Controllable Negative Thermal Expansion by Mechanical Pulverizing

37 mins ago - Controllable Negative Thermal Expansion by Mechanical Pulverizing in Hexagonal Mn0.965Co1.035Ge Compounds. Sheng Yang† ...
0 downloads 0 Views 2MB Size
Download PDF
Article Cite This: Inorg. Chem. 2018, 57, 14199−14207

pubs.acs.org/IC

Controllable Negative Thermal Expansion by Mechanical Pulverizing in Hexagonal Mn0.965Co1.035Ge Compounds Sheng Yang,†,# Shengcan Ma,*,†,# Kai Liu,† Yongfeng Hu,‡ Kun Yu,† Xingqi Han,† Zhishuo Zhang,† Ying Song,† Changcai Chen,† Xiaohua Luo,† Dunhui Wang,§ and Zhenchen Zhong†

Inorg. Chem. 2018.57:14199-14207. Downloaded from pubs.acs.org by UNIV OF SOUTH DAKOTA on 11/19/18. For personal use only.

†

Jiangxi Key Laboratory for Rare Earth Magnetic Materials and Devices/Institute for Rare Earth Magnetic Materials and Devices (IREMMD), Jiangxi University of Science and Technology, Ganzhou 341000, People’s Republic of China ‡ Canadian Light Source, University of Saskatchewan, Saskatoon, Saskatchewan S7N 2V3, Canada § National Laboratory of Solid State Microstructures & Jiangsu Key Laboratory for Nano Technology, Department of Physics, Nanjing University, Nanjing 210093, People’s Republic of China S Supporting Information *

ABSTRACT: Negative thermal expansion (NTE) material as a compensator is very important for accurately controlling the thermal expansion of materials. Along with the magnitude of the coefficient of thermal expansion, the operating temperature window of the NTE materials is also a major concern. However, only a few of the NTE materials possess both a large operating temperature range and a large thermal expansion coefficient. To explore this type of new NTE material, the Mn0.965Co1.035Ge fine powders were prepared by mechanical ball milling (BM). These fine powders show a largely extended NTE operation temperature window simultaneously possessing a giant thermal expansion coefficient. For samples treated with different BM times, such as the BM-0.5h, BM-4h, and BM-12 h samples, the operating temperature window (ΔT) and linear thermal expansion coefficient (αL) are 167 K (222−389 K) and ∼ −63 ppm/K, 221 K (140−360 K) and ∼ −41.3 ppm/K, and 208 K (234−442 K) and ∼ −40 ppm/K, respectively, which are larger than most wellknown NTE materials. More strikingly, all BM samples have a large constant linear NTE coefficient with an ultrawide temperature window covering room temperature. For these three samples, these values are ∼ −52 ppm/K (140 K), ∼ −58.3 ppm/K (110 K), and ∼ −65 ppm/K (80 K), respectively. The origin of the excellent NTE properties is discussed based on the thermomagnetic measurements and X-ray absorption spectroscopic results.

1. INTRODUCTION It is general knowledge that a vast majority of solid materials increase in volume upon heating and shrink upon cooling, that is, the so-called positive thermal expansion (PTE). In many high-precision devices, including space devices, high-precision optical mirrors, printed circuit boards, and machinery parts, thermal expansion is an important parameter for the thermal stability of materials.1,2 The shape change, very small it may be, which is brought about by PTE, would worsen the excellent performance of materials. However, only a very limited number of materials exhibit zero thermal expansion (ZTE, neither expanding during heating nor shrinking during cooling) behavior. Therefore, the materials with the negative thermal expansion (NTE, shrinkage upon heating and expansion upon cooling) property have attracted great interest. By forming composites with PTE materials, NTE materials as a compensator would precisely control thermal expansion behavior of materials, for which the desired coefficient of thermal expansion (CTE) could be acquired.3 During the past few years, the NTE effect has been observed in a number of © 2018 American Chemical Society

materials, such as the well-known ZrW2O8 family, with the NTE effect originating from the flexible framework in crystal structure.1,4 Very recently, many materials undergoing phase transition have been found to have a large NTE effect. For example, it has been reported that the (Hf,Nb)Fe 2 compounds,5 antiperovskite manganese nitrides,6−10 La(Fe,Si)13-based,11,12 La(Fe,Si,Co)13,13 La(Fe,Al)13-based,14 and La(Fe1−xCox)11.4Al1.6 compounds15 show the NTE effect across the magnetic phase transition. There are two important parameters concerning the NTE properties of a material: (1) the magnitude of the CTE and (2) operating temperature range. Profoundly important to fundamental research and the practical application is the manipulation to search for materials simultaneously possessing a large coefficient of NTE and a wide operation temperature window. In general, there are two strategies to obtain the thermal expansion properties: one is the direct measurement Received: August 8, 2018 Published: November 7, 2018 14199

