cvd3

Page 1

E3

IEa

b

c

d

a

ARR2A

BAfhMdtT

mT

0d

Nuclear Engineering and Design 251 (2012) 191– 202

Contents lists available at SciVerse ScienceDirect

Nuclear Engineering and Design

j ourna l ho me page: www.elsev ier .com/ locate /nucengdes

ffects of phosphorous-doping and high temperature annealing on CVD grownC-SiC

.J. van Rooyena,b,c,∗, J.H. Neethlingb, A. Henryd, E. Janzénd, S.M. Mokoduwec, A. Janse van Vuurenb,. Olivierb

CSIR, National Laser Centre, PO Box 395, Pretoria 0001, South AfricaDepartment of Physics, Nelson Mandela Metropolitan University, PO Box 77000, Port Elizabeth 6031, South AfricaFuel Design, PBMR, 1279 Mike Crawford Avenue, Centurion 0046, South AfricaDepartment of Physics, Chemistry and Biology, Semiconductor Materials, Linköping University, Linköping 58183, Sweden

r t i c l e i n f o

rticle history:eceived 17 January 2011eceived in revised form2 September 2011ccepted 22 September 2011

a b s t r a c t

The integrity and property behavior of the SiC layer of the Tri-isotropic (TRISO) coated particle (CP) forhigh temperature reactors (HTR) are very important as the SiC layer is the main barrier for gaseous andmetallic fission product release. This study describes the work done on un-irradiated SiC samples pre-pared with varying phosphorus levels to simulate the presence of phosphorus due to transmutation. 30Sitransmutes to phosphorous (31P) and other transmutation products during irradiation, which may affectthe integrity of the SiC layer. The P-doping levels of the SiC samples used in this study cover the rangefrom 1.1 × 1015 to 1.2 × 1019 atom/cm3 and are therefore relevant to the PBMR operating conditions.Annealing from 1000 ◦C to 2100 ◦C was performed to study the possible changes in nanostructures andvarious properties due to temperature. Characterization results by X-ray diffraction (XRD), secondaryion mass spectrometry (SIMS), scanning electron microscopy (SEM), transmission electron microscopy(TEM) and high resolution transmission electron microscopy (HRTEM), are reported in this article. Asgrain boundary diffusion is identified as a possible mechanism by which 110mAg, one of the fission acti-vation products, might be released through intact SiC layer, grain size measurements is also included inthis study. Temperature is evidently one of the factors/parameters amongst others known to influencethe grain size of SiC and therefore it is important to investigate the effect of high temperature annealingon the SiC grain size. The ASTM E112 method as well as electron back scatter diffraction (EBSD) was usedto determine the grain size of various commercial SiC samples and the SiC layer in experimental PBMR

◦ ◦

Coated Particles (CPs) after annealing at temperatures ranging from 1600 C to 2100 C. The HRTEM micro-graph of the decomposition of SiC at 2100 ◦C are shown and discussed. Nanotubes were not identifiedduring the TEM and HRTEM analysis although graphitic structures were identified. The preliminary con-clusion reached is that the P-content at these experimental levels (1.1 × 1015 to 1.2 × 1019 atom/cm3)does not have a significant influence on the nanostructure of SiC at high temperatures withoutirradiation.

Published by Elsevier B.V.

Abbreviations: ACF, advance coater facility; AFM, atomic force microscopy;F, bright field; CP, coated particle; CVD, chemical vapor deposition; EAG, Evansnalytical Group; EBSD, electron back scatter diffraction; FTIR, Fourier trans-

ormed infrared; HRTEM, high resolution transmission electron microscopy; HTR,igh temperature reactors; LiU, Linköping University; NMMU, Nelson Mandelaetropolitan University; ORNL, Oakridge National Laboratory; SAD, selected area

iffraction; SEM, scanning electron microscopy; SIMS, secondary ion mass spec-rometry; TBP, tertiary butyl phosphine; TEM, transmission electron microscopy;RISO, tri-isotropic; XRD, X-ray diffraction.∗ Corresponding author. Present address: Fuel Performance and Design Depart-ent, Idaho National Laboratory, PO Box 1625, Idaho Falls, ID 83415, USA.

el.: +1 2085337199.E-mail address: [email protected] (I.J. van Rooyen).

029-5493/$ – see front matter. Published by Elsevier B.V.oi:10.1016/j.nucengdes.2011.09.066

1. Introduction

The integrity and property behavior of the SiC layer of the tri-isotropic (TRISO) coated particle for high temperature reactors(HTR) are very important. The SiC is the main barrier for gaseousand metallic fission product release. Although most fission prod-ucts are retained by the combination of SiC and pyrolytic carbonlayers, previous observations of silver release during fuel testingand operations identified the release of silver by so-called intact

SiC layers (Nabielek et al., 1977; Yurko et al., 2008). Nabielek andBrown (1975) proposed grain boundary diffusion as a mechanismfor 110mAg release through intact SiC layers. Petti et al. (2003)also microscopically compared US manufactured- with German

