2. Materials and Methods
The Vit105 (Zr52.5Cu17.9Ni14.6Al10Ti5 at. %) BMG plates with a size of 60 mm × 10 mm × 2 mm were used in this study. The plates were produced by squeeze casting into a massive copper mold with a melt cooling rate of K/s.
HPT processing was carried under a pressure GPa on ⌀10 mm anvils with a 0.3 mm depth groove, with a rotation speed of 1 rpm, at room temperature (RT). Samples with a diameter of 10 mm and a thickness of 1 mm were cut out from the as-cast BMG plates for HPT processing.
Part of the BMG disks were subjected to conventional HPT for
,
and
full revolutions. Some disks were subjected to the accumulative HPT (ACC HPT) procedure, which was reported in [
38]. During the ACC HPT procedure, at the first stage, a disk-shaped sample with a thickness of
mm and ⌀10 mm were subjected to HPT for two anvil revolutions,
, under a pressure of 6 GPa. As a result, a disk with
mm and ⌀
mm was obtained. At the second stage, the HPT disk produced at the first stage was cut into four segments, and these segments were stacked on the HPT-anvils on top of each other. As a result, a stack with a total height
mm was obtained. A pressure of 6 GPa was imposed on this stack, HPT processing for
revolutions was performed again. The cycles of “accumulative HPT for
” were repeated three times. At the last stage, the cycle was repeated, but with 4 turns of the anvils, and the result was a monolithic sample. The total number of revolutions of the anvils during ACC HPT processing was
. At each stage, the shear strain produced by HPT processing for
was summed with the strain produced by the compression of segment stack, and the total strain was the sum of strains produced at all four stages. Very approximate estimates show that the total strain produced during “accumulative HPT for
”, due to the summation of compressive and torsional HPT strains, was about 2–3 times larger than that during conventional HPT for
.
The XRD analysis of the samples was performed on the Rigaku Ultima IV (Rigaku Corporation, Tokyo, Japan) diffractometer using CuK monochromatized radiation in the reflected beam, and a parabolic graphite monochromator was used. The diffractometer X-ray tube operated at a voltage and current of 40 kV and 40 mA, respectively. The Bragg–Brentano scheme of goniometer was used. Measurements were performed at a field size of mm, located in the area that is 2.5 mm from the disc center. The relationship between the intensity of reflected X-ray quantum’s and diffraction angle was within 10 to 80 in increments of 0.02 and with 6 s of exposure time per point. XRD data were analyzed using PHILIPS ProFit software. Netzsch DSC 204 F1 Phoenix calorimeter (NETZSCH-Gerätebau GmbH, Selb, Germany) was used for DSC testing, and Netzsch Proteus software (ver. 4.8.5) was used for DSC analysis; the heating temperature was 500 C, which is higher than the crystallization temperature for Vit105. The heating rate was 20 K/min. In order to determine more precisely the relaxation energy, the DSC tests were conducted in the following manner: first, heating to a temperature slightly above (420 C), then cooling to 80 C; then a second heating to a temperature slightly above the temperature of the crystallization finish (C), cooling; then a third heating to the temperature . The relaxation energy was found from the difference in the enthalpies of the first and second heatings in a region of 80–420 C. The crystallization energy, , was found from the difference in the enthalpies of the second and third heatings in the crystallization region. Characteristic points of the curves were determined as described in ISO 11357-1:2009.
The structure of both as-cast and HPT-processed specimens was investigated on a ZEISS Libra 200FE (Carl Zeiss Microscopy GmbH, Oberkochen, Germany) transmission electron microscope (TEM) at an acceleration voltage of 200 kV, in both high-resolution TEM (HRTEM) and dark field TEM modes. The samples for TEM were prepared by the focused ion beam (FIB) technique using an ZEISS Auriga Crossbeam SEM-FIB (Carl Zeiss Microscopy GmbH, Oberkochen, Germany), and finished using Gatan PIPS (Pleasanton, CA, United States) ion polishing system. The selected area electron diffraction (SAED) data were collected from an area of 80 nm in diameter.
The hardness of specimens was measured using a microhardness tester Emco-Test Durascan 50 (EMCO-TEST Prüfmaschinen GmbH, Kuchl, Austria) with a Vickers-type indenter. The static load was 1 N (100 g) and the dwell time of loading was 10 s. Five indentations were conducted at each measurement point (the center of the HPT-processed disk-specimen and the edge of the HPT-processed disk-specimen) for each specimen.
