SEARCH

SEARCH BY CITATION

Keywords:

  • aluminium;
  • corrosion;
  • electrochemical corrosion investigation;
  • magnesium

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

To improve the mechanical properties, the castability as well as the corrosion resistance of magnesium alloys, aluminium is frequently employed. As a rule, technically relevant alloys contain between 3 and 9 wt% aluminium. Normally, aluminium contents under 3 wt% is said to have no corrosion reducing effect. In order to investigate the influence of specific additions of aluminium, various methods are employed in practice and different electrolytes or concentrations of saline solutions are used as testing media. In doing this, it is rarely possible to compare the results. For this reason, the current work both discusses the influence of aluminium and the effect of different saline solutions on the corrosion behaviour of low alloy magnesium–aluminium alloys as well as compares two different measuring procedures (electrochemical and gravimetric).

1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

With a density of 1.738 g/cm3, magnesium is deemed to be the lightest of all structural metals. However, despite substantial weight savings and excellent weight specific strength properties, magnesium's use as a structural material in lightweight structures is severely limited owing to its strong tendency to corrosion [1-3]. By means of alloying elements such as aluminium, an attempt is made to elevate the corrosion resistance in which the influence of aluminium has not been explicitly explained. In essence, aluminium is said to have a corrosion reducing effect [3-9], although it is possible to demonstrate that aluminium contents below a specific concentration lead, if anything, to a deterioration [8].

Aluminium is a typical alloying element for magnesium (inter alia [3, 6, 8, 9]). Besides lowering the notch sensitivity and elevating magnesium's low toughness, only a small increase in the specific density occurs for the magnesium–aluminium alloy. In addition to this, the strength and hardness values are elevated via solid solution and precipitation hardening [3]. A specific fraction of aluminium in magnesium alloys leads to an improvement in the corrosion resistance by means of the ability to passivate the surface. However, aluminium increases both the tendency to induce micro porosity as well as the risk of local corrosion by forming the intermetallic phase Mg17Al12. Aluminium contents from 3 to 9 wt% are employed in technically relevant alloys due to the described advantages (inter alia [3, 6-9]). The addition of aluminium below a specific content leads to a deterioration of the corrosion properties. For this reason, the general influence of aluminium on the corrosion behaviour is not considered as explicitly conclusive.

Based on the propositions described in the literature, the aluminium content of the magnesium–aluminium alloys is varied between 0.5 and 3.0 wt% in the current tests. Polarization measurements and immersion tests are carried out in which the concentration of the saline test solutions is also varied. The results from both measuring procedures are compared with each other and thereby contribute to the present discussion to basically correlate the electrochemical measurements with other analytical procedures [10-12]. In addition to this, a quantitative analysis is performed of the corroded specimens' loss in volume by means of a micro-computed tomographic (µCT) measuring technique.

2 Materials and procedures

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

2.1 Materials

The cylindrical specimens used in the test were manufactured from magnesium–aluminium alloys possessing various aluminium contents including a reference of pure magnesium. The chemical compositions were determined by using a GDOES at the Institute of Materials Science (IW) and can be seen in Table 1.

Table 1. GDOES chemical analysis of the magnesium specimens; mean values from three measurements
AlloyMg (%)Al (%)Zn (%)Cu (%)Fe (%)Mn (%)Ni (%)Si (%)Sn (%)Ca (%)Rest
Mgpure99.5080.0000.1010.0550.0250.0390.0670.0170.0190.0020.167
Al0.599.1350.4540.0720.0410.0240.0360.0450.0110.0190.0020.161
Al1.098.7650.9610.0380.0240.0200.0350.0250.0100.0100.0010.111
Al1.598.1261.5400.0600.0350.0220.0370.0340.0100.0140.0010.121
Al2.097.6741.9360.0730.0450.0240.0390.0490.0100.0170.0020.131
Al3.096.4183.1890.0740.0450.0240.0420.0480.0100.0170.0020.131

2.2 Sample preparation

Two specimen types were produced for carrying out the corrosion tests. On the one hand, cylindrical specimens (d = 10 mm, h = 5 mm) were turned for the immersion tests and, on the other, cylindrical specimens were manufactured possessing a diameter of d = 8 mm and a height of h = 20 mm. To perform the polarization measurements, these specimens were embedded in cold-curing epoxy resin and ground (2500 paper grade) such that a flat measuring surface of 0.503 mm2 was provided. For both corrosion tests, each of the specimens was taken from a dried atmosphere (desiccator) and immediately cleaned in acetone in order to remove machining and grease residues.