Page 2

1 neerin

mt1

ofcittiapTiTSoaShuWoooiogpr

ttt(mtpettt(p

tpntprSaTVwuaT2bt

ttC

92 I.J. van Rooyen et al. / Nuclear Engi

anufactured TRISO CPs and found that the grain size and struc-ure influence the 110mAg release. It was found that 90% of the10mAg was released from the large columnar grained SiC, whereasnly 30% was released from the small grained SiC particles. There-ore this study was undertaken to determine the grain size ofhemical vapor deposition (CVD) grown SiC. The reason for thisnvestigation is to determine the change in grain size after highemperature annealing as grain size may influence the strength ofhe CP and further may influence the silver release from the CP dur-ng operation. It was decided to initially do grain size evaluationfter annealing from 1600–2100 ◦C on flat commercial 3C-SiC sam-les, due to the relatively ease of sample handling and preparation.his work was followed by the annealing of two selected exper-mental PBMR CP batches at temperatures from 1600 to 2000 ◦C.hese very high annealing temperatures were chosen becausenead et al. (2007) found that no significant strength degradationf CVD SiC occurs for temperatures up to 1500 ◦C. Long durationnnealing (>100 h) studies, although important for the effect oniC strength (Van Rooyen et al., 2010) as it is an indication ofow the grain size will be modified during the reactor operationnder specific conditions, is not included in this phases of work.ork by Byun et al. (2009) also makes reference to the influence

f grain size on the SiC strength. This temperature range is alsof interest as the fuel sphere manufacturing process make usef a 1 h sintering process at 1950 ◦C and this study will indicatef any grain size changes could be expected due to the sinteringperation. This work forms part of a larger experimental pro-ramme of the PBMR Fuel Design department to test the designarameters of the CPs for application in very high temperatureeactors.

Additionally an investigation was also proposed to determinehe effect that different Si isotopes may have on the SiC crys-al structure during the CVD manufacturing process and afterransmutation due to irradiation. 30Si transmutes to phosphorous31P) and other transmutation products during irradiation, which

ay affect the integrity of the SiC layer. This study describeshe work done on unirradiated SiC, but the SiC samples wererepared with varying phosphorous levels to “simulate” the pres-nce of P due to transmutation. Heinisch et al. (2004) reportedhat 36 appm P are produced by neutron irradiation-inducedransmutation in a modular pebble bed reactor (265 MWth,otal neutron flux 1.25 × 1014 n/cm2 s for 10 full power years4.4 displacements per atom (dpa)). This amounts to 8.2 appm/dpahosphorous.

Calculations of material activation due to neutrons were done byhe PBMR’s Nuclear Engineering Analysis (NEA) group using Euro-ean Activation System-2007 (EASY) comprising of the EAF-2007uclear data and the FISPACT-2007 inventory code. These calcula-ions were done for the different isotopic compositions of SiC asart of the Fuel Design study, but only the results for the natu-al Si-isotopic composition (92.2% 28Si, 4.6% 29Si, and 3.1% 30Si) foriC are described in this paper. Density of SiC used is 3.20 g/cm3

nd the total volume of SiC layer irradiated is 7.13 × 10−05 cm3.he fuel residence time in the reactor core was determined bySOP and is 925 days, meaning that one cycle through the PBMRill take in average 152.16 days (925/6). Thus the irradiation timesed for this calculation is for the fuel residence time of 925 daysnd is applicable to the 6 passes for the 400 MWth PBMR design.he cooling period applied for this activation is 30 days (Maage,009). These calculations revealed that 10 appm P are producedy neutron irradiation-induced transmutation under PBMR condi-ions.

The main purpose of this work is therefore to investigatehe effects of high temperature annealing and P-doping onhe micro- and nanostructure of the SiC layer of PBMR TRISOPs.

g and Design 251 (2012) 191– 202

2. Materials and characterization methods

2.1. Materials

Three sample sets are used in this study namely bulk CVD 3C-SiC, two PBMR CP batches and P-doped CVD 3C-SiC. The first twosample sets are used for the grain size determination and the thirdsample set is used to study the effect of P-doping on the SiC micro-and nanostructure after high temperature annealing. The first twosample sets for the grain size characterization are shown in theExperimental Research Plan flow diagram in Fig. 1.

The bulk CVD-grown polycrystalline 3C-SiC (flat samples) usedfor the first phase of this study was prepared by Rohm and Haas,Advanced Materials Co., USA. This commercial available SiC waschosen since it provided similar grain sizes to that used in the PBMRfuel SiC layer. Five small square samples were cut from the orig-inal sample received from Oakridge National Laboratory (ORNL)and prepared for grain size determination. Four flat CVD-grownSiC samples were annealed at temperatures 1600 ◦C, 1700 ◦C,2000 ◦C, 2100 ◦C in a Webb 89 Vacuum furnace at Nelson MandelaMetropolitan University (NMMU), South Africa, and one samplekept as reference.

