2. Experimental
Figure 1 shows the schematic view of ATHENA, showing the layer-by-layer assembly of electric probes in yellow, as well as the specimen in blue. From the top to the bottom of the anvil in
Figure 1, the inner electrode electrodes are shown in yellow, the blue rectangle is the sample, and the red strips are the outer electrodes. The disc between the anvils is the gasket, where the green part is composed of cBN and a central hole serves as the sample chamber.
The shapes of both the specimen and electric probes are controlled by a mask with a specific pattern. Typically, the mask is made of gold foil of 10 μm in thickness. The pattern is fabricated by laser cutting automatically with computer-assisted design (CAD).
The Pt electrodes that serve as the inner electrodes were deposited on the surface of the culet of the upper diamond anvil with a dimension of 300 micrometers, as shown in
Figure 2a. The La specimen was further deposited directly on top of the Pt electrodes of the same anvil, as shown in
Figure 2b, before being clasped with the lower anvil, where the outer electrodes of 10 micrometer Au wires were located.
The size of the specimen was 60 μm (L) × 30 μm (W) × 1 μm (T), and it could be well controlled by using the mask technique during deposition. The deposition was performed with a home-fabricated magnet sputtering system with a 10−6 Tor vacuum. The thickness of the Pt electrodes and La specimen could be well tuned from tens of nanometers to several micrometers by changing the sputtering current and depositing time. The typical thickness was about 300 nanometers for Pt electrodes and 3 μm for La specimens in the experiments.
All the processes were conducted in a glove box with flowing Ar gas to prevent the oxidation of the La metal. The electronic transport properties of La under high pressure as a function of temperature were investigated via four probe electrical conductivity methods in a diamond anvil cell (DAC) made of CuBe alloy, as described in [
8]. Pressure was generated by a pair of diamonds with a 300 μm diameter culet. A gasket made of T301 stainless steel was pressed from a thickness of 250 μm to 20 μm, and a hole was drilled at the center with a diameter of 200 μm. The hole was filled and compressed with fine cubic boron nitride (cBN) powder mixed with epoxy. A center hole of 70 μm in diameter was further drilled at the cBN layer to serve as the sample chamber. The fine cubic boron nitride (cBN) powder mixed with epoxy was pressed on the surface of the gasket to protect the electrode leads and insulate them from the metallic gasket. Au wires of 18 μm in diameter were used as the outer electrodes. The sample chamber was filled with dry NaCl powder as the pressure-transmitting medium. The pressure was determined via the ruby fluorescence method at room temperature before and after each cooling using the method described in [
12]. Three ruby balls were mounted around the sample with pressure errors of about ~2 GPa at 60 GPa. The diamond anvil cell was placed inside a MagLab system to perform the experiments. The temperature was automatically controlled by a program of the MagLab system that could provide synergetic extreme environments with temperatures from 300 K to 1.5 K and a magnetic field of up to 9 Tesla. A thermometer was mounted near the diamond in the diamond anvil cell to monitor the exact sample temperature. The X-ray diffraction (XRD) of the sample was performed on a Rigaku Ultima VI (3 KW) diffractometer using Cu Kα radiation generated at 40 kV and 40 mA. The X-ray diffraction patterns were collected with a scanning rate of 10° per minute and a scanning step length of 0.01°. The sample surface was covered with Mylar (there were no diffraction peaks for the covered materials except a background at a low-angle range) to protect samples from degradation in air. In addition to the La metal, there were also some La
2O
3 phases that we were unable to avoid during the process of the manipulation of the samples for the measurements of XRD patterns (please see the
Supplementary Materials) at room temperature as La is very easy to oxidize. However, La
2O
3 is highly insulated, and thus did not contribute to the electric conductivity measurements.
3. Results and Discussion
Figure 3a shows the evolution of resistivity (
ρ) as a function of temperature for La at various pressures up to 63.3 GPa, showing very sharp superconducting transitions. La is superconductive at ambient pressure, with a critical transition temperature (Tc) of around 5 K [
13]. Previous reports showed the change of Tc of La under pressure through AC susceptibility measurements. The maximum pressure previously reported was about 50 GPa [
14]. The superconducting transition temperature is defined through the differentiation of
ρ over temperature, as shown in
Figure 3b. The temperature of onset T
conset and the temperature for zero resistivity T
czero can be easily obtained, as shown in
Figure 3b. The superconducting transition width ΔT
c, which corresponds to the difference between T
conset and T
czero, is rather small (less than 1 K), as shown in the inset of
Figure 3b, indicating the high quality of sample as well as a homogenous pressure environment. Hereafter, we refer to T
conset as T
c.