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

times to ensure homogeneity. The as-prepared sample was first ground with agate mortar and then sealed with carbide balls and alcohol in a carbide ball milling jar in an Ar atmosphere. The weight ratio of ball-to-powder-to-alcohol was 5:1:0.6. High-energy planetary ball milling (Pulverisette 5, FRITSCH, Germany) process was carried out subjecting to a different milling time (0.5, 3, 4, 5, 12 h) with a constant speed of 200 rpm. Hereafter, these samples are denoted as BM-0.5h, BM-3h, BM-4h, BM-5h, BM-12h, respectively. Measurements. The thermomagnetic M(T) curves were measured by using the PPMS DynaCool Cryogen-free System (Physical Property Measurement System DynaCool, Quantum Design, USA) equipped with a vibrating sample magnetometer (VSM) module. Before the M(T) measurements, the samples must first be cooled to 10 K under the zero applied field, and then the heating M(T) curves from 10 to 400 K were measured under the field of μ0H = 0.1 T. This is defined as the zero field cooling (ZFC) model. After that, the cooling curves were obtained from 400 to 10 K under μ0H = 0.1 T, that is, in the field cooling (FC) model. The calorimetric properties were determined using a differential scanning calorimeter (DSC, TA Instruments Q2000) with the heating and cooling rate of 10 K min−1. The RT and temperature dependent (130−450 K) powder X-ray diffraction (PXRD) measurements were performed, respectively, with Cu-Kα radiation (λ = 1.54 Å) for BM samples. The temperature-dependent XRD measurements were carried out in the ZFC model: first, the samples were cooled down to 100 K via liquid nitrogen under the zero applied field; second, the samples are gradually warmed to each measurement temperatures by a heating equipment. The Rietveld refinements of XRD patterns are performed by using the FullProf software.25 The Mn and Co K-edge X-ray absorption spectroscopy was elucidated to study the local structure of these atoms for samples with different ball mill times. These measurements were performed using the SXRMB beamline at the Canadian Light Source (CLS), Saskatoon, Canada. Total electron yield and fluorescence yield, using a 4-element Si(Li) drift detector, was used to record the K-edge spectra of Mn and Co. X-ray absorption data were processed using Athena package 90.26 Multiple scan spectra were first averaged, and the spectra were normalized to unity using a linear pre-edge subtraction and quadratic polynomial as the post edge line for background subtraction.

by dilatometry or by using the strain gauge technique in the physical property measurement system (PPMS).3,13,14 For example, the linear CTE (αL) of bulk LaFe13‑xAlx (x = 1.8, 1.9, 2.1, 2.3, 2.5, and 2.7) and LaFe13−xSix (2.8 ≤ x ≤ 3.2) compounds, 1 3 , 1 4 and bonded MnCoGe-based and (Ga0.7Cu0.3)1−xMnxNMn3 (x ≤ 0.4)3,10 composites with binders, e.g., epoxy resin, were measured using the strain gauge technique. The other is the characterization by the variation of lattice parameters and cell volume with temperature, which is obtained via refining the temperaturedependent X-ray diffraction (XRD) or neutron diffraction data.8,11 In general, due to the existence of microscopic related effects during the compacted process, such as pores, the two methods are not expected to give the same results. For powder samples, furthermore, it should be noticed that their thermal and electrical conductivity would be deteriorated to some extent by forming the composites with binders.3,16 Therefore, the second method is very effective and significant in determining the NTE performance in powder materials. As a member of magnetic transition materials, MnCoGebased compounds have been recently studied due to the large room temperature (RT) magnetocaloric effect (MCE) arising from the ferromagnetically martensitic transformation (FMT) from the paramagnetic (PM) austenite with the hexagonal Ni2In-type structure (space group P63/mmc) to ferromagnetic (FM) martensite with the orthorhombic TiNiSi-type structure (space group Pnma).17−19 The volume of these alloys shows an abnormal contraction as large as ΔV/V −3% ∼ −5% across the reverse martensite transformation (MT).3,20−22 As mentioned above, the materials, such as HfFe2-based, antiperovskite manganese nitrides and LaFe(Si,Al)-based, possess a large NTE effect associated with the magnetic phase transition.5−15 Therefore, a large NTE is expected to be achieved in MnCoGe-based alloys. In 2015, Hu et al. reported giant NET in optimized MnCoGe-based compounds bonded with epoxy resin due to the martensitic structural transformation.3 The excellent NTE performance shows the bonded MnCoGebased compounds have potential as a NTE material.3 In 2016, Tong et al. first reported colossal NTE with an extended temperature interval covering RT in fine-powdered Mn0.98CoGe.21 In their work, about 141 ppm/K of αL with 90 K of operation temperature window (ΔT) is obtained in Mn0.98CoGe powders after 10 times of thermal cycling treatments, which indicate both thermal cycling and ball milling (BM) methods can effectively optimize the NTE performance of MnCoGe-based alloys.21 However, there has been only a few corresponding investigations so far. Meanwhile, the ΔT of NTE is still very limited for these alloys due to the steepness of coincident MT, which is a very serious drawback for practical applications. Hence, it is greatly desired to broaden the ΔT of NTE and simultaneously retain large αL. Recently, Hu et al. reported that the temperature range of MT would be effectively broadened by introducing the residual strain (RS) in Mn-basehexagonal systems.23,24 In this work, therefore, the RS is introduced into the fine-powdered Mn0.965Co1.035Ge prepared via high energy BM. The MT temperature interval, and accordingly ΔT, is significantly broadened, while the very large αL is retained simultaneously.

3. RESULTS AND DISCUSSION The XRD patterns recorded at RT for the Mn0.965Co1.035Ge BM samples with different times are shown in Figure 1a. As can be seen, all the BM samples reveal the coexistence of hexagonal Ni2In-type austenite and orthorhombic TiNiSi-type martensite structure, all with the dominant fraction of the hexagonal phase, implying that an MT occurs around RT. It is also clear that the XRD patterns were sharpest for the BM-0.5h sample, becoming weaker and less well resolved as a function of ball mill time (Figure 1a), suggesting that the sample becomes appreciably amorphous with increasing BM time. The crystalline structure of orthorhombic TiNiSi-type martensite and hexagonal Ni2In-type austenite is shown in Figure 1, panels b and c, respectively. It is demonstrated that the TiNiSitype structure can be regarded as an orthorhombic distortion of the hexagonal Ni2In-type structure from the crystallographic point of view.17 The temperature-dependent XRD measurements are performed in the temperature range 130−450 K during the heating run for Mn0.965Co1.035Ge BM-0.5h (Figure 2a), BM-4h (Figure 2b), and BM-12h (Figure 2c) samples. Here, circle and diamond symbols denote the orthorhombic TiNiSi-type martensite and hexagonal Ni2In-type austenite structure, respectively. For the sample BM-0.5h (Figure 2a), the dominant phase is the orthorhombic structure with a tiny trace of hexagonal Ni2In-type structure at 171 K. When the temperature increases to 211 K, the peaks of the hexagonal