Two experimental CP batches used for the second phase of thisstudy were produced by PBMR Fuel Development Laboratories atNECSA during 2007. These two batches were chosen as it gave arange of different manufacturing conditions manufactured in the“Advance Coater Facility” (ACF). It is noted that the SiC depositiontemperatures (1450 ◦C, 1510 ◦C) and the SiC thickness are differentfor these two batches. The SiC thicknesses of the two batches (39and 32 �m, respectively) represent also the highest and the lowestof the SiC thickness specification values. The SiC deposition rate is∼0.23 �m/min for both batches.

The third sample set consist of phosphorous-doped 3C-SiC lay-ers deposited, at the Linköping University’s (LiU) facility, on Si (100)substrates using the concept of the hot wall chemical vapor depo-sition with silane (SiH4) and propane (C3H8) as precursors dilutedin hydrogen (H2). Propane was introduced in the reactor chamberprior to the silane addition during the heating-up cycle to preparea thin layer of SiC on the silicon substrate. This was to ensure epi-taxial growth of mono-crystalline layers once silane is added atthe growth temperature. However by changing growth procedurespolycrystalline layers were also obtained. Phosphorus doping ofthe layers was done during epitaxy using tertiary butyl phosphine(TBP) (C4H9PH2) as a donor source (Henry and Janzén, 2005). TheSiC deposition temperature varies between 1210 ◦C and 1320 ◦C forthe 13 samples manufactured.

2.2. Grain size determination

Grain size determination was carried out using two methodsnamely the Heyn Lineal Intercept procedure in accordance withASTM E112-96 (2004) and electron back scatter diffraction (EBSD)analysis (Tan et al., 2008; Helary et al., 2004).

2.2.1. Grain size determination using Heyn Lineal Interceptmethod

The flat SiC samples were polished with 30, 9, 6, 3 and 1 �mpolishing cloths to obtain a mirror finish side to be analyzed.The TRISO CPs under investigation was prepared by grinding theparticles to half-way through in order to expose the differentlayers. The particles were polished with the final polishing donewith colloidal silica, containing particles with an average size of

0.05 �m. All prepared samples were then etched using a modifiedMurakami’s etchant. The modified Murakami’s etchant was pre-pared by adding 25 g of both KOH and NaOH to 100 ml of distilledwater. After the fore mentioned dissolved and the liquid cleared,

Page 3

I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202 193

plan

2twTmfrtbf

tcc

tsmlte

S

ne

G

tT

in an epoxy resin and mechanically thinned and polished using aBuehler-Beta grinder polisher down to a 0.05 �m colloidal silicafinish.

Fig. 1. Experimental research

5 g of K3[Fe(CN)]6 (potassium-hexa-ferricyanide) was added tohe mixture and stirred while heated until it dissolved. The reagentas kept at approximately 200 ◦C throughout the etching process.

he CPs were then added and left in the etchant for approxi-ately 1 h and separately, the CVD-grown samples were etched

or 30 min. The samples were then removed from the etchant andinsed repeatedly in distilled water containing KOH and then in dis-illed water only. Finally the samples were placed in an ultrasonicath for a few minutes and then mounted and placed in the SEMor investigation of the grains and grain boundaries.

The average grain size of the flat CVD SiC samples and thewo batches CPs were determined using the Heyn Lineal Inter-ept method on 10 SEM micrographs for each sample and it wasalculated as follows.

Every micrograph has a demonstration line at the bottom withhe size of the line, SD, given in microns. The length of the demon-tration line, LD can be measured in cm. Draw 15 test lines on theicrograph, see example in Fig. 2, measure the length of one test

ine, LT in cm (the measurement will apply to all the test lines). Now,he size of the test lines, ST can be calculated using the followingxpression:

T = SDLT

LD(1)

Grain size, G, is therefore the size of the line, ST, divided by theumber of grains cut by the test line, N, and mathematically it isxpressed as:

= ST (2)

N

Follow the same procedure for all 10 micrographs and for everyest line drawn on each micrograph and take the average grain size.he % relative accuracy (RA) as determined according to ASTM E112

for grain size measurements.

(2004) is in all cases calculated to be lower than 10% and is consid-ered to be acceptable precision. However, the grain size of the SiClayer of the TRISO CPs are not equiaxed in shape and therefore thegrain sizes determined in this study, should be clearly indicate themeasurement direction for comparison purposes with results fromother researchers.

2.2.2. Grain size determination using EBSDThe coated particles containing the SiC layers were embedded

Fig. 2. Example of a SEM micrograph used for grain size determination.

Page 4

194 I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202

oped p

eebcT0bfi

2

tTtd

uwtpirStvfpo0

(

Fig. 3. Typical example of the schematic presentation of LiU P-d

The analysis of the prepared samples was done using a HKLlectron backscattered detector coupled to a JEOL 7001F scanninglectron microscope. The sample was tilted to 70◦ within the cham-er with fine calibration of the detector geometry done using therystal structure for 3C-SiC, the expected SiC polytype to be present.he area for the EBSD analysis was selected with a step size of.1 �m for the mapping chosen. The scan duration was typicallyetween 9 and 11 h for each scan. In addition the crystal structureor the 6H-SiC phase was also included during the analysis to checkf any grains of this structure were present.

