**3. Results**

The data collected during the extrusion trials are summarized in Table 1. This sets out the data regarding the most meaningful specimens that were analyzed in this study. It should be underlined that the samples numbered from 1 to 15 were processed through a conventional extrusion, and then, the liquid nitrogen valve was opened and the specimens numbered up to 25 were subjected to transient conditions. Thus, only the semi-products numbered from 26 to 40 were extruded with the liquid nitrogen mold cooling system working at full capacity.

**Table 1.** Data featuring the extrusion trials of the specimens. Samples 5, 9, 13 were extruded conventionally. Samples 21 and 25 were processed during the transition step. Specimens 33 and 37 were extruded when the liquid nitrogen mold cooling worked at full capacity. The thermocouples (TC) were positioned within the mold as reported in Figure 3.


The data regarding the characterization of defects are reported below. Two different species of defects were detected on the surface in the junction area of the extruded semi-products. Their SEM-BSE images are shown in Figure 5, and the acquired chemical compositions are quoted in Table 2.

**Figure 5.** SEM-BSE images of extrusion defects detected in the junction area of the billets: sample 5 (**a**), sample 13 (**b**). A, B, C are the point in which the chemical analysis was performed (Table 2).

**Table 2.** SEM-EDS chemical analysis of the areas highlighted in Figure 5, the element concentrations are expressed as wt %.


Therefore, the analyzed defects can be classified as pick-up and dye pick-up. Pick-ups (Figure 5a—point C) are intermittent score lines of varying lengths, which end in a fleck of aluminum debris. They can be easily identified by their chemical composition, which is the same as that of the extruded aluminum alloy [23]. The exogenous deposited layers were composed by a different material and they can be recognized as dye pick-up (Figure 5b—point A, Figure 5a—point A), which are particles worn-out of the extrusion dye and deposited on the semi-product surface [23–27].

Thereafter, the products, subjected to the whole production cycle, including painting and chromium plating, were also analyzed. The typical section, recorded via SEM-BSE, of the surface defects after these processes, is shown in Figure 6. It is worth highlighting that the paint layer does not follow the defect profile; it would thus cover and hide the defects featured by heights up to 15 μm.

**Figure 6.** SEM-BSE images of sections of the final products defects: sample 12 (**a**), sample 25 (**b**).

Through optical microscopy, a metallographic analysis was performed on the tested samples. Figure 7 sets out two micrographs obtained after threating the samples via Barker electrochemical etching. However, this analysis did not highlight any detail about the microstructure, which could be relevant or have influence on the generation of defects. Indeed, at the grain boundaries no secondary phase or trace of segregation phenomena is present [28].

**Figure 7.** Micrographic analysis of the pick-ups presents within the defect area the specimen 5 (**a**) and specimen 33 (**b**).

Moreover, the microstructures appeared homogeneous across the junction area sections (Figure 8a). Thus, no relationship between the microstructure and the surface defects was identified. Moreover, no differences in the morphologies of the crystal grains were detected; all the junction areas of the tested specimens featured coarse equiaxial microstructures, without been affected by any variations in the working parameters.

However, few millimeters away from the junction areas, the microstructures of the specimens were very different (Figure 8b). The grains displayed a lamellar structure, oriented along the extrusion direction, usually featuring the extruded semi-products. This occurred because the process arrest leaves the junction area for a much longer time in contact with the mold after being deformed. Thus, a recrystallization phenomenon was promoted, as testified by the loss of the extrusion deformation pattern of the grains [29].

**Figure 8.** Micrographic analysis of sample 9 in the junction area (**a**) and in a different area representative of the rest of the semi-product (**b**).

Moreover, the grain size number G, an index measuring the average grain dimensions, was calculated through the Heyn intercept method and the results are reported in Table 3.


**Table 3.** Grain size number (G) index.

Furthermore, a SEM-SE morphological analysis wasperformed, to qualitatively assess the differences between the different mold cooling methodologies. The most interesting results concerned billets extruded with the same working parameters. In Figure 9, the junction areas of billets 5-2 (air cooled dye) and 25-4 (liquid nitrogen cooled die) are displayed in detail. Compared to the 25-4 sample, the 5-2 specimen showed much larger defects and a large quantity of the material adhered onto the surface.

**Figure 9.** SEM-SE morphological analysis of the junction areas of the billets comparing one produced through an air cooled die 5-2 (**a**) and one produced via the liquid nitrogen cooled die 25-4 (**b**).

The results of the roughness tests, performed on the surface of the body of the extruded semi-product—hence not in the junction areas—are shown in Table 4. The mean roughness grew with the mold cooling liquid nitrogen valve aperture during the extrusion trials. On the other hand, the measured maximum roughness displayed a decreasing trend under the same operative conditions.

**Table 4.** Roughness tests results (Ra: Roughness Average; Rz: Average Maximum Height of the Profile; Rmax: Maximum Roughness Depth).


The surface texture data, collected via microscopy focus-variation technology, are set out below—both the parameters describing the roughness (Table 5) and the height histogram of the surfaces and its statistics (Figure 10 and Table 6). The results are similar to those for the profile roughness. The mean values of the surface roughness slightly increased with use of the liquid nitrogen cooling system. On the other hand, the parameters describing the maxima and the minima of the surface, underwent a decreasing trend with the use of the liquid nitrogen cooling system.

In detail, the skewness values indicated that the height distribution of the liquid nitrogen-cooled sample was more "valley-tailed" than the traditionally extruded surface. For both distributions, the kurtosis parameter highlighted a leptokurtic behavior with a positive excess kurtosis. The higher value of the sample traditionally extruded testifies the presence of larger tails in its height distribution, due to the occurrence of larger surface defects in the sample produced through this technology, as observed via other characterization techniques as well.


