Journal of Nuclear Materials 420 (2012) 405–408

Contents lists available at SciVerse ScienceDirect

Journal of Nuclear Materials

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

Deuterium permeation of amorphous alumina coating on 316L prepared by MOCVD

Shuai Li ⇑, Di He, Xiaopeng Liu, Shumao Wang, Lijun JiangDepartment of Energy Materials and Technology, General Research Institute for Nonferrous Metals, Beijing, China

a r t i c l e i n f o a b s t r a c t

Article history:Received 7 September 2011Accepted 24 October 2011Available online 4 November 2011

0022-3115/$ - see front matter � 2011 Elsevier B.V. Adoi:10.1016/j.jnucmat.2011.10.040

⇑ Corresponding author. Tel.: +86 10 8224 1238.E-mail address: shuail@kth.se (S. Li).

The deuterium permeation behavior of the alumina coating on 316L stainless steel prepared by metalorganic chemical vapor deposition (MOCVD) was investigated. The alumina coating was also character-ized by X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and scanning electron micro-scope (SEM). It was found that the as-prepared coating consisted of amorphous alumina. This aluminacoating had a dense, crack-free and homogeneous morphology. Although the alumina coating was amor-phous, effective suppression of deuterium permeation was demonstrated. The deuterium permeability ofthe alumina coating was 51–60 times less than that of the 316L stainless steel and 153–335 times lessthan that of the referred low activation martensitic steels at 860–960 K.

� 2011 Elsevier B.V. All rights reserved.

1. Introduction

Tritium permeation through structural materials is a key issuein the development of DEMO fusion reactors [1–3]. The relativelyhigh tritium pressure produced in PbLi breeder blanket and thelarge surface area of coolant channels raise the problem of high tri-tium permeation rate through materials. Tritium loss due to highpermeation rate reduces the economic viability of fusion reactorsand causes radiological hazards. It is widely recognized that cera-mic coatings provide an attractive solution to lower tritium perme-ation in the structural materials [4–16].

Among all the ceramic materials, alumina has low hydrogen per-meation rate, excellent corrosion resistance and high mechanicalhardness, thus it is a promising material for tritium permeation bar-rier application [10–16]. Forcey et al. [10] found that the aluminizedlayer on 316L offered a reduction of hydrogen permeation rate of3–4 orders of magnitude. The effectiveness of this permeation bar-rier was attributed to the surface oxide layer consisting of alumina.Levchuk et al. [11–13] prepared the alumina coatings by filtered arcdeposition. The 1 lm thick alumina coating had a-alumina struc-ture, and it exhibited a deuterium permeation reduction factor of103 in the temperature range of 973–1073 K [11]. Although aluminahas a rather low hydrogen permeation rate, the effectiveness of per-meation barrier also depends much on the integrity of the coating[16,17]. It has been shown by Pisarev et al. [17] that the permeationreduction factor of a protection coating could be reduced to 10 witheven small cracking (0.001% of surface is open) if the permeationthrough the bare coating took place in the pure diffusion limited re-gime. Although the permeation through ceramic coatings did not al-ways take place in the pure diffusion limited regime [11–13], it is

ll rights reserved.

believed that cracks and pores should be highly avoided for theeffective tritium permeation barriers [16]. Therefore, the prepara-tion of crack-free and dense coatings with good adhesion to sub-strates is essential for the application of alumina coatings astritium permeation barriers.

Metal organic chemical vapor deposition (MOCVD) is an inter-esting technique to prepare fine alumina coatings [18–23]. The con-ventional physical vapor deposition of alumina coatings hasdifficulties to process complex shape surfaces [11–13], while chem-ical vapor deposition needs high deposition temperature, e.g.1100–1200 �C [24–26]. MOCVD is characterized by the pyrolyticdecomposition of precursor and alumina can be deposited on com-plex shape substrates at relatively low temperatures. The composi-tion, microstructure, and crystallinity of the coating can be easilyadjusted by tuning deposition conditions, such as reactor geometry,precursor, reactive atmosphere, flow rate, pressure and depositiontemperature [18]. Although the alumina coatings prepared byMOCVD have been widely investigated in terms of chemical, elec-trical and mechanical properties [18], knowledge of the deuteriumpermeation behavior of the coatings is still limited. In this work, thedeposition of the alumina coating on 316L stainless steel was per-formed by MOCVD. The deuterium permeation behavior of thiscoating was examined in order to provide better understanding ofthe alumina coatings for the tritium permeation barrier application.