2. EXPERIMENTAL INFORMATION Synthesis. A polycrystalline sample of Mn0.965Co1.035Ge alloy was prepared by the arc-melting technique at Ar atmosphere in a watercooled copper crucible. The samples were flipped and remelted 3−4 14200

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

increase at 180 and 162 K for BM-4h and BM-12h samples, respectively, which is lower than that of sample BM-0.5h. Furthermore, the orthorhombic phase still exists until 390 and 422 K for BM-4h and BM-12h samples, respectively. All these results suggest that more and more hexagonal phase is retained at even very low temperatures, and the temperature interval of MT is broadened with the increasing BM time. In addition, it is pronounced that the diffraction peaks widen from BM-0.5h to BM-12h, as shown in Figure 2a−c. This implies the grain size reduces with the elongating BM time. Meantime, the wider, weaker, and more asymmetric peaks suggest that the sample is becoming appreciably amorphous with increasing BM time (see Figure 2a−c), which is consistent with Figure 1a. Displayed in Figure 2d−f are the evolutions of lattice parameters with increasing temperature for both orthorhombic and hexagonal structures, which were obtained from the refining of temperature-dependent XRD patterns for three BM samples (see Figure 3a), respectively. Previous studies manifested that the axes of these two structures are related as aorth = chex, borth = ahex, and corth / 3 = ahex .22,27,28 As can be seen from Figure 2d−f, the lattices for all three samples reduce suddenly from the aorth to chex and elongate from the borth to ahex with the increasing temperature during the MT process, while there is almost no change from corth / 3 to ahex, for which the volume of unit cell has a large negative expansion from Vorth. to 2Vhex (shown in Figure 4a−c).3,27,28 Figure 3a shows a Rietveld fit to the diffraction data of BM0.5h sample at T = 171 K. From the Rietveld refining, the temperature-dependent hexagonal and orthorhombic phase fraction are estimated and displayed in Figure 3b for BM-0.5h, BM-4h, and BM-12h sample, respectively. As can be seen, with the increasing BM time, more and more fractions of the hexagonal phase cannot martensitically transform to the orthorhombic phase even during the whole temperature range. The change ratio is increasingly lowered, being ∼75%, ∼60%, and ∼53% for these three BM samples, respectively.

Figure 1. (a) The XRD patterns for BM Mn0.965Co1.035Ge samples with different times. Circles and diamonds indicate orthorhombic TiNiSi-type martensite and hexagonal Ni2In-type austenite, respectively. Panels (b) and (c) give the crystalline structure of orthorhombic TiNiSi-type and hexagonal Ni2In-type structure, respectively.

structure begin to appear, and meanwhile, the orthorhombic structure is suppressed. With the increase of temperature to 361 K, the orthorhombic structure becomes less and less at the expense of the hexagonal phase. Above 361 K, almost no orthorhombic structure exists. Obviously, the temperatureinduced MT occurs between 211 and 361 K. With the increasing BM time, the trace of the hexagonal structure at the lowest measurement temperature, i.e., T ≈ 140 and 132 K for the BM-4h and BM-12h, respectively, becomes more and more obvious. And then, the hexagonal phase begins to remarkably

Figure 2. Temperature-dependent XRD patterns for sample (a) BM-0.5h, (b) BM-4h, and (c) BM-12h, respectively. The evolution of lattice parameters with the increase of temperature for (d) BM-0.5h, (e) BM-4h, (f) BM-12h sample, respectively. 14201

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

volume has a contraction along with the FMT, which can be as large as ΔV/V ≈ −4.4, −4.4, −4.5% with the increasing temperature for the samples BM-0.5h, BM-4h, BM-12h, respectively. Here, the difference in lattice volume is defined as ΔV/V ≈ (2Vhex.− Vorth.)/Vorth..3,27,28 On the basis of the refined values of cell volume and phase fraction (Figure 3b), we calculated the average volumetric NTE coefficient (αv) and accordingly the linear expansion αL in view of the equation αL ≈ 1/3αv for the polycrystalline materials.29 For the BM-0.5h sample, as seen in Figure 4d, the ΔV/V shows an obvious NTE effect with a very broad operation temperature window of ΔT = 167 K (from 222 to 389 K) covering RT. Here, the operation temperature window is defined as the temperature range in which a negative slope of ΔV/V-T occurs (Figure 4d− f). The average αv is about −189 ppm/K, and accordingly, the corresponding αL value reaching up to about −63 ppm/K (Figure 4d). When the BM time is elongated to 4 h and further 12 h, as displayed in Figure 4e,f, ΔT is further broadened to 221 K (140−361 K) with αv ≈ −124 ppm/K (αL ≈ −41.3 ppm/K) and 208 K (234−442 K) with αv ≈ −120 ppm/K (αL ≈ −40 ppm/K) for BM-4h and BM-12h sample, respectively. As illustrated in Figure 5, we compare the |αL| and ΔT values of the present alloys with those of selected MnCoGe-based and other well-known NTE materials.1,3,5−10,12,21,30−33 As seen from Figure 5, some materials have a large |αL|, but limited ΔT. For example, Bi0.95La0.05NiO3 has a limited ΔT of 60 K, though its |αL| is 82 ppm/K.32 Furthermore, ΔT is only 20 K, although |αL| is 57 ppm/K for Mn3.3Ag0.7N,9 and up to 145.4 ppm/K for (Ga0.7Cu0.3)0.85Mn0.15NMn3.10 Such narrow ΔT is disadvantageous for practical applications. In contrast, other materials have a large ΔT, but very small |αL|, which is also unfavorable. For example, the ΔT = 1049.7 K for the famous NTE material, ZrW2O8, is unprecedented, but its |αL| is only 9 ppm/K.1 Similarly, |αL| is only 3.9 ppm/K and even 2.74 ppm/K for LaFe10.1Si2.914 and Hf0.85Nb0.15Fe2,5 respectively, though their ΔT can reach up to 220 and 150 K. We divide these materials in Figure 5 into two areas along ΔT = 150 K (green dash line).

