Stabilization of an ambient pressure, collapsed tetragonal phase in CaFe As and tuning of the orthorhombic / antiferromagnetic 2 2 transition temperature by over 70 K by control of nano-precipitates. 1 1 S. Ran,1 S. L. Bud’ko,1 D. K. Pratt,1 A. Kreyssig,1 M. G. 0 2 Kim,1 M. J. Kramer,2 D. H. Ryan,3 W. N. Rowan-Weetaluktuk,3 n Y. Furukawa,1 B. Roy,1 A. I. Goldman,1 and P. C. Canfield1 a J 4 1Ames Laboratory, US DOE and Department of Physics and Astronomy, 2 Iowa State University, Ames, Iowa 50011, USA ] n 2Ames Laboratory, US DOE and Department of Materials Science and Engineering, o c Iowa State University, Ames, Iowa 50011, USA - r p 3Centre for the Physics of Materials and Physics Department, u s . McGill University, Montreal, Quebec H3A 2T8, Canada t a m (Dated: January 25, 2011) - d n o c [ 1 v 5 9 5 4 . 1 0 1 1 : v i X r a 1 Abstract We have found a remarkably large response of the transition temperature of CaFe As single 2 2 crystals grown out of excess FeAs to annealing / quenching temperature. Whereas crystals that are annealed at 400◦ C exhibit a first order phase transition from a high temperature tetragonal to a low temperature orthorhombic and antiferromagnetic state near 170 K, crystals that have been quenched from 960◦ C exhibit a transition from a high temperature tetragonal phase to a low temperature, non-magnetic, collapsed tetragonal phase below 100 K. By use of temperature dependent electrical resistivity, magnetic susceptibility, X-ray diffraction, M¨ossbauer spectroscopy and nuclear magnetic resonance measurements we have been able to demonstrate that the tran- sition temperature can be reduced in a monotonic fashion by varying the annealing / quenching temperature from 400 to 850◦ C with the low temperature state remaining antiferromagnetic for transition temperatures larger than 100 K and becoming collapsed tetragonal / non-magnetic for transition temperatures below 90 K. This suppression of the orthorhombic / antiferromagnetic phase transition and its ultimate replacement with the collapsed tetragonal / non-magnetic phase is similar to what has been observed for CaFe As under hydrostatic pressure. Transmission elec- 2 2 tron microscopy studies indicate that there is a temperature dependent, width of formation of CaFe As with a decreasing amount of excess Fe and As being soluble in the single crystal at lower 2 2 annealing temperatures. For samples quenched from 960◦ C there is a fine (of order 10 nm), semi- uniform distribution of precipitate that can be associated with an average strain field whereas for samples annealed at 400◦ C the excess Fe and As form mesoscopic grains that induce little strain throughout the CaFe As lattice. 2 2 PACS numbers: 74.70.Xa; 61.50.Ks; 64.75.Nx; 75.30.-m; 72.15.-v 2 I. INTRODUCTION CaFe As manifests an extreme example of the coupled magnetic / structural phase tran- 2 2 sitionthatepitomizesthephysicsoftheundopedparentsoftheFeAs-basedsuperconductors.1,2 The strongly first order transition at ambient pressure from a high temperature, tetragonal, paramagnetic phase to a low temperature, orthorhombic, antiferromagnetic phase takes place near 170 K in single crystals grown out of Sn flux and manifests a hysterisis of several degrees as seen in thermodynamic, transport, and microscopic measurements.1–5 CaFe As is also the most pressure sensitive of the AFe As and 1111 compounds with its 2 2 2 2 structural/magneticphasetransitionbeinginitiallysuppressedbyover100KperGPa.2,6–10 As pressure increases a non-magnetic, collapsed tetragonal phase that is stabilized by ∼ 0.3 GPa intersects and terminates the lower pressure orthorhombic / antiferromagnetic phase line near 100 K and 0.4 GPa and rises to 300 K by ∼ 1.5 GPa.6–8 In addition to this extreme pressure sensitivity, CaFe As is also very sensitive to non-hydrostaticity.2,7–12 If 2 2 the pressure medium solidifies before the structural phase transitions, then the anisotropic changes in the unit cell lead to non-hydrostatic (by definition) stress, which in turn leads to dramatically broadened transitions and a structurally mixed phase sample in the 0.4 GPa pressure region. This mixed phase includes a small amount of strain stabilized, high temperature tetragonal phase which superconducts at low temperature.2,6–10,12,13 The use of helium as a pressure medium allows for a minimization of these non-hydrostatic effects and has allowed for the determination of the T −P phase diagram.2,7–9 Sn grown single crystals