**Table 5.** Surface roughness parameters.

**Table 6.** Statistics of the height histograms of specimens' surfaces.


**Figure 10.** Height histogram of the samples' surfaces: traditionally extruded (specimen 5) and the liquid nitrogen cooled (specimen 37).

#### **4. Discussion**

The experimental data from the current study testify that the liquid nitrogen cooling effect is not instantaneous and a transition time is required before a steady working regime temperature can be reached and maintained. This is related to the mold's thermal inertia. Furthermore, when using this technology, the working temperature becomes lower relative to conventional air cooling of the extrusion mold. However, the differences between temperatures measured at the different dye holes broaden. This phenomenon may be related to the inhomogeneity of the heat removal, linked to the non-optimal sizing of the cooling channel. Therefore, this aspect should be taken into account during the design of the mold, to achieve easy and uniform temperature control [30].

The experimental data regarding the defects analysis in the junction area of the semi-finished extruded billets, found the presence of two varieties of defects. The first featured aluminum flecks of debris adherent to the surface. The chemical composition confirmed that the origin of this debris was the extruded material, since it is identical. The adhesion between the dye and the extruded material generated these defects, as testified by their peculiar morphologies. These defects can thus be recognized as pick-ups [23].

Further to this, a second category of defects was detected and it identified as dye pick-up. This observation was proven by the chemical composition analysis of the deposited material. Indeed, this composition is similar the composition of the AISI H13 tool steel, which is reported in Table 7. Since the extrusion dye is made in this material, its provenance is clearly proved [23–27].

Moreover, any metallurgical origin of the analyzed defects can be excluded. Indeed, both the SEM/BSE analysis and optical microscopy did not detect the presence of any other phases or intermetallic precipitates, any abnormal microstructural features or any abnormal chemical element concentrations. It follows that the origin of the detected defects is linked to wear issues and, in detail, it can be recognized as relating totally to the processing parameters and the tools [23–27].

**Table 7.** Extrusion dye material (AISI H13 tool steel) chemical composition, as designed by standards, the element concentrations are expressed as wt %.


It should be highlighted that no correlation between the variation of extrusion parameters during the production trials and the species defects were identified. The generation of these kinds of defects would not be totally prevented using liquid nitrogen cooling of the extrusion dye. On the other hand, the occurrence of new defect classes can be excluded by using this technology.

In addition, the liquid nitrogen cooling of the extrusion dyes resulted in a beneficial effect on the amount and on the size of the extrusion defects with patent improvements. Indeed, although the analyzed extrusion defects were few and topologically isolated, the use of this technology further lowered their occurrence and dimensions [11,12].

However, considering their application, in which the aesthetic aspect plays a predominant role, even isolated and very small defects would cause the rejection of the extruded semi-product (Figure 11). Taking into account the further steps involved in the realization of the finished product, we can set the defect acceptance threshold at a height of 15 μm. Indeed, since the paint layer does not follow the defect profile, it could cover and hide these small defects.

**Figure 11.** Semi-finished (**a**) and finished (**b**) products: the production cycle, after extrusion, is completed by brushing, acid degreasing, painting and chromium plating.

Considering this finishing process, the most significant roughness data were those describing the maxima of the roughness measurements of the surfaces and the distribution of the height histogram. These data (Tables 4–6 and Figure 10) demonstrate the beneficial effect of the liquid nitrogen cooling of the extrusion dye, and an improvement in this key aspect can be appreciated. In addition, the surface roughness maps of the reconstructed surfaces likewise highlight this aspect (Figure 12). The surface of the liquid nitrogen cooled specimen was very smooth and featured valleys along the extrusion direction. On the other hand, on the surface of the conventionally extruded semi-product, peak-like defects were detected emerging from the surface, harming the subsequent finishing processes.

Liquid nitrogen dye cooling technology improves the extrusion process through a complex mechanism, which can be described as follows: the debris, which generates the previously analyzed defects, is in both cases related to the adhesion phenomena occurring during extrusion between the dye and the flowing aluminum. However, this debris is much less present when the liquid nitrogen cools down the mold, even if the measured temperature is the same as for the air cooling. Hence, the improvement in the surface finishing, obtained through the liquid nitrogen cooling technology, is not only related to the lowering of the working temperature.

A different kind of contribution, which could increase the surface finishing, is given by the mechanical action that the nitrogen exerts on the flowing extruded material. Indeed, while absorbing the heat cooling the dye, nitrogen undergoes a phase transformation from the liquid to the gaseous state. This phase transition is known to be accompanied by a very large increase in volume. However, since the cooling circuit is pressured, the phase transformation generates a detaching pressure on the flowing material, which keeps it separated from the mold.

**Figure 12.** Surface roughness maps. The surface of the conventionally extruded semi-product (**a**) is featured by peak-like defects emerging from the surface. The surface of the liquid nitrogen cooled specimen (**b**) features the presence of valleys along the extrusion direction.

Moreover, liquid nitrogen cooling has another beneficial effect for surface finishing. In detail, it acts as an inert atmosphere enveloping the flowing material, avoiding its oxidation. This plays a central role in wear phenomena, in which aluminum is involved, since its oxide is very easily generated and is extremely hard. On the other hand, in comparison, aluminum is extremely soft and easily damageable with respect to its oxide [4].

Therefore, these beneficial effects on surface finishing, related to the use of liquid nitrogen mold cooling, result from a combination of the cooling effect, the inertizing effect and a mechanical detaching pressure generated by the liquid-to-gaseous transition.