Figure 3. (a) Rietveld fit to the diffraction data of BM-0.5h sample at T = 171 K. (b) The temperature dependence of hexagonal and orthorhombic phase fraction for BM-0.5h, BM-4h, and BM-12h sample, respectively. The change ratio between the orthorhombic martensite and hexagonal austenite is ∼75%, ∼60%, and ∼53% for these three samples, respectively. A, B, and C denote the hexagonal austenite fraction at T = 300 K for these three samples, respectively.

Especially, when the BM time increases to 12 h, the coexistence of hexagonal and orthorhombic structures can be observed during the whole measured temperature range (even up to ∼440 K, 97% hexagonal phase, +3% orthorhombic phase). All these indicate that the temperature range of MT broadenss with increasing BM time. Shown in Figure 4a−c is the temperature dependence of cell volume for three BM samples. It should be noticed that the

Figure 4. Temperature dependence of cell volume for the samples (a) BM-0.5h, (b) BM-4h, (c) BM-12h, respectively. The temperature-dependent volumetric thermal expansions ΔV/V (441 K) (the reference temperature is 441 K) for the samples (d) BM-0.5h, (e) BM-4h, (f) BM-12h, respectively. The NTE temperature region is identified for all samples. The linear fit part signifies the constant thermal expansion coefficient. 14202

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

Figure 5. Schematic illustration of the |αL| value and the magnitude of ΔT for the well-known NTE materials (the construction of the value of the αL and ΔT is based on the reports in the corresponding literature): 1-nano CuO (ref 30), 2-SrCu3Fe4O12 (ref 31), 3Bi0.95La0.05NiO3 (ref 32), 4-ScF3 (ref 33), 5-LaFe10.5Co1.0Si1.5 (ref 12), 6-LaFe10.1Si2.9 (ref 14), 7-ZrW2O8 (ref 1), 8-Hf0.95Nb0.05Fe2 (ref 5), 9-Hf0.85Nb0.15Fe2 (ref 5),10-Mn3(Ga0.5Ge0.4Mn0.1)N (ref 6), 11Mn3(Zn0.4Sn0.6)N (ref 7), 12-Mn3Cu0.5Ge0.5N (ref 8), 13-Mn3.3Ag0.7N (ref 9), 14-(Ga0.7Cu0.3)0.85Mn0.15NMn3 (ref 10), and 15-Mn0.98CoGe (ref 21), 16-MnCo0.98Cr0.02Ge (ref 3), 17-Mn0.965Co1.035Ge (this work). The inset indicates the partial enlargement.

Figure 6. (a) The thermomagnetic M(T) curves under μ0H = 0.1 T in ZFC (solid symbols) and FC (open symbols) modes, and (b) the dM/dT(T) curves in ZFC mode for the Mn0.965Co1.035Ge BM samples. The long arrows denote the temperature evolution. The inset indicates the comparison of M(T) behavior between bulk and BM0.5h samples.

It is fascinating to find that the overwhelming majority of NTE materials are concentrated on the side of ΔT < 150 K. In other words, the NTE materials with the ΔT larger than 150 K are very few. Even rarer are the ones possessing both large ΔT and appreciable αL. Therefore, for samples in this work, the excellent NTE properties combining giant αL and ultrawide ΔT should be more suitable for the practical application. As shown in the region of ΔV/V-T curves, marked by the red line in Figure 4d−f, even if for NTE which is nearly independent of temperature,3 the giant αL value in a very wide temperature range is still to be obtained. The average αv value (and accordingly αL value) can reach up to −156 ppm/K (αL=-52 ppm/K), −175 ppm/K (αL=-58.3 ppm/K), and −195 ppm/K (αL=-65 ppm/K) for the BM-0.5h, BM-4h, and BM12h samples, respectively. Meanwhile, it is intriguing that wide ΔT near RT is still obtained, being ΔT = 140, 110, and 80 K for these three samples, respectively. These ΔT values are twice or almost thrice as large as ΔT value in bonded MnCo0.98Cr0.02Ge compound.3 As mentioned above, the excellent NTE properties of Mn0.965Co1.035Ge fine powders originate from the abnormal volume expansion around MT. Additionally, it has been reported that the NTE properties are controlled by the thermomagnetic M(T) behaviors.3,11,12,15 To better comprehend the mechanism of NTE properties, Figure 6a presents ZFC and FC M(T) curves under μ0H = 0.1 T for the Mn0.965Co1.035Ge fine powders with different BM times. It should first be observed from the inset of Figure 6a that the BM-0.5h sample exhibits a coupled FMT with a thermal hysteresis (Thys) of ∼22 K. It is evident that, however, this FMT is not as sharp as that of Mn0.965Co0.035Ge bulk counterpart.22 During the BM process, the collisions between samples and balls and ones among sample particles may change significantly the stress distribution and possibly introduce residual strain (RS).23 We roughly estimate the strain (%) in the BM samples from the XRD data at RT via the jade software. The value is about 0.124, 0.165, 0.355 for BM-0.5h, B-4h, and BM-12h samples, respectively. Evidently, the strain value increases with the elongation of BM time. It is wellknown that the FMT is very closely related to the stress.3,19,23,24,34 For bulk samples, annealing induces the stress

