Bainite

from Wikipedia, the free encyclopedia

Bainite (named after the US metallurgist Edgar C. Bain ) is a structure that can be formed during the heat treatment of carbon steel through isothermal transformation or continuous cooling.

The term interstage structure is also used synonymously for bainite in German-speaking countries. It forms at temperatures and cooling speeds that lie between those for pearlite and martensite formation . In contrast to the formation of martensite, folding processes in the crystal lattice and diffusion processes are coupled here, making various conversion mechanisms possible. Due to the dependence on the cooling rate, carbon content, alloying elements and the resulting formation temperature, the bainite has no characteristic structure. Bainite, like perlite, consists of the phases ferrite and cementite (Fe 3 C), but differs from pearlite in shape, size and distribution. Basically, a distinction is made between two main structural forms, the upper bainite (also granular bainite) and the lower bainite .

Bainitizing or isothermal conversion in the bainite stage is austenitizing with subsequent quenching to temperatures above the martensite start temperature M s . The cooling rate must be chosen so that no conversion can take place in the pearlite stage. When held at the temperature above M s , the austenite changes as completely as possible to bainite.

Slowly turning over the austenite, starting from the grain boundaries or imperfections , produces ferrite crystals that are heavily oversaturated with carbon and have a body-centered cubic crystal lattice ( krz lattice ). In the case of the lower bainite, the carbon is deposited in the form of spherical or ellipsoidal cementite crystals within the ferrite grain due to the higher diffusion speed in the krz lattice. In the case of the upper bainite, the carbon can diffuse into the austenite area and form carbides there.

The upper bainite arises in the upper temperature range of the bainite formation, it has a needle-shaped structure that is very reminiscent of martensite. Due to the favorable conditions for diffusion, the carbon diffuses to the grain boundaries of the ferrite needles. Irregular and interrupted cementite crystals develop here. Because of the random distribution, the structure often has a grainy appearance. If the metallographic analysis is insufficient , the structure can easily be confused with perlite or the Widmanstätten structure .

The lower bainite is created with isothermal and continuous cooling in the lower temperature range of the bainite formation. As a result of the ferrite formation, the austenite is enriched in carbon; with further cooling, the austenite areas transform into ferrite, cementite, acicular bainite and martensite. By austempering be internal stresses is reduced and the toughness is increased, is that this process for crack-sensitive steels and complicated shaped components serving.

The schematic representation of the temperature- and time-dependent transformation areas (here for the production of bainitic cast iron with spheroidal graphite ) also shows the area of ​​bainite formation:

(1) quenching to martensite
(2) intermediate tempering
(3) cooling to bainite
(4) pearlite area
(5) Bainite area

Bainite morphologies

Fig. 1: Bainite morphologies

Bainite is a structure that is created from austenite at temperatures below pearlite formation up to martensite formation, both isothermally and with continuous cooling. A distinction is made between upper and lower bainite. Upper bainite consists of needle-shaped ferrite arranged in packages. Between the individual ferrite needles there are more or less continuous films of carbides parallel to the needle axis. Lower bainite, on the other hand, consists of ferrite plates within which the carbides are formed at an angle of 60 ° to the needle axis. Under certain transformation conditions, other bainite morphologies such as inverse, granular or long-needle bainite can also arise, as illustrated in Figure 1 (bainite morphologies).

Definitions of bainite

There are currently three different definitions of bainite in the literature, which lead to considerable misunderstandings. One distinguishes

  • the microstructural definition,
  • the kinetic definition and
  • the surface relief definition,

which are linked to special phenomena of the phase transition, so that their general validity or invalidity cannot be easily decided.

The microstructural definition

According to this, bainite is regarded as a non-lamellar product of the eutectoid decay of ferrite and carbide in iron-based materials. The two product phases are formed one after the other in a diffusion-controlled manner, with the carbides either precipitating in the ferrite formed first or at its interface. If the separation of the second phase is missing for thermodynamic or kinetic reasons, as is possible with the conversion of silicon-containing steels, one should actually no longer speak of bainite according to this definition. The stipulations made, on the other hand, allow us to speak of bainitic transformations even in the case of non-ferrous metals.

The kinetic definition

This definition is based on the assumption that in the isothermal and continuous ZTU diagram for the beginning and end of the bainite transformation, curves that can be distinguished from those of the pearlite transformation occur, and thus the bainite has its own kinetics of formation. The bainite transformation is supposed to be separated from the pearlite transformation by an area with slow transformation, the expansion of which is strongly influenced by alloying elements. Since bainite can be detected in some steels despite the lack of the inert area, this definition proves to be unsuitable.

The surface relief definition

The relationship between the bainitic transformation and the martensitic transformation can be seen in the appearance of a surface relief. This is compatible with viewing bainite as a plate-shaped phase that is created above the martensite start temperature M s by shearing from the austenite lattice. The conversion takes place through a coordinated, non-thermally activated atom transfer across the moving phase interface . The kinetics of the transformation is determined by the diffusion of interstitial atoms in the austenite, which can take place both before and after the shear. This surface relief definition is currently the most common bainite definition.

Nucleation

Fig. 2: Subdivision of a bainite needle into sub-units

Bainite needles (sheaves) are elongated plates, the thicker ends of which begin at grain boundaries. They include ferritic subunits that are more or less completely separated from each other by carbides or retained austenite, as indicated in Figure 2. The contiguous subunits are separated from each other by small-angle grain boundaries and, in turn, show an elongated slat or plate shape, which, according to Nabarro, is most favorable for phases formed in a stress field (see also the electron microscopic representation in Figure 3). It is currently agreed that in unalloyed hypoeutectoid and silicon-containing hypereutectoid steels, the formation of the lower and upper bainite begins with a carbon-oversaturated ferrite nucleus. Only in silicon-free unalloyed hypereutectoid steels can cementite be the first phase formed at higher transformation temperatures. One then speaks of inverse bainite.