2.3 Corrosion tests

The corrosion tests are divided into two procedures. Firstly, the corrosion current density icorr of the individual alloys is determined via dynamic polarization measurements and the corrosion rate Pi is computed in mm/y (Equation (1)) [10-12]. Secondly, immersion tests are carried out over a time period of 7 days in order to measure the corrosion rate Pw from the loss of mass ΔW (Equation (2)) [10-12]. The non-corroded volume is determined in an additional analysis of the specimens using the µCT. These values were then correlated with the loss of mass from the immersion tests.

  • display math(1)
  • display math(2)

Icorr is the current density in milliampere per square centimetre based on the Tafel evaluation as well as on the Butler–Volmer equations. ΔW is the weight loss in milligram per square centimetre per day, which can be calculated directly from the immersion tests. The data for the individual electrolytic solutions are depicted in Table 2.

Table 2. Data for the testing solutions
Measuring methodSodium chloride-concentration in %
Polarization measurement0.050.090.500.902.505.00
Immersion test0.000.090.500.902.505.00

To measure the corrosion current density, a potentiostat (IM6e) from the company Zahner elektrik© is employed using the software Thales Flink and the measurements are performed in a conventional 3-electrode set-up. A platinum and a saturated calomel electrode represent the counter and the reference electrodes, respectively. During the initial step, the specimens' free corrosion potential is measured for 10 min after which the polarization is carried out using a scan rate of dE/dt = 500 µV/s and a sampling interval of Δt = 500 ms. A Matlab-Program was available to evaluate the measured results with whose help the free corrosion potential's minimum value was initially defined. Using this value, a defined limit value, which is based on the Tafel evaluation as well as on the Butler–Volmer equations, concurs with the current density-potential curves in the linear ranges of the semi-logarithmic graphical representation. As a reproducible result, one obtains the exchange current density as well as the averaged free corrosion potential with the standard deviation normalized to the normal hydrogen electrode (NHE). The tests are each performed in 265 ml electrolyte which is made using deionized water for each test and is buffered by means of 10 ml TRIS-Buffer (Tris(hydroxylmethyl)-aminomethan, C4H11NO3, pH 7.4).

The immersion tests are carried out in a 15 l polymer basin, which is covered during the tests using a polymer lid. As an electrolyte, 10 L of deionized water were used to which sodium chloride concentrations are added according to Table 2 and initially buffered with 100 ml of TRIS-Buffer (pH 7.4). In addition to this, the pH value is measured every 24 h. The specimens are mounted on polymer holders in the basin at an angle of between 15° and 25° (following the arrangement in the salt spray test; DIN EN ISO 9227) and the electrolyte is continuously circulated via immersion pumps. Prior to and after each test, the specimens' masses are gravimetrically measured in order to be able to compute the corrosion rate via the loss of mass. For this purpose, each of the corroded specimens are etched in chromic acid solution (H2CrO4: 200 g CrO3, 10 g AgNO3, 20 g B9(NO0)2) at room temperature for 2 min in order to remove the corrosion deposits.

2.4 Methods of analysis

As an additional analytical and evaluation method, a micro-computed tomograph (µCT, Scanco Medical µCT80) is employed which enables the corroded specimen to be represented in three dimensions. The corroded specimens are scanned in the µCT using a voxel size measuring accuracy of 20 µm. The integration time during the measurement is 400 ms for a tube voltage set to 55 kV. The results of this investigation are density dependent grey-scaled tomographic images of corroded magnesium specimens. These data form the basis for determining the specimen's remaining volume. The grey scale threshold value, for delimiting the specimen volume to be computed from the measured volume, is manually set for every measurement since the X-ray beam is differently attenuated by various parameter ratios between the corroded edge zone and the bulk parent material for individual specimens. Owing to its lower density, the corroded edge zone's linear attenuating coefficient is smaller than that of the non-corroded material. For this reason, the specimen edges in the µCT images are represented as darker grey scales. Here, the threshold value is chosen such that the overall remaining specimen, including the less dense edge zone, completely enters into the volume computation. As examples, the Figs. 1 and 2 depict the evaluation procedure (grey scaled images and threshold value). The result is a 3D-model, which is composed of the tomographic images, with which the volume of the corroded specimen is quantitatively represented.

