A mössbauer investigation of the surface of α-iron-I. Corrosion and passivation in H2O2 solution

A mössbauer investigation of the surface of α-iron-I. Corrosion and passivation in H2O2 solution

Corrosion Science, Vol. 22, No. 9, pp. 831-844, 1982 Printed in Great Britain. 0010-938X/82/070831-14 $03.00/0 Pergamon Press Ltd. A MOSSBAUER INVES...

788KB Sizes 0 Downloads 3 Views

Corrosion Science, Vol. 22, No. 9, pp. 831-844, 1982 Printed in Great Britain.

0010-938X/82/070831-14 $03.00/0 Pergamon Press Ltd.

A MOSSBAUER INVESTIGATION OF THE SURFACE OF ~-IRON--I. CORROSION A N D PASSIVATION IN H20,~ SOLUTION* G. BELOZERSKII,J~ C. BOHM, T. EKDAHL a n d D. L1LJEQUIST University of Stockholm, Department of Physics, Vanadisvfigen 9, 113 46 Stockholm, Sweden.

Abstract--a-iron foils previously exposed to air were studied by DCEMS (Depth Selective Conversion Electron MOssbauer Spectroscopy). The foils were found to be covered by a thin ( ~ 4 rim) passive layer not giving rise to any MOssbauer signal, but detectable by DCEMS. ESCA studies showed that it contained ferric ions and -OOH but no a-iron. The first stages of corrosion of tt-iron in water and the passivating influence of H20,~ were studied by DCEMS. Oxide thicknesses and distributions in the range of 5-t00 mn were determined. At room temperature, corrosion is inhibited at H~O2 concentrations of 0.006 and 16.8 700,while intermediate concentrations lead to the formation of 7-FeOOH. At 85°C the corrosion is not fully inhibited. INTRODUCTION THIS PAPER is an investigation o f the properties o f iron surfaces which have undergone no special treatment, b u t which are able to retain, for a considerable time, a characteristic grey metallic lustre with no visible sign o f corrosion. Alloys belonging to the austenite class, with a high c o n c e n t r a t i o n o f alloying elements, are n o w often substituted for cheap pearlite steels. The wider a p p l i c a t i o n o f the latter is h a m p e r e d due to their high rate o f corrosion. In neutral water (e.g. the m o d e r n developments o f steam a n d a t o m i c energy technology), a particularly i m p o r t a n t p r o b l e m is how to p r o m o t e the f o r m a t i o n o f a protective oxide layer o11 the surface. Peroxide is often used as an oxidizer.I, ~- The p u r p o s e o f the oxidizer is to p r o m o t e the conversion from ferrous to ferric ions a n d the t b r m a t i o n o f i r o n complexes. Changes occurring on the surface o f e-iron when exposed to neutral water have been studied in the present paper. Results o b t a i n e d with peroxide a d d e d in differt:nt concentrations are reported. The similarity o f the chemical properties o f pearlite steels a n d c~-iron enable us to p r o p o s e that i n f o r m a t i o n o b t a i n e d a b o u t the p r o d u c t s (however not the rate) o f the corrosion o f ~-iron in water with peroxide may contribute to the u n d e r s t a n d i n g o f c o r r e s p o n d i n g processes for pearlite. T h e state o f the c~-iron before this corrosion t r e a t m e n t was also examined together with results concerning a thin passive surface layer. F o r the surface investigation o f polycrystalline alloys, E S C A a n d conversion electron M r s s b a u e r s p e c t r o s c o p y are very suitable methods. W i t h E S C A , thin surface *Manuscript received 2 March 1981 ; in amended form 15 February 1982. tPresent address: Leningrad State University, Department of Chemistry, 199164 Leningrad, USSR. 831