The ferrite nucleation of the bainite usually takes place at the austenite grain boundaries due to thermal lattice vibrations through cooperative lattice shear and, more rarely, at other lattice defects. Once a critical radius has been exceeded, the germ becomes capable of growth and forms a subunit. New nucleation sites (sympathetic nucleation) form at the interfaces of the first bainite nucleus. Nucleation in austenite is possible despite the increased carbon content there, since a high-energy α-γ interface is replaced by a low-energy α-α interface so that the energy required for nucleation is available. The rate of nucleation increases with increasing subcooling below the equilibrium temperature. In return, the subunits become smaller and more numerous, because the growth of the subunits stops as soon as new ones germinate at their phase boundaries. The size of the subunits is independent of the austenite grain size and the bainite needle size. The latter is limited by the austenite grain boundaries and existing needles. In contrast, Olson, Bhadeshia and Cohen assume in a more recent work that the nucleation of bainite, like that of martensite, is based on the presence of preformed nuclei. Viable embryos of critical size are assumed so that the problem of nucleation is reduced to the onset of germ growth. The sympathetic nucleation is explained by the fact that the growing bainite needles lead to adaptive deformations in the austenite with dislocation arrangements in the vicinity of the growing needle that correspond to preformed nuclei.

Germ growth

Fig. 3: Structure of 80 Si 13 after 2 h conversion at T u = 350 ° C (21,000 x)

In the temperature range of the bainitic transformation there is practically no diffusion of the matrix atoms, while at the same time there is a high diffusivity of the carbon and nitrogen atoms. The phase interface between austenite and ferrite is partly coherent and can be viewed as being made up of interfacial dislocations. The conversion takes place through thermally activated sliding of this interface through the atomic lattice, with larger movements of the matrix atoms taking place without any change of place. This martensite-like transformation caused by shear is controlled by the diffusion of the interstitial atoms, which is slow compared to the speed of a sliding interface.

Bhadeshia considers the coupled processes of carbon diffusion and lattice shear in connection with the thermally activated movement of the transformation front. During the waiting time of the transformation front in front of an obstacle until the next activating event, diffusion processes can take place that lower the free enthalpy of the product phases and thus increase the driving force for the interface movement. After overcoming the obstacle, the transformation front then runs free again, without being hindered by diffusion processes, until it hits the next obstacle.

This idea is contrasted with a diffusion model in which the growth of the bainitic ferrite is attributed to the diffusion-controlled movement of steps (ledges) in the α-γ interface, i.e. the same mechanism that was also used in connection with the formation of pre-eutectoid ferrite with Widmanstätten- Structure is discussed. Sandvik found, however, that the deformation twins that occur in the deformed austenite are overrun by growing bainite needles and are found as lattice defects in the ferrite. A conversion through diffusion-controlled movement of steps would have to stop at twin boundaries, since the necessary lattice coherence is disturbed there. The adoption of the lattice defect in the ferrite also contradicts a diffusion-controlled conversion. According to the evidence provided by Dahmen, it is important that a surface relief can also result from a diffusion-controlled transformation and is therefore not a clear indication of a shear-controlled transformation.

thermodynamics

The driving force of a conversion is given by the difference between the free energies of the starting phases and the product phases. The equilibrium phases that have the greatest difference between the free enthalpy and the initial phases do not necessarily have to be established. Both martensitic and bainitic transformation lead to a metastable state. The energy content of these states lies above the equilibrium state in a relative minimum and, under certain conditions, can shift towards equilibrium with the release of energy. Such metastable states can be achieved in the bainitic transformation, for example, by means of carbon-rich ferrite in equilibrium with ε-carbide. The occurrence of concentration gradients, as a result of which the differences in the free enthalpy can be very different within the phases, lead to metastable states.

Figure 4: Free enthalpy of the alpha-gamma conversion in equilibrium

Figure 4 shows the dependence of the free enthalpy of phases α and γ on their carbon content. An equilibrium reaction of γ with the carbon concentration X γ takes place to α with the carbon concentration X γα and γ with the carbon concentration X αγ . The two equilibrium concentrations are on a tangent to the equation

which is applied to both the α and the γ parabola.

In order to achieve the equilibrium concentration of X γα in α or X αγ in γ, a strong carbon diffusion is necessary. The free enthalpy of the γ phase decreases somewhat from G γ to G αγ , while the free enthalpy of a volume element converted into α is greatly reduced to G γα . The overall system reduces its free enthalpy by the amount ΔG. The driving force for the conversion is given by ΔG α .

If the conversion conditions are changed, provided that there is sufficient driving force, a non-equilibrium reaction can take place in which different carbon concentrations are set in the product phases of X γα and X αγ . In Figure 5 it is assumed that austenite with concentration X γ converts into ferrite with a concentration X α > X γα . If the conversion took place in a purely diffusion- controlled manner , the driving force ΔG α would be dissipated solely by the movement of the diffusion field in front of the conversion front (ΔG = ΔG α ), at which the concentration X m <X αγ would then be established. If, however, a portion ΔG s is required for the movement of the phase boundary and the shear-related, cooperative atom transfer across the moving boundary, then the concentration X i <X m is established there .

Figure 5: Breakdown of the free enthalpy into shear and diffusion-related parts

The division of ΔG α into ΔG d and ΔG s results from the condition that the diffusion must take place at the same speed as the shear. Due to this coupling of diffusion and shear, the carbon concentrations develop dynamically in front of the moving interface, as Figure 14 shows. The highest carbon concentration of the austenite X i occurs in the interface. From there, the carbon diffuses into the austenite, which increases the carbon content of the austenite X γ (dashed curve in Figure 14). If X γ approaches the value X m , a further reaction becomes impossible, since the enthalpy reduction in the overall system ΔG no longer occurs. The bainitic transformation stops and can only be achieved by lowering X γ z. B. continued by carbide formation or by lowering the temperature.

Retained austenite

Carbide formation from the austenite is a prerequisite for a complete bainitic transformation. Since carbides absorb large amounts of carbon, they represent carbon sinks that suck carbon out of the austenite. Carbon enrichments in the austenite, which - as shown above - would bring the transformation to a standstill, are then no longer possible. If carbide formation is prevented or delayed, for example by using silicon as an alloying element, larger amounts of austenite are not converted. After quenching to room temperature, they are wholly or partially present as retained austenite . The amount of remaining austenite depends on how far the martensite start temperature has decreased in the remaining austenite.

Lower bainite

Fig. 6: Structure of 80Si10 after 4 h transformation at T u = 250 ° C (1200 x)

The lower bainite is formed at relatively low transformation temperatures from below the transition temperature to the upper bainite to below the martensite start temperature. Theoretically, lower bainite can form up to the martensite finish temperature. Figure 6 shows the structure of lower bainite in silicon steel 80Si10.