Studies by small-angle neutron scattering were performed at the Joint Institute for Nuclear Research, Dubna, Russia. Small-angle neutron scattering (SANS) experiments were performed on the YuMO time-of-flight spectrometer at high flux pulse IBR-2 reactor [
41,
42]. YuMO is an instrument with two movable detector systems placed at sample-to-detectors distances of 5.28 and 13.4 m, resulting in a
Q range of 0.006–0.5 Å
, where
and
is the scattering angle.
The measured neutron scattering spectra were corrected for the transmission and thickness of the sample, background scattering on the film substrate and on the vanadium reference sample using SAS software [
43], yielding a neutron scattering intensity in absolute units of cm
.
3. Results and Discussion
As revealed by transmission electron microscopy, the as-cast BMG is amorphous (
Figure 1a). The BMG after HPT processing is also predominantly amorphous (
Figure 1b). However, the appearance of the microstructure slightly changes. Fragmentation appears on rings on SAED images, taken from an area of 60 nm in diameter (
Figure 1b, inset), while HRTEM (
Figure 1b) photos reveal some fringes that might mean some kind of ordering in the structure.
The appearance of the dark field also changes slightly (
Figure 1e). Individual regions of about 5 nm in size are visible in the dark field. Apparently, some strain-induced nanocrystallization begins in Vit105 during HPT processing. However, the areas that are visible in the dark field in HPT-treated Vit105, if they are crystallites, are very small—less than 5 nm—and have an internal structure, because they consist of many luminous points, which is not typical for equilibrium nanocrystals. The luminous inhomogeneous regions observed in the dark field are more likely to indicate the formation/growth of amorphous clusters. The structure and composition of clusters/crystallites have not yet been identified due to their small amount. It should be noted that in other areas of the same sample, the diffraction pattern shows the amorphous state of the material, and strain-induced nanocrystallization is not observed (
Figure 1c).
Strain-induced nanocrystallization has also been observed earlier for some other amorphous alloys [
15,
16,
17,
19,
24], but in Zr-based BMG, it is apparently less distinct and was not noted in many works [
21]. Cluster formation along with nanocrystallization in HPT amorphous alloy was observed earlier in [
44,
45]. It can be assumed that the formation of clusters in HPT amorphous alloy is the initial stage of nanocrystallization. Shear bands were not observed in any of the HPT-processed Vit105 samples by TEM, which is related to the small contrast of shear bands in Zr-based BMG; shear bands could not be detected by TEM in many other works either [
13,
14,
21]. Perhaps, more detailed and precise TEM studies will enable revealing the features in the structure of the HPT-processed Vit105, but at the moment the interpretation of the fine features of structure requires further experimental investigations.
XRD shows that the structure of the original alloy, and after HPT, is amorphous (
Figure 2). Crystalline peaks, even if observed in HPT-processed samples, are very small, and their intensity indicates an extremely low content of the crystalline phase. Here we note that, according to [
46], even 10% content of the crystalline phase can give a clearly visible peak in the diffractogram of the amorphic alloy. Therefore, in this case, SPD-induced nanocrystallization does not play a decisive role in the transformation of the structure, unlike, for example, an amorphous NdFeB alloy and an amorphous Al-based alloy, where SPD-induced nanocrystallization plays a crucial role [
15,
16,
17,
24]. The position of the maximum first amorphous halo of the Vit105 BMG shifts towards the lowest angles after HPT processing (
Table 1). According to [
34,
35], the radius of the first coordination sphere
and the relative changes in free volume
were found from the position of the maximum first amorphous halo. The values of
and
obtained in accordance with the mentioned procedure are undoubtedly very approximate, since the position of the averaged maximum of the first amorphous halo can change, due the decomposition of the amorphous phase into two amorphous phases of slightly differing compositions, or due to partial nanocrystallization, which is not taken into consideration in this procedure. However, this procedure enables evaluating the variation intensity of the amorphous structure resulting from external action, using conventional XRD.