release and structural relaxation, which produce the sharp FMT and accordingly narrow FMT temperature interval (inset of Figure 6a).22,34 For BM samples, the uneven distribution of RS enforces the defects involved redistribution in the grain and grain boundaries and stabilizes the hexagonal austenite, pushing MT to lower temperature.23,24,34 A certain proportion of austenite loses MT with temperature reducing.23 And even in the temperature range far below MT temperature Tt, e.g., 170 K for BM-0.5h sample, the austenite can still be observed (Figure 2a and Figure 3). This should be responsible for the obvious broadening of the FMT temperature interval, and accordingly the widened ΔT of NTE,12 for the BM-0.5h sample in comparison with the bulk (inset of Figure 6a).3,23,24 In fact, a two-step phase transition is discerned in BM-0.5h sample (Figure 6 and Figure S1). Around Tt ≈ 332 K, shown in Figure 6b, there is a very small negative peak in the dM/ dT(T) curve, which should correspond to the MT.24 With the further descending temperature, another negative peak in the dM/dT(T) curve is observed at Thex C ≈ 278 K (Figure 6b), which is very close to the Curie temperature (∼275 K) of hexagonal austenite in stoichiometric MnCoGe.35,36 Thus, it should correspond to the single magnetic transition of hexagonal Ni2In-type state, which further demonstrates that a certain amount of parent phase loses MT and can be retained even very low temperature.23,24 The two-step phase transition can also clearly be seen in the DSC heating curve (Figure S1a). When the BM time increases to 4 h and further 12 h, two-step phase transition is more distinguishable, as displayed in M(T) and dM/dT(T) curves (Figure 6a,b) as well as DSC curves (Figure S1b,c). Meanwhile, the phase transition becomes more gradual, which would explain the broader NTE ΔT for BM-4h and BM-12h samples. Almost invariable Thex C (∼278 K and ∼272 K for BM-4h and BM-12h, respectively) further verifies the simple magnetic transition of retained austenite phase, while Tt increases markedly from ∼332 K for BM-0.5h to ∼362 K for BM-4h sample first, and then remains unchanged for the BM-12h sample. In addition, Thys around Tt remains constant for BM-4h and BM-12h samples, ∼23 K. It is worth noting that the appreciable Thys can still be observed around Thex C , ∼20 K and ∼14 K for BM-4h and BM-12h sample, 14203

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

XANES spectra are divided into two parts: one is the pre-edge structure and the other is defined as the main edge (or main line) above the pre-edge part.22,37−39 The pre-edge structure is assigned as the quadrupole transition from the 1s core state to the empty 3d states, that is, 3dN → 1s1 3dN+1,38,39 while the main edge describes the dipole transitions from the 1s core state to the 4p unoccupied density of states of 3d-band.22,40 For Co K-edge, as seen Figure 8a, the intensity of main edge of bulk and BM samples are almost identical. Therefore, we must consider the impact of preparation condition on the phase fraction from the pre-edge structure. As displayed in Figure 8a and its inset, the enlarged view of pre-edge, the intensity of pre-edge unambiguously increases from the bulk to BM samples and continues to increase from BM-0.5h to BM4h sample. Strikingly, a visible pre-edge peak appears at ∼7709.6 (position A) for BM-4h sample.38 As reported, the pre-edge intensity is in connection with the occupancy of the available 3d states.37−39 Thus, the increase of pre-edge intensity implies the growth of the 1s 3d quadrupole excitation and accordingly the strengthening of 4p(Ge)-3d(Co) mixing. 38,39 In addition, from bulk to BM samples and continuously from BM-0.5h to BM-4h sample, there is a lowering energy position, indicating the increasing electron densities for these samples. Because the BM samples are not oxidized (no impurity phases appear, as seen in Figures 1 and 2), this increase of the electron densities should be due to the charge transfer from Ge ligands to Co, as both Co and Mn are metallic, and thus the charge has to come from Ge. That is, the lowering of the charge-transfer energy leads to an increase of the 4p(Ge)-3d(Co) mixing.39 In the MnCoGe hexagonal structure, it has been reported that Co−Ge is a covalent bond via the formation of the strong orbital covalent hybridization between (sp)3 of Ge atoms and (sd)3 of Co atoms.41 Hence, hexagonal structure is stabilized from the bulk to BM samples. From BM-0.5h to BM-4h samples, more 4p(Ge)-3d(Co) hybridization should result in the retention of more fraction of hexagonal austenite. For BM12 h sample, however, the pre-edge intensity does not continue to increase but decline back to almost the same level as the BM-0.5 h. This fact can be explained by the percentage of hexagonal phase at RT (e.g., at T = 300 K) for three BM samples, as demonstrated by A, B, and C points in Figure 3b. Interestingly, the change of Co−Ge bond length, which is obtained from the EXAFS of Co K-edge for bulk and BM samples (Figure 8c), also supports the aforementioned conclusion. From bulk to BM samples, first, the reduction of Co−Ge interatomic distance indicates the strengthening of 4p(Ge)-3d(Co) mixing.39 Second, for three BM samples, the Co−Ge separation at ∼4.43 Å reduces to ∼4.40 Å first and then increases to ∼4.42 Å, implying the sp orbital overlap first enhances and then weakens, which accord well with the discussion above. Finally, it is interesting that for the Co−Ge bond at ∼2.46 Å for BM samples, though it has the almost unchanged Co−Ge distance, the BM-4h has the lowest peak intensity (Co as the central atom), and thus its coordination number is lowest, implying it is less ordered than other BM samples, which is consistent with the XRD results discussed above. In addition, the results obtained from the XANES and EXAFS of Mn K-edge corroborate the variation of hexagonal phase proportion for bulk and BM samples (Figure 8b,d). First, the intensity of pre-edge grows from bulk to BM samples, and the growth follows from BM-0.5h to BM-4h (inset of