Education rate

Vasudevan, Graham and Axon determine a change in the conversion kinetics for bainite formation when the temperature falls below 350 ° C and identify the conversion product as lower bainite. This grows with an activation energy of 14,000 cal / mol (0.61 eV), which is discussed as a rate-determining process in connection with carbon diffusion in the oversaturated ferrite. Due to the increasing carbon content, the jump in volume in the α → γ conversion increases with decreasing conversion temperatures.

Radcliffe and Rollason give values ​​of 7,500 to 13,000 cal / mol (0.33 to 0.56 eV) as the activation energy for the formation of the lower bainite, J. Barford those of 14,500 to 16,500 cal / mol (0.63 to 0.72 eV). It is assumed that the lower bainite has its own conversion mechanism.

Carbon split on the transformation front

Figure 7: Carbon gradient near the interface in the lower bainite

At the low transformation temperatures, because of the low diffusibility of the carbon in the austenite and the high transformation rates measured, no appreciable proportion of the carbon can diffuse from the ferrite into the austenite. So initially a martensitic transformation of the austenite takes place with almost full carbon supersaturation, so that the carbon content of the ferrite formed remains almost the same as that of the austenite. Figure 7 illustrates this fact. The high carbon content in the ferrite can be reduced after the conversion either through carbide formation in the ferrite or through diffusion into the retained austenite.

Carbide formation

Figure 8: Schematic representation of the carbide arrangement in the lower bainite
Figure 9: Schematic representation of the formation mechanism of the lower bainite

At first it was thought that when the lower bainite is formed, the carbides are precipitated from the austenite directly at the interface in such a way that the interface energy is minimized. Bhadeshia was able to prove, however, that the carbides separate from the ferrite during the conversion.

Similar to tempered martensite, the carbides form inside the ferrite needles in the same crystallographic directions with angles to the needle axis of around 60 ° (see Fig. 8). This is mostly ε-carbide (Fe 2.4 C), which changes to cementite after a longer transformation period. The precipitation of carbides behind the transformation front lowers the carbon supersaturation of the ferrite and thus the free enthalpy of the structure. The carbide form corresponds to the state of minimal strain energy. The number and fine distribution of the carbides are responsible for the good mechanical properties of the lower bainite.

In connection with the position of the precipitated ε-carbides at an angle of 60 ° to the ferrite needle axis, it was assumed that the precipitates form on deformation twins. However, no correspondence could be established between the orientation of the carbide precipitates and the twin planes in the ferrite. It was therefore assumed that the carbide precipitation takes place in an oriented manner for energy reasons.

However, it has been shown that the growing ferrite creates twins when the austenite is deformed. These twins in austenite are sheared by the transformation front and "transferred" into the krz lattice. Carbides are formed at these lattice defects in the further course of the transformation. This explains why the habit level of the carbide precipitation does not correspond to a twin level in the ferrite.

The mechanism of carbide formation developed by Spanos, Fang and Aaronson is based on long ferrite nuclei (1), as outlined in Figure 9, on which further ferrite units are formed in the second step through sympathetic nucleation (2). The austenite enclosed between the ferrite units is heavily enriched in carbon by diffusion from the ferrite until carbide is formed from the austenite (3). In the last step, the gaps around the carbides close, since further ferrite formation can now take place in the austenite areas, which are now depleted in carbon. Moving small-angle grain boundaries compensate for existing differences in orientation between the ferrite units, so that their former boundaries almost disappear (4).

Orientation relationship

According to Bhadeshia, the Kurdjumov-Sachs orientation relationship predominantly occurs between the austenite and ferrite of the lower bainite.

(2.6)

At the same time there are orientational relationships according to Nishiama-Aquarius.

(2.7)

The two orientational relationships differ by only about 5 °. The orientation relationship between ferrite and cementite applies to the lower bainite

(2.8)

In a more recent study, however, one finds the orientation relationship according to Bagaryatski

(2.9)

Fulfills. Finally, Shackleton and Kelly do not succeed in demonstrating an orientation relationship between cementite and austenite for the lower bainite. This leads to the conclusion that the cementite in the lower bainite is precipitated within the ferrite and not from the austenite.

According to Dorazil, Podrabsky and Svejcar, the ε-carbides show orientational relationships to austenite as well as to ferrite, which are

let describe. After that, it cannot be decided for the ε-carbide whether it is precipitated from the bainitic ferrite or from the austenite.

Retained austenite stabilization

Since there is hardly any carbon partitioning at the temperatures in the lower bainite range, the bainitic reaction can usually take place completely, so that little or no residual austenite remains. However, if the reaction is terminated prematurely by quenching, the austenite that has not yet been converted to bainitic converts to martensitic and, depending on the carbon content and alloy composition, residual austenite can remain.

By adding silicon to the alloy, carbide formation in the C-oversaturated ferrite is suppressed. The carbon therefore diffuses into the as yet unconverted austenite and increases the carbon content there until the bainitic conversion comes to a standstill. The austenite, which has not yet been converted, is so enriched with carbon that it remains as retained austenite even after quenching to room temperature.

Transition temperature from lower to upper bainite

Fig. 10: Influence of the carbon content on the transition temperature from the lower to the upper bainite

Another controversial aspect of bainite formation is the transition temperature from lower to upper bainite. It is assumed that this - as shown in Figure 10 - increases with increasing carbon content from 400 ° C to around 550 ° C at 0.5% by mass. If the carbon content continues to rise, and the conversion rate remains the same, the ferrite formed becomes more oversaturated, so that the carbon diffuses out into the austenite more and more slowly. Accordingly, ever higher transformation temperatures are required for sufficient carbon diffusion into the austenite so that carbide precipitates can form there. If the state of the alloy exceeds the extrapolated Acm line of the Fe-Fe 3 C diagram , the alloy becomes quasi hypereutectoid and carbide is precipitated from the austenite, which corresponds to the formation of the upper bainite. Therefore, the transition temperature drops above 0.7 mass% C to 350 ° C. Below this temperature, carbide precipitation from the austenite is slower than that from the ferrite and lower bainite is formed.