According to these calculations, HPT processing for
leads to a ≈2% increase in the
and also leads to an increase in the values of full width at half maximum (
) by 4.2%. The increase in
resulting from HPT processing could be related both to an increase in free volume [
36] and a possible separation of the initially homogeneous amorphous phase into two or more constituents, slightly different in composition [
35,
47]. In the XRD pattern, the diffusion peaks from them usually merge and form a common halo [
47]. Consequently, the parameters
and
, found from the position of the common halo’s gravity center, together with
can be used only tentatively to estimate the degree of the structural transformations in the amorphous alloy during HPT processing. Conventional HPT processing for
also leads to a slight increase in the values of
—up to 5%—which is supposed to indicate a further growth of the non-homogeneity of Vit105. According to the calculations, conventional HPT processing for
leads to a 0.3% decrease in
. However, the negative value of
(i.e., a decrease in free volume) could be the result of a large error in determining
and
from the position of the averaged maximum of the first amorphous halo, additionally broadened after HPT processing. HPT processing for
leads to an increase in the values of
by 3%, and to a ≈4% increase in
. Such a large obtained value of
could also be the result of a large error in finding
and
from the position of the averaged maximum first amorphous halo, but it confirms the tendency for
to grow as a result of HPT processing. ACC HPT processing for
leads to an increase in
by ≈2%, and the increase in the
is greater—up to 17% (
Table 1), which indicates a more intensive structural transformation of Vit105 during ACC HPT processing.
According to the DSC data, the relaxation energy
, as a result of HPT processing, also steadily increases with an increase in
n—from ≈1.5 J/g for the as-cast state to 11 J/g after HPT processing for
(
Table 2)—which indicates the growth enthalpy and the free volume of amorphous alloys [
13,
14]. This is correlated with the XRD data. As a result of ACC HPT processing, the increase in the
of Vit105 is greater, up to 29 J/g, which indicates a more intensive structural transformation and growth of the
of the amorphous alloy during ACC HPT. The heat, which is released during relaxation and determined using DSC as the difference between the first and the second heatings, appears at 150
C for the ACC HPT-processed samples, which is noticeably lower than in the case of the as-cast Vit105 and Vit105 after conventional HPT. It can be assumed that ACC HPT processing leads to the appearance of some types of defects/structural elements in the amorphous phase, relaxing at a relatively low temperature. Such defects in the amorphous phase were reported earlier, in particular, in [
48]. Relaxation energy and free volume are connected by the relationship
, where
A is a certain constant [
37]. According to [
37], the constant
A for Zr-based amorphous alloys can be estimated as
J/g. Therefore,
. The free volume,
, calculated from
confirms the growth of free volume of the BMG during HPT processing, and especially after ACC HPT processing. It should be noted that, in the accepted calculation procedure, the “free volume”
is determined with respect to a fully relaxed state, while
, determined from the XRD data, is relative to the as-cast state of the BMG. The data on the variation of
obtained from relaxation energy apparently correspond better to the actual picture than the values of
, obtained from the XRD data.
The crystallization of Vit105 takes place in two stages, as indicated by two peaks in the DSC diagram (
Figure 3). This was observed earlier in similar BMGs [
49]. The temperature,
, of the Vit105 BMG does not change very much as a result of HPT processing. The view of crystallization DSC peak changes considerably as a result of ACC HPT processing. The peaks of the two crystallization stages merge, the temperature of crystallization onset slightly increases, by 10
C, which is apparently related to a change in the first stage of crystallization. As a rule, HPT deformation on various BMGs leads to a decrease in the temperature of crystallization onset [
32,
50,
51]. However, in the case of two-stage crystallization, which takes place in this alloy, HPT deformation, on the contrary, increases it (
Figure 3). An increase in the crystallization temperature as a result of HPT was observed earlier in [
14]. To analyze the changes in the crystallization processes as a result of HPT and ACC HPT processing, further experimental studies are required.
Microhardness studies show that, as a result of HPT processing, the
of Vit105 declines relative to the as-cast state (
Table 3). The decrease in the
of BMGs as a result of HPT processing was also observed earlier in other works [
21,
52]. The decrease in the
of BMGs as a result of HPT processing is related to an increase in free volume. The decrease in
is especially intense at the edge of the HPT-processed samples, which is attributed to the larger strain at the edge of the HPT-processed samples [
11]. After ACC HPT processing, microhardness at the edge of the Vit105 samples decreases even further. It is also known that the value of
at different points of an HPT-processed sample is influenced in a complex manner by the pattern of microstress distribution [
39], which can explain, in particular, the inconsistent variation in
in the center of the HPT-processed sample.
Low resolution of the SANS technique cannot give the detailed structure at the atomic level but is useful to describe structural features typically at distances from 1 to 100 nm [
53]. In the case of disordered systems, such as metallic glasses, in the absence of strong correlations between inhomogeneities (fluctuations in the scattering density), the scattered intensity can follow the power law of
Q:
where
A and
B are constants, and
B is the background. It was shown [
54] that for objects with complex branched surfaces (surface fractal of dimension from
)
, scattering can be given by:
where
is the measure of the fractal surface and
is the difference between coherent scattering density of the inner part of amorphous cluster and its boundary (surface) layer, which have the different composition of elements or less density packing. The scattering densities of elements Zr
52.5Cu
17.9Ni
14.6Al
10Ti
5 were calculated using [
55] and are given in
Table 4.
