Applications of Surface Analytical Techniques in Corrosion Research (Mainly High Temperature Corrosion) Uporaba površinskih analiznih tehnik v raziskavah korozije H.Viefhaus1, Max-Planck-lnstitut, Dusseldorf, Germany Prejem rokopisa - received: 1996-10-01; sprejem za objavo - accepted for publication: 1996-11-04 The application of materials under various ertvironmental conditions strongly depends on the corrosion properties of those materials. This is in particularly true for high temperature materials to be used in povver plants, petrochemical, chemical and automobil industry, where during application high temperatures and aggressive environments may cause great problems. At lovv temperatures the corrosion of metals is often inhibited by a passive layer on the metal surface. To understand the phenomenon of passivity the formation and nature of this surface film, quite often only a few nm's thick, has to be characterized. Corrosion protecting layers on high temperature materials, either grown during application or precovered before application, have a much larger thickness. In order to study the growth mechanisms, nature and properties of corrosion protecting layers thin films have to be characerized spreading over a quite large range of thicknesses, from a few nm's for the passive layers up to several nm's for the protecting layers on high temperature materials. Different methods for thin film analysis using surface analytical methods will be presented and illustrated by examples from different areas of corrosion research. Key vvords: corrosion, high temperature corrosion, surface analytical techniques Uporaba materialov v različnih okoljih zavisi od njihovih korozijskih lastnosti. To je posebno pomembno pri uporabi materialov pri visokih temperaturah in agresivnih medijih npr. v elektrarnah, petrokemijski, kemijski in avtomobilski industriji, kjer lahko pride do hudih industrijskih havarij. Pri nizkih temperaturah se na površini kovin tvori tanka pasivna plast, ki zavira korozijo. Razumevanje pojava nastanka in narave tanke pasivne plasti, ponavadi debele le nekaj nanometrov je mogoče samo s karakterizacijo teh plasti. Protikorozijske zaščitne plasti materialov, ki se uporabljajo pri visokih temperaturah in nastajajo med samo uporabo ali pa so bile predhodno nanesene, so debelejše. Študij mehanizma rasti, narave in lastnosti protikorozijskih prevlek je mogoč z raziskavami protikorozijskih prevlek. Te so različnih debelin, od nekaj nanometrov debelih pasivnih tankih plasti do nekaj mikronov debelih prevlek za zaščito na visokih temperaturah. Za analizo protikorozijskih plasti se uporabljajo različne metode površinske analize, ki so prikazane v članku, kakor tudi primeri z različnih področij korozije. Ključne besede: korozija, visokotemperaturna korozija, metode površinske analitike 1 Introduction A metal is normally described as beeing passive, if for the existing surrounding atmosphere a high corrosion rate would be expected, instead of the very lovv corrosion rate to be observed. This passivity is caused by very thin dense oxide (and/or hydroxide) layers vvhich are formed on the metal by the corrosion process. In order to get a better understanding on the effect of those passive layers they have to be analysed vvith respect to composition and thickness by very surface sensitive methods. High temperature oxidation and corrosion can cause great problems in povver plants, petrochemical and chemical indus-try. A very important precondition for the practical application of a metalic material at high temperature is its oxidation or high temperature corrosion resistance. This precondition may be fulfilled if on the surface of the material protecting oxide layers are formed. These oxide layers can grovv under use of the material by reaction of the material elements vvith the surrounding oxy-gen atmosphere or by a specific preoxidation at suitable temperatures in oxygen containing atmospheres. The protecting effect of those oxide layers relys on their property to act as a diffusion barrier betvveen the metallic ' Dr. Sc. H. VIEFHAUS Max-Planck-Inslilul fiir Eisenforschung GmbH 40074 Dusseldorf. Postfach 140 444 Germany and the corrosive atmosphere surrounding it. Oxyde lay-ers therefore have a key function for the application of materials in high temperature technology and there is a great need for doing research and testing the materials for such applications. 