.3. Characterization of P-doped SiC samples (sample set 3)

The P-doped SiC layers were characterized after high tempera-ure annealing from 1600 ◦C to 2100 ◦C using various techniques.he LiU samples were further divided in smaller samples andhe specific characterization work executed on the sub-samples isetermined and a typical example is schematically shown in Fig. 3.

Annealing was performed in a resistively heated Webb 89 vac-um furnace supplied by R.D. Webb Company USA. The samplesere loaded in graphite or ceramic holders at room tempera-

ure and heated to the required temperature at a rate of 25 ◦Cer minute. After completion of annealing at the required hold-

ng time and temperature, the samples were furnace-cooled tooom temperature. As the Linkoping samples were SiC grown oni-wafers, argon annealing atmosphere were chosen to ensure thathe vaporized silicon be removed from the sample surfaces to pre-ent contamination of the SiC samples itself. However, due to theurnace maximum temperature limitations, the annealing of sam-le X377CB at 2100 ◦C were conducted under vacuum for 10 min

nly. The rest of the samples were annealed under argon for 1 h and.5 h for the 1600 ◦C and 2000 ◦C cycles, respectively.

The results of only five techniques namely X-ray diffractionXRD), secondary ion mass spectrometry (SIMS), scanning electron

olycrystalline X377 sample and characterization identification.

microscopy (SEM), transmission electron microscopy (TEM) andhigh resolution transmission electron microscopy (HRTEM), arebriefly described in this paper. The results obtained from the othertechniques like the atomic force microscopy (AFM) and Fouriertransformed infrared (FTIR) will be described in separate articlesat a later stage. The basic principles of the specific five techniquesare not described in this paper.

It needs to be noted that although different structure morpholo-gies were found and briefly discussed in this paper, the relevantsubsamples used for the TEM observations, exhibited similar mor-phologies. The annealing studies on X364 subsamples which showglobular structures were therefore not considered in reaching con-clusions in this study. A large number of TEM micrographs (>250)were analyzed for this study to facilitate the conclusions reached.

3. Results and discussion

3.1. Grain size determination using Heyn Lineal Intercept method

The grain size measurement of the annealed bulk CVD 3C-SiC results indicates that the grain size increases with increasingannealing temperature. This can be seen in Fig. 4, which shows themicrographs of etched grains and in Fig. 5. The average grain sizeincreases from 7.6 �m for the unannealed sample to 10.1 �m forthe sample annealed at 2100 ◦C. The observation made from theseresults is that the average grain size seems to reach an upper limitafter annealing at 2000 ◦C. This however needs to be confirmed withmore samples to increase the statistical confidence. Further exami-nation reveals that the shape of the grains also change significantlyfrom elongated in the unannealed sample, to more spherical grains

after annealing as shown in Fig. 4. The high concentration of twinsand stacking faults present in the SiC grains contributes to the dif-ficulty to differentiate between grain boundaries and this is partlyresponsible for the relatively large standard deviations obtained

Page 5

I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202 195

Fig. 4. (a–e) SEM micrographs showing the average grain size increase and shape changes occurred with increased annealing temperature.

Fig. 5. Influence of high temperature annealing on the CVD ORNL polycrystalline 3 C-SiC.

Page 6

196 I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202

olycry

fdtut

tawifliaisguaasf

FDIf

ied the grain sizes across the layer progressively varied from beingsub-micron at the inner-PyC and SiC interface to being severalmicrometers in size at the outer PyC and SiC interface. Fig. 9 shows

Fig. 6. Influence of high temperature annealing on the p

or the measured grain sizes. Another possible reason for the largeeviation in measured grain sizes is most probably due to the facthat the samples were only prepared in one direction (perpendic-lar to the plane of deposition), therefore the shape factor is notaken into account.

The results of grain size determination of the SiC layer of thewo experimental PBMR TRISO CP batches are presented in Fig. 6nd although the results suggest also a slight increase in grain sizeithin experimental error, after high temperature annealing, the

ncrease in grain size is smaller than that reported in Fig. 4 for theat SiC samples. It should be noted however, that the grain size

n the flat SiC samples exhibited a definite distribution of largend small grains which is most likely the reason for the largerncrease in average grain size upon annealing measured in theseamples. From the data shown in Fig. 6 it is clear that the averagerain size of batch D is larger than that of batch E for temperaturesp to 1600 ◦C. For annealing temperatures above 1600 ◦C the aver-

ge grain sizes seems to remain the same. This phenomenon is ingreement with the findings by Van Rooyen et al. (2010) where thetrength measurements of CPs showed also that no significant dif-erence in strength values of these two batches exists at 2000 ◦C.

ig. 7. Micrograph showing a typical etched image of the SiC layer of the CP batch after annealing at 2000 ◦C with an average grain diameter of 3.7 �m (Heyn Lineal

ntercept method). The etching process also accentuated the high density of stackingaults.

stalline 3 C-SiC layer of PBMR TRISO CP batches D and E.