image

Figure 1. Magnification of the edge region in the 2D tomographic image grey scales

Download figure to PowerPoint

image

Figure 2. Magnification of the edge region in the 2D tomographic image using the set threshold value (145)

Download figure to PowerPoint

3 Results

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

3.1 Micrographs

The input material was casted, extruded as well as machined at the IW. Typical micrographs after etching the samples with a mixture containing 95 ml ethanol, 3 ml hydrochloric acid and 2 ml nitric acid are given in Fig. 3 (transverse section) and Fig. 4 (cross section). Grain refinement is detected for higher aluminium contents as well as a grain orientation towards the pressing direction, which is more distinctive for aluminium contents above 2.0 wt%. Aluminium contents below 1.0 wt% have less influence on the grain structure and orientation.

image

Figure 3. Transverse sections of the Mg[BOND]Al alloys, reflected light, light-field, etched with a mixture containing 95 ml ethanol, 3 ml hydrochloric acid and 2 ml nitric acid

Download figure to PowerPoint

image

Figure 4. Cross sections of the Mg[BOND]Al alloys, reflected light, light-field, etched with a mixture containing 95 ml ethanol, 3 ml hydrochloric acid and 2 ml nitric acid

Download figure to PowerPoint

3.2 Electrochemical measurements

Figure 5 shows some typical polarization curves for binary Mg[BOND]Al alloys and the pure magnesium reference for four different sodium chloride solution concentrations (0.05%, 0.50%, 2.50% and 5.00% sodium chloride). It can be seen, that there is a significant growth of the current density for high concentrations of sodium chloride as well as a slight growth with an increasing aluminium content, especially for 3.0 wt% Al, which has the lowest resistance to corrosion. For low sodium chloride contents, the cathodic polarization branches of the binary Mg[BOND]Al alloys are quite similar, in their position as well as in their general behaviour of the curve. Again for 3 wt% Al particular variations can be found. Both, the cathodic and anodic branches, were used to evaluate the corrosion rates, which was difficult because the branches are not linear in all cases. Therefore, the data was fitted using the Matlab-Program. The current densities icorr evaluated from the icorr data for all measurements are summarized in Table 3. Here, it can be clearly seen that the current density as well as the corrosion rate rises with increasing concentrations of sodium chloride. Simultaneously, there is a tendency that elevating the aluminium content leads to a significant deterioration in the corrosion resistance (Fig. 6).

image

Figure 5. Polarization curves for binary Mg[BOND]Al alloys and the pure magnesium reference for four different sodium chloride solution concentrations

Download figure to PowerPoint

Table 3. Current density icorr for the calculation of Pi (Equation (1)) of Mg alloys in different electrolytic solutions in mA/cm2
Alloy0.05% NaCl0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure0.170.180.420.490.801.00
Al0.50.180.220.580.550.790.90
Al1.00.170.210.510.590.700.91
Al1.50.200.210.600.612.200.89
Al2.00.170.300.700.722.201.20
Al3.00.510.201.301.995.0010.00
image

Figure 6. Graphic representation of the binary Mg[BOND]Al alloys' corrosion rate Pi and the pure magnesium reference in different sodium chloride solution concentrations

Download figure to PowerPoint

Compared with the pure magnesium reference, the corrosion rate depends only marginally on the aluminium content for sodium chloride solutions below 1%. With increasing fractions of aluminium, the corrosion rate is over-proportionally elevated for rising concentrations of sodium chloride.