2 Methods In order to studv the grovvth mechanisms, nature and properties of corrosion protecting layers thin films have to be analysed spreading over a large range of thicknesses, from a fevv nm's for the passive layers up to several tenth of m(i's for the protecting Iayers on high temperature materials. To analyse thin films vvith respect to layer composition and thickness different depth profiling methods can be applied depending on the thickness of the layer under study and on various sample preparation methods. Fol-lovving the mainly applied surface analytical methods to be used for depth profiling of homogeneous and inhomogeneous surface layers are listed. Depthprofiling homogeneous layers inhomogeneous layers AES AES XPS S IMS SIMS SNMS GDOES AES is Auger electron spectroscopy, XPS is X-ray excited photoelectron spectroscopy, SIMS is secondary ion mass spectroscopy, SNMS is secondary neutrals mass spectroscopy and GDOES is glow discharge optical emission spectroscopy. To analyse inhomogeneous surface layers laterally re-solving methods like AES and SIMS have to be applied. For this paper a restriction to the electron spectroscopic methods will be carried out. More detailed information on the application of the remaining methods may be found elsevvhere1. To get a depth profile in most cases destruction of the layer to be analysed has to be performed by one of the follovving methods: a) Sputter depth profiling b) Angle lapping c) Crater edge profiling d) bali cratering For sputter depth profilling the layer is decomposed by bombardement with noble gas ions and parallel or successive analysis of the momentary interface by a surface analytical method is carried out. Angle lapping means that by using normal metal pol-ishing equipment a layer covered sample is polished un-der a very flat angle. Using polishing angles down to about 10 a spreading of the surface layer up to a factor of about 100 is possible and this spreaded part of the layer may be analysed after transfer into a surface analytical system by AES point or line analysis for example. For crater edge profiling the crater edge resulting from ion beam etching accompanying a single sputter depth profile is exploited. The crater edge of such a profile exposes the strata of the interface in a manner related to that produced by angle lapping, but the striking differ-ence is that for crater edge profiling the resulting angles between surface and interface are 3 orders of magnitude less. In the čase of bali cratering a rotating, spherical, steel bali coated vvith fine diamond paste, is used to grind a spherical crater into the sample surface and after that the sample is transferred into the surface analytical system to analyse the sputter cleaned crater vvalls. Advantages and disadvantages of the different methods are discussed in some detail in2. 3 Results a) Very thin surface layers (passive layers) The only nondestructive method to depth profile very thin surface films is by angle dependent XPS - or AES -measurements. The first example to be presented con-cerns an oxide layer vvhich vvas formed at room temperature in air on a pure zine sample. The thickness of this layer vvas assumed to be only a fevv nm's and it should be less than the information depth of the applied electron spectroscopic methods AES or XPS. The next figure 1 compares the Zn - LMM Auger spectrum of a clean zine sample (sputter cleaned by Ar+ ion bombardment) vvith the same sample after oxidation at room temperature in air. The comparison makes clear that for the oxidized sample additional features can be recognized if the analyser energy resolution is adequate (0.05%). XPS studies on the same sample lead to the results that also for the Zn 3d- and Zn 2p- photoelectron signals a distinetion betvveen metal and oxide is possible, but the difference is most pronounced for the Zn - LMM Auger signal (in this čase X-ray excited), figure 2. If the information depth for the Zn - LMM Auger