832

G. BELOZERSKIIet al.

layers (in the o r d e r o f 2 n m ) a r e i n v e s t i g a t e d f l giving i n f o r m a t i o n on different e l e m e n t s p r e s e n t a n d t h e i r c h e m i c a l state. H o w e v e r , E S C A gives n o direct i n f o r m a t i o n o n p h a s e s t r u c t u r e , a n d in o r d e r to i n v e s t i g a t e t h i c k e r layers s p u t t e r i n g m u s t be used, w h i c h m a y affect the s u r f a c e a n d t h u s i n t r o d u c e s y s t e m a t i c errors. M 6 s s b a u e r s p e c t r o s c o p y d o e s n o t suffer f r o m these d r a w b a c k s , b u t is o n t h e o t h e r h a n d restricted to c e r t a i n e l e m e n t s . F o r t u n a t e l y , i r o n is a m o n g these, m a k i n g M 6 s s b a u e r s p e c t r o s c o p y a p o w e r f u l t o o l for p h a s e s t r u c t u r e analysis o f i r o n oxides. F o r this r e a s o n Depth Selective C o n v e r s i o n Electron M6ssbauer Spectroscopy (DCEMS) m e t h o d o l o g y d e v e l o p e d in ref. 4 - 7 was c h o s e n ; t h e m e t h o d o l o g y is briefly d e s c r i b e d in t h e n e x t section. A p r e l i m i n a r y r e p o r t o f the p r e s e n t w o r k has been g i v e n recently, s DCEMS METHODOLOGY The most common way of performing M6ssbauer spectroscopy is with a transmission geometry, detecting the gamma radiation transmitted through the sample of interest2 Instead of detecting the resonant absorption of gamma, one may detect the conversion and Auger electrons emitted from the atomic shells of the resonantly excited nucleus during its de-excitation. This yields what is known as a conversion electron M6ssbauer spectrum. 1° As the electrons have a limited range in the absorber, this M6ssbauer spectrum derives from a surface layer, the thickness of which is of the order of the electron range. For the M~hssbauer isotope 57-Fe, whose electron yield in this context is dominated by the 7.3 keV K-conversion electrons, this thickness is approx. 200 nm. A further improvement is obtained if emerging electrons with a particular kinetic energy are selected. This is done by means of an electron spectrometer, which is combined with the usual MOssbauer instrumentation. 4 The basic idea is that there is a correlation between the electron energy loss and the depth from which the electron starts; thus, by varying the electron spectrometer setting, it is, roughly speaking, possible to obtain M6ssbauer spectra from selected depths in the absorber. This is the DCEMS method. The development of quantitative DCEMS is based on a knowledge of the conversion electron scattering in the absorber. 6 Figure 1 shows some weight functions, determined for our DCEMS instrumental arrangement by Monte Carlo simulation of electron scattering, 5 including the effect of the electron spectrometer profile. 5,7 The weight function Tv(x) for the electron spectrometer setting V (corresponding, within resolution, to a particular electron energy), is defined as the relative contribution per 57-Fe de-excitation at depth x in the absorber to the conversion electron M6ssbauer

I

/

~

//V

V/

.d

V =5850 volts = 5800 volts

~ [ 5 6 0 0

o

aoo

volts

4o0

600 x, Feb

coo

Iooo

FIG. I. 57-Fe DCEMS weight functions Tv(x) (properly, the 7.3 keV K-conversion electron weight functions), i.e. the probability that a 57-Fe de-excitation at depth x in the absorber contributes to a M6ssbauer spectrum measured at electron spectrometer setting V. V is related to electron energy E roughly as V = 0.795 E. The unit FeA is the mass/area thickness of 1 A of iron.

