TP during one revolution of substrate, which is the same for

Fig. 7. Column inclinations in RP at different positions on the substrate estimated by

the equations in this paper for n = 1, n = 2 and n = 5. Part of the experimental data is

from Ref. [16].

Y. Sun et al. / Applied Surface Science 255 (2009) 5831–58365836

morphology is determined by the competition between substraterotation speed and deposition rate, and the adatom diffusivity istaken into account [17–19]. Based on the model, considering theabove assumption, it is taken that the morphology of the depositswith rotating substrate was initially formed to be helical. Undercertain deposition rate and with certain adatom diffusivity, if thesubstrate rotation speed is high, then the growth of depositduring one revolution is short; it is possible for the adatoms tomove from peaks to the valleys and form column morphology. Sowith high deposition rate, low substrate rotation speed, and lowatom diffusivity or substrate temperature, it will be relatively easyto form helical morphology.

4.2. Effect of substrate rotation on crystallographic orientation

Both the helical microstructure corresponding to low substraterotation speed and the columnar microstructure corresponding tohigh substrate rotation speed exhibit no preferential orientation. It

implies that the morphology does not necessarily correspond withthe crystallography.

5. Conclusion

The substrate rotation speed affects the microstructure ofEBPVD fabricated metal sheet. Helical microstructure is formedwhen the substrate rotation speed is low. And columnar micro-structure is formed under high substrate rotation speed. IntegrateVIA was found to be effective in evaluating the column inclinationunder asymmetric vapor incidence pattern.

Acknowledgement

The authors wish to thank Prof. Hu Jin for her helps in SEMobservation. We also very appreciate PhD Li Xiao for his valuablesuggestions and helps. This work is supported by Program forChangjiang Scholars and Innovative Research Team in University ofChina.

References

[1] B.A. Movchan, G.F. Badilenko, A.V. Demchishin, Thin Solid Films 63 (1979) 67–75.[2] B.A. Movchan, Surf. Eng. 22 (2006) 35–46.[3] R.F. Bunshah, J. Vac. Sci. Technol. 11 (1974) 633–638.[4] J.A. Thornton, Ann. Rev. Mater. Sci. 7 (1977) 239–260.[5] B.A. Movchan, F.D. Lemkey, Surf. Coat. Technol. 165 (2003) 90–100.[6] J. Singh, D.E. Wolfe, J. Mater. Sci. 40 (2005) 1–26.[7] S. Lichter, J. Chen, Phys. Rev. Lett. 56 (1986) 1396–1399.[8] P. Meakin, Phys. Rev. A 38 (1988) 994–1004.[9] R. Fiedler, G. Schirmer, Thin Solid Films 167 (1988) 281.

[10] R.N. Tait, T. Smy, M.J. Brett, Thin solid Films 226 (1993) 196–201.[11] A.G. Dirks, H.J. Leamy, Thin Solid Films 47 (1977) 219–233.[12] D.J. Paritosh, Srolovitz, J. Appl. Phys. 91 (2002) 1963–1972.[13] K. Wada, N. Yamaguchi, H. Matsubara, Surf. Coat. Technol. 191 (2005) 367–374.[14] U. Schulz, S.G. Terry, C.G. Levi, Mater. Sci. Eng. A 360 (2003) 319–329.[15] S. Schiller, U. Heisig, S. Panzer, Electron Beam Technology, Wiley, New York, 1982.[16] X. Lin, Y. Sun, G.P. Song, X.D. He, Key Eng. Mater. 353–358 (2007) 1637–1640.[17] F.C. Akkari, M. Kanzari, B. Rezig, Physica E 40 (2008) 2577–2582.[18] F. Liu, M.T. Umlor, L. Shen, J. Weston, W. Eads, J.A. Barnard, G.J. Mankey, J. Appl.

Phys. 85 (1999) 5486–5488.[19] R. Figueroa, T.G.S. Cruz, A. Gorenstein, J. Power Sources 172 (2007) 422–427.