of CaFe As are highly deformable and join the RSb and RBi 2 2 2 2 compounds14,15 as rare examples of a malleable intermetallic compounds. Single crystal plates can be bent, and, to some extent, even rolled by simply grasping with tweezers and applying minor torques across the sample length by pressing one end of the crystal down- ward on the surface of the lab bench, or microscope stage. This malleability can lead to extreme broadening of features in ground samples, as were seen in early attempts at powder X-ray diffraction.1 CaFe As samples were initially grown out of Sn and characterized in single crystal 2 2 3 form.1,5 Sn grown crystals are well formed, faceted plates that generally have planar di- mensions of several mm and thicknesses between 0.1 and 0.5 mm.1,2 For measurements that require larger sample volumes pseudo-polycrystalline7 or oriented single crystalline assem- blies (see Figure 1 in Ref. 16) can be used. Larger single crystals of CaFe As have been 2 2 grown out of ternary melts rich in FeAs.8 In order for these larger crystals to manifest a structural / magnetic phase transition similar to that seen in the smaller Sn grown crystal they were annealed at 500◦ C for 24 hours (a temperature similar to the decanting temper- ature of the Sn grown samples). Without this annealing the larger, FeAs-grown, samples had dramatically suppressed transition temperatures. Given recent observations of small shifts in the structural and magnetic transition tem- peratures of BaFe As samples, and of the superconducting transition in doped BaFe As , 2 2 2 2 as well as sharpenings of their signatures in thermodynamic and transport data,17,18 we undertook a systematic study of the effects of post growth thermal treatment of FeAs grown single crystals of CaFe As . We have discovered that once again CaFe As is the extreme 2 2 2 2 case in the AFe As series, manifesting a surprisingly large suppression of the structural / 2 2 magnetic transition temperature in as grown samples (nearly 50%) that, even more remark- ably, can be systematically changed from ∼ 170 K to below 100 K with the lowest transition temperature samples having a transition into the non-magnetic, collapsed tetragonal state, but at ambient pressure. In order to characterize and understand the effects of temperature treatment, as well as thenatureofthelowtemperaturestate,wehaveperformedawidevarietyofthermodynamic, transport, microscopic, and spectroscopic measurements. Temperature dependent electrical resistivity and magnetic susceptibility measurements were used to determine a transition temperature (T∗) annealing temperature (T ) phase diagram as well as identify similari- a ties between the collapsed tetragonal phase and the low temperature state of FeAs-grown CaFe As crystals quenched from temperatures between 850 and 960◦ C. For annealing tem- 2 2 peratures T (cid:38) 400◦ C the T∗ −T phase diagram is found to be remarkably similar to the a a T∗-pressure (P) phase diagram, bringing up the question of what could the relationship be- tween T and P be? Temperature dependent single crystal X-ray diffraction measurements a were then employed to unambiguously show that the crystallographic phase transition in as 4 grown samples quenched from 960◦ C is one to a collapsed tetragonal state that is in qual- itative as well as quantitative agreement with what is found for Sn-grown samples under applied pressures of ∼ 0.4 GPa. Temperature dependent Mo¨ssbauer spectroscopy measure- ments showed that the low temperature magnetic state of annealed FeAs-grown CaFe As 2 2 single crystals remains antiferromagnetic until the transition temperature is suppressed to below 100 K when the low temperature ground state becomes non-magnetic, a result con- firmed by nuclear magnetic resonance (NMR) measurements. Finally, transmission electron microscopy (TEM) measurements revealed that there is a small, temperature dependent width of formation for CaFe As , allowing for a solid solubility of excess Fe and As in the 2 2 single crystals that decreases with temperature. As the quenching temperature is reduced from 960◦ C to 400◦ C the initially fine precipitate coarsens, decreasing the degree of strain detected in the sample. II. EXPERIMENTAL METHODS Single crystals of CaFe As were grown out of excess FeAs by rapidly cooling a melt of 2 2 CaFe As from 1180◦ C to 1020◦ C over 3 hours, slowly cooling from 1020◦ C to 960◦ C over 4 4 35 hours and then decanting off the excess