Fig. 11: Schematic Fe-Fe 3 C diagram with drawn metastable Fe-ε-carbide diagram

However, the increase in the transition temperature for small carbon contents, as has been observed, stems from the definition that the transition temperature is the highest temperature at which carbide still precipitates from the ferrite. Since in the course of the formation of the upper bainite, especially after long conversion times, due to the carbon enrichment in the austenite and the increasing supersaturation of the ferrite, carbide can also be deposited in the ferrite, the curve found does not represent the transition of the formation mechanisms. Rather, the transition from the upper to the lower bainite is traced back to the hypothetical Fe-ε-carbide phase diagram. Figure 11 shows that below a transformation temperature of 350 ° C, ε-carbide precipitates out of the ferrite. Accordingly, the transition temperature is fixed at 350 ° C regardless of the carbon content. According to this theory, the excretion of ε-carbide is the most important mechanism for the formation of the lower bainite. The metastable ε-carbide then transforms into stable cementite after longer conversion times.

Figure 12: The starting temperature of the lower bainite in relation to the bainite and martensite starting temperature

Another approach to the transition temperature is proposed as follows: It is assumed that when the transition temperature is undershot, a change in the conversion mechanism takes place, which has its own kinetics and its own operating temperature, which is between the bainite and martensite starting temperature (see Fig 12). Like the other two curves, the transition temperature increases with decreasing carbon content, since the driving force required for the formation of lower bainits and thus the supercooling decreases with the carbon content. The experimentally observed drop in the transition temperature at low carbon contents is assessed here as a hardenability problem. Austenitz falls after a very short time, so that upper bainite is formed when it cools down to the transition temperature. The samples only cool down quickly enough at lower transition temperatures. The precipitation of ε-carbide from the supersaturated ferrite is represented as a race between precipitation and the diffusion of carbon into the austenite. Accordingly, the carbon present in the ferrite is only sufficient for ε-carbide formation in steels with a higher carbon content, which has been confirmed experimentally.

Upper bainite

Fig. 13: Structure of 80Si10 after 4 h transformation at T u = 450 ° C (1200 x)

At transition temperatures below the range of pearlite formation and above the range of formation of the lower bainite, upper bainite is formed. The carbon diffusion in austenite is of crucial importance for this phase transition. Figure 13 shows the structure of the upper bainite in the silicon steel 80Si10.

Education rate

In the temperature range between 350 ° C and 400 ° C, an activation energy of 34,000 cal / mol (1.48 eV) is found for the conversion, which corresponds approximately to that for carbon diffusion in γ-iron (1.34 eV). Above 350 ° C, a constant carbon content of 0.03% is observed in the ferrite, which comes close to the equilibrium concentration. At the same time, a linear change in length of the sample that decreases with increasing transition temperature is observed.

Alternatively, values ​​of 18,000 to 32,000 cal / mol (0.78 to 1.39 eV) or 22,000 to 30,000 cal / mol (0.95 to 1.30 eV) are found for the activation energy of the formation of upper bainite.

Carbon split on the transformation front

Fig. 14: Carbon gradient near the moving phase interface during the bainitic transformation

The ferrite of the upper bainite has a lower carbon content than the austenite from which it was formed, but is still oversaturated. This supersaturation decreases with increasing transformation temperature due to the increasing diffusion into the austenite, which is strongly enriched in carbon by this mechanism. At low transformation temperatures, a carbon content of X m is quickly reached in the vicinity of the interface (see Fig. 14), since the carbon diffusion into the austenite is delayed. The bainitic reaction quickly comes to a standstill and can only continue through renewed sympathetic nucleation. This explains the decreasing width and increasing number of bainite aggregates as the transformation temperature decreases. The high concentration of carbon in austenite is reduced by the formation of carbide. Is carbide formation impossible, e.g. B. due to high Si contents, large amounts of retained austenite remain in the structure.

Carbide formation

Fig. 15: Schematic representation of the carbide precipitation in the upper bainite

If the austenite is enclosed by growing ferrite needles, it accumulates so much that carbides can separate out of the austenite. It is always cementite that is precipitated directly from the enriched austenite. The carbides of the upper bainite are always in the form of more or less continuous carbide films between the ferrite needles (see Fig. 15). As the carbon content of the alloy increases, the ferrite needles become thinner and the carbide films discontinuous and occur more frequently. It is found that the nucleation of the carbides is facilitated by the stresses that arise as a result of the forming of the growing ferrite needles in the surrounding austenite. From the studies of the orientation relationship between carbide, austenite and ferrite, it is concluded that the carbides in the upper bainite are also caused by lattice shear. Aaronson contradicts this theory and shows that both the formation of bainitic ferrite and carbides can be explained by a diffusion-controlled transformation.

Orientation relationship

The orientation relationship according to Nishiyama-Wassermann is observed between austenite and ferrite of the upper bainite, which is also valid for the lower bainite. The Kurdjumov-Sachs relationship can also be valid within the framework of the accuracy of the diffraction images generated. For the orientation between cementite and austenite, Pitsch suggests the relationship

Pickering on the other hand

in front.

According to Pickering, no orientation relationships between ferrite and cementite are observed, from which he concludes that the cementite does not precipitate from the ferrite but from the austenite.

Retained austenite stabilization

If the austenite becomes heavily enriched with carbon, the bainite formation can come to a standstill if the enrichment is not reduced by the formation of carbides. In the context of the kinetic definition of bainite, this phenomenon is referred to as the “phenomenon of incomplete conversion”. In the temperature range of this incomplete conversion, the nucleation of the cementite is hindered. This can be achieved by adding chromium or silicon. In both cases, the enriched austenite proves to be very stable against quenching to room temperature, so that considerable amounts of residual austenite can remain, which significantly influence the mechanical properties of the alloy.

Influence of the alloying elements on bainite formation

Estimating the influence of the alloying elements on bainite formation is relatively complex, as the kinetics of the reactions that occur often do not change proportionally to the proportions of alloying additions. To make matters worse, the elements influence each other in terms of their effect. Alloy components that form a substitution solid solution with the iron phases only have an indirect influence on the bainitic transformation, since no substitution atom diffusion occurs in this temperature range of bainite formation. The growth kinetics of the bainite can be changed by influencing the diffusion rate of the carbon through the alloy element. From a qualitative point of view, the elements manganese , nickel , chromium and silicon lower the bainite start temperature and extend the transformation time. The elements chromium, molybdenum , vanadium and tungsten lead in the ZTU diagram to a separation of the pearlite area from the bainite area and to the formation of an area that is slow to transform.