signal corresponding to the oxidized state of Zn is less than the thickness of the oxide layer, angle dependent measurements should reveal a more pronounced oxide signal at lovver angles of analysis. This is illustrated by figure 3. The results of an angle dependent measurement on the X-ray excited Zn - LMM Auger signal are depieted in figure 4 and clearly demonstrate, for angles of analysis ranging from 25° to 90°, that the information deepth is less than the thickness of the oxide layer. For a smooth, homogeneous, contamination free thin oxide layer the measured signals of the metallic and ox-ide components may be used to determine the oxide thickness according to equation (1): L = Nni \m exp[-d / \m sinft] 1„* No* 1—exp[-d / sinp] (D vvhere lm, lox are the intensities of the metal and oxide signals, Nm, Nox are the densities of the metal and the oxide, Xm, \ox are the inelastic mean free pathes of the Auger electrons in the metal and in the oxide, (3 is the angle of analysis in relation to the sample surface, see figure 3. 9B0 990 1000 Kinetic Energy / eV 1010 1020 Figure I: Comparison of the Zn - LMM - Auger spectrum for a clean zine sample and after oxidation at room temperature in air for 2 vveeks Slika 1: Primerjava AES spektrov čistega Zn - LMM pred in po oksidaciji dva tedna na zraku pri sobni temperaturi The (2): above equation (1) may be changed to equation sin(3 N, N,„ k, ■X? (2) zine zine oxide zine hydroxide ^LMM 17.7 20 21 6.7 8 7 According to equation (2) a plot of 1 /sin(3 against the term on the right hand side of equation (2) should give a straight line and from the slope of this straight line the thickness may be derived. angle of analysis /\ surface layer bulk information depth I 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 10J binding energy Figure 2: Zn - 3d-. Zn - 2p - photoelectron and Zn - LMM - Auger -signal for clean and oxidized zine sample (see figure 1) Slika 2: Zn - 3d, Zn - 2p XPS signal in Zn - LMM AES signal za čisti in oksidirani cink (glej sliko 1) For a determination of the layer thickness values for the A,'s are necessary. There are no experimental data for the X - values of zine, therefore the corresponding X's of zine and zine oxide (of zine hydroxide as well, which will be needed lateron) were calculated according to a relation derived by Tanuma et al.3. The calculated values are (in A): Table 1: The calculated X values (in A) Figure 3: Shematic illustration of the higher surface sensitivity at lower angles of analysis Slika 3: Shematični prikaz višje površinske analizne občutljivosti pri nizkih kotih The angle dependence of the peak areas of the Zn -LMM Auger signal and the Zn - 2p photo electron signal are plotted in figure 5 for the metallic and the oxide con-tributions. From this plot and by using additionally the known values for Nm, Nox and X0* figure 6 can be derived. The observed straight line for the dependence of the Zn - LMM Auger signal leads to a thickness of 15A for the oxide. If we have a closer look to the well resolved oxygen O - ls photo electron signal corresponding to the oxide layer, figure 7, we can detect that the assumption of a pure oxide layer was not correct, additionally to the ox-ide signal centered at about 530 eV peak energy a second signal caused by some hydroxide contribution is observed around 532 eV peak energy. Results of angle de-pendent measurements on the oxygen O - ls signal in figure 8 indicate that we have an inner oxide layer and an outer hydroxide layer. - Zn-LMM metal i - oxide Analysenwlnkel - 24,5' 33.5' - - 42.5' 51.5' metal / 60.5- 82.5' _ 90.0' - _3-m 250 255 260 265 270 Binding Energy / eV 275 280 Figure 4: Angle dependence of the Zn - LMM Auger signal for an oxidized zine sample (as for figure 1) Slika 4: Kotna odvisnost Zn - LMM AES signala za oksidiran cink (kot na sliki 1) metal contribution in% 100 60 40 20 oo°o Zn - 2Po, ,oo O O o o o o o o o O Zn - 2pm„, o Zn - Auger O* Zn - Auger mat °o o o o OOo + + 70 50 30 angle of analysis 10 L L '-ZnOH [l-exp(dZn0H/XZn0H cos©)] / / XZn cZn exp(- A,Zn0 cos© ) exp(- I cos© (3) + o O CvJ t— _c CD CO 22 20 18- 16- 14- 12 10 - d„ 15A Zn - LMM 1,0 1,2 —I-'-1— 1,4 1,6 1 / sirili 2,0 Figure 