respectively. This indicates that MT accompanies all along, further corroborating the coexistence of orthorhombic martensite and hexagonal austenite within a broad temperature range (Figure 3b), which slows down the phase transition. With the BM time increasing, the average particle size of pulverized samples keeps reducing, as presented in Figure S3. From BM-0.5h to BM-4h sample, different particle sizes and different stress relaxation processes give rise to more uneven distribution of RS.23,24 Higher percentage of hexagonal austenite loses MT in BM-4h sample,23,24 which agrees well with the temperature-dependent XRD results discussed above (Figure 2 and Figure 3b). The hexagonal parent phase and orthorhombic martensite coexist in a broader temperature range, which widens the phase transition width.23,24 As a consequence, the ΔT of NTE becomes wider from BM-0.5h to BM-4h sample. When the BM time further elongates, it is reasonably inferred that the RS relaxation process possibly reaches saturation. This is supported by the very similar variation trend of phase fraction with temperature changing and very close phase volume fraction in the low temperature range (near 150 K) for BM-4h and BM-12h samples, as exhibited in Figure 3b. Therefore, the values of NTE αL and ΔT do not increase further, nor the values of Tt and Thys from BM-4h to BM-12h sample (Figure S6). In order to further support the aforementioned analysis and discussion, M(H) curves at T = 5 K are measured and plotted in Figure 7 for bulk and BM samples. It is evident that the

Figure 7. M(H) curves measured at T = 5 K for bulk and BM samples. Evidently, the saturation magnetization reduces before and after ball milling, and continually decreases with the elongating BM time.

saturation magnetization reduces before and after ball milling, and continually decreases with the elongating BM time. The value is 111.3, 107.2, 92.6, and 81.7 Am2 kg−1 for bulk and BM samples (μ0H = 5 T), respectively. This further demonstrates more and more hexagonal phase with relatively small magnetic moment are retained at even very low temperature from bulk to BM samples, which validates our conclusion. As discussed above, a fraction of hexagonal austenite loses MT by BM treatment. To further corroborate this, we examined the bulk and BM Mn0.965Co1.035Ge samples at the ambient temperature using the X-ray absorption spectroscopy, including X-ray absorption near edge spectra (XANES) and extended X-ray absorption fine structure (EXAFS), of Co and Mn K-edge,26 as shown in Figure 8a−d, respectively. For 3d transition-metals, the K-edge XANES correlates with the excitation process from 1s core electron to unoccupied bounded states, which can provide the electronic state of Xray absorbing atom and its surrounding local structure.37 Considering the single electron excitation model, the 1s 14204

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

Figure 8. Co (a and c) and Mn (b and d) K-edge XANES (a and b) and EXAFS (c and d) spectra of bulk (dash),22 BM-0.5h (solid), BM-4h (dot), and BM-12h (short dash) Mn0.965Co1.035Ge samples recorded at RT, respectively. The inset of (a) and (b) denotes the enlarged view of pre-edge structure.

Figure 8b). Similarly, this growth manifests the increase of 3dN → 1s1 3dN+1.38,39 It has been reported that the magnetic moment of the MnCoGe compound is mainly localized on Mn atoms due to the almost vanishing of Co moments originating from the strong Co−Ge covalent hybridization as mentioned above.36,41 So, the increase of 3d empty density states near the Fermi level suggests more electrons would not be able to contribute to the total magnetic moment of the Mn atoms.37 This indicates that the higher fraction of the hexagonal phase is retained for BM-0.5h and further to BM-4h samples in contrast with the bulk sample owing to smaller saturation magnetic moment of hexagonal austenite than that of orthorhombic martensite.35,36 Similarly, the intensity of pre-edge of the BM12 h sample reduces back to almost the same level as the BM0.5 h, which is also interpreted by the percentage of the hexagonal phase at RT discussed above (Figure 3). Likewise, it is interesting that the increase of the Mn−Mn bond length from bulk to BM-0.5h and further to BM-4h, obtained from the EXAFS of the Mn K-edge (Figure 8d), indicates the weakening of the Mn−Mn interaction and accordingly the reduction of magnetic moment for samples. When the BM time further increases to 12 h, the Mn−Mn distance decreases back to almost an identical degree with that of BM-0.5h, which indicates that the magnetic moment of BM-12h increases slightly, implying a smaller hexagonal structure percentage at RT (Figure 3b).

approach saturation, for which the NTE properties accordingly reach saturation. The results obtained from the XANES and EXAFS of Co and Mn K-edge corroborate these conclusions. These results demonstrated a new strategy to achieve the excellent NTE properties, including both broad ΔT and giant ΔT, which highlights the potential applications of NTE materials in thermal compensators.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.8b02195. The DSC curves for BM-0.5h, BM-3h, BM-4h, and BM12h samples (Figure S1), M(T) curves under μ0H = 0.1 T applied field and dM/dT(T) curves for BM-3h and BM-5H samples (Figure S2), SEM images for BM-0.5h, BM-3h, BM-4h BM-5h, and BM-12h samples and average particle size for these samples (Figure S3), enlarged view of pre-edge structure of Co K-edge (Figure S4), comparison between the dM/dT(T) and ΔV/V(T) for BM samples (Figure S5), and NTE and MT parameters (Figure S6) (PDF)