liquid, essentially quenching the samples from 960◦ C to room temperature. These samples will be referred to as, ”as grown samples”. Post growth, thermal treatments of samples involve the following variables: annealing tempera- ture, annealing time, and annealing environment. Annealing environment refers to either (i) annealing a whole, unopened, decanted growth ampoule, or (ii) annealing individual crystals that have been picked from a growth and resealed in evacuated silica tubes. For studies of the effects of annealing temperature, we seal several crystals into an evacuated silica tube and anneal for 24 hours in a furnace stabilized at the specified temperature. The sample is placed into the hot furnace and, after annealing it is quenched to room temperature. Longer time anneals (seven days) were used to prepare whole, unopened batches of samples. In order to study the effects of annealing on FeAs grown samples with transition temperatures (and features) like the Sn-grown crystals, samples that had been annealed for a week at 400◦ C were subsequently sealed into an evacuated silica tube and annealed for 24 hours in a furnace stabilized at the specified temperature. Although a detailed study of the annealing time dependence of sample changes will need to be done in the future, we found that, for 5 example, at 450◦ C a one hour anneal is not enough to effect complete change, but anneals longer than 4 hours do not lead to any further significant changes in sample behavior; at 800◦ C annealing appears to be completed in under 0.5 hour. Temperature dependent magnetization measurements were made in a Quantum Design (QD) MPMS unit. Temperature dependent electrical resistivity was measured in a four probe configuration, with Pt wires attached to the samples by Du Pont 4929N Ag-paint (cured at room temperature), in a QD PPMS unit. Although normalized resistivity values are plotted, the resistivity values of samples did not vary outside of the uncertainty asso- ciated with a combination of geometric error (associated with measuring dimensions of the sample) and difficulties associate with sample exfoliation. The average, room temperature resistivity of as grown, 700◦ C annealed and 400◦ C annealed samples was 3.75±0.75 mΩ cm (a 20% variation). In order to identify the nature of the structural transition in the as-grown CaFe As 2 2 crystal (quenched from 960◦ C) and to determine the temperature dependence of the lattice parameters, high-energy X-ray diffraction measurements (E = 99.62 keV) using an area detector were performed on the 6-ID-D station in the MUCAT Sector at the Advanced Photon Source. At this high energy, X-rays probe the bulk of a crystal rather than just the near-surface region and, by rocking the crystal about both the horizontal and vertical axes perpendicular to the incident X-ray beam, an extended range of a chosen reciprocal plane can be recorded.19 For the measurements, the horizontal angle, µ was scanned over a range of ±3.6◦ for each value of the vertical angle, η, between ±3.6◦ with a step size of 0.4◦. The two-dimensional scattering patterns were measured by a MAR345 image-plate positioned 1503 mm behind the sample. The crystal was mounted on the cold finger of a closed-cycle refrigerator surrounded by a beryllium heat shield and vacuum containment. Additionally, the crystal was mounted such that there was access to (hk0), (h0l), and (hhl) reciprocal lattice planes. The (008) and (220) peaks were fit for lattice parameter determinationand, forthesemeasurements, thetotalexposuretimeforeachframewas383s. The M¨ossbauer absorbers were prepared by attaching several single crystal plates to a 12 mm diameter disc of 100 µm thick Kapton foil using GE-7031 varnish. The spaces between 6 the crystals were filled with a radio-opaque paint prepared by mixing 1-5 µm tungsten powder (obtained from Alpha-Aesar) with diluted GE-7031 varnish. The absence of gaps in the completed mosaic was confirmed first visually and then by looking for transmission of the 6.4 keV Fe-K X-ray from the 57Co Mo¨ssbauer source. In the configuration used, the α crystalline c -axis was parallel to the M¨ossbauer γ-beam. The Mo¨ssbauer spectra were collected on a conventional spectrometer using a 50 mCi 57Co/Rh source mounted on an electromechanical drive operated in constant acceleration mode. The spectrometer was calibrated against α-Fe metal at room temperature. Tem- peratures down to 5 K were obtained using a vibration-isolated