3.3 Immersion test

Table 4 shows a summary of the mean mass losses ΔW of the investigated magnesium–aluminium alloys with increasing aluminium content for different concentrations of the testing media's sodium chloride. The corrosion rates computed from the mass loss of the individual alloys significantly rise with increasing aluminium content (Table 4 and Fig. 7). One exception to this is the alloy Al2.0 whose corrosion rate lies above that of Al3.0. This is confirmed by the literature; only a small addition of aluminium leads to the deterioration of the corrosion resistance for magnesium–aluminium alloys. One can assume from this that small additions of aluminium lower the tolerance limit for iron and generates effective cathodes in the magnesium matrix via the formation of Al3Fe. These cathodes lead to a significant increase in the corrosion rate (inter alia [8]).

Table 4. Mean weight loss ΔW for the calculation of the corrosion rate Pw (Equation (2)) of Mg alloys in different electrolytic solutions in mg/cm2/day
Alloy0.0% NaCl0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure2.191.972.052.842.282.83
Al0.53.5211.8619.9320.1920.4917.01
Al1.01.7913.7720.5425.1225.6927.49
Al1.55.8421.3127.6929.6230.2231.71
Al2.018.1026.1034.7034.6441.4341.80
Al3.013.5524.0829.4632.9636.7642.14
image

Figure 7. Graphic representation of the binary Mg[BOND]Al alloys' and the pure magnesium reference's corrosion rate Pw in different sodium chloride solution concentrations of the immersed samples

Download figure to PowerPoint

Elevating the sodium chloride concentration in the testing solution leads to an increase in the pH value after exposure in sodium chloride. All solutions were initially buffered (7.38 < pHstart < 7.60) by means of a TRIS buffer so the original conditions used to be almost the same. Due to the reaction of the buffering solution resp. the chemical ingredients with sodium chloride there are slight differences in the pH value after buffering the solutions. In the final state, the pH value differs between pH 7.71 and pH 9.81 dependent on the test parameters. It is clear that for high sodium-chloride additions in the testing solution, a significant increase in the mass loss as well as the corrosion rates occur independently of the alloy's aluminium content (Table 4 and Fig. 7).

If one compares the corrosion rates Pi and Pw, it becomes clear that a discrepancy is found between the differently measured corrosion rates. This is confirmed by the ratio Pw/Pi (Table 5). No explicit relationship can be established between the size of the relative deviation and the testing parameters aluminium or sodium chloride content. However, the essential tendencies to maintain a significant elevation of the corrosion rate with increasing aluminium content or the electrolyte's sodium chloride concentration can be verified from the employed measuring method.

Table 5. Ratio Pw/Pi of the different measuring methods
Alloy0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure1.010.450.530.260.26
Al0.54.962.683.372.381.74
Al1.06.023.703.913.372.78
Al1.59.334.244.461.263.27
Al2.07.994.564.421.733.20
Al3.011.072.081.520.680.39

3.4 Computed tomography-based analysis

Volume models of the non-corroded specimen's volume are calculated via the computed tomographic analysis. This enables the tested specimens to be virtually represented in three dimensions in the computer and the volume to be computed. Using alloy Al1.0 as an example; Fig. 8 shows how the specimens behave in the different electrolyte solutions possessing 0.0% to 5.0% sodium chloride. Each of the non-corroded specimen volumes is represented after 7-day test period. Whereas only a slight corrosive attack can be observed after 7 days in deionized water (0.0% sodium chloride), first and foremost around the periphery of the cylindrical volume, a significant decrease in volume over the entire specimen surface can already be established in 0.09% sodium chloride. As expected, the corrosive attack strengthens with increasing sodium chloride concentration in the electrolytes; a significant deviation from the original geometry can already be discerned above 0.5% sodium chloride (Fig. 8). The visual impression is confirmed by means of the non-corroded volume's computation and the mass loss based on these volumes (Table 6). Each sample analysed with the µCT was taken from the immersion tests before. Therefore, a direct comparison between the results by weighing after exposure to sodium chloride and the calculations based on the results achieved from the virtual models of the µCT is possible. The mass loss from the immersion tests (Table 7) was used to calculate the corrosion rate Pw and the corroded volumes based on the virtual models (Table 6) were used to compute the mass loss and finally the corrosion rate PwµCT. Since individual specimens had already completely dissolved during the exposure to sodium chloride, it was not possible to scan these using the µCT so that these results are missing.