■

4. CONCLUSIONS In summary, we prepared the fine powders by BM and studied the NTE behaviors in these samples. These BM samples are shown to have NTE properties of both ultrawide ΔT and giant αL. The fundamental change of stress distribution and accordingly the uneven distribution of RS stabilize the hexagonal parent phase and pull down the MT temperature. With the increasing BM time, a higher percentage of austenite loses MT, which broadens the coexistence temperature range of hexagonal austenite and orthorhombic martensite, for which the ΔT of NTE becomes broader. However, when the BM time elongates more than 4 h, the RS relaxation process would

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Shengcan Ma: 0000-0001-6364-9059 Dunhui Wang: 0000-0002-3317-8572 Author Contributions #

S.Y. and S.M. contributed equally to this work.

Notes

The authors declare no competing financial interest. 14205

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry

■

(12) Huang, R. J.; Liu, Y. Y.; Fan, W.; Tan, J.; Xiao, F. R.; Qian, L. H.; Li, L. F. Giant Negative Thermal Expansion in NaZn13-Type La(Fe,Si,Co)13 Compounds. J. Am. Chem. Soc. 2013, 135, 11469− 11472. (13) Li, W.; Huang, R. J.; Wang, W.; Zhao, Y. Q.; Li, S. P.; Huang, C. J.; Li, L. F. Abnormal thermal expansion properties of cubic NaZn13-type La(Fe,Al)13 compounds. Phys. Chem. Chem. Phys. 2015, 17, 5556−5560. (14) Li, S. P.; Huang, R. J.; Zhao, Y. Q.; Li, W.; Wang, W.; Huang, C. J.; Gong, P. F.; Lin, Z. S.; Li, L. F. Broad Negative Thermal Expansion Operation-Temperature Window Achieved by Adjusting Fe-Fe Magnetic Exchange Coupling in La(Fe,Si)13 Compounds. Inorg. Chem. 2015, 54, 7868−7872. (15) Zhao, Y. Q.; Huang, R. J.; Li, S. P.; Wang, W.; Jiang, X. X.; Lin, Z. S.; Li, J. T.; Li, L. F. Effect of cobalt doping on the structural, magnetic and abnormal thermal expansion properties of NaZn13-type La(Fe1‑xCox)11.4Al1.6 compounds. Phys. Chem. Chem. Phys. 2016, 18, 20276−20280. (16) Wang, Y. X.; Zhang, H.; Liu, E. K.; Zhong, X. C.; Tao, K.; Wu, M. L.; Xing, C. F.; Xiao, Y. N.; Liu, J.; Long, Y. Outstanding Comprehensive Performance of La(Fe, Si)13Hy/In Composite with Durable Service Life for Magnetic Refrigeration. Adv. Electron. Mater. 2018, 4, 1700636. (17) Ren, Q. Y.; Hutchison, W. D.; Wang, J. L.; Studer, A. J.; Campbell, S. J. Magnetic and Structural Transitions Tuned through Valence Electron Concentration in Magnetocaloric Mn(Co1−xNix)Ge. Chem. Mater. 2018, 30, 1324−1334. (18) Si, X. D.; Shen, Y. L.; Ma, X. X.; Chen, S. J.; Lin, J.; Yang, J.; Gao, T.; Liu, Y. S. Field dependence of magnetic entropy change and estimation of spontaneous magnetization in Cd substituted MnCoGe. Acta Mater. 2018, 143, 306−317. (19) Caron, L.; Trung, N. T.; Bruck, E. Pressure-tuned magnetocaloric effect in Mn0.93Cr0.07CoGe. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84 (R), 020414. (20) Liu, Y.; Qiao, K. M.; Zuo, S. L.; Zhang, H. R.; Kuang, H.; Wang, J.; Hu, F. X.; Sun, J. R.; Shen, B. G. Negative thermal expansion and magnetocaloric effect in Mn-Co-Ge-In thin films. Appl. Phys. Lett. 2018, 112, 012401. (21) Lin, J. C.; Tong, P.; Zhang, K.; Tong, H. Y.; Guo, X. G.; Yang, C.; Wu, Y.; Wang, M.; Lin, S.; Chen, L.; Song, W. H.; Sun, Y. P. Colossal negative thermal expansion with an extended temperature interval covering room temperature in fine-powdered Mn0.98CoGe. Appl. Phys. Lett. 2016, 109, 241903. (22) Ma, S. C.; Ge, Q.; Hu, Y. F.; Wang, L.; Liu, K.; Jiang, Q. Z.; Wang, D. H.; Hu, C. C.; Huang, H. B.; Cao, G. P.; Zhong, Z. C.; Du, Y. W. Driving higher magnetic field sensitivity of the martensitic transformation in MnCoGe ferromagnet. Appl. Phys. Lett. 2017, 111, 192406. (23) Wu, R. R.; Shen, F. R.; Hu, F. X.; Wang, J.; Bao, L. F.; Zhang, L.; Liu, Y.; Zhao, Y. Y.; Liang, F. X.; Zuo, W. L.; Sun, J. R.; Shen, B. G. Critical dependence of magnetostructural coupling and magnetocaloric effect on particle size in Mn-Fe-Ni-Ge compounds. Sci. Rep. 2016, 6, 20993. (24) Liu, Y.; Shen, F. R.; Zhang, M.; Bao, L. F.; Wu, R. R.; Zhao, Y. Y.; Hu, F. X.; Wang, J.; Zuo, W. L.; Sun, J. R.; Shen, B. G. Stress modulated martensitic transition and magnetocaloric effect in hexagonal Ni2In-type MnCoGe1‑xInx alloys. J. Alloys Compd. 2015, 649, 1048−1052. (25) Rodriguez-Carvajal, J. FULLPROF: A Program for Rietveld Refinement and Pattern Matching Analysis; Toulouse, 1990. (26) Ravel, B.; Newville, M. Athena, Artemis, Hephaestus: data analysis for X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537−541. (27) Johnson, V. Diffusionless Orthorhombic to Hexagonal Transitions in Ternary Silicides and Germanides. Inorg. Chem. 1975, 14, 1117−1120. (28) Kanomata, T.; Ishigaki, H.; Suzuki, T.; Yoshida, H.; Abe, S.; Kaneko, T. Magneto-volume effect of MnCo1−xGe(0⩽x⩽0.2). J. Magn. Magn. Mater. 1995, 140−144, 131−132.