A MOssbauer investigation of a-iron

833

spectrum measured at setting V. These weight functions are the basic tools in the interpretation of DCEMS spectra. 4-7 Methods of analysis are described in ref. 6-7. For a given sample, Mfissbauer spectra are usually recorded at 2-3 settings. The simultaneous analysis of these spectra gives information on the distribution of various phases (such as u-iron and its various corrosion products) as a function of the depth under the surface within a range ~ 0-100 nm. The analysis begins by standard Lorentzian fitting of the spectra, in order to determine the phase signals, i.e. the M6ssbauer spectral areas of the various phases at the different settings used. The phase signals are used as input data for the DCEMS analysis. 6,7 The final result of the analysis is a model of the phase distribution in the absorber, such that the phase signals calculated from this model (by means of the appropriate weight functions) agree with the phase signals obtained from the Lorentzian fitting. In some cases, namely those where the spectra are strongly dispersed or distorted, due to relaxation, the phase signals of the relaxation part have been estimated by subtraction of background and Lorentzian peaks from the experimental spectrum. The phase distribution model is quantitatively defined by a number of arbitrary parameters p(i), the number of such parameters being less than the number of phase signals. When a single electron spectrometer setting has been used, one is usually restricted to a calculation of the average thickness of the oxide on the sample surface. With two settings and two phases, the model may lbr example be specified as follows: p(1) ~ of the absorber surface is covered by an oxide of thickness p(2). This is a simple model, but it is not meaningful to use a more complicated model unless one has definite information to support it. With three or more settings, more complicated model structures may be used. Typically measurable oxide thicknesses in 57-Fe DCEMS range fi'om several to about 100 rim. The final model (giving the best fit, in z" sense, to the experimental phase signals) is found by computerized variation of the p(i) parameters; the procedure is usually rapid and easy. 7 A greater difficulty may lie in calculating the phase signals accurately from complex experimental spectra, in particular those involving relaxation. In the present analysis it has generally been assumed that the Debye -Waller factors of the variou~ corrosion products are approximately the same as for iron. If the Debye-Waller factor or the number of 57-Fe nuclei per unit volume somewhere in the absorber departs strongly from the expected value, this will be apparent in the analysis by a deviation from the sum rule for the phase signals. ~,7 (What is actually observed is the "effective" 57-Fe density, which is the product of these two factors.) This requires that two or more settings have been used. In particular, one is with D C E M S able to detect the presence of nonresonant regions, i.e. regions giving no M/Sssbauer signal (e.g. through absence of resonant nuclei, or because the Debye-Waller factor is very small). This is, of course, due to the fact that such regions in the absorber, although giving no phase signal of their own, will affect the electron scattering and therefore the phase signals due to the other (resonant) phases. This is applied in the present work. With the present technique it is necessary to use absorbers enriched in the Mtissbauer isotope 57-Fe, if good statistical accuracy is required within short measuring times (about 1 day/absorber). This is however not a principal difficulty. Improved luminosity and detector efficiency should make a more general use of natural (non-enriched) absorbers feasible. E X P E R I M E N T A L RESULTS A N D DISCUSSION

Investigation of the passive layer on ~-h'on The samples were pure iron foils (99.9Y/o) enriched to 9 0 ~ in the M6ssbauer active isotope Fe-57. The foils were cold rolled to a thickness of 25 ~m and thereafter annealed at 900°C for 6 h in hydrogen at atmospheric pressure. They were then allowed to cool to r o o m temperature in the same atmosphere. Afterwards the foils were exposed to air for a period exceeding one month. Depth selective M6ssbauer spectra were obtained at room temperature in conventional constant acceleration mode for a number of different electron spectrometer settings. Before being mounted in the spectrometer, the foils were cleaned in trichloroethylene and redistilled water. The first measurement of a passive layer was made on such an 0~-iron foil, Some typical spectra from this absorber are shown in Fig. 2. There is no trace of any signal but that of ~-iron. The spectral areas (normalized with respect to time of measure-

834

G. BELOZERSKIIet aL

A

c I

-5



I

-1

mm/s

o

~



°

i

1 mn~/~

FIG. 2. Spectra from or-iron prior to the reduction, recorded at the spectrometer setting 5850 V, and measured with two different velocity ranges. The spectrum in (b) shows the two innermost lines (3 and 4) of the a-iron spectrum.

ment) at various settings are shown in Fig. 3 (as circles); these constitute the "resonance profile" of this sample, i.e. the Mrssbauer signal as function of setting V. It may be noted that the relative line intensities in Fig. 2 differ from those encountered in ordinary transmission Mrssbauer spectroscopy. This is due to the surface sensitivity of the present method; i.e. the outermost layers have a magnetic field parallel to the surface, at right angles to the incident gamma radiation. 11 The sample was then reduced in the spectrometer in a hydrogen atmosphere at 2.10 -3 torr and 600°C for 1 h, to remove possible oxide layers. After this, the spectrometer was brought to a vacuum of 0.5.10 -5 torr. M6ssbauer spectra were measured at several different settings, each showing only 0~-iron. They were then analysed with respect to spectral area, giving the resonance profile of the sample after reduction (Fig. 3). It was assumed that the foil was not oxidized during the measurement. The difference between the resonance profiles before and after reduction (Fig. 3)

A M6ssbauer investigation of u-iron

835

I I

v

I

I

z

I

E

o

z I

I

I

5400

I

I

±_

5600 Setting,

I

5800

t

I 6000

votts

FIG. 3. A comparison between the resonance profile of an u-iron foil reduced in hydrogen atmosphere, and the resonance profile of the same foil before reduction. The profile measured before reduction is indicated by open circles.