2. Experimental

2.1. Coating preparation

Deposition of alumina coatings was performed by MOCVD in ahorizontal hot wall reactor. The schematic diagram of the MOCVDsystem is shown in Fig. 1. The alumina coatings were single-sidedeposited on 316L stainless steel disks. Aluminum acetylacetonate

Fig. 1. Schematic diagram of the MOCVD system. (1) Mass flow meter, (2) waterbubbler, (3) ball valve, (4) heater, (5) precursor, (6) electric furnace, (7) substrate,(8) vacuum gauge, (9) rotary pump.

406 S. Li et al. / Journal of Nuclear Materials 420 (2012) 405–408

(Al(acac)3, P99.8%, Xingye Chemical) was used as precursor. Al(a-cac)3 was heated at 120 �C to evaporate. H2 was used as carriergas with a flow rate of 20 sccm. The carrier gas H2 was mixed withwater vapor by flowing through a water bubbler before arriving atthe precursor evaporation zone. The alumina coatings were depos-ited at 350 �C for 4 h with pressure of 1.2–1.4 kPa in the chamber.The coatings were further thermal annealed at 700 �C for 2 h in Aratmosphere after deposition to eliminate structural water in alu-mina [19]. After annealing at 700 �C for 2 h, the uncoated side of316L disk was slightly oxidized. In order to exclude the influenceof this oxide layer, the uncoated side of 316L disk was polishedand cleaned before the deuterium permeation measurement.

2.2. Coating characterization

The compositional analysis of the alumina coating was carriedout by X-ray photoelectron spectroscopy (XPS, PHI QuanteraSXM) with a monochromatic Al Ka radiation. The phase analysisof the coating was performed by grazing incidence X-ray diffrac-tion (XRD, Rigaku-D/max2500). The coating morphology wasexamined by scanning electron microscope (SEM, Hitachi-S4800).

Deuterium permeation measurements of the alumina coatedand bare 316L samples were conducted by the permeation appara-tus shown in Fig. 2. The permeation chamber was divided by sam-ple into two parts, e.g. the upstream chamber and the downstreamchamber. Samples were sealed with 316L gaskets. Deuterium wasintroduced into the upstream chamber at 40–100 kPa using a nee-dle valve. The pressure of deuterium was monitored by a quartzvacuum gauge (10–100 kPa, DL-10, Beijing Xinhengjiu Tech.). Thedownstream chamber was pumped by a turbo-molecular pump,and the background pressure of the downstream chamber wasmonitored by an ionization gauge (6 � 10�8–10�1 Pa, DL-7, BeijingXinhengjiu Tech.). Measurement of the permeated deuteriumthrough a sample to the downstream chamber was performed bya quadrupole mass spectrometer (QMS, Hiden HPR30). Calibrationfor the QMS from deuterium permeation flux to ion current wasconducted using a standard leak (5.2 Pa m3/s, 296 K, 100 kPa).

The procedure of permeation measurements was as follows.Deuterium was introduced stepwise into the upstream chamber.

Fig. 2. Schematic diagram of the deuterium permeation apparatus. (1) Needlevalve, (2) quartz vacuum gauge, (3) diaphragm valve, (4) turbo-molecular pump, (5)rotary pump, (6) electric furnace, (7) sample, (8) ionization gauge, (9) quadrupolemass spectrometer.

The next portion of deuterium was introduced, when the steadystate of deuterium permeation flux was reached. The downstreamchamber was continuously pumped with a base pressure of 2 �10�5 Pa. Deuterium permeation flux through a sample was mea-sured by QMS. The 316L disks had a diameter of 29 mm, and diskswere mechanically polished down to thickness of 0.40 mm. Themeasured surface area of the sample was 3.46 cm2. The aluminacoated sample was installed in such a way that the coated surfacefaced the upstream chamber to avoid potential oxidation of the un-coated side due to oxygen traces in the deuterium gas [5,11]. Whilefor the bare 316L disks, Pd films were deposited on surfaces to avoidoxidation at elevated temperatures. The alumina coated and bare316L samples were heated by an electric furnace. After each perme-ation measurement at a fixed temperature, the coated sample waskept at 973 K for 1.5 h to release the deuterium accumulated in sam-ple, while the bare 316L sample was heated at 773 K for 1.5 h.