5: Angle dependence of the Zn - LMM Auger signal and Zn -2p photoelectron signal for an oxidized zine sample (as for figure 1), metal and oxide contributions are plotted Slika 5: Kotna odvisnost Zn - LMM AES signala in Zn - 2p XPS signal za oksidiran cink (kot na sliki 1), prikazana sta deleža kovine in oksida Looking more detailed to the angle dependence of the Zn - LMM Auger signal it can be recognized, figure 9, that also for this signal the "oxide" - component ex-hibits some peak shape variations in dependence on the angle of analysis. If we try to fit the Zn - LMM Auger signal for the oxidized sample at first, figure lOb, by a metal contribution according to figure lOa and the re-maining peak area for the oxide contribution by a single peak this is not possible. We need two different peaks as shown in figure lOc, to get an acceptable fit of the whole Zn - LMM spectrum. We can novv try evaluate the thickness of the oxide and the hydroxide separately. Again we assume that the oxide and hydroxide layer are homogeneous layers. The intensity ratios of the sig-nals of the different layers with respect to the signal of the zine substrate are given according to S. Lecuyer et al.4 by: Figure 6: Determination of the oxide thickness according to equation (2), see text Slika 6: Določitev debeline oksidne plasti po enačbi (2), glej tekst MPI,DUSSELDOHF XPS - Spectrum ZNC148.DAT Region i / 1 Level 1 / 1 Rad iat ion Mg Kalpna Max Count nate 3637 CPS Ana lyser 5 eV Step Size 0.05 cV Dwell Time 100 ms No of Channels 401 No of Scans 30 Time for Region 1203 Sec Acquired 14: 32 08-Jan-96 Plotted 59 OS-Jan-96 532 534 Binding Energy / eV 538 Figure 7: XPS - O - ls signal for an oxidized zine sample (as for figure 1) Slika 7: XPS - O - ls signal za oksidiran vzorec cinka (kot na sliki 1) T2 (®) = ^no CZn0 [ l-exp(- , dzn" J] / I cz„ exp(- ^ZnO COS0 dZnO ^ZnO COS0 (4) where Ci is the concentration within layer i and dj is the layer thickness. For ali the angle dependent measurements of the Zn -LMM Auger signal recorded for the oxidized zine sam- O - 1s - Signal Hydroxid 528 532 Binding Energy eV 536 250 35 30 k 25 C 0 20 u n ib t 10 s 5 0 Zn LMM "A 20° - Me A Ok OH \ - 75° 256 258 260 262 264 266 268 Binding Energy / eV 270 272 274 Figure 9: Zn - LMM Auger signal for an oxidized zine sample (as for figure 1) recorded at 20° and 75° angle of analysis Slika 9: Zn - LMM AES signal oksidiranega cinka (kot na sliki 1) posnet pri kotih analize 20° in 75° ple a peak fitting for the total LMM signal was per-formed in a similar way as discussed before. The measured peak area ratios lox/lme and loH/lme for the oxide and the hydroxide layer in dependence on the angle of analy-sis ZN1 = 0 are listed in the table 2 together with the same peak area ratios calculated according to equation (3) and (4) by using a thickness of 13A for the oxide layer and a thickness of 4A for the hydroxide layer. if 200 C o 150 Figure 8: Angle dependent measurements for the XPS - O - ls signal of an oxidized zine sample (as for figure 1) Slika 8: Meritve XPS - O - ls signala za oksidiran vzorec cinka (kot na sliki 1) v odvisnosti od kota 100 11 t 50 S Q k 10 G 8 0 6 u n 4 f 2 s oJ k 10 C 8 0 6 u n 4 t 2 s 0 /\y metal signal metal signal fitted \ a / \ z // / metal + oxide signal metal signal fitted --b '-■"ST" • -1---- ' Zn LMM metal + oxide signal metal,oxide and /a\ hydroxide fitted /A\ Ox \\ 1 OH M e j \\ --J C Me I// \ i // y - 256 258 260 262 264 266 268 Binding Energy / eV Figure 10: a) Zn - LMM Auger signal for a clean zine sample, metal signal fitted; b) Zn - LMM Auger signal for an oxidized zine sample (as for figure 1), metal part fitted; c) Zn - LMM Auger signal for an oxidized zine sample (as for figure 1), metal oxide and hydroxide parts fitted Slika 10: a) Zn - LMM AES signal za čisti cink; b) Zn - LMM AES signal za oksidiran cink (kot na sliki 1) kovinski del se prilega; c) Zn -LMM AES signal za oksidiran cink (kot na sliki 1) kovinski in hidroksidni del se prilega krivulji Table 2: The angle dependent measurements of the Zn - LMM Auger signal for the oxide and hydroxide layer © lox/lme lox/lmc l0H/lme l0H/lme (exp.) (calc.) (exp.) (calc.) 78.1 13 14.7 11.7 10.5 76 9 