DOI: 10.1002/sia.1281

NON-DESTRUCTIVE SURFACE ANALYSIS APPLIED TO ATMOSPHERIC CORROSION OF TIN

S. Jouen, B. Hannoyer, O Piana

Université de Rouen – L.A.S.T.S.M - I.U.T. 76821 Mont Saint Aignan Cedex France

Abstract:

Tin samples have been exposed in industrial, rural and urban outdoor atmosphere.

Corroded surfaces were studied with FTIRAS, XRD, SEM with EDS analysis and Mössbauer

spectroscopy. Corrosion is continuous and heterogeneous during the first ten months of

exposure, and the corrosion products are mainly insoluble and non-protective. Corrosion rate

differs between each site of exposure, being increasing from urban to rural area. Analysis

performed reveal that corrosion layers are mainly composed of poorly crystallised and

hydrated stannic oxide SnO2.xH2O. Lower proportions of Sn(II) and Sn(IV) corrosion

products are also identified as carbonates, sulfates and chlorides compounds. FTIRAS and

Mössbauer spectroscopy analysis remain the most adaptable to perform the characterisation of

tin corrosion products. They give reliable information about the chemical species, the valence

state of tin nuclei and the evolution of the corrosion layer with time.

Keywords: tin; atmospheric corrosion; Mössbauer spectroscopy; stannic oxide; tin salt

INTRODUCTION

Degradation of metals in the atmosphere is known to be an electrochemical process. It

involves an aqueous electrolyte formed on the surface with atmospheric pollutants. Corrosion

rate and corrosion layers composition of mainly metals and alloys were often studied since the

beginning of the century. Nevertheless atmospheric corrosion of tin and tin alloys remains not

well known and understood again in spite of important degradations often observed for

instance on bronze sculptures exposed in urban atmosphere or on electrical contacts when

they are used as protective coating. A thin protective film of SnO2 covers rapidly the surface,

and increases with moisture. Except chlorine, usual atmospheric pollutants have no significant

effects on tin and main corrosion products detected are tin oxides1,2. The rarest published

results about tin corrosion have pointed out that only SnO2 is detected in urban and marine

area, SnO2, SnO and 5SnO.2H2O in industrial area and 5SnO.2H2O with traces of

Sn4(OH)6Cl2 near an electrolyse factory3,4. Analysis of corrosion products on outdoor bronze

has revealed also numerous copper compounds and the tin hydrated oxide SnO2.nH2O5.

During ten years of exposure in rural, industrial and marine sites, Finkelday has observed a

continuous weight gain of tin sample6, reaching respectively 100, 600 and 800 µg/cm².

The purpose of this present work is to contribute to a better knowledge of tin corrosion

layers composition. As part of a field exposition of several metals, our results on tin coupons

allowed to conclude that composition of the corrosion layers is more complex than published

data and is in fact a mixture of chemical compounds not always easily identified.

SAMPLE PREPARATION AND INVESTIGATION

METHODS

Single- and double-sided tin coupons (98.8% purity) sized 30 x 30 x 0.5 mm were used as

samples for the field exposure. The surfaces were cleaned ultrasonically in acetone and

isopropyl alcohol. As regards the single-sided sample, the side not exposed to precipitation

was masked off with a special varnish.

Field exposure program started in December 1997 in the northwest of France, around

Rouen city (urban, industrial and rural area). The test sites were chosen near the “Air-

Normand” institute monitoring stations, where some environmental conditions (NO2, NO,

SO2, O3, Temperature, Relative Humidity,…) are regularly followed. The average values of

pollutant concentration during the twelve months of exposure are reported in table 1.

Exposed samples were mounted facing south with an inclination angle of 45° on a

Plexiglas fixture. Duplicate and triplicate double- and single-sided samples respectively were

used in order to obtain weight gain data.

Tin samples weight measurements were carried out before and after exposure with a

Mettler Toledo AG 245 balance having a readability of 0.01 mg.

The analysis of corrosion layers were performed on one panel after 15, 30, 45, 60, 75, 90,

120, 150, 210, 300 and 365 days of exposure, using several non destructive spectroscopic

methods of analysis. Fourier Transform Infrared Reflection-Absorption (FTIRAS) spectra

were recorded with a Nicolet 710 instrument equipped with a Spectra-Tech reflectance

accessory and with a microscope allowing microanalysis from surfaces upper than 40 x 40

µm. This technique allows to identify easily the chemical bounding present in crystalline and

non-crystalline corrosion products.