The steady state permeation rate J of hydrogen through materialfollows the equation [26]:

J ¼ P � pn

d

where P is the permeability of material, p is the driving pressure, nis the pressure exponent, and d is the sample thickness. The pres-sure exponent n is evaluated on the basis of the pressure depen-dence of hydrogen permeation. A half power (n = 0.5) dependenceimplies that the rate limiting process for permeation is hydrogenatom diffusion through material, while n = 1 holds for the case ofsurface reaction as the rate limiting process.

3. Results and discussion

The chemical composition of coating was investigated by XPS.The single broad peaks of Al 2p and O 1s were illustrated for thealumina coating after thermal annealing at 700 �C for 2 h. Thebinding energies of Al 2p and O 1s were 74.7 eV and 531.2 eVrespectively, which are the typical values for a pure alumina[27,28]. Fig. 3 shows the XRD pattern of the alumina coating on316L substrate. As shown in Fig. 3, the diffraction peaks can beindexed with austenitic Cr–Ni–Fe–C (JCPDS 31-0619), which isfrom the 316L substrate. No crystalline phase of alumina is ob-served in Fig. 3, implying that the MOCVD deposited alumina coat-ing on 316L substrate is amorphous in this work. The evolution ofcrystalline structures of the MOCVD deposited alumina coatingshas been widely investigated [19–22]. The low temperature(6600 �C) deposition of alumina coatings from Al(acac)3 precursoroften yields amorphous alumina [19–21]. Moreover, it was

20 40 60 800

200

400

Inte

nsity

2θ (º)

Cr-Ni-Fe-C

Fig. 3. XRD pattern of the alumina coating on 316L substrate.

S. Li et al. / Journal of Nuclear Materials 420 (2012) 405–408 407

reported that the crystalline alumina could be obtained at elevateddeposition temperatures [20–22] or by post-deposition thermalannealing [19,23]. Ito et al. [22] found that the alumina coatingdeposited at 900 �C was amorphous, while c- and a-alumina wereobtained at higher deposition temperatures, e.g. >1050 �C. In addi-tion, Pflitsch et al. [19] reported that the alumina coating wasamorphous after thermal annealing at 797 �C, and the coatingtransformed to c- , h- and a-alumina in sequence with increasingthermal annealing temperature. In this work, the 700 �C annealedalumina coating was found to be amorphous, in agreement withthe literature findings [19,23]. Furthermore, it should be pointedout that the amorphous alumina coating was stable at tempera-tures 6700 �C, for example, the coating was found to be amor-phous even after longer annealing time of 10 h at 700 �C (XRDdata are not shown here).

The morphology of the alumina coating on 316L substrate wasinvestigated by SEM in Fig. 4. Fig. 4a revealed the surface morphol-ogy of the alumina coating. No cracking was found on the coatingsurface, and the coating possessed a dense and homogeneous mor-phology. Furthermore, no gap or pore was detected between thecoating and 316L substrate as shown in Fig. 4b, illustrating theexcellent adhesion of the alumina coating. The integrity of thecoating is of vital importance for the performance of tritium per-meation barriers. It was found that the detachment of the aluminacoating from Eurofer substrate could significantly deteriorate theperformance of the coating [16]. Haanappel et al. [23] proposedthat the post-deposition thermal annealing at mild temperatures,e.g. 700 �C, could reduce significantly the intrinsic stress in

Fig. 4. SEM showing the morphology of the alumina coating on 316L substrate. (a)Surface morphology; (b) cross section micrograph.

alumina coatings deposited by MOCVD. This can explain the excel-lent adhesion of the alumina coating to substrate in this work.

Fig. 5 shows the permeation rate of the alumina coating on 316Lsubstrate in terms of deuterium pressure at temperatures of 860 K,910 K and 960 K. The double logarithmic plots exhibit perfect lin-ear relationship. The pressure exponents n of deuterium perme-ation were between 0.67 and 0.73 as illustrated in Fig. 5. Thisimplied that the deuterium permeation process in alumina coated316L was not pure diffusion limited (n = 0.5), and the permeationwas also influenced by the surface effect due to the presence of alu-mina coating.