can be explained as due to the presence o f a surface layer on the non-reduced sample. This layer gives no discernible M6ssbauer signal, but affects the iron M6ssbauer signal due to its effect on the electron scattering. The detailed methodological aspects o f this analysis are described elsewhere, r' They indicate that the average mass/area thickness o f this layer is equivalent to that of 2 . 5 : 5 1 tam o f iron; in a convenient unit ~ we refer to this as 2.5 Fe nm (25 Fe.A). This experiment was repeated with another ~-iron foil, giving the same result. The statement that there is no trace o f any signal but that of ~-iron needs some elaboration. We assume in the following that the Debye-Waller factor is approximately the same for a-iron and the corrosion products. 1~ There is a n u m b e r o f typical iron oxide M6ssbauer spectra, and, consequently, typical positions in the M6ssbauer spectrum where one m a y expect iron oxide signals to appear. The first type o f oxide spectra are those made up o f sextets. In this case, the outermost lines of the oxide spectrum are outside the outermost lines of the c~-iron spectrum. This may, for ferric iron at r o o m temperature be 7 - F e O O H (~,y)-FezO3, or F%O4. One can see from Fig. 2(a) that an upper limit for this signal is in the order of 1 °/ o/ o f the amplitude o f the second or fifth line o f the or-iron. The second type o f oxide spectra are made up o f singlets or doublets. At r o o m

836

G. BELOZERSKIIet

al.

temperature the typical position for ferric iron is a doublet with an isomer shift 6 = 0.36 m m s -1 and a quadrupole splitting ~ = 0.5 -- 1 mm s -1. This is typical for certain other hydroxides, for amorphous oxides and for superparamagnetic particles of ~-FeOOH and ~-Fe2Os. One line of this doublet is approximately in the center of the ~-iron spectrum. The absence of a spectrum of this kind may be concluded with the precision better than 0.5 ~ (Fig. 2b). The third type of spectra are those due to superparamagnetic relaxation. These spectra are difficult to describe in simple terms. For relatively large particles, about 10-20 nm, we have smeared-out spectra with a decreased value of He~r- For smaller particles, down to about 5 nm, the spectrum is more dispersed and it is impossible to discern any lines. For even smaller particles, the spectra collapse to singlets in the case of magnetite or maghaemite (*), or to doublets in the case of haematite and ~-FeOOH. The fourth type of spectra are due to ferrous iron. At room temperature, these give a doublet with an isomer shift of about 1 mm s-1. This doublet is located in the neighbourhood of the line No. 4 of the ~-iron spectrum. From Fig. 2(b) one can see that with a precision of about 0.5 ~ , there are no compounds with ferrous iron present, i.e. there is no significant difference in amplitude or width between line No. 3 and line No. 4. An ESCA investigation of a non-reduced foil showed the presence of ferric ions and OOH, but no trace of ferrous iron. The absence of ~-iron (bulk material) in the ESCA signal shows that the whole surface is covered by a layer which is at least in the order of 4 nm thick 3, i.e. about 20 FeA which is in agreement with the DCEMS measurement. The ESCA measurement shows that the passive layer does not only consist of adsorbed molecules. From DCEMS methodology 5'x2 it is known that for high settings, the outermost 20 FeA of a homogeneous absorber contributes some 1 0 - 2 0 ~ of the total signal. On the basis of the thickness of the observed layer it is proposed that if there is magnetic ordering in the layer, then the spectra should be of the third (superparamagnetic) type. Due to fast relaxation, at least 50 ~ of the spectrum should consist of a singlet or a doublet between the third and the fourth line of c~-iron. As above, any signal with an amplitude of more than 0.5 ~ would be detected in this position (Fig. 2b). It is concluded therefore that the observed surface layer has unusual properties and gives essentially no M6ssbauer signal at room temperature. For ~-iron, the Debye-Waller factor f ~ 0.7. The absence of a measurable M6ssbauer signal from the passive surface layer shows that for this layerf.a, where a is the relative Fe abundance in the layer, is very small compared to the value expected for a bulk oxide. Moreover, since the ESCA Fe signal is comparable with that of OOH, the result indicates that the f value of the layer should be small compared to values typical for bulk. It can be assumed that the layer consists of very hydrous ferric oxide. It may be suggested that it is similar in structure to liquid crystals. The existence of the layer prevents oxidation and corrosion of our foil. *Magnetite is discussed later but it should be kept in mind that in the ease of very fast relaxation magnetite and maghaemite cannot be distinguished from each other. This case occurs in the following in spectra measured before annealing.