SANS data of Vit105 in the as-cast state, after HPT
and after ACC HPT
are presented in
Figure 4. The results of fits with (
2) are given in
Table 5.
corresponds to the intersection of the model curve with the background from the sample and approximately gives an estimate of the minimal size inhomogeneity of the scattering density in the sample, as
.
At this stage, it is impossible to determine the size distribution and volume fraction of clusters, since their composition and difference in elements are unknown, but only the general elemental composition of BMG is available. Since the scattering intensity is proportional to the square of the difference between the scattering densities at the boundary and inside the cluster (contrast factor), to the volume and concentration of clusters, then in the formula for determining the particle size distribution, one can obtain a large difference in scattering either due to a change in the volume concentration of clusters or due to a combination various elements forming phases at the interface. In this regard, the analysis of the Porod’s invariants admits great arbitrariness and does not allow us, at the moment, to estimate the size distribution of clusters.
Figure 4 shows comparative changes in scattering behavior before and after different HPT modes. All analyzed states are characterized by the same
scattering law-scattering from a highly branched surface, not necessarily closed. This approach was used to analyze the minimum size of clusters/regions with sections existing in the amorphous phase. At the same time, Porod’s approach to the analysis of scattering curves is based on the
scattering law from a smooth surface closing the cluster. The analysis of the scattering curves within the framework of the
scattering law showed that, in all considered states, the morphology of the structures (the hierarchy of the sizes and distribution of clusters) remains identical.
SANS data analysis shows that, in the as-cast state, the amorphous phase consists of some structural inhomogenities (conditionally clusters), with a minimal size approximately beginning from nm.
It should be noted that these clusters have not yet been able to be revealed on existing equipment using the TEM method, as well as in [
49] on same Vit105 BMG.
After HPT , the minimal structural inhomogenities are observed as beginning from 6.3 nm and after ACC HPT from 3.3 nm. It is interesting to note that the scattering for the as-cast state and after treatment procedure follow to the same power law with the exponent (so-called surface fractals, objects with complex rough surfaces).
The structures of these samples are similar on large sizes. More intensive treatment gives changes in the cluster interface at smaller scales, while the surface area between the clusters increases.
The significant increase in scattering intensity (see
Table 5) can be qualitatively explained by two factors. The first is a change in the contrast of the scattering density between the amorphous cluster and the boundary layer due to the release of weakly coupled elements to the cluster surface. For example, the release of
,
,
to the surface can increase the contrast several times (
Table 4). Accordingly, the contrast in the second power can give an increase in intensity by an order of magnitude. Perhaps this is the result of the emergence on the interface of clusters of elements loosely coupled in the amorphous phase. This separation is not significant, occurs at the nanoscale and is not revealed by other structural research methods, only by SANS.
The second factor is the increase in surface area, the
parameter in (
3) as a result of cluster size reduction after HPT processing. A decrease in the size of the amorphous cluster in an amorphous alloy leads to an increase in the interface between amorphous clusters and, as a result, to an increase in free volume, which is also observed by DSC. Unfortunately, it is impossible to unambiguously determine the role of the first or second parameter due to the lack of additional information. Moreover, ACC HPT leads to stronger grinding of SANS-registered clusters than ordinary HPT and, accordingly, to a larger increase in free volume, which is confirmed by other data (DSC, XRD, HV). The results of the parameters are presented in
Table 5.
We shall note that the SANS results on some changes in the cluster structure as a result of HPT are confirmed by the previously reported TEM data. In general, the structural transformations observed in HPT Vit105 are quite complex. On the one hand, there is evidence of the beginning of some strain-induced nanocrystallization and cluster growth (TEM), which is rather a relaxation process and should lead to a decrease in the free volume. On the other hand, XRD and DSC indicate an increase in the free volume and internal energy of the amorphous phase. An increase in the free volume simultaneously with the formation of clusters and nanocrystals as a result of HPT should lead to the multiplication of shear bands under loading and an increase in the plasticity of the amorphous alloy, as was observed earlier [
14,
56]. Thus, as a result of HPT, the structure of the amorphous phase becomes more complex and heterogeneous, as evidenced by SANS data, but the nature of these changes requires additional research and analysis.