  • Carbon is the main factor influencing the morphology of bainite. With increasing carbon content, the growth in width of the bainite needles comes to a standstill earlier because of the difficult carbon diffusion. Accordingly, the bainite needles are becoming thinner and more numerous. An increasing carbon content also promotes carbide formation from both the ferrite (in the case of the lower bainite) and from the austenite (in the case of the upper bainite). As the carbon content increases, the incubation time is extended and the bainite start temperature (B s ) is lowered.
  • The incubation time is also extended and B s is reduced by adding chromium . The increase in austenite resistance can lead to the fact that in certain temperature ranges no more conversion takes place over a long period of time and a conversion-slow area occurs.
  • Silicon raises the A C1 and A C3 temperatures of the metastable Fe-Fe 3 C diagram and shifts the eutectoid concentration to lower carbon contents . The kinetics of pearlite and bainite formation is only marginally influenced by silicon. Silicon is practically insoluble in cementite.
  • Manganese greatly increases the austenite stability both in the pearlite and in the bainite stage, which can lead to high retained austenite contents in manganese steels and the transformation times in the bainite stage become relatively long. This improves the through-hardening capability , also for the bainitic transformation . Manganese is soluble in cementite and forms Mn 3 C with carbon with a structure isomorphic to cementite .
  • An addition of nickel leads, such as chromium or manganese to a decrease of B S . In the case of high nickel contents, the range of complete bainitic transformation is strongly constricted, for example to the temperature range up to 10 ° C. above the martensite start temperature with the addition of 4% nickel.
  • Molybdenum increases the A C3 temperature without affecting the A C1 temperature. Above all, it delays the pre-eutectoid ferrite excretion and the formation of pearlite. In the case of large components, this makes it easier to cool down to the temperatures of the bainitic transformation without pre-precipitation of ferrite or pearlite.
  • The formation of ferrite and pearlite is also greatly delayed by boron . The pearlite area shifts to longer transformation times, while the bainite formation remains unaffected. In this way, purely bainitic structures can be created, especially with continuous transformation. It is important that the existing nitrogen is bound by aluminum or titanium , since the boron nitrides that otherwise arise cause embrittlement .

The bainitic transformation in silicon steels

Figure 16: Course of the Si and C concentrations X (left part image) as well as the Si and C activities A (right part image) during the nucleation of cementite

In the case of steels containing silicon, in comparison to the mechanisms of bainitic transformation already described in silicon-free steels, some peculiarities arise, since silicon suppresses the formation of cementite. Since carbide formation is a prerequisite for a complete bainitic transformation, incomplete transformations with high residual austenite contents occur in silicon steels. Investigations on silicon steels can provide important information for the elucidation of the formation of bainitic ferrite, since the conversion products are not changed by a subsequent carbide formation.

Silicon is practically insoluble in cementite. The growth of a cementite nucleus therefore presupposes the diffusion of silicon, which can only take place very slowly at the transition temperatures of bainite formation. A silicon gradient builds up around the cementite nucleus, which locally greatly increases the activity of the carbon (see Fig. 16). This reduces the carbon inflow to the cementite nucleus, so that the nucleus cannot grow any further.

The conversion in the area of ​​the upper bainite takes place in silicon steels in two phases because of the difficult carbide formation. In the first phase, bainitic ferrite is formed at a relatively high rate of formation, with the surrounding austenite being heavily enriched with carbon. In the second phase, which takes a very long time in silicon steels, carbides are then formed from this enriched austenite. By lowering the carbon content in the austenite, ferrite formation can continue, and secondary ferrite is formed as the ferrite needles grow laterally. In the area of ​​the lower bainite, ε-carbides separate out within the ferrite after only short transformation times, since silicon has little effect on ε-carbide formation. Only the conversion of the ε-carbide into cementite is suppressed by the silicon present. Due to the existing carbide formation, the lower bainite has lower amounts of retained austenite than the upper bainite. The carbides found cannot be identified as cementite as they contain significant amounts of silicon. Röhrig and Dorazil also report the occurrence of silicocarbides after a longer transformation in the temperature range of the upper bainite.

With a higher silicon content and transformation temperatures between 350 ° C and 400 ° C, large areas of retained austenite can arise that are only slightly enriched with carbon and have a negative effect on the mechanical properties of the alloy. In the austenite, which is enclosed between growing ferrite needles, there are deformation twins that indicate the locally high carbon content of the austenite between the ferrite needles.

Incomplete conversion phenomenon

One observes that the bainitic transformation proceeds more and more incompletely as it approaches B S , until it comes to a standstill at B S. After a while, in which nothing happens, pearlite begins to form. If, by adding alloying elements, the temperature range of pearlite formation is shifted to higher or bainite formation to lower temperatures, a temperature range arises in which transformations take place only after very long times, if at all. This phenomenon is explained by the suppressed formation of carbide at higher temperatures. The austenite quickly becomes enriched with carbon, so that the transformation comes to a standstill after a short time.

The phenomenon of incomplete transformation also ignites the controversy surrounding the mechanism of bainite formation. Bradley and Aaronson attribute the sluggish area to a "Solute Drag Like Effect" (SDLE). This model assumes that substitution atoms cannot freely diffuse through the atomic lattice in the temperature range of bainite formation, but that they accumulate in the moving phase boundary. If these are elements that lower the carbon activity, the driving force for the diffusion of the carbon from the ferrite into the austenite decreases. This effect lowers the rate of transformation and, in extreme cases, can bring the phase interface moving during the transformation to a standstill after a short time due to the formation of carbides within this interface. In a direct statement, Bhadeshia and Edmonds disagree, since there are examples of alloying elements that lower the carbon activity but do not cause an inert range. Furthermore, the SDLE can only explain the area between bainite and perlite that is slow to transform, but not the second area that is slow to transform, which was found between the lower bainite and the upper bainite.

Mechanical properties of bainitic iron-based alloys

Solidification mechanisms

The most important occurring in the bainitic microstructure solidification smechanismen are the grain boundary strength , the displacement solidification, the solid-solution strengthening and dispersion strengthening .

In the case of grain boundary strengthening, the question arises of how a grain size in the bainitic structure is to be defined. One possibility is the former austenite grain size, which indirectly determines the length of the bainite needles and the size of the packages, which are composed of several needles. Edmonds and Cochrane found no relationship between the austenite grain size and the strength properties, while they found the relationship for the package size

Find.

The width of the individual bainite needles is defined as the grain size and determined

which corresponds to the Hall-Petch relationship. Since the ferrite needles become smaller and more numerous as the transformation temperature falls, the observed increase in strength can be justified.