A M6ssbauer investigation of a-irot"

837

Corrosion at tow H~Oz concentration orrosion 25"C .006% H20 z

)fCorrosion ~.0.012%

25"C HzOz

~'~

~ C o r r o s i o n 85"C ~ . 0 , 0 0 6 % H20 z

-) J

I

I

I

SingLe Line oxide*

SingLe Line'oxide*

V=5850 D..=6

V= 5800, 5400 D = 2 2 0 + - 30 (8~-- t%)

~

I

I

FeOOH B=0.32, • = 0 3 0 , F=0.5 Fe304 (reLaxation) V : 5800, 5400 D = 280+-40 (63+-4%)

~

nneaLing 200"C

I

I nnealing 200"C

)

I

Fe304 H,fr = 46.4 T

SingLe Line oxide* Fe30, (traces of)

V = 5850 Dov =13+-2

V= 5800, 5400 D = !80 + - 5 0 (17+-4%)

FeOOH ~, • as above, lP= 0 6 5 F%04 (reLax present) H,. ~ 44.8T V = 5800 D.v = 1 0 0 + - 2 0

I (~nneoLing 4 0 0 %

I A ~ a L i n g 400"C

I F%04

H,f~ = 48 T

H e. = 48 T £=0.4 (12th Line}

* This Lineis inour opinionrelated to the reLaxationspectrum of magnetite.Its position is near to thatof the first Line in the ferricdoublet. Both oxides, i.e. there

isno discernible distribution difference.

~fSnnealingT *c

)

I

Fe~04 V = 5800, 5600, 5400 D = 150+-30 (21+-4%)

)

I

V : 5800, 5600, 5400 D:20+-5 (~75%] D~550+200(~6%}

~

F%0. [brood Lines, reLaxation) H,.-- 4 4 . 2 + - Q 5 T V= 5800, 5600, 5400 D =500 (80%)

I nneoling 600"C

I Fe~Oa H..= 48T, 1~=0.5 (12th Line) Wustite 8=0.86+-0.02 • = 0.75 +- 0.03 F= 0.60 + - 0.05 V = 5800, 5600, 5400 D = 4 0 + - 5 (~100%) D : IO00÷ - 400 (~10%) t

FIG. 4. Flowchart showing the low HzO~ concentration treatment scheme and presenting the results from the corresponding analysis. 5, e and F are oxide isomer shifts, quadrupole splittings and line widths, respectively, all in mm s -1. The parameters of a-iron were practically the same in all cases and are not written. Spectrometer settings V used in the measurements are indicated. The results of the DCEMS phase distribution analysis are indicated, e.g. as D = n n n (NN %), which means that N N % of the absorber surface is covered by the oxide to a depth o f n n n Fe,~. Day is average thickness, i.e. assuming that the whole absorber surface is evenly covered by the oxide.