image

Figure 8. Volume models of the non-corroded specimen's volumes of the alloy Al1.0 after 7-day immersion in saline solutions

Download figure to PowerPoint

Table 6. The Mg alloys' mass loss ΔW in different electrolyte solutions in mg/cm2/day calculated from the computed tomographic analysis of the non-corroded specimen's volume
Alloy0.0% NaCl0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure1.841.641.552.171.892.32
Al0.53.066.387.948.598.126.96
Al1.01.075.768.3310.6210.6311.14
Al1.51.398.9412.1612.7212.91
Al2.07.4511.63
Al3.05.8710.6712.78
Table 7. The “reference mass loss” ΔW (weighted) in different electrolyte solutions in mg/cm2/day for the comparison with the computed tomographic analysis of the non-corroded specimen's volume
Alloy0.0% NaCl0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure2.252.082.132.822.272.80
Al0.51.987.9615.4718.6918.7114.22
Al1.01.5913.7219.9923.6025.6927.00
Al1.51.8820.1727.4430.0529.55
Al2.017.7426.80
Al3.012.4622.6729.44

Based on the computation of the corroded specimen's volume, the loss in mass can be determined. With the aid of the loss in mass, the corrosion rate PwµCT is calculated in accordance with Equation (2). In doing this, the observations of the electrochemical and gravimetric investigations are confirmed in as far that the trends in the corrosion behaviour correlate with each other (Figs. 9 and 10). However, in comparison to the corrosion rate Pw, a deviation of about a factor 2 is shown which is reflected in the ratio Pw/PwµCT (Table 8).

image

Figure 9. Graphic representation of the binary Mg[BOND]Al alloys' and the pure magnesium reference's corrosion rates Pw in different sodium chloride solution concentrations of one sample taken from the immersion tests for the comparison with the computed tomographic analysis

Download figure to PowerPoint

image

Figure 10. Graphic representation of the binary Mg[BOND]Al alloys' and the pure magnesium reference's corrosion rate PwµCT in different sodium chloride solution concentrations

Download figure to PowerPoint

Table 8. Ratio Pw/PwµCT of different measuring methods
Alloy0.0% NaCl0.09% NaCl0.5% NaCl0.9% NaCl2.5% NaCl5.0% NaCl
Mgpure1.231.271.371.301.211.21
Al0.50.651.211.952.182.242.04
Al1.01.492.392.402.222.422.42
Al1.51.362.262.242.362.29
Al2.02.382.30
Al3.02.122.122.30

It is not simple for this series of tests to directly compare the corrosion rates Pw and PwµCT resulting from the different measurements of the respective loss in mass (Pw via weighing after the immersion tests and PwµCT calculated via the corroded volume after the exposure to sodium chloride). Nevertheless, the corrosive attack can be visually assessed and characterized via the virtual representation of the non-corroded specimen's volume.

4 Discussion

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

All three methods of measuring lead to differences in the results concerning the corrosion rates, but the overall tendencies can be seen for each method. The reproducibility of the results concerning the electrochemical measurements, the immersion tests as well as the computed tomographic analysis is quite good. There are only small derivations about 0.4% for low sodium chloride concentrations up to 10% for electrolytes containing 2.5% sodium chloride and 5.0% sodium chloride. These derivations are similar to those found in each series of the immersion tests analysed in the presented experiments.

Looking at these results a few sources of error might cause the scatter:

First of all, there is a significant discrepancy between the corrosion rate Pi based on the electrochemical measurements and the corrosion rate Pw found in the immersion tests. There is a problem in the interpretation of the current-density-potential-curves where fitting lines in the linear parts of the cathodic and anodic branches can be difficult. Especially when the linear region is very small, the Matlab-Program is limited. On the other hand during exposure to sodium chloride, the electrolyte amount has an influence on the overall corrosion reaction as well. For the polarization measurements, a chamber containing 256 ml sodium chloride buffered with 10 ml TRIS was used, the immersion tests were carried out in 10 L sodium chloride buffered with 100 ml TRIS. So the ratio electrolyte:TRIS was not the same throughout both measurements what could lead to slight differences in the corrosion behaviour. In comparison to the methods of analyzing (Tafel evaluation and Bulter–Volmer equation), this argument is negligible. The discrepancy between results based on the Tafel evaluation and results by weighing is already known from literature (inter alia [11, 12]).