The temperature dependence of deuterium permeability of thealumina coating on 316L substrate is shown in Fig. 6. Since thepressure exponents of the alumina coated 316L did not correspondto 0.5, the permeability calculated based on equation J = P�p0.5/dwas a function of the driving pressure. It was proposed that thesurface influence of hydrogen permeation in materials was lesspronounced with high driving pressures [11]. Therefore, the per-meability of the alumina coating was calculated at deuterium pres-sure of 100 kPa in this work. The permeation reduction factors(PRFs) of the alumina coating on 316L substrate were 51 at860 K, 60 at 910 K and 51 at 960 K, respectively. The PRF of the alu-mina coating is related to the substrate material. Chikada et al. [6]reported that the difference of PRFs of Er2O3 coatings was about 1order of magnitude on various substrates, such as austenitic 316Land martensitic F82H. Previous works also found that the hydro-gen permeation rates could be different of 1 order of magnitudefrom 523 K to 873 K among austenitic and martensitic steels[6,29]. It was believed that the deuterium permeation quality ofthe coatings was same on different type of steels [6]. Thus, it is rea-sonable to evaluate the permeation behavior of the alumina coat-ings on various substrates. In this work, the permeation data oflow activation martensitic steels, such as Eurofer [11], MANET[29] and F82H [30], were provided as well for comparison inFig. 6. It was found that the permeability of the alumina coatingwas 153–335 times less than that of Eurofer, MANET and F82Hat temperatures of 860–960 K, which meets the requirement ofPRF > 100 compared with the bare martensitic steels in thefusion blanket system [3]. Thus, the amorphous alumina coatingprepared by MOCVD is qualified as the tritium permeation barrier.

Furthermore, it is noted that the PRF of the alumina coating inthis work is still inferior to the best reported data [7,8,11]. Forexample, Levchuk et al. [7] has reported the high deuterium PRFof 2000–3500 at 973 K for a thin Al–Cr–O coating on Eurofer. Theinferior performance of the alumina coating in this work may be

4x104 6x104 8x104 105 1.2x10510-7

10-6

860K (0.69) 910K (0.73) 960K (0.67)

J (m

ol /

m2

s)

Pressure (Pa)

Fig. 5. Deuterium permeation flux of the alumina coating on 316L substrate as afunction of driving deuterium pressure at different temperatures. Numbers in theparenthesis are the pressure exponents n.

1E-13

1E-12

1E-11

1E-10

1E-9alumina

316L

P (m

ol /

msP

a0.5 )

MANET [29]

Eurofer [11]F82H [30]

1000/T (1/K)1.0 1.2 1.4 1.6

Fig. 6. Arrhenius plots of the deuterium permeability of the alumina coated andbare 316L samples. Data of Eurofer [11], MANET [29] and F82H [30] are referred.

408 S. Li et al. / Journal of Nuclear Materials 420 (2012) 405–408

due to the amorphous nature of the coating. Previous worksshowed that the crystalline structure of coating affected the perfor-mance of tritium permeation barrier [9,11]. The crystalline aluminacoating was reported to have higher PRF than the amorphous one[11,13]. Therefore, it is expected that the alumina coating can offerbetter hydrogen permeation suppression performance if the coat-ing consists of crystalline alumina. As discussed above, the crystal-line alumina can be obtained at elevated deposition temperature orby post-deposition thermal annealing. The deuterium permeationbehavior of the crystalline alumina coatings prepared by MOCVDwill be examined in near future.

4. Conclusion

The deuterium permeation behavior of the alumina coating on316L was investigated. The amorphous alumina coating depositedby MOCVD on 316L was crack-free, dense and homogeneous.Moreover, the deuterium permeability of this amorphous aluminacoating was 51–60 times less than that of the 316L stainless steeland 153–335 times less than that of the referred low activationmartensitic steels at 860–960 K, offering efficient suppression ofdeuterium permeation in structural materials. Furthermore, it is

expected that the crystalline alumina prepared at elevated deposi-tion temperature or by post-deposition thermal annealing can fur-ther improve the performance of alumina coatings.

References

[1] G. Benamati, C. Chabrol, A. Perujo, E. Rigal, H. Glasbrenner, J. Nucl. Mater.271&272 (1999) 391–395.