closed-cycle refrigerator with the sample in a partial pressure of helium to ensure thermal uniformity. Spectra were fitted using a conventional, non-linear, least-squares minimization routine to a sum of equal-width Lorentzian lines. The line positions for the magnetic sextets observed in the ordered state were calculated assuming first-order perturbation in order to combine the effects of the magnetic hyperfine field and the electric field gradient. As the samples were oriented mosaics rather than powders, the line intensities were constrained to be in the ratio 3 : R : 1 : 1 : R : 3 (following the conventional practice of labeling the lines from negative to positive velocity)20 with the intensities of the two ∆m = 0 lines being variable (R) to allow I for the expected magnetic texture. R = 0 would correspond to the moments being parallel to the M¨ossbauer γ-beam, whereas R = 4 indicates that the moments are perpendicular to the beam. Nuclear magnetic resonance (NMR) measurements were carried out on 75As (I = 3/2; γ/2π = 7.2919 MHz/T) by using a homemade phase-coherent spin-echo pulse spectrometer, to investigate the magnetic and electronic properties of differently treated CaFe As crystals 2 2 from a microscopic point of view. 75As-NMR spectra at a resonance frequency of 51 MHz were obtained by sweeping the magnetic field. TEM samples were prepared by mechanically polishing the single crystal to ∼ 10 µm thick along the c-axis and then ion milling to perforation using 3 keV ∼ 18◦ incident angle and following up with 30 min at 500 eV at 10◦ to remove milling damage. All milling was performed using a liquid N cooled stage (sample T ∼ 120 K). Samples were analyzed 2 7 using a Philips CM30 TEM operated at 300 keV. Energy dispersive spectroscopy (EDS) and selective area diffraction patterns (SADP) were also performed on the samples in the TEM. III. DATA PRESENTATION Figure 1 presents the resistivity and magnetic susceptibility for CaFe As single crystals 2 2 grown out of Sn and for CaFe As single crystals grown out of excess FeAs. Two data sets 2 2 are shown for FeAs grown crystals: one data set shows measurements on an as grown crystal that was decanted at, and quenched from, 960◦ C; the other data set shows measurements on a sample from a batch that was subsequently annealed at 400◦ C for a week. The Sn-grown single crystal and the FeAs grown sample that has been annealed at 400◦ C are quite simi- lar, both manifesting similar, modest increases in resistivity and decreases in susceptibility associated with the phase transition near 170 K.1,2 On the other hand, the FeAs sample that was quenched from 960◦ C shows a significantly larger, very sharp drop in magnetization occurring well below 100 K. The electrical resistivity also drops discontinuously at this temperature, associated with the sample suddenly undergoing a violent structural phase transition that often (usually) leads to shattering along the length and width of the bar, as well as loss of contacts. InadditiontothequantitativedifferencesshowninFig. 1, thereisaqualitativedifference between the as grown, CaFe As single crystals from FeAs solution and the single crystals 2 2 grown from Sn. Whereas the Sn-grown single crystals are malleable and can easily be bent and deformed, the crystals quenched from a 960◦ C FeAs solution are brittle and tend to shatter if bending is attempted. The FeAs grown crystals that have been annealed at 400◦ C, however, recover some of the malleability of the Sn grown ones and can deform a little without shattering. Giventhedramaticdifferenceintransitiontemperature, aswellasthedifferentsignatures of the transition in resistivity and magnetization, several questions arise. Among them we consider: (i) what is the nature of the phase transition in the as grown sample and (ii) can the transition in annealed samples be varied from near 170 K to below 100 K in a systematic 8 manner? We will address the latter question first and return to the former after the creation of a transition temperature (T∗) - annealing temperature (T ) phase diagram. a In order to assess the extent to which the 170 K phase transition that occurs in Sn-grown, as well as annealed FeAs-grown, samples of CaFe As can be systematically shifted down 2 2 to below 100 K we measured the temperature dependent susceptibility and resistivity of as grown samples that were annealed for 24 hours at temperatures ranging from 250 to 850◦ C. Figure 2 presents