What is more surprising is the deviation between the corrosion rate based on the immersion test and the corrosion rate calculated from the computed tomographic analysis of the non-corroded specimen's volume. This is astonishing because the sample analysed with the µCT was taken from the immersion tests before. It was expected that the corrosion rates are nearly identical. A source of error could be the voxel size measuring accuracy of 20 µm. A higher resolution should minimize this issue but increase the measuring time significantly. During the time between etching the sample in chromic acid solution and scanning the sample, the specimen is in contact to the environmental air and could react in a corrosive way. Furthermore, the grey scale threshold value, for delimiting the specimen volume to be computed from the measured volume, is manually set for every measurement since the X-ray beam is differently attenuated by various parameter ratios between the corroded edge zone and the bulk parent material for individual specimens. Here, the threshold value was chosen such that the overall remaining specimen, including the less dense edge zone, completely enters into the volume computation. If the transition is not precise some derivations can occur.

Nevertheless, the comparison between the corrosion rates from the immersion tests and the ones from the computed tomographic analysis of the non-corroded specimen's volume leads to a deviation of about a factor 2, which is compared with the electrochemical measurements a good agreement.

5 Conclusions

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

In the electrochemical and gravimetric measurements carried out, it was possible to show that:

  • A low aluminium content significantly lowered the corrosion resistance of binary magnesium–aluminium alloys.
  • Independent of the alloy's aluminium content, the testing electrolytes' sodium chloride-concentration elevates the corrosion rate.
  • A high deviation of the corrosion rate Pi from the corrosion rate Pw exists based on the specific corrosion behaviour of the base magnesium material (negative difference-effect).
  • The computed tomographic analysis and evaluation is eminently suitable for visually assessing and virtually representing the non-corroded specimen's volume.
  • Determining the corrosion rate PwµCT in accordance with the corrosion rate Pw's computation leads to a deviation of about a factor 2. For this reason, this evaluation methodology must initially be supported by using gravimetric measurements and can therefore currently only be concomitantly employed.

Acknowledgement

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References

The authors would like to thank the colleagues at the Institute of Materials Science involved in the investigations and the analytical techniques.

References

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Materials and procedures
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. Acknowledgement
  9. References
  • 1
    S. Bender, E. Boese, A. Heyn, J. Göllner, S. Schultze, K. Schiebold, H. Stöckigt, D. Witte, presented at Werkstoffprüfung, Berlin, Germany, Dezember 01–02, 2005, pp. 492498.
  • 2
    M. Kühlein, U. Galovsky, Mater. Corros. 2004, 55, 6.
  • 3
    C. Kammer, Magnesiumtaschenbuch, Aluminium-Verlag, Düsseldorf 2000.
  • 4
    G. Song, A. Atrens, Adv. Eng. Mat. 2003, 5, 12.
  • 5
    G. Song, A. Atrens, Adv. Eng. Mat. 2007, 9, 3.
  • 6
    E. Aghion, D. Eliezer, Magnesium Alloys – Science, Technology and Applications, The Israel Consortium for the Development of Magnesium, Technology, Haifa 2004.
  • 7
    H. E. Friedrich, B. L. Mordike, Magnesium Technology – Metallurgy, Design Data, Application, Springer Verlag, Berlin, Heidelberg 2006.
  • 8
    S. Bender, Berichte aus der Werkstoffkunde, Shaker Verlag, Aachen 2010.
  • 9
    R. Arrabal, A. Pardo, M. C. Merino, S. Merino, M. Mohedano, P. Casajú, Mater. Corros. 2011, 62, 326.
  • 10
    M. Liu, P. Schmutz, P. J. Uggowitzer, G. Song, A. Atrens, Corr. Sci. 2010, 52.
  • 11
    Z. Shi, M. Liu, A. Atrens, Corr. Sci. 2010, 51.
  • 12
    Z. Shi, A. Atrens, Corr. Sci. 2011, 53.