Figure 4 shows the evolution of T
c as a function of pressure. The T
c of La increases strongly with increasing pressure and reaches about 12 K at 24.3 GPa. Then, T
c begins to slowly decrease with a further increase in pressure. The increase in T
c up to 24.3 GPa and the subsequent fall are the result of the interplay of the interdependence among the values of the electron phonon coupling, the average phonon frequency and the Hopfield parameter [
15]. We plot our results together with the previous reports in
Figure 4; previous reports and experimental data are from [
5,
13,
14], respectively. There are two structures of lanthanum at ambient conditions: (a) the stable αLa, with a double hcp (dhcp) structure; and (b) the metastable βLa, with an fcc structure. The superconducting transitions of αLa and βLa are near 5 and 6 K, respectively, at equilibrium ambient pressure. The pressure dependence of superconducting transition temperatures (Tc) of αLa and βLa metal have been reported previously. The Tc(P) of αLa shows a rather complicated variation. There is an abrupt increase in Tc near 2 GPa, and two peaks at 12 GPa and 39 GPa, respectively, as reported in [
14]. The Tc of βLa increases rapidly with pressure and reaches almost 12.9 K at about 18 GPa. Thus, the evolution of Tc as a function of pressure is structure-dependent. There are two La phases in our sample, as usual—i.e., the ambient αLa and the βLa phase—as also observed in a previous report [
14].
The lattice parameters for diamond and αLa are 3.56 Å and 5.32 Å, respectively, with an apparent lattice mismatch. The lattice mismatch usually results in strain. The strain can be of a GPa level sometimes and thus will affect the properties of the specimen [
16]. The apparent lattice mismatch between the La specimen and diamond subtract will gives rise to compressive strain; i.e., the specimen will be at compression. This is why the high-pressure phase βLa appeared in the specimen. The βLa phase takes place at ~2 GPa, as reported in [
14]. The deposition of a thin film usually causes disorder because of the rapid cooling effect, which is different from that of the bulk samples for which crystallization can be guaranteed. It is well known that disorder is harmful to the stabilization of the Cooper Pair, which is the base for superconductivity. As a consequence, the disorder will decrease the superconducting transition temperature [
17]. This is the likely reason for the slightly lower Tc observed in our sample than those of the bulk samples for which the crystallization can be guaranteed studied in previous experiments. On the other hand, the experimentally observed peaks of Tc are at about 13.8 GPa, 18 GPa and 12 GPa, respectively [
5,
13,
14]. Thus, the difference of the peak pressure of Tc can be about 6 GPa, even for previously reported bulk samples. This is reasonable if such factors as the purity or quality of samples, the nohydrostatic effects of different pressure media or the pressure calibration deviation etc. are taken into account in each experiment. Additionally, the theoretically calculated peak of Tc is at 18.3 GPa, meaning that the lattice parameters and bulk moduli are in good agreement with experimental observations [
15]. In our experiment, the peak of Tc was at ~20 GPa, which is close to the value of 18 GPa reported in [
13] and the calculated value of 18.3 GPa in [
15].
The dependence of Tc on pressure in our sample is generally in agreement with that of previous reports, taking into account the usual difference between film and bulk samples.
Figure 5 shows that Tc shifts at 63.3 GPa upon applying magnetic fields. A perpendicular magnetic field H was applied to the surface of the La specimen and changed from 0 Tesla to 5 Tesla with steps of 0.5 Tesla. It can be seen clearly that T
c decreases with the increasing applied magnetic field H at 63.3 GPa, indicating that the transition is superconductive. The upper critical field H
c2(0) calculated from the Werthamer–Helfand–Hohenberg (WHH) relation of the La superconductor at 63 GPa is estimated to be 8 Tesla.
Figure 6 shows the normal state resistivity at various pressures. In order to analyze the normal state behavior under pressure, we applied the power law formula
ρ(
T) =
ρ0 +
ATn to the resistivity of La, where
ρ0 and
A are the residual resistivity and the temperature coefficient, respectively. The value of
n reflects the scattering mechanism. At a low temperature, if
n equals 2, it is evidence of a Fermi liquid [
18]. For example, we plot the experimental curve of
ρ over
T at 38.9 GPa and the fitted curve in
Figure 7. The fitting parameters are listed in
Table 1. The adjusted R-square (degree-of-freedom adjusted coefficient of determination), which is a measure used to judge the fitting quality, is 0.99988, indicating very good reliability.
Figure 8 illustrates the phase diagram of the superconducting transition temperature; normal state behavior is observed in terms of the fitted exponent
n as function of pressure. The samples show superconductivity in the whole pressure range for high-pressure phases up to 63.3 GPa. The value of
n moves from smaller to larger than 2 within the pressure range of experiments, indicating that normal state gradually transformed from a correlated electron system to a Fermi liquid (FL) and then to a more free electron-like system. This can be well understood as electrons become itinerant when energy bands widen at high pressures.