Fig. 7 shows a typical etched structure used for the grain size deter-mination of the SiC layer of the TRISO CP. This figure also illustratesthe difficulty to identify the grain boundaries due to the presenceof high density stacking faults which also are accentuated duringthe etching process.

3.2. Grain size determination using EBSD

Fig. 8 shows an example of an IPF coloured crystal orientationmap of the 3C-SiC structure across the layer of the reference sam-ple of CP batch D. No 6H-SiC areas could be detected during themapping of any of the layers investigated. In all the cases stud-

Fig. 8. An example of an IPF coloured orientation map of the 3C-SiC structure acrossthe layer of the reference sample of CP batch D. The grain sizes across the layerprogressively varied from being sub-micron at the inner-PyC and SiC interface tobeing several micrometers in size at the outer PyC and SiC interface.

Page 7

I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202 197

Fr

as

tiowcd

Fig. 10. This histogram showed the recalculated grain size statistics of the referencesample of CP batch D after removing the grains with a minimum grain size of 0.4 �m

TE

Fo

ig. 9. A histogram of grain size distribution and grain size statistics of the analyzedegion of the reference sample of CP batch D.

histogram of grain size distribution and Table 1 the grain sizetatistics of the analyzed region.

The grain size statistics for D-REF show an average grain diame-er of 0.368 �m, however if the histogram of grain size distributions considered, a large contribution towards the grain size statistics

riginate from grains smaller in size or close to 0.1 �m in diameterhich is comparable to the step size used. Thus these grains

annot be considered for the analysis. Setting a threshold grainiameter value of at least 0.4 �m to be included for the analysis

able 1BSD grain size statistics of the analyzed region of the SiC layer of the CP batches D and E

Measurement D-REF D-2

Average, expectation (EX), �m 0.368 (3.0)a

Variance, dispersion (D2X) 0.340

Standard deviation (s) 0.583

Coefficient of variation (s/EX) 1.587

Minimum value (Xmin), �m 0.113

Maximum value (Xmax), �m 6.502

Size of the data set (N) 1533 27

a Determined ASTM E 112.b Not available.

ig. 11. Batch 361 XRD diffractogram of the 3C-SiC grown on Si(1 0 0). The epilayer is grorientation.

(two adjacent indexed spots in the x and y directions) which would minimize theincorrect statistical evaluation.

and removing all the grains less than this would remove anyuncertainty. Thus choosing a minimum grain size of 0.4 �m (twoadjacent indexed spots in the x and y directions) would minimizethe incorrect statistical evaluation. Removing these grains form

the data set for the case under discussion, the size is being reducedfrom 1533 grains to 355. The recalculated grain size statistics isshown in Table 1 with the re-plotted histogram in Fig. 10.

.

000 ◦C E-REF D-REF re-calculated

0.307 (3.6)a 0.343 (2.7)a 1.1580.276 0.279 b

0.525 0.528 0.9281.710 1.539 b

0.113 0.113 0.40710.466 6.651 6.50286 1564 355

wn heteroepitaxially and only the (h 0 0) peaks are visible since there is a preferred

Page 8

198 I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202

LiU P

3

3

mwm1dd

Fig. 12. Average grain size distribution in the

.3. Characterization of the P-doped SiC samples

.3.1. Chemical analysisThe P-content in selected LiU P-doped prior to annealing was

easured by the Evans Analytical Group (EAG), USA, using SIMSith a detection limit of 1 × 1013 atom/cm3 and these measure-

ents are shown in Table 2. The doping level varied between

.1 × 1015 and 1.2 × 1019 atom/cm3. Sample X377 has the highestoping level at 1.2 × 1019 atom/cm3 and was therefore chosen toiscuss further in this paper.

-doped polycrystalline 3C-SiC sample X364.

The PBMR calculated value for the typical PBMR reactor yieldsa P concentration of 9.8 × 1017 atom/cm3. The P-doping levels ofthe selected samples indicated in Table 2, showed that theseLiU prepared samples are relevant to the PBMR conditions asit gives a representative P concentration spread around typi-cal PBMR conditions. This phosphorous concentration is also in

the same order found by neutron-transmutation studies done byBaranov et al. (2007) for a neutron dose of ∼1 × 1020 cm−2 for30Si enriched SiC. Wellmann (2005) found that P-doping concen-trations of up to 1.3 × 1018 atom/cm3 were achieved and further

Page 9

I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202 199

LiU P

sp1

tfwdan

Fig. 13. Average grain size distribution in the

uggested that much higher doping levels are achievable. The workresented in this paper showed achieved concentrations up to.19 × 1019 atom/cm3.

Hendriks et al. (1982) found that for low doping concentra-ions, the P atoms will be trapped at the grain boundaries, but thator higher doping concentrations only a small fraction of P atoms

ill be trapped at the grain boundaries and therefore only a smallepleted zone next to the grain boundaries are expected. The SIMSnalysis of the LiU samples was done on bulk areas and thereforeo detail profile is available to confirm or reject this statement.

-doped polycrystalline 3C-SiC sample X377.