Depending on the transformation temperature, high dislocation densities of 10 9 to 10 10  cm −2 exist in bainitic ferrite . The dislocation density decreases because of the decreasing deformation of the ferrite with increasing transformation temperatures. It is higher, the more carbides are present.

Only some of these dislocations take part in the plastic deformation as sliding dislocations. Their movement through the metal lattice is hindered by the spatial structure of the non-slip dislocations, the dissolved foreign atoms, the carbides as well as grain and phase boundaries. The proportion of dislocation hardening can be quantified

estimate. α 1 is a constant, the shear modulus, the amount of the Burgers vector and ρ the total dislocation density .

The interaction between slip dislocations and interstitial or substitution atoms in the respective slip planes lead to a stress component

where α 2 and are constants and the impurity concentration. The carbon dissolved in the bainitic ferrite increases as the transformation temperature falls, which leads to increased solid solution strengthening.

The carbides in the upper bainite only influence the strength properties to the extent that they promote the formation and propagation of cracks. They do not interact with the sliding dislocations because they are located at the ferrite needle boundaries. In the lower bainite, the carbides formed in the ferrite cause precipitation hardening, which reduces the stress component

supplies. It is n e , the number of carbide particles per mm 2 and , constants.

The mixing rule is used to determine the strength properties of mixtures of different phases

suggested. It turns V i the volume fraction of the microstructure i and σ i represents the resistance characteristic value. This estimate has been found suitable for the mixture of upper bainite and martensite. However, greater inaccuracies occur when mixing lower bainite with martensite. The mixture of bainite with retained austenite can be assessed according to this formula, as long as the retained austenite does not convert to martensite.

Influence of retained austenite on mechanical properties

It can be seen that the amount and morphology of retained austenite has a strong effect on the toughness properties of steels with different levels of silicon content due to the high ductility and transformability of retained austenite. During the deformation of states with a higher carbon concentration, the retained austenite transforms into martensite, while during the deformation of states with a lower carbon content, twinning is observed in the austenite. The amount of retained austenite at which the greatest elongation at break occurs is given as 33 to 37% by volume. Samples with a higher retained austenite content (up to 50% by volume) again show poorer toughness properties. The reason for this behavior lies in the morphology of the retained austenite. With lower retained austenite contents, the retained austenite lies in a film-like manner between the ferrite needles and acts as a sliding film between the harder ferrite aggregates, thereby improving the deformability of the structure. The contribution of the retained austenite to the overall deformation is very high because of the expansion-induced martensite formation, so that an increase in the retained austenite amount is equivalent to an improvement in the elongation at break. With higher amounts of retained austenite, the retained austenite is arranged in blocks and its deformation mechanism changes from the expansion-induced martensite formation to deformation through the formation of twins. Since the proportion of retained austenite, which is arranged in blocks, increases with a further increase in the retained austenite content, this leads to lower elongation at break from a retained austenite amount of 37% by volume. This relationship is also responsible for the K IC value , which decreases with increasing conversion temperature .

Deformation and strength behavior

The isothermal bainite conversion offers a number of advantages. In the area of ​​the lower bainite, in addition to high strengths, very good toughness properties are achieved, as shown for steels with a carbon content of 0.1 to 1%. The chromium content was varied from 0 to 1% and the silicon content from 0.1 to 0.6%. At transformation temperatures of 400 to 600 ° C, a yield strength ratio of 0.6 to 0.8 was determined. For tensile strengths above 850 N / mm 2 , the steels converted in the bainite stage showed a superior ductility compared to normally tempered steels. These very good mechanical properties of bainite are retained down to the lowest temperatures. In addition, the elongation at break, necking at break and notched impact strength are higher than with comparable strength after normal tempering. The creep resistance, fatigue swinging strength and fatigue life are positively influenced by this thermal treatment process.

The transition from the lower to the upper bainite causes a jump in the transition temperature of the impact strength. The upper bainite shows the higher transition temperatures, which is due to the different carbide structure. The facet size of the split fracture surfaces corresponds to the size of the bainite colonies. Any martensite that may be present leads to a reduction in the facet size.

Sometimes bainitic steels show a very low yield strength . Schaaber blames an incomplete conversion for this. According to his investigations, the yield point only reaches its maximum when the highest possible degree of conversion is achieved. In addition to the yield point, the fatigue strength is particularly sensitive to incomplete conversion.

Materials with a bainitic structure are successfully used for valve and disc springs, as the bainitic structure has advantages in terms of fatigue strength and fatigue strength of these components. It can be shown that the fatigue strength of bainitically converted samples is higher than that of tempered samples with comparable tensile strength. It is important to ensure that the bainitic transformation is as complete as possible. The bainitic structure is characterized by the fact that it can effectively reduce stress peaks generated by internal or external notches and cracks.

The conversion in the bainite is not only interesting because of the good mechanical properties, but also in terms of a low distortion and virtually crack-free hardening heat treatment. As a result of the relatively high transformation temperatures, both the quenching and the transformation stresses are very much lower than with conventional hardening. In addition, the transformation in the bainite stage is associated with considerably smaller changes in volume than the martensitic transformation.

Cyclic deformation behavior at room temperature

According to Machera, a distinction can be made between four stages of fatigue in the cyclical loading of steels: the elastic-plastic alternating deformation stage, the microcrack formation stage, the stage of stable crack propagation and finally the fatigue fracture . In hardened steels, the alternating deformation stage predominates and microcracking only occurs shortly before fatigue failure. With normalized or quenched and tempered steels, the stable crack propagation can cover a considerable part of the service life, depending on the level of stress.

Fig. 17: Nominal stress total expansion hysteresis loop with the parameters of the alternating deformation

In the case of elastic-plastic alternating deformation, the stress-total strain relationship provides hysteresis loops, from which various parameters can be derived from a sufficiently stabilized material behavior according to Figure 17. With a stress-controlled test, the total strain amplitude ε a, t and the plastic strain amplitudes ε a, p can be determined as a function of the number of load cycles N. Cyclical solidification (softening) is associated with a decrease (increase) of ε a, p and thus also of ε a, t . When the test is carried out under total strain control, however, the stress amplitudes σ a and the plastic strain amplitudes ε a, p appear as dependent variables. A cyclical solidification (softening) is associated with an increase (decrease) in σ a and a decrease (increase) in ε a, p . If the dependent variables are plotted as a function of the logarithm of the number of load cycles for a given stress amplitude, so-called alternating deformation curves result. If one takes from these associated value pairs of σ a and ε a, p or ε a, t and plots them against each other, one obtains the cyclic stress-strain curve. Cyclical yield and elongation limit values ​​can be taken from this like a stress-strain curve of the tensile test.