X-ray diffraction (XRD) was used to identify crystalline phases and was conducted with a

Seifert XRD 7 instrument under low incidence conditions (3°), using CuK radiation.

Elemental composition of the corrosion layers obtained from SEM/EDS measurements

were performed with a Hitachi S-2460N equipped with a Kevex analytical system.

The tin corrosion products were moreover investigated using Conversion Electron

Mössbauer Spectroscopy (CEMS). Mössbauer spectra were acquired with the use of a

Ba119SnO3 source at room temperature. This nuclear technique provides information on the

local electronic environment of the probe nuclei.

RESULTS AND DISCUSSION

Surface analysis of unexposed tin coupons with CEMS has revealed the presence of a thin

layer of Sn(IV) compound, identified as native SnO2.

The corrosion of tin is heterogeneous. After a few days of exposure, the metal surface is

recovered with white and grey clusters (Fig. 1). Number and size of these clusters increase

with exposure time until they completely cover the metal surface. Visuals observations clearly

point out the difference of corrosion rate relative to the different sites. Corrosion is increasing

from urban to industrial and finally to the rural area. After 150 days of exposure, we estimated

than approximately 53%, 63% and 81% of the tin surface is recovered by corrosion products

in urban, industrial and rural atmosphere respectively.

Characterisation of tin corrosion products

Compositions of corroded layers are very similar in the three environments, and mainly

consist of tin (IV) oxide.

Broad peaks on the XRD spectra are typical of poorly crystallised SnO2. Few small peaks

indicate also the presence of other crystalline compounds without allowing their unambiguous

identification. As well as O, traces of S and Cl were detected by EDS analysis.

The use of IR microspectroscopy on area of 50 x 50 µm permits to improve analysis of

clusters on surface sample and to point out the multiphases system existing in the layers such

as oxide, sulfate, carbonate and water, as illustrated by Fig. 2. Absorption bands localised at

980, 1040 and 1135 cm-1 are unambiguously assigned to sulfates stretching modes. The broad

band near 700 cm-1 indicates the presence of stannic oxide in the layer, the band at 1650 cm-1

is assigned to vibrational mode of water, indicating hydrated phases, whereas the broad band

and 1453 cm-1 is probably originating from carbonated compounds. At our knowledge, it is

the first time that sulfates and carbonates phases are detected among corrosion products of tin

exposed in outdoor conditions.

The originality of this work is the use of Mössbauer spectroscopy to study the corrosion

products of tin. This nuclear method provides a clear distinction between Sn (II) and Sn (IV)

corrosion products. Typical spectra are shown in Fig.3a. They are analysed in terms of two

main components. Two Mössbauer hyperfine parameters can be deduced from the spectra, the

isomer shift (), which is sensitive to the atomic valence states, bond angles and atomic

spacings ( Sn(IV) < Sn(0) < Sn(II)), and the quadrupole splitting (), which depends on the

symmetry of the charge surrounding the probe nucleus. The single peak S1 located at = 2.6

mm/s (relative to BaSnO3 reference isomer shift) can be identified as tin substrate. The

strongest component is a doublet D1 consistent with average parameters ( = 0 mm/s and =

0.6 mm/s) closed to hydrated stannic oxide SnO2.xH2O values7. This attribution is consistent

with XRD results but this doublet is not always as the doublet expected for usual SnO2

doublet (amorphous or crystalline compounds). We assume that other Sn(IV) phases are

present in the layers. A clear identification of these phases is not possible because of the weak

subspectra area and the poor litterature Mössbauer data about tin corrosion products. Both

adjustments, with one average doublet or one doublet and singlets can be considered from

least-squares fits. To conclude, new experiments at lower temperature have to be performed in

order to use Debye-Waller factor variations expected for different compounds.

Average of Sn(IV) species versus the corresponding relative Mössbauer area is reported

on Fig.3b. The decrease of versus relative area, so versus exposure time, can be correlated

to some modifications in the average composition of the Sn(IV) compounds in the mixture,

but this point can not be discussed until now as we said above. A second explanation could be

an evolution of stannic oxide crystalline state and/or hydration with the duration. In this way,

large of the first step would be associated to the formation of an amorphous and non-

hydrated stannic oxide, and the decrease of with time to an hydration of the oxide and a

partial crystalline state. Then, the lowest values observed for the long time would be the

expression of crystallisation and dehydration of the oxide.