838

G. BELOZERSKIIet aL

The influence of annealing at temperatures of 200, 400 and 600°C at 10-~ torr was investigated. The annealing was made in the spectrometer by means of a tungsten heating element. In all cases there were no detectable oxide signals after the annealing. It is concluded that the annealing does not convert the inert layer into an active (resonant) one (in Mrssbauer sense). H~O2 Low concentrations of H~O2 in water are often used to promote the formation of a protective oxide layer. For this reason two corrosion experiments at 25°C were performed in re-distilled oxygen-free water with a concentration of peroxide 0.006 and 0.012 ~. A third experiment was made with the same concentration as the first one but at 85°C. The corrosion time was 22 h. The results of the first and second experiments were similar (see Figs. 4 and 5). After corrosion one can see only a very weak oxide line with an area of about 4 for both spectra at the setting 5800 V. DCEMS analysis on basis of this single setting shows that the average thickness of the oxide formed in the first experiment is ca. 6 FeA; however, this oxide covers probably at most about 20 ~ of the surface, which means that it is about 30 FeN thick. In the second experiment the oxide seems to be thicker and more concentrated to a small part of the surface (cf. Fig. 4). The Mrssbauer parameters of different ferric oxides and hydroxides at room temperature may be very similar, especially in the case of superparamagnetism or deficient structure. As only room temperature measurements were performed annealing was carried out to obtain additional information for the phase analysis. The annealing was in all cases performed during 1 h. The treatment scheme is shown in Fig. 4. After annealing at 200°C one can detect lines from magnetite. The quantity of magnetite is given in Fig. 4. The low He~ value suggests presence of relaxation or imperfect structure of the magnetite. The annealing at 400°C leads to an increase of the amount of magnetite and of Herr. The outer lines are now sharp. The sample used in the second experiment was annealed at 600°C. After this, the Mrssbauer spectrum changed drastically. The amount of magnetite decreased and in the center of the spectrum, around line 4 of the e-iron spectrum, a doublet arised with parameters similar to the M(Sssbauer parameters of wiistite. The line 4 of F%O4 became broader after annealing at 600°C, from 0.25 to 0.43 mm s -1, i.e. the stoichiometry of the magnetite was destroyed. The DCEMS analysis of this sample--after annealing at 600°C and also after the annealing at 400°C--indicates a more structured depth distribution, which may have been present to some extent already at an earlier stage. Thus, practically the whole surface is covered by a thin (,,~ 40 FeA) oxide layer. On about 10~ of the surface --with due consideration to the crudeness of the distribution model--the oxide is however very thick, ca. 500 FeA or more. Corrosion at 85°C is more intensive (cf. Fig. 6). In the original spectrum before annealing there is a typical doublet, supposedly originating from a hydroxide, presumably 7-FeOOH. There is also a small trace of superparamagnetic magnetite. After annealing at 200°C the magnetite content increases, and after 400°C practically all the hydroxide is converted into magnetite. The spectra are however complicated due to superparamagnetic relaxation. In the DCEMS analysis after 400°C annealing, it C o r r o s i o n in

A M6ssbauer investigation of ct°iron 5800v

5600V

839 5400V

~°Fe t-

'l

"-r --1~'1 A

!

200%

m > •~

o

Fen0, ~

g: 400"C

o

FIG. 5. Some spectra obtained at differentTelectron spectrometer settings (5800, 5600 and 5400 V--arranged as columns) afte~corrosion at 25°C in a 0.012~ H~O~ solution (A and B) and spectra obtained after subsequent annealings at 200°C (C and D), 400°C (E-G) and 600°C (H-J) (cf. Fig. 4).

appears that a region of decreased effective 57-Fe density is present near the surface. From data of Figs. 4 and 5 it is deduced that at low temperatures peroxide promotes the formation of a magnetite layer on the surface of 7-iron. This layer has a stoichiometry close to that of normal magnetite. The annealing at 400°C leads to the formation of a good magnetite structure. The superparamagnetic relaxation is now absent. Additional annealing at 600°C leads to the conversion of magnetite into wtistite

840

G. BELOZERSKIIet 5800V

al. 5400V

5600V

=-Fe I

I

I

'1

B

~'-~OOH

Fe~O~ ~i.

li

-5

i,

ii

o

~'

;

i

~ml.

-s

o

s

mm/j

~

o

s

m

/s

FIG. 6. Some spectra obtained after corrosion at 85°C in a 0.006% H=O~solution (A and B) and spectra after subsequent annealings (C-F) (cf. Fig. 4).

and to the increase of the line width of magnetite, which indicates a deterioration of the structure. From the DCEMS analysis it is found that there is no significant difference between the depth distributions of the magnetite and the wiistite. At 85°C the corrosion at 0.006% H2Oz concentration leads essentially to the formation of y-FeOOH layers, the structure of which is very different from the structure of ~-iron or ferrites, and which does not form a protective layer. As a consequence of this, annealing at the same temperatures as above does not lead to the formation of ferrites with good structure (cf. Figs. 5 and 6) .We would like to emphasize that the identification of the magnetite is uncertain, since the lines are wide and since the magnetic field is unusually small. The oxide must however have ferrite structure. For comparison three series of corrosion measurements were made at higher concentration, at 25, 25 and 85°C respectively. Treatment scheme and results are shown in Fig. 7. The first measurement was made with 6.8 % H~O~. After 18 h, when corrosion products could be seen on the surface, the exposure was interrupted. The M6ssbauer spectra showed the presence of y-FeOOH. A DCEMS analysis determined

A MiSssbauer investigation of miron

841

Corrosion at high HzOz concentration

~6.o8rrosion 25"C % H~Oz

)