The alternating deformation curves allow conclusions to be drawn about the material behavior during cyclic loading. Normalized steels usually show, after a quasi-elastic incubation period, a load cycle interval of strong alternating softening, which is followed by a service life range with alternating hardening. The alternating softening observed can be attributed to the occurrence of elongation inhomogeneities that run as fatigue Lüders bands over the measuring section.

Quenched and tempered steels also show a strong alternating softening after an incubation period, which lasts until cracks form. As the voltage amplitude increases, both the incubation time and the service life decrease. Since the formation of new dislocations is unlikely due to the very high density of dislocations, the plastic deformations that occur must be generated by rearranging the existing dislocation structure. In hardened material states, there are increased opportunities for the dislocations to interact elastically with the carbon atoms dissolved in an increased non-equilibrium concentration, which leads to alternating hardening. Since the proportion of dissolved carbon is reduced by tempering, the possibilities of interaction of the dislocations with the carbon atoms are reduced and the transformation of the dislocation structure leads to alternating softening.

The cyclic plastic deformations at the crack tip are decisive for stable crack propagation. The crack propagation is determined by the range of stress intensity ΔK. The increase in crack length per load change is due to

where c and n are constants. A double logarithmic plot of da / dN over ΔK results in a linear relationship. Crack growth no longer occurs below a threshold value of ΔK. At very high ΔK values, unstable crack propagation leads to breakage.