magnetic susceptibility and resistivity data for representative annealing temperatures. The decrease in susceptibility (or increase in resistivity) can be shifted down in temperature by choosing an appropriate annealing temperature between 400 and 800◦ C. For annealing between these temperatures the transitions, particularly as seen in the resistivity data, remain quite sharp and shift in a systematic manner. Whereas the size of the jump in the magnetization remains fairly constant in the samples annealed in this temperature region, there is a monotonic increase in the magnitude of the increase in the resistivity (see Fig. 7 below). Such a clear temperature dependence of the effects of annealing, over such a wide temper- ature range, begs the question of what the annealing time dependence of these effects is. In other cases of clear annealing effects, both time and temperature cuts through phase space are needed to establish unambiguous annealing protocols.21 In Fig. 3 we show the evolution of the magnetic susceptibility for different annealing times. At 450◦ C, 0.5 hr is insufficient time to effect any significant change; 1.0 hr leads to split, broadened features with drops in susceptibility below both 170 and 100 K; 3.0 hr leads to a single, sharp feature near 170 K, comparable to what is seen for 24 hour anneals. This progression shows that for 450◦ C, 24 hours is longer than the salient time scale for annealing. As would be expected, for higher temperatures the salient time scale is even shorter. In Fig. 3(b) samples from a batch that had been annealed for a week at 500◦ C, with a transition temperature above 150 K, were annealed at 800◦ C for representative times. As can be seen, even a 0.5 hour anneal causes the sample to behave in a manner similar to the as grown (quenched from 960◦ C) samples. Figure 2 also demonstrates that 24 hour anneals at temperatures of 300◦ C or lower temperatures do not change the temperature dependence of the as grown samples. The 9 data from the sample annealed at 350◦ C for 24 hours shows somewhat broadened drops in susceptibility near both 170 and 100 K, similar to what was seen for a 1.0 hr anneal at 450◦ C (Fig. 3a), indicating that at 350◦ C 24 hours is comparable to, but less than the salient time scale. Although longer annealing times for T (cid:46) 350◦ C may lead to a sharp, single transition near 170 K (as is seen for the 400 and 500◦ C, 24 hr anneals) the time needed to achieve this state is anticipated to become exponentially long. The one other data point we can add to this is the fact that 20◦ C (room temperature) anneals approaching 104 hours have not led to significant changes in behavior of as grown samples. A 24 hour anneal at 850◦ C does not significantly change the transition temperature from that measured for the as grown samples quenched from 960◦ C (perhaps not too surprisingly since 850◦ C is approaching the 960◦ C quench temperature); the resistivity data for this sample, though, can be collected below the transition temperature, showing that the low temperature state has a lower resistivity, leading to a downward jump in resistivity when cooling through the transition temperature. In order to see if similar changes in transition temperature could be induced by annealing samples that started with transitions near 170 K (i.e. started with transitions similar to those found in Sn grown CaFe As ) we annealed an entire batch of crystals at 400◦ C for 2 2 a week. The resistivity and susceptibility data for these samples are also shown in Fig. 2 and are essentially the same as that found for the 24 hr anneal of individual crystals. Single crystals from this, ”400◦ C anneal for one week” batch were then separately sealed in silica ampoules and annealed for 24 hours at temperature ranging from 500◦ C to 800◦ C. The temperature dependent resistivity and susceptibility for these samples are shown in Fig. 4. As was the case for the as grown samples, sharp features in both resistivity and susceptibility systematically shift to lower temperature when the sample is annealed at higher temperature. The sample annealed at 800◦ C shows the larger drop in susceptibility and broke on cooling through its transition, making it appear to be similar to the as grown, quenched from 960◦ C, samples. Figure 5 presents the transition temperature - annealing temperature, T∗ − T , plot. a Figure 6 illustrates how values for T∗, as well as the error bars, were inferred from the 10