Few spectra exhibit a third component, a weak doublet D2, identified as Sn(II) compounds

according to its isomer shift (3,3 to 3,7 mm/s). Although their intensity may be low, its

existence can be unambiguously concluded from the combination of all experimental results.

Because of the electronic configuration of bivalent tin (4d105s²5p²) a correlation between

isomer shift and quadrupole splitting has been established8. In a previous work, we used this

relation to try an identification of Sn (II) corrosion products9. We concluded that Sn4(OH)6Cl2

could be the main Sn(II) product after 2, 3 and 12 months of exposure in rural site, SnSO4 as

main Sn(II) product after 2 and 3 months in industrial area and another Sn(II) compound in

urban area after 14 days, 2, 3 and 7 months of exposure, which attribution was not suggested

according to the lack of information in the litterature. The presence of SnSO4 in industrial

corrosion products is not surprising. As a matter of fact, FTIRAS analysis clearly indicated

the presence of sulfate in all corrosion layers.

Mössbauer spectroscopy permits quantitative approach. In a mixture of phases, the relative

subspectra area Ai is proportional to the atomic concentration Ci of Sn in the compound i,

according to the relation Ai = fiCi. The atomic fraction f, or Lamb-Mössbauer factor, of tin

nuclei affected by the Mössbauer effect, is correlated to the Debye-Waller factor which

depends on the nature of compounds. The relative value of each f factor in a mixture of

corrosion products can provide an high sensitivity to some compounds and on the contrary a

very low sensitivity to others. It is particularly important for tin compounds. The f factor of

stannic oxide is 9 times greater than that of metallic tin (f(Sn) = 0.05 and f(SnO2) = 0.45),

leading a relative area of stannic oxide 9 times greater for a same atomic concentration of the

two species10. f factor values relative to stannic oxide factor, were determined for SnO, SnSO4

and SnCl2.2H2O available commercial powders, f(SnO2) / f(SnO) 2.3 is in good agreement

with literature data11 and therefore valid the experimental method for others compounds

(f(SnO2) / f(SnSO4) 5.95 and f(SnO2) / f(SnCl2.2H2O) up to 25).

These data have important consequences for the interpretation of Mössbauer spectra

because it seems not possible at room temperature to distinguish a small amount of SnSO4

when it is mixed with high amount of stannic oxide. Improved experiments are necessary.

Quantitative evolution of the corrosion layer

Weight gain evaluations were performed on triplicate and duplicate panels for the single-

and double-sided samples respectively. The results are displayed in Fig. 4a, which shows the

average weight gain of samples versus exposure time.

A rather linear increase of mass gain is observed with time whatever the side of the

sample, in agreement with the finding of Finkelday in rural, industrial and marine

atmospheres after ten years of exposure6. This increase is attributed to the heterogeneous

corrosion mechanism and the growth of main insoluble and non-protective corrosion

products.

Different variations are observed according to the exposure site which one can classify by

increasing mass gain: urban, industrial and rural, respectively 190, 330 and 510 µg/cm²/y for

the single-sided samples and 345, 455 and 595 µg/cm²/y for the double-sided samples. These

values are different from those of Finkelday but this last does not mention the composition of

the atmosphere whereas the results of the indoor UN/ECE exposure programm reveal a very

great dispersion according to the site12.

Precipitation data obtained by the organism “cercle des partenaires du patrimoine”

indicate a dry period between 1.5 and 3 months of exposure, that corresponds to the abrupt

increase of mass gain observed on all the curves of the Fig.4a. This increasing is attributed to

the formation of corrosion products, being in position to be solubilised with rain afterwards.

The thickness of the corrosion layer was evaluated and followed from Mössbauer spectra,

supposing that stannic oxide is the only constituent of the layer9. When the thickness of the

layer increases, relative area of metallic tin and stannic oxide subspectra, respectively

decreases and increases, because of the low range of the electrons in a solid. The Fig.4b

presents the evolution of relative area of stannic oxide with time in urban, rural and industrial

sites. The results are in very good agreement with weight increase and spectroscopic analysis.

Thickness of the oxide layers in the different environments is increasing from urban to rural

area. The decrease observed on the three curves in the dry period between 1.5 and 3 months of

exposure is associated to the increase of weight gain observed in the same dry period (see Fig.

4a). Then, it clearly appears that this abrupt weight increase is mainly induced by soluble

corrosion salts, as SnSO4.

Until know, tin corrosion rate have not been correlated with the surrounding environment.