3.3.2. X-ray diffractionX-ray diffraction measurements were conducted using the

Philips PW1729 diffractometer at the LiU Physics department andthe Philips PW1840 diffractometer at NMMU. The diffractogramsfor all samples showed the 3C-SiC phase grown on Si(1 0 0). A typ-ical diffractogram of the single crystal sample X361 measurement

is shown in the diffractogram (2�-scan) in Fig. 11 and were usedto identify the SiC phase and the orientation of the epilayer. Theepilayer is grown heteroepitaxially and only the (h 0 0) peaks arevisible since there is a preferred orientation.

Page 10

200 I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202

Table 2Bulk SIMS analysis of P in SiC.

Sample no. EAG SIMS P concentration (atom/cm3)

Analysis 1 Analysis 2 Average

X380-C 6.46 × 1018 6.72 × 1018 6.59 × 1018

X377-E 1.20 × 1019 1.18 × 1019 1.19 × 1019

X383-BA 5.23 × 1017 5.45 × 1017 5.34 × 1017

X387-BA 5.36 × 1015 5.35 × 1015 5.36 × 1015

X382-G 1.65 × 1018 1.57 × 1018 1.61 × 1018

X391-B 1.08 × 1015 1.13 × 1015 1.11 × 1015

3

sm

XassaTs

Fsagdtwnrpotiatc

ci

Fpt

used for the TEM analysis. A large amount of TEM micrographs weretaken and evaluated to determine the effect of temperature and P-doping level on the nanostructures of 3C-SiC. Only selected images

X392-B 1.75 × 1015 1.71 × 1015 1.73 × 1015

X393-B 6.68 × 1015 6.74 × 1015 6.71 × 1015

.3.3. Scanning electron microscopic evaluationA XL30 Philips SEM at NMMU was used to investigate the top

urface morphology of the samples prepared at LiU and to deter-ine the grain sizes.The grain sizes of selected samples from X364, X377, X380 and

383 were determined using the Heyn Lineal Intercept method inccordance with ASTM E112 (2004). The SEM micrographs in Fig. 12how globular top surface structure for sample X364 and the grainize measurement shows a large variation in grain size, which prob-bly indicated a large variance in the SiC deposition temperature.he grain sizes of sample X377 showed approximately even averageizes and fairly homogeneous morphologies (Fig. 13).

From the grain size determination schematically presented inig. 14, it is shown that grain sizes of the P-doped polycrystallineamples are fairly evenly sized whereas large grain size variancesre visible for the un-doped sample X364. This fairly homogeneousrain size distribution of the P-doped samples, showed that theeposition temperature were in a narrow range as in contrast withhose of sample X364. During the preparation of sample X364 itas also further observed that the substrate sample was locatedear the input end of the chamber and that the temperature in thategion was most probably higher as reported for this growth. Theresence of molten Si substrate was observed during the removalf the sample which is also an indication of much higher tempera-ure. The yellow optical colour of the top edge of sample X364 alsondicate that the substrate temperature for sub-samples X364ABnd X364AC may have been higher as the recorded 1210 ◦C. Deduc-ions from work by Chin et al. (1977) showed that the yellow optical

olour may indicate a higher substrate temperature.

López-Honorato et al. (2008) indicated that the top surfacehanged from a globular to a faceted structure as stoichiometrymproves. Fig. 15 is included to show the difference between typical

ig. 14. Schematic presentation of the grain size distribution measured on differentositions on the sample. The grain size distribution curve of un-doped polycrys-alline sample X364 is shown for comparison.

Fig. 15. SEM micrographs showing the difference between typical globular andfaceted top surface structures.

globular and faceted top surface structures. The stoichiometry ofthese samples is however not confirmed at time of this study. Theeffect of deposit temperature on these three examples are notas clear as described by Lee et al. (2001) and Chin et al. (1977),but it is important to notice that the doping levels are varyingand therefore the manufacturing conditions are different foreach sample as well. It is recommended that these morphologiestogether with the rest of the 13 LiU samples be investigated inconjunction with the respective manufacturing parameters.

3.3.4. Transmission electron microscopic evaluationA 200 kV Philips CM20 transmission electron microscope was

Fig. 16. Bright field TEM micrograph and SAD pattern of sample X364CA afterannealing at a temperature of 1600 ◦C. The SAD is consistent with the cubic 3C-SiC phase with the beam along the 〈1 1 0〉 direction. The streaks in the SAD are dueto the thin twin platelets visible in the bright-field TEM image.

Page 11

I.J. van Rooyen et al. / Nuclear Engineering and Design 251 (2012) 191– 202 201

Fah

af

2i

6TtP

dt(tPft

Fa1

Fig. 19. HRTEM image (250,000×) of the SiC–graphite interface region of sample

ig. 17. Bright field TEM micrograph and SAD of sample X377I after annealing at temperature of 2000 ◦C. The SAD indicates that a phase transformation to 6H-SiCexagonal phase occurred. The beam is along the 〈1 0 0〉.

re shown in this paper as it represents the typical results obtainedrom the full TEM evaluation.

High resolution TEM images were acquired using a JEOL200MCO aberration corrected (S)TEM operated at 200 kV with an

n-column omega filter (Oxford University).The single crystal LiU 3C-SiC samples (P concentration from 0 to

.71 × 1015 atom/cm3) were annealed at 1600 ◦C and 2000 ◦C. TheEM evaluation of these samples showed that no phase transforma-ion, no nanotubes and no decomposition were observed for these-doping levels up to an annealing temperature of 2000 ◦C.

The annealing of the polycrystalline LiU 3C-SiC samples wereone at 1600 ◦C, 2000 ◦C and 2100 ◦C respectively where phaseransformation from 3C- to 6H-SiC were observed for samples X364un-doped) and X377 (highest P-doped sample) at 2000 ◦C. As both

hese samples showed a phase transformation, it indicates that the-doping level is not a contributing fact towards this phase trans-ormation, but it is temperature dependant. Figs. 16 and 17 showhe bright field (BF) TEM image and corresponding selected area

ig. 18. Bright Field TEM micrograph and SAD of sample X377CB after annealing at temperature of 2100 ◦C. The SAD shows the four rings of the graphite namely 002,01, 004 and 112.

X377CB after annealing at 2100 ◦C.

diffraction (SAD) of the samples X364 and X377 after annealing at1600 ◦C and 2000 ◦C, respectively. Fig. 18 shows the BF TEM imageand SAD of the sample X377CB after annealing at 2100 ◦C. Evidenceof decomposition of SiC is clearly visible. The diffraction ring pat-tern in Fig. 18 shows that the SiC transformed to polycrystallinegraphite. The first four rings of graphite 0 0 2, 1 0 1, 0 0 4 and 1 1 2,are visible. No nanotubes were observed in any of the polycrys-talline LiU samples.

The HRTEM image in Fig. 19 shows the SiC–graphite interfaceregion of sample X377CB after annealing at 2100 ◦C. Randomlyorientated graphite lamella which has the (0 0 2) parallel to theelectron beam, display the typical HRTEM fringe image of the basalplanes. The SiC–graphite interface region and the graphite lamella

are shown in Fig. 20 at a higher magnification.

Fig. 20. HRTEM image of sample X377CB after annealing at a temperature of 2100 ◦Cshowing the basal planes in the graphite lamella at higher magnification (800,000×).

Page 12

2 neerin

4

s1vtti

etStttbs

lat

tna

od

tH

5

pvtqr

ub

•

••

pmeet

02 I.J. van Rooyen et al. / Nuclear Engi

. Conclusions

The average grain size determined on the flat CVD SiC samples,uggests an increase from 7.6 �m for the unannealed sample to0.1 �m for the sample annealed at 2100 ◦C. An interesting obser-ation made from these results is that the average grain size seemso level off after annealing at 2000 ◦C. Further examination revealshat the shape of the grains also change significantly from elongatedn the unannealed sample, to more spherical grains after annealing.

The increase of average grain size of the SiC layer of the twoxperimental PBMR TRISO CP batches with increasing annealingemperature is smaller than that determined for the flat CVD 3C-iC. Although there is a difference in grain size noted between thewo batches after annealing at 1600 ◦C (3.9 and 3.0, respectively),he average grain size of the two batches levels off at tempera-ures above 1600 ◦C. This result is in agreement with the findingsy Van Rooyen et al. (2010) where the compression strength mea-urements of CPs showed this same trend.

The average grain size values from EBSD are approximately 10×ower than those measured by the Heyn Lineal Intercept methodnd this is most likely a consequence of the different approxima-ions used for the quantification schemes.

The preliminary conclusion reached is that the P-content athese experimental levels (1.1 × 1015 to 1.2 × 1019 atom/cm3) doesot have a significant influence on the nanostructure of SiC afternnealing at high temperatures without irradiation.

Although phase transformations from 3C- to 6H-SiC werebserved after annealing at 2000 ◦C, it is not attributed to the P-oping concentrations.

Decomposition of the SiC is only observed at 2100 ◦C annealingemperature and no nanotubes were identified during the TEM andRTEM analysis although graphitic structures were identified.

. Recommendations

The observed grain size increase with increasing annealing tem-erature for both the flat SiC and SiC layer of the CPs needs to beerified with a larger statistical sample. Furthermore, it is impor-ant to include the shape factor in these determinations by doing aualification measurement test on different oriented sections withegards to the growth direction.

As no grain size specification for SiC layers for TRISO CPs is doc-mented, it is recommended that a statistical sample of CPs peratch needs to be investigated with the following in mind:

The grain size of actual CPs are investigated with specific ref-erence to long duration (>100 h) annealing as a function oftemperature to relate the grain size modification with specificreactor operational requirements.The grain size and strength correlation be determined.The influence of grain size on the Ag transport rate be determined.

It is also recommended that the EBSD results be further inter-reted with regards to the grain boundary orientation. The EBSD

ap could also give valuable information with regards to the ori-

ntation with respect to the growth direction and needs to bevaluated further to do comparisons between the two batches andhe influence of this orientation on the strength of the CP.

g and Design 251 (2012) 191– 202

It is also recommended that the morphologies and stoichiome-try of the three samples mentioned in this paper together with therest of the 13 LiU samples be investigated in conjunction with therespective manufacturing conditions and influence of the P-dopingon the resistivity of SiC. It is also recommended that the influence ofP-doping level on the heat conductivity and mechanical propertiesof SiC be determined.

Acknowledgements

This research was sponsored by PBMR’s Fuel OptimizationTechnology Programme. The use of the NMMU and PBMR FuelDevelopment Laboratory facilities are gratefully acknowledged.Johannes Mahlangu (PBMR), Ellen Nquma (PBMR), Jaco Olivier(NMMU) and Jacques O’ Connell (NMMU) are thanked for theannealing operations. The HRTEM images were produced by Dr.Sarah Haigh at Oxford University.

References

ASTM E112-96, 2004. Standard Test Methods for Determining Average Grain Size.Baranov, P.G., Ya, B., Ber, B., Ilyin, I.V., Ionov, A.N., Mokhov, E.N., Muzafarova,

M.V., Kalieevskii, P.S., Kopév, M.A., Kalieevskii, A.K.K., Godisov, O.N., Lazebnik,I.M., 2007. Pecularities of neutron-transmutation phosphorous doping of 30Sienriched SiC crystals: electron paramagnetic resonance study. J. Appl. Phys. 102,063713.

Byun, T.S., Hunn, J.D., Miller, J.H., Snead, L.L., Kim, J.W., 2009. Evaluation of fracturestress for the SiC layer of TRISO-coated fuel particles using a modified crush testmethod. Int. J. Appl. Ceram. Technol., doi:10.1111/j. 1744-7402.2009.02462.x.

Chin, J., Gantzel, P.K., Hudson, R.G., 1977. The structure of chemical vapor depositedsilicon carbide. Thin Solid Films 40, 57–72.

Heinisch, H.L., Greenwood, L.R., Weber, W.J., Williford, R.E., 2004. Displacementdamage in silicon carbide irradiated in fission reactors. J. Nucl. Mater. 327,175–181.

Helary, D., Bourrat, X., Dugne, O., Maveyraud, G., Perez, M., Guillermier, P., 2004.Second International Topical Meeting on High Temperature Reactor Technology, Beijing, China, September 22–24, Paper No. B07.

Hendriks, M., Radelaar, S., De Keijser, T.H., Delhez, R., 1982. Morphology andresistivity of CVD polycrystalline silicon layers containing carbon. EDP Sci.,http://dx.doi.org/10.1051/jphyscol:1982141.

Henry, A., Janzén, E., 2005. Epitaxial growth and characterization of phosphorousdoped SiC using TBP as precursor. Mater. Sci. Forum 483–485, 101–104.

Lee, K.S., Park, J.Y., Kim, W., Hong, G.W., 2001. Effect of microstructure of SiC layeron the indentation properties of silicon carbide-graphite system fabricated byLPCVD method. J. Mater. Sci. Lett. 20, 1229–1231.

López-Honorato, E., Meadows, P.J., Tan, J., Xiao, P., 2008. Control of stoichiometry,microstructure, and mechanical properties in SiC coatings produced by fluidizedbed chemical vapor deposition. J. Mater. Res. 23 (6).

Maage, S., 2009. Inventory products of SiC layer under PBMR conditions. PBMRDesign Information Transmittal DIT001316.

Nabielek, H., Brown, P.E., 1975. The release of silver-110m in high temperaturereactors: technical note. In: OECD Dragon Project, vol. 657, p. 370.

Nabielek, H., Brown, P.E., Offerman, P., 1977. Silver release from coated particle fuel.Nucl. Technol. 35, 483.

Petti, et al., 2003. Key differences in the fabrication, irradiation and high temper-ature accident testing of US and German TRISO-coated particle fuel, and theirimplications on fuel performance. Nucl. Eng. Des. 222, 281–297.

Snead, L.L., Nozawa, T., Katoh, Y., Byun, T.S., Kondo, S., Petti, D.A., 2007. Handbook ofSiC properties for fuel performance modelling. J. Nucl. Mater. 371, 329–377.

Tan, L., Allen, T.R., Hunn, J.D., Miller, J.H., 2008. EBSD for microstructure and propertycharacterization of the SiC-coating in TRISO fuel particles. J. Nucl. Mater. 372,400–404.

Van Rooyen, I.J., Neethling, J.H., Van Rooyen, P.M., 2010. The influence of anneal-ing temperature on the strength of TRISO coated particles. J. Nucl. Mater. 402,136–146.

Wellmann, P.J., 2005. Additional pipework opens up transistor applications for

SiC, http://compoundsemiconductor.net/articles/magazine/11/3/3/1 (accessiondate 2007/01/17).

Yurko, J.P., Sorensen, K.M., Kadak, A., Yan, X.L., 2008. Effect of helium injection onnatural circulation onset time in a simulated pebble bed reactor. Paper 58230.In: Proc. HTR 2008, Washington, DC, USA, 28 September–1 October.