ACKNOWLEDGMENTS XAS experiments were performed at the Canadian Light Source, which is supported by NSERC, the National Research Council Canada, the Canadian Institutes of Health Research, the Province of Saskatchewan, Western Economic Diversification Canada, and the University of Saskatchewan. We acknowledge the technical help of Aimee Maclennan during the XAS measurements. We also appreciate Prof. Yunzhi Tang (Jiangxi University of Science and Technology) for the help of DSC measurements. This work has been financially supported by the National Natural Science Foundation of China (Grant Nos. 51561023 and 51671097), the Jiangxi Provincial Education Department’s Key Research Project of Science and Technology (Grant No. GJJ170487), the Ganzhou Science and Technology Innovation Talent Plan (Grant No. 3208000033), the Training Project of Outstanding Doctoral Dissertations in Jiangxi University of Science and Technology (Grant No. YB2017010), and the Graduate Student Innovation Special Funds Project of Jiangxi University of Science and Technology (Grant Nos. ZS2017-S005 and XS2017-S006).

■

REFERENCES

(1) Evans, J. S. O.; Hu, Z.; Jorgensen, J. D.; Argyriou, D. N.; Short, S.; Sleight, A. W. Compressibility, Phase Transitions, and Oxygen Migration in Zirconium Tungstate, ZrW208. Science 1997, 275, 61− 65. (2) Hao, Y. M.; Gao, Y.; Wang, B. W.; Qu, J. P.; Li, Y. X.; Hu, J. F.; Deng, J. C. Negative thermal expansion and magnetic properties of Y2Al3Fe14‑xMnx compounds. Appl. Phys. Lett. 2001, 78, 3277−3279. (3) Zhao, Y. Y.; Hu, F. X.; Bao, L. F.; Wang, J.; Wu, H.; Huang, Q. Z.; Wu, R. R.; Liu, Y.; Shen, F. R.; Kuang, H.; Zhang, M.; Zuo, W. L.; Zheng, X. Q.; Sun, J. R.; Shen, B. G. Giant Negative Thermal Expansion in Bonded MnCoGe-Based Compounds with Ni2In-Type Hexagonal Structure. J. Am. Chem. Soc. 2015, 137, 1746−1749. (4) Mary, T. A.; Evans, J. S. O.; Vogt, T.; Sleight, A. W. Negative Thermal Expansion from 0.3 to 1050 K in ZrW2O8. Science 1996, 272, 90−92. (5) Song, Y. Z.; Chen, J.; Liu, X. Z.; Wang, C. W.; Gao, Q. L.; Li, Q.; Hu, L.; Zhang, J.; Zhang, S. T.; Xing, X. R. Structure, Magnetism, and Tunable Negative Thermal Expansion in (Hf,Nb)Fe2 Alloys. Chem. Mater. 2017, 29, 7078−7082. (6) Takenaka, K.; Takagi, H. Giant negative thermal expansion in Ge-doped anti-perovskite manganese nitrides. Appl. Phys. Lett. 2005, 87, 261902. (7) Takenaka, K.; Asano, K.; Misawa, M.; Takagi, H. Negative thermal expansion in Ge-free antiperovskite manganese nitrides: Tindoping effect. Appl. Phys. Lett. 2008, 92, 011927. (8) Song, X. Y.; Sun, Z. H.; Huang, Q. Z.; Rettenmayr, M.; Liu, X. M.; Seyring, M.; Li, G. N.; Rao, G. H.; Yin, F. X. Adjustable Zero Thermal Expansion in Antiperovskite Manganese Nitride. Adv. Mater. 2011, 23, 4690−4694. (9) Lin, J. C.; Tong, P.; Tong, W.; Lin, S.; Wang, B. S.; Song, W. H.; Zou, Y. M.; Sun, Y. P. Tunable negative thermal expansion related with the gradual evolution of antiferromagnetic ordering in antiperovskite manganese nitrides Ag1‑xNMn3+x (0 ≤ x ≤ 0.6). Appl. Phys. Lett. 2015, 106, 082405. (10) Guo, X. G.; Tong, P.; Lin, J. C.; Yang, C.; Zhang, K.; Wang, M.; Wu, Y.; Lin, S.; Song, W. H.; Sun, Y. P. Large negative thermal expansion in (Ga0.7Cu0.3)1‑xMnxNMn3 (x ≤ 0.4), compensating for the thermal expansion of cryogenic materials. Scr. Mater. 2017, 128, 74−77. (11) Li, S. P.; Huang, R. J.; Zhao, Y. Q.; Wang, W.; Han, Y. M.; Li, L. F. Zero Thermal Expansion Achieved by an Electrolytic Hydriding Method in La(Fe,Si)13 Compounds. Adv. Funct. Mater. 2017, 27, 1604195. 14206

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207

Article