In this study, the importance of pollutant concentration or climatic parameters in the corrosion

rate is difficult to point out, because of the low number sites of exposure. Therefore, regarding

the pollutants concentration, it appears that O3 concentration could be an influenced

parameter. The amount of precipitation or the time of wetness could be also important

parameters to understand tin corrosion mechanism and corrosion rate.

Conclusion Atmospheric corrosion of tin is not well known. By means of the use of several methods of

analysis, this work allows a better knowledge of corrosion layers composition. Corrosion of

tin is relatively important and continuous and increases from urban to rural site of exposure.

The corrosion layers consist of a complex mixture of corrosion products, with a poorly

crystallised hydrated stannic oxide SnO2.xH2O as main constituent. It is obvious that an

evolution of the structure or the composition of Sn(IV) corrosion products occurs with the

exposure time. Others Sn(IV) and Sn(II) phases were also clearly detected, as sulfates,

carbonates and chlorides compounds.

Acknowledgement The authors would like to thank the Air-Normand Institute for the pollutant concentration

measurements.

References

1. Tompkins H.G. J. Electrochem. Soc. 1973 ; 120(5), 651

2. Yasuda K.I, Umemura S., Aoki T. IEEE Transactions on components, Hybrids, and

Manufacturing Technology, 1987 ; 10(3), 456

3. Peters V.S, Paintaske R, Hecht G. Korrosion. 1987 ; 18, 39

4. Biestek T, Drys M. Powloki Ochr. 1981 ; 1, 5

5. Robbiola L, Fiaud C. ICOM Committee for Conservation 1993 ; 2, 796

6. Finkelday W.H. Proc. American Society for Testing and Materials 1943 ; 43, 137

7. Cook P.S, Cushion J.D, Cussidy P.J. Fuel 1985 ; 64, 1121

8. Lees J.K, Flinn P.A. Phys. Letters 1965 ; 19, 186

9. Jouen S. Thesis, Rouen, France, 2000, 187

10. Huffmann G.P, Dunmyre G.R. J. Electrochem. Soc. 1978 ; 125(10), 1652

11. Stjerna B, Granqvist C.G, Seidel A, Haggstrom L. J. Appl. Phys. 1990 ; 68(12), 6241

12. Tidblad J, Kucera V. UN/ECE International Cooperative Programme on effects on

materials, including historic and cultural monuments, Stockholm, Sweden, 1998, report 26

Tables and Figures captions

Table 1. Average pollutant concentrations and temperature during the 12 months of exposure in industrial, urban and rural area (France).

Figure 1. Scanning electron micrograph of tin sample exposed 1 month in the industrial area. Figure 2. Infrared microspectroscopy (50 x 50 µm²) spectrum from tin exposed 5 months in the industrial area. Figure 3. (a) CEMS spectrum of the tin corroded surface after 14 days, 1, and 2 months of exposure in the urban area. (b) Average of Sn(IV) species versus the corresponding relative Mössbauer area. Figure 4. (a) Weight gain of single-sided tin samples versus exposure time in industrial, urban and rural atmospheres. (b) Evolution of the relative area of stannic oxide calculated from Mössbauer spectra (CEMS) versus exposure time in industrial, urban and rural area

Site of exposure

Average pollutant concentrations (µg/m3) Temperature (°C) SO2 NO NO2 O3

industrial 29 9.7 24.2 47.1 12.8 urban 22.3 41.1 50.1 25.8

rural 12.2 2.9 8.7 60.1

50 µm

Wavenumbers (cm-1)

2

4

1000 800 1200 1400

Ref

lect

ance

(%)

H2O

CO32-

SnO2

SO42-

Relative velocity (mm.s-1)0-10 +10

1.00

1.14

Rel

ativ

e in

tens

ity (

%)

1.00

1.14

1.00

1.19

14 days

1 month

3 months

0

Substract S1

Sn(II)Sn(IV)

D1

D2

D2

D2

+10-10

Sn

(IV)

30 40 50 60 70 80 90 1000,50

0,55

0,60

0,65

0,70

Relative Mössbauer area of Sn(IV) doublet D2

Urban Industrial Rural+

(a)

(b)

0

100

200

300

400

500

0 2 4 6 8 10 12Exposure period (months)

Wei

ght g

ain

(µg/

cm²)

Dry periodrural

industrial

urban

0

20

40

60

80

100

0 2 4 6 8 10 12

Rural

Urban

Industrial

Exposure period (months)

Rela

tive

area

of s

tann

ic o

xide

(%) (b)(a)