C6Orrosion 25"C B% H~Oz

I

I T - FeOOH 8 =0 3 9 ÷ - 0 . 0 2 =092+-004 ['=0 52

V =5800

)

I

I (AnneoLing400*C }

(AnneaLing 200"C

Fe:~04

H~47T

)

I

I

I V; 5800, 5400 D-- 270+-50 (56+-4%)

I

V= 5800, 5400 D =260+-40 (42+-8%)

I

Y - FeOOH (8, ~, Cos above) WusLite

k2 °+++

FeOOH 8 = O.36 + - 0 . 0 1 ~= 1.06 +- 0.02 F = 0.59

a - Ir0n

V = 5850, 5400 D=300 (40%) (AnneaLing200°C)

(/~orrosion 85"C

FeOOH Wustite

F ~ 0.6 V=5800

D=15+-2

V: 5800, 5400 D(FeOOH)= 200+-20 D(FeO) =200 - 250÷- 40 (53+-7%) *

I

)

I Wustite = 0.96 ÷ - 0 05 =0.82+- 0 0 6 V ~58 O0

I AnneaLing 600°C

)

I * Reduced effective 57-Fe density was found near the surface.

Wustite

(8, { as above ) V; 5800, 5400 D:275+-50 (40+-9%)

Fio. 7. Flowchart showing the high H~Oa concentration treatment scheme and presenting the results from the corresponding analysis. Notations are defined as in Fig. 4.

that the 7-FeOOH formed a surface layer with thickness 30 nm. After annealing at 200°C there was no trace of magnetite, but the hydroxide was converted directly into wiistite. Concerning the formation of wiistite at these low temperatures, one must note the

842

G. BELOZERSKIIet al.

5400 V

5800 V ~.- Fe

c

E

FeO

| 1

1

I-'-I

6000C ~ c

mmA

i

mmA

FIG. 8. Some spectra obtained after corrosion at 85°C in a 13.6~o H202 solution (A and B) and spectra after subsequent armealings (C-G) (cf. Fig. 7). (G) shows the same spectrum as (E) but is measured with a larger velocity range.

A M0ssbauer investigation of ct-iron

~43

following. It is usually concluded that wiistite cannot form at temperatures below 570°C. a4,1'5 However, first it may be noticed that our M6ssbauer parameters for the wtistite identified here agree well with those of ref. 14. Second, we note that we are not dealing with bulk wiistite, but rather with a very small quantity of wiistite on the surface of ~-iron. We suggest that the unusual properties encountered here for the wiistite are due to the reducing effect of the ~-iron. At the 6.8 ~ concentration the peroxide has a smaller suppressing effect than in the case of lower concentration. We therefore propose that the hydroxide formed under this high concentration of H202 has another structure than the hydroxide formed at lower concentration, i.e. such that the hydroxide converts very easily into wiistite. The conversion begins at 200°C and goes directly to wiistite without an intermediate step of magnetite (compare Figs. 4 and 7). It must be stressed, however, that the M6ssbauer parameters of the wiistite produced in the different processes are different. Corrosion at very high concentration, 16.8~, at the same temperature is fully suppressed, so that after 3 days no visible products are found. After 400°C annealing for 1 h a 9°o magnetite signal appeared in the spectrum. This might be due to the passive layer becoming thicker during the H20,, treatment, and subsequently being converted to magnetite during annealing. The third experiment was made at 85°C at 13.6}~0 H~Ov After 30 h traces of corrosion were visible on the surface and the exposure was terminated. The spectra are shown in Fig. 8. The 7-FeOOH has the same isomer shift as above but the quadrupole splitting is larger; ~ -~ (1.06 :]: 0.04) m m s -~. One can see that the quadrupole splitting is different from that formed under different conditions, which is not surprising since it is known that the quadrupole splitting of 7-FeOOH is dependent on layer thickness? 6 As in the case of corrosion at 6.8 °/o H202 at room temperature, hydroxide converts directly into wiistite. After annealing at 200~C one has both hydroxide and wiistite. The DCEMS analysis shows that the wiistite is situated beneath the hydroxide. The activation at 400°C for 1 h is enough for full conversion of the hydroxide. Additional annealing at 600°C does not change the spectrum significantly. It is interesting to note that the parameters of the w~istite is different from the two previous cases. The DCEMS analysis also indicated a region of lower 57-Fe effective density near the surface, after the 85°C, 13.6 °/H.~O,, corrosion treatment. This effect decreased during the successive annealings. SUMMARY The existence of a passive layer on the surface of ~.-iron exposed to air, which can be seen by ESCA, was independently confirmed by DCEMS. It was found that the thickness of the layer was in the order of 40 •, and that it contained ferric but not ferrous iron. A surprising result was the absence of any detectable M6ssbauer signal from the layer. At room temperature, low concentration of H202 inhibits the process of corrosion. It is difficult to estimate the quantity of oxide, and also, with our statistics, to decide whether the product is magnetite or maghaemite. It is however obvious that this oxide layer grows during the corrosion treatment. At higher temperature, the rate of corrosion is higher and leads to the formation of ~,-FeOOH. The phase transformation during annealing depends on the temperature of oxidation, i.e. magnetite which was formed on samples corroded at high temperature have:

844

G. BELOZERSKIIet al.

very bad structure and very small H~fr. Magnetite formed at low temperature has m u c h better structure and converts, at 600°C annealing, to wtistite. Corrosion at very high concentration of peroxide, i.e. 16.8 %, is practically fully suppressed, and leads only to the growth of the protective layer. Corrosion at intermediate concentration, on the other hand, leads to the formation o f ~,-FeOOH with rather large quadrupole splitting, which is typical for thin layers o f 7 - F e O O H . le The fact that the transformation from hydroxide to wtistite is direct and requires very low activation energy is unusual. Corrosion at higher temperature is not fully suppressed and leads to 3,-FeOOH with larger quadrupole splitting than at lower temperature, which m a y be due to a m o r e dispersed structure. One must emphasize that the wtistite which was formed, in all cases had different parameters depending on the prehistory. Acknowledgements--We are grateful to Prof. N. G. Varmerberg at the Chalmers University of

Technology and University of G6teborg, Sweden, for carrying out the ESCA analysis, and to Prof. L. N. Moskwin and Dr. S. B. Tomilov, Leningrad State University, USSR, for helpful and stimulating discussions. REFERENCES 1. R. K. FREIER, VGB Kraftwerkteehnik Speisewassertagung (1969). 2, L. N. MOSKWIN,A. A. EFIMOV,S. B. TOMILOV,E. P. BREDIKHINAand A. I. GORSHKOV,Thermal Engineering 27, 305 (1980). 3. C. J. POWELL,Surf. Sci. 44, 29 (1974). 4. U. B.KVERSTAM,B. BODLUND-RINGSTROM,C. BOHM,T. EKDAHLand D. LILJEQUIST,Nucl. Instrum. Meth. 154, 401 (1978). 5. D. LIL1EQUIST,T. EKDAHLand U. B.~VERSTAM,Nucl. Instrum. Meth. 155, 529 (1978). 6. D. LILT~QUISTand B. BOOLUND-RINGSTR/3M,Nuel. h~strum. Meth. 160, 131 (1979). 7. D. LILrEQtnS%C. BOHMand T. EKDAHL,NucL lnstrum. Meth. 177, 495 (1980). 8. G. BELOZERSKH,C. BOHM,T. EKDAHLand D. LILJEQUIST,Report at the Symposium on Recent Chemical Applications of M6ssbauer Spectroscopy, 179th National ACS Meeting, Houston, Texas (1980). 9. A. M. PRITCHARDand C. M. DOBSON,Nature, Wash. 224, 1295 (1969). 10. A. SETTECAMARAand W. KEUNE,Corros. Sei. 15, 441 (1975). | I. N. N. GREENWOODand T. C. GraB, Mossbauer Spectroscopy. Chapman and Hill, London (1971). 12. G. BELOZERSKII,C. BOHM,T. EKDAHLand D. LILJEQUIST,Nuc. lnstrum. Meth. 192, 539 (1982). 13. W. MEISELand G. KREYSA,Z. anorg, allg. Chem. 395, 31 (1973). 14. N. N. GREENWOODand A. T. HOWE,Dalton Transeations 1, 110 (1972). 15. H. SHECHTER,P. HILLMANand M. RON, J. appl. Phys. 37. 3043 (1966). 16. A. MINKOVAand J. P. SCHtJNK,C.r. dcad. bulg. Sci. 28, 1171 (1975).