literature

Web links

Commons : Bainit  - collection of images, videos and audio files

Individual evidence

  1. ↑ RF Mehl: The Physics of Hardenability - the Mechanism and the Rate of the Decomposition of Austenite . Hardenability of Alloy Steels Symposium 20th Annual Convention of the American Society for Metals, Detroit 1938.
  2. ↑ JM Oblak, RF Hehemann: Structure and Growth of Widmannstaetten - Ferrite and Bainite . In: Transformation and Hardenability in Steels . Climax Molybdenum Symposium, 1967.
  3. ^ FB Pickering: Structure and Properties of Bainite in Steels . In: Transformation and Hardenability in Steels . Climax Molybdenum Co, Michigan 1967, pp. 109-132.
  4. ↑ LJ Habraken, M. Econopoulos: Bainitic Microstructures in Low Carbon Alloy Steels and Their Mechanical Properties . In: Transformation and Hardenability in Steel . Climax Molybdenum Co, Michigan 1967, pp. 69-108.
  5. ↑ G. Spanos, HS Fang, DS Sarma, HI Aaronson: Influence of Carbon Concentration and Reaction Temperature upon Bainite Morphology in Fe-C-2 Pct Mn Alloys . Metallurgical Transactions A Vol 21A, 1990, pp 1391-1411.
  6. ^ HI Aaronson, HJ Lee: Another Visit to the Three Definitions of Bainite . Scripta Metallurgica, 1987, pp. 1011-1016.
  7. ↑ HI Aaronson, WT Reynolds Jr., GJ Shiflet, G. Spanos: Bainite Viewed Three Different Ways . Metallurgical Transactions A vol 21A, 1990, pp. 1343-1380.
  8. ↑ HI Aaronson: Bainite Reaction . Encyclopedia of Materials Science and Engineering Vol 1, 1986, pp. 263-266.
  9. ^ SK Liu, WT Reynolds, H. Hu, GJ Shiflet, HI Aaronson: Discussion of "The Bainite Transformation in a Silicon Steel" . Metallurgical Transactions A vol 16A, 1985, pp. 457-467.
  10. ^ SJ Matas, RF Hehemann: The Structure of Bainite in Hypoeutectoid Steels . Transactions of the Metallurgical Society of AIME Vol 221, 1961, pp 179-185.
  11. ^ GR Srinivasan, CM Wayman: The Crystallography of the Bainite Transformation . Acta Metallurgica 16, 1968, pp. 621-636.
  12. ↑ HKDH Bhadeshia, JW Christian: Bainite in Steels . Metallurgical Transactions A Vol 21A, 1990, pp 767-797.
  13. ^ FRN Nabarro: The Influence of Elastic Strain on the Shape of Particles Segregating in an Alloy . Proceedings of the Physical Society 52, 1940, pp. 91-104.
  14. ^ HI Aaronson, C. Wells: Sympathetic Nucleation of Ferrite . Transactions of the Metallurgical Society of AIME, 1957, pp. 1216-1223.
  15. ↑ GB Olson, HKDH Bhadeshia, M. Cohen: Coupled Diffusional / Displacive . Transformations Acta Metallurgica Vol 37 No 2, 1989, pp 381-389.
  16. ^ JW Christian: Simple Geometry and Crystallography Applied to Ferrous Bainits . Metallurgical Transactions A Vol 21A, 1990, pp 799-803.
  17. ↑ HKDH Bhadeshia: A Rationalization of Shear Transformation in Steels . Acta Metallurgica Vol 29, 1981, pp. 1117-1130.
  18. ^ RF Hehemann, KR Kinsman, HI Aaronson: A Debate on the Bainite Reaction . Metallurgical Transactions Vol 3, 1972, pp 1077-1094.
  19. ↑ BPJ Sandvik: The Bainite Reaction in Fe-Si-C Alloys: The Primary Stage, the secondary stage . Metallurgical Transactions A Vol 13A, 1982, pp 777-800.
  20. ↑ U. Dahmen: Surface Relief and the Mechanism of a Phase Transformation . Scripta Metallurgica Vol 21, 1987, pp 1029-1034.
  21. ^ J. Burke: The Kinetics of Phase Transformations in Metals . Pergamon Press, London / New York / Paris 1965.
  22. ↑ JN Hobstetter: Decomposition of Austenite by Diffusional Processes . Interscience Publishers, New York / London 1962.
  23. ^ GB Olson: Interphase Kinematics and the Roles of Structure and Composition in Solid-State Transformations . Scripta Metallurgica Vol 21, 1987, pp 1023-1028.
  24. ↑ HKDH Bhadeshia: Diffusional and Displacive Transformations . Scripta Metallurgica Vol 21, 1987, pp 1017-1022.
  25. ↑ P. Vasudevan, LW Graham, HJ Axon: The Kinetics of Bainite Formation in a Plain Carbon Steel . Journal of the Iron and Steel Institute, 1958, pp. 386-391.
  26. ^ SV Radcliffe, EC Rollason: The Kinetics of the Formation of Bainite in High-Purity Iron-Carbon Alloys . Journal of the Iron and Steel Institute, 1959, pp. 56-65.
  27. ^ J. Barford: Kinetic Aspects of the Bainite Reaction . Journal of the Iron and Steel Institute, 1966, pp. 609-614.
  28. ↑ HKDH Bhadeshia, D. Edmonds: The Bainite Transformation in a Silicon Steel, Part I + II . Metallurgical Transactions Vol 10A, 1979, pp. 895-907.
  29. ↑ HI Aaronson, MR Plichta, GW Franti, KC Russel: Precipitation at Interphase Boundaries . Metallurgical Transactions Vol 9A, 1978, pp. 363-371.
  30. ↑ a b H.KDH Bhadeshia: The Lower Bainite Transformation and the Significance of Carbide Precipitation . Acta Metallurgica Vol 28, 1980, pp 1103-1114.
  31. ↑ G. Spanos, HS Fang, HI Aaronson: A Mechanism for the Formation of Lower Bainite . Metallurgical Transactions A Vol 21A, 1990, pp 1381-1390.
  32. ↑ DN Shackleton, PM Kelly: The Cristallography of Cementite Precipitation in the Bainite Transformation . Acta Metallurgica Vol 15, 1967, pp. 979-992.
  33. ↑ E. Dorazil, T. Podrabsky, J. Svejcar: Investigation of the bainite transformation in silicon steel . Archiv für das Eisenhüttenwesen 53/7, 1982, pp. 289–293.
  34. ↑ RWK Honeycombe, FB Pickering: Ferrite and Bainite in Alloy Steels . Metallurgical Transactions Vol 3, 1972, pp. 1099-1112.
  35. ↑ KJ Irvine, FB Pickering: High-Carbon Bainitic Steels Special Report 93: Physical Properties of Martensite and Bainite . The Iron and Steel Institute, 1965, pp. 110-125.
  36. ^ HI Aaronson: Discussion of "The Bainite Reaction in Fe-Si-C Alloys: The Primary Stage" and "... the Secondary Stage". Metallurgical Transactions A Vol 17A, 1986, pp. 1095-1100.
  37. ↑ Pitsch: The orientational relationship between cementite and austenite . Acta Metallurgica 10, 1962, p. 897.
  38. ↑ HI Aaronson, HA Domian: Partition of Alloying Elements Between Austenite and Proeutectoid Ferrite or Bainite . Transactions of the Metallurgical Society of AIME 236, 1966, pp. 781-797.
  39. ^ I. Stark, GDW Smith, HKDH Bhadeshia: The Distribution of Substitutional Alloying Elements During the Bainite Transformation . Metallurgical Transactions A Vol 21A, 1990, pp 837-844.
  40. ^ Habraken: silicon in steel, manganese in steel. In: De Ferri Metallurgraphica II . Volume II: Structure of the steels . Editions Berger-Levrault, Paris / Nancy 1966.
  41. ↑ R. Chatterjee-Fischer: Overview of the transformation in the bainite stage and its application . Draht-Fachzeitschrift 26, 1975, pp. 618-622.
  42. ↑ W. Eilender, R. Mintrop, W. Lutz: Investigations into the interstage tempering of hot-work steels . Stahl and Eisen 72, 1952, pp. 1149-1156.
  43. ↑ A. Massip, L. Meyer: Heavy plate and hot strip made of bainitic steels with a very low carbon content . Stahl and Eisen 98/19, 1978, pp. 989-996.
  44. ↑ S. Ehrlich: Transformation behavior and structure of the bainitic-austenitic steel 80 Si 10 . Diploma thesis University of Karlsruhe, 1990.
  45. ↑ BPJ Sandvik: An Experimental Study of Bainite Formed in Fe-Si-C alloys . American Society of Metals. Material Science Division. Transformation Committee phase. International Conference on Solid-Solid Phase Transformations. Pittsburgh 1981, pp. 1023-1027.
  46. ↑ K. Röhrig: Isothermal conversion of cast iron with spheroidal graphite in the bainite stage . Härterei Technische Mitteilungen 39/2, 1984, pp. 41-49.
  47. ↑ E. Dorazil, J. Svejcar: Investigation of the upper bainite on silicon steel . Archiv für das Eisenhüttenwesen 50/7, 1979, pp. 293-297.
  48. ↑ JR Bradley, HI Aaronson: Growth Kinetics of Grain Boundary Ferrite Allotriomorphs in Fe-CX Alloys . Metallurgical Transactions A Vol 12A, 1981, pp. 1729-1741.
  49. ↑ HKDH Bhadeshia, DV Edmonds: Authors' Reply . Metallurgical Transactions A Vol 16A, 1985, pp 466-468.
  50. ^ DV Edmonds, RC Cochrane: Structure-Property Relationships in Bainitic Steels . Metallurgical Transactions A Vol 21A, 1990, pp 1527-1540.
  51. ↑ D. Eifler: Inhomogeneous deformation phenomena when vibrating a differently heat-treated steel of the type 42 Cr Mo 4 . Dissertation, University of Karlsruhe, 1981.
  52. ↑ O. Schaaber: About influencing factors in the isothermal austenite transformation in the intermediate stage (bainite area I and II) . Draht, Coburg No. 1, 1952, pp. 7-13 (I) and Draht, Coburg No. 5, 1952, pp. 127-137 (II).
  53. ↑ H. Tauscher: The influence of the intermediate level tempering on the fatigue strength of steel . Draht 15, 1964, pp. 519-521.
  54. ^ E. Macherauch: Internship in materials science . 6th edition, Vieweg Verlag, Braunschweig 1985, pp. 252-291.
  55. ↑ D. Munz, K. Schwalbe, P. Mayr: Fatigue behavior of metallic materials . Vieweg-Verlag, Braunschweig 1971.
  56. ↑ E. Macherauch, P. Mayr: The structural mechanics of the fatigue of iron-carbon alloys . Habilitation thesis, University of Karlsruhe, 1979.
  57. ↑ D. Eifler, E. Macherauch: Inhomogeneous deformation phenomena when the heat-treated steel 42 Cr Mo 4 is exposed to vibrations . Zeitschrift für Werkstofftechnik 13, 1982, pp. 395-401.
This version was added to the list of articles worth reading on January 25, 2008 .