[2] D.L. Smith, J.H. Park, I. Lyublinski, V. Evtikhin, A. Perujo, H. Glassbrenner, T.Terai, S. Zinkle, Fusion Eng. Des. 61 (62) (2002) 629–641.

[3] C.P.C. Wong, V. Chernov, A. Kimura, Y. Katoh, N. Morley, T. Muroga, K.W. Song,Y.C. Wu, M. Zmitko, J. Nucl. Mater. 367 (370) (2007) 1287–1292.

[4] A. Perujo, K.S. Forcey, Fusion Eng. Des. 28 (1995) 252–257.[5] T. Chikada, A. Suzuki, C. Adelhelm, T. Terai, T. Muroga, Nucl. Fusion 51 (2011)

063023.[6] T. Chikada, A. Suzuki, Z. Yao, D. Levchuk, H. Maier, T. Terai, T. Muroga, Fusion

Eng. Des. 84 (2009) 590–592.[7] D. Levchuk, H. Bolt, M. Döbeli, S. Eggenberger, B. Widrig, J. Ramm, Surf. Coat.

Technol. 202 (2008) 5043–5047.[8] P.J. McGuiness, M. Cekada, V. Nemanic, B. Zajec, A. Recnik, Surf. Coat. Technol.

205 (2011) 2709–2713.[9] T. Chikada, A. Suzuki, T. Kobayashi, H. Maier, T. Terai, T. Muroga, J. Nucl. Mater.

(2011), doi:10.1016/j.jnucmat.2010.12.283.[10] K.S. Forcey, D.K. Ross, C.H. Wu, J. Nucl. Mater. 182 (1991) 36–51.[11] D. Levchuk, F. Koch, H. Maier, H. Bolt, J. Nucl. Mater. 328 (2004) 103–106.[12] F. Koch, R. Brill, H. Maier, D. Levchuk, A. Suzuki, T. Muroga, H. Bolt, J. Nucl.

Mater. 329 (333) (2004) 1403–1406.[13] D. Levchuk, F. Koch, H. Maier, H. Bolt, Phys. Scr. T108 (2004) 119–123.[14] Y. Ueki, T. Kunugi, N.B. Morley, M.A. Abdou, Fusion Eng. Des. 85 (2010) 1824–

1828.[15] E. Serra, J. Am. Ceram. Soc. 88 (2005) 15–18.[16] A. Aiello, I. Ricapito, G. Benamati, A. Ciampichetti, Fusion Eng. Des. 69 (2003)

245–252.[17] A. Pisarev, I. Tsvetkov, S. Yarko, Fusion Eng. Des. 82 (2007) 2120–2125.[18] A.N. Gleizes, Chem. Vap. Deposition 13 (2007) 23–29.[19] C. Pflitsch, D. Viefhaus, U. Bergmann, B. Atakan, Surf. Coat. Technol. 201 (2006)

73–81.[20] C. Pflitsch, D. Viefhaus, U. Bergmann, B. Atakan, Thin Solid Films 515 (2007)

3653–3660.[21] S.K. Pradhan, P.J. Reucroft, Y. Ko, Surf. Coat. Technol. 176 (2004) 382–384.[22] A. Ito, R. Tu, T. Goto, Surf. Coat. Technol. 204 (2010) 2170–2174.[23] V.A.C. Haanappel, Oxid. Met. 43 (1995) 459–478.[24] A. Lasson, S. Ruppi, Int. J. Refract. Met. Hard Mater. 19 (2001) 515–522.[25] S. Ruppi, A. Larsson, Int. J. Refract. Met. Hard Mater. 23 (2005) 306–316.[26] K.S. Forcey, A. Perujo, F. Reiter, P.L. Lolliceroni, J. Nucl. Mater. 200 (1993) 417–

420.[27] J.H. Eun, J.H. Lee, S.G. Kim, M.Y. Uma, S.Y. Park, H.J. Kim, Thin Solid Films 435

(2003) 199–204.[28] J.R. Lindsay, H.J. Rose, W.E. Swartz, P.H. Watts, K.A. Rayburn, Appl. Spectrosc.

27 (1973) 1–5.[29] K.S. Forcey, D.K. Ross, J.C.B. Simpson, J. Nucl. Mater. 160 (1988) 117–124.[30] A. Pisarev, V. Shestakov, S. Kulsartov, A. Vaitonene, Phys. Scr. T94 (2001) 121–

127.