9.05 5.8 4.1 73 5.6 5.44 2.8 2.2 70.5 3.5 3.9 1.8 1.65 68 3.15 3.1 1.25 1.05 64 2.3 2.25 0.8 0.7 60.5 1.9 1.8 0.6 0.54 58 1.5 1.6 0.5 0.42 45.4 1 1 0.25 0.28 32.9 0.79 0.77 0.18 0.18 21.4 0.65 0.67 0.15 0.17 13 0.66 0.63 0.14 0.14 For the first thickness determination, assuming a single oxide layer, a thickness of 15A resulted, which is in very good agreement with a thickness of 17A for the double layer consisting of a 13A thick oxide and a 4A thick hydroxide layer. b) Thick layers (high temperature oxidation) For future power plants high strength 9% Cr steels are being considered as construction materials for steam piping, headers and superheater tubes up to 600°C. For those materials it vvas found that the destruction of the protective oxide scale occurs during eposure in simulated combustion gas by the presence of vvater vapour. Al-though this detrimental effect of vvater vapour on the oxi-dation resistance of ferritic Cr - steels is knovvn already for a long time no conclusive mechanism has been eluci-dated. According to figure 11, illustrating schematically the variation vvith alloy Cr content of the oxidation rate and oxide scale structure, several different oxides are ex-pected to appear for a Fe 9% Cr alloy and the above mentioned oxidation conditions. The Fe 9% Cr alloys vvere oxidized isothermally in N2 - 1% O2 vvith and vvithout various H2O contents at 650°C. Using the crater bali equipment a crater vvas ground into the different oxidized samples. The crater bali process is illustrated in figure 12. After transferring Alloy chromium content, wt% Figure 11: Schematics of the variation vvith alloy chromium content of the oxidation rate and oxide scale structure (based on isothermal studies at 1000°C in 0.13 atm oxygen) Slika 11: Shematičen prikaz vpliva različnih vsebnosti kroma na stopnjo oksidacije in struktura oksida (izotermna oksidacija pri 1000°C in 0.13 atm kisika) the samples into the scanning Auger system the samples vvere sputter cleaned and SEM images recorded. Tvvo ex-amples vvill be presented to shovv the possibilities of this method. The first sample vvas oxidized in N2 - 1% O2 -4% H20 at 650°C for 3 hours and the second one for 10 hours, figure 13 to 16. For the application of high temperature materials the formation of even, slovv grovving and well adherend ox-ide layers are desired. AI2O3 layers vvould be the thermo-dynamically most stable layers for nearly ali conditions occuring during the use of the material. Quite often hovvever the nucleation and adherence of the AI2O3 layers are not satisfactory. The improve the adherence of those lay-ers oxidation of a model alloy Fe - 6% Al - 0,5% Ti and 50 to 100 ppm C vvere studied. After oxidation of this alloy the samples vvere inves-tigated by surface analytical methods. The oxide layers vvere fine grained, vvell adherent and represented an ex-cellent protection against carburizing atmospheres. This vvas tested by long term investigations. In order to find out the reason for this improvement of the protecting properties Auger depth profiles of the oxide layer on top of the model alloy vvere recorded. A typical example is shovvn in figure 17. The most striking feature of this depth profile is the simultaneous enrich-ment of carbon and titanium at the oxide metal matrix interface. This observation leads to the assumption that by formation of a TiC layer at the interface this improvement of the corrosion protecting properties could be real-ized. By further detailed surface analytical investigations on a single crystal model alloy of the same composition and applying LEED (lovv energy electron diffraction) and AES it could be found out that on several lovv in-dexed crystal surfaces this TiC layer grovvs epitaxially. d =--------- 8R Figure 12: Scheinatic illustration of the crater bali etching process Slika 12: Shematski prikaz kraterja dobljenega s procesom jedkanja s kroglico 400 500 600 700 400 500 600 700 Figure 13: SEM of a part of the crater etched area of the Fe 9% Cr sample oxidized for 3 h (see text) and characteristic Auger point spectra for the different areas Slika 13: SEM posnetek dela jedkalnega kraterja vzorca zlitine Fe 9% Cr, po 3 urah oksidacije (glej tekst) in karakteristični AES spektri, posneti na označenih mestih Metal dusting is known to be a dangerous high temperature corrosion phenomenon in petrochemistry and in reformer and direct reduction plants. In strongly car-burizing atmospheres and temperatures from 400 to 800°C Iow alloyed Fe, Ni and Co base alloys are subject to a catastrophic carburization leading into a desintegra-tion of the material into a dust composed of fine metal particles and carbon. For the reaction mechanism of the metal dusting process it was assumed Ihat instable car-bides form as an intermediate before they decompose to metal and carbon dust. In order to get a more detailed picture of the metal dusting process an iron sample from the initial stages of the metal dusting process (680°C, 78% H2, 15% CO and 0.5% H2O) vvas removed from the carburizing atmos- Figure 14: a) oxygen; b) chromium; c) and iron - images of the same surface area as shovvn in the SEM image in figure 13 Slika 14: a) kisik; b) krom; c) in železo - slike delov površin prikazanih na SEM posnetku slike 13 phere and using the crater bali equipment a crater vvas etched into the sample surface. After transferring the etched sample into the scanning Auger system a clean surface of the crater area vvas produced by Ar ion bom-bardement. Immediately aftervvards Auger spectra and Auger images vvere recorded. The different chemical states of carbon vvithin graphite and in carbide may eas-ily be distinguished by Auger electron spectroscopy be-cause of a characteristic Auger signal peak shape for the individual compounds. This is demonstrated by the Auger spectra in figure 18, vvhich vvere recorded for different areas of the etched crater. Because of the different peak shape and a slight difference in peak energy graphite and carbide may be imaged separately. This is illus-trated by the follovving figures, figure 19a to figure 19d, vvhich shovv additional to a SEM image Auger elemental maps for carbidic carbon, carbon in graphitic form and iron of the crater region. 400 500 600 "00 400 500 600 700 Figure 15: SEM of a part of the crater etched area of the Fe 9% Cr sample oxidized for 10 h (see text) and characteristic Auger point spectra for the different areas Slika 15: SEM posnetek dela jedkalnega kraterja zlitine Fe 9% Cr, po 10 urah oksidacije (glej tekst) in karakteristični AES spektri, posneti na označenih mestih Figure 16: a) oxygen; b) chromium; c) and iron - images of the same surface area as shown in the SEM image in figure 15 Slika 16: a) kisik; b) krom; c) in železo - slike istih delov površin prikazanih na SEM posnetku slike 15 The sum up the results from different kinds of inves-tigations the following development of surface layers during the metal dusting process can be derived (sche-matically): surface surface surface surface metal (Fe) +C diss. Fe3C graphite coke(Fe+C) metal(Fe) +C diss. Fe3C graphite metal(Fe) +C diss. Fe3C metal (Fe) +C diss. Well adherent and corrosion protecting oxide layers are of great importance for high temperature alloys. The adherence of the oxide layers is affected by the morphol-ogy and the chemical composition of the oxide/metal in-terface. In this study scanning Auger microscopy (SAM) is used to investigate the oxide/alloy interface of oxi-dized Fe-Cr-Al alloys (undoped or doped with Ti, Ce and C'i 20)' C (2) 0 • , . A A AAA AA A AA • . h 4» ■ A ' • • "i." . •'fe "' . Al A ■ * -1-,-T^-1-1- , ~ 20 40 60 80 100 120 KO Sputlertime (min) Figure 17: AES depth profile of a polycrystalline Fe-6Al-0,5Ti-0,01C sample oxidized for h at 1000°C and 10"19 bar oxygen partial pressure Slika 17: AES profilni diagram polikristalne zlitine Fe-6Al-0,5Ti-0,01 C po 1/2 urni oksidaciji na temperaturi 1000°C in parcialnemu tlaku kisika 10'19 bar HPI DUSSELD0RF Auger - Spectrum V.G.Scientific CLU100.DAT Region / 1 Level 1 / 1 Point i / 7 92 p 1 graph K 90 r ea ■-'graph u \ n 84 ^ 80 - 240 250 260 270 280 290 Kinetic Energy / eV HPI DUSSELDORF Auger - Spectrum V.G.Scientific CLU100.0AT Region / i Level 1 / 1 Point i / 7 Rad lat Ion Elactron Man Count Rata 1373161 CPS CRfl 9tap Sira 0.70 aV Dwall Tina 92 K 90 C 88 0 86 Bcarb/ D cart) 2015 No of Scana n 84 Tl«a for Raglon 302 Sac Acgulrad 11:31 04-Aug-94 Plottad 07: 42 08-Aug-94 t s02 80 250 260 270 280 Kinetic Energy / eV Radlatlon Elactron M«x Couot Rate 1373161 CPS CRR Stap Sira 0.70 aV D»all T1m of Channala S of Scana TlM for Raglon 302 Sac Acgulrad 11:31 04-Aug-94 Plot tad 07: 42 08-Aug-94 Figure 18: 'graphitie' (upper spectrum) and 'carbidic' carbon - KVV Auger signal of a crater bali etched Fe metal dusting sample indicating the different peak shapes and illustrating the peak (P,) and background (Bi) energy positions for recording the Auger images Slika 18: 'grafitni' (zgornji spekter) in 'karbidni ogljik' - KVV Augerjev signal s kroglico jedkanega Fe prašnatega vzorca z različnimi oblikami vrhov prikazuje vrh (Pj) in ozadje (Bi) in energijske pozicije za posnete AES spektre Figure 19: a) to d) SEM and SAM images of a crater bali etched and sputter cleaned Fe metal dusting sample Slika 19: a) do d) SEM in SAM posnetki s kroglico jedkanega Fe prašnatega vzorca in z Ar+ ioni jedkan Fe prašnat vzorec Figure 20: a) to c) SEM images of surface areas where the oxide layer is partly (or completely) removed Slika 20: a) do c) SEM posnetki površine, kjer je bila oksidna plast delno (ali popolnoma) odstranjena Figure 21: Sulfur image of the same surface area as shown in the SEM image of figure 20 c) Slika 21: Posnetek žvepla na površini, prikazani na sliki 20 c) 200 400 600 800 1000 1200 1400 Elektronen-Energie [eV] Figure 22: Auger point spectra of a) a void area; b) a rugged surface area Slika 22: AES spekter posnet a) v vrzeli; b) na hrapavi površini Y) after partly removing the oxide layer by in situ bend-ing. Thin Fe-Cr-Al ribbons, doped and undoped, were produced by meltspinning. Rectangular specimens were cut from the ribbons and ultrasonically cleaned in ace-tone. The samples were oxidized at 1273 K in a control- 200 400 600 800 1000 1200 1400 Elektronen-Energie [eV] led He - O2 gas mixture at an oxygen partial pressure of 133 mbar. Bending of the specimens was performed in UHV at a residual pressure of 5 x 10~9 Pa to spali off parts of the oxide layer. The stripped oxide/metal interface was in-vestigated by in situ SEM and SAM. The metal surface shows individual voids and rugged parts, indicating imprints of the removed oxide, see SEM figures 20a to 20c. For the undoped alloy poor adher-ence of the oxide layer is observed. Sulphur is strongly enriched at the surface of the voids, figure 21 shows a sulphur image of the sample area as for the SEM image in figure 20c and 22a and 22b Auger point spectra of a void surface and a rugged part of the interface. On the Ti containing alloys the oxide layer is again poorly adherent. Sulphur is also strongly enriched at the surface of voids. On the Y- and Ce- containing alloys the oxide layer is vvell adherent and the sulphur concentration is belovv the detection limit. The poor adherence of the oxide layers on undoped and Ti doped Fe-Cr-Al alloys is correlated to the pres-ence of sulphur at the alloy surface. Sulphur enrichment is explained by sulphur segregation to the free alloy surface and by additional sulphide formation for the Ti containing alloys. The positive effect of Y and Ce on the ad- herence of the oxide layers is explained by sulphide pre-cipitation in the bulk and thus preventing sulphur segregation to the free surface of the alloy. 4 Summary Depending on the thickness of the surface layers to be analysed by surface analytical methods vvith respect to composition and thickness, different methods of depth profiling have to be applied. Valuable information may be derived from the results of the surface analytical in-vestigations, leading to a better understanding of the grovvth mechanisms and the corrosion protecting properties of oxide layers on metal surfaces. 5 References 'Guidelines for methods of testing and research in high temperature corrosion, European Federation of Corrosion Publications 14, The Institute of Materials, London, 1995, 189 2 in Practical Surface Analysis, eds. D. Briggs and M. P. Seah, Wiley, Chichester, 1990 3S. Tanuma, C. J. Povvell and D. R. Penn, Surf. Interface Anal., 20a, 1993, 77 4 S. Lecuyer, A. Quemerais and G. Jezequel, Surf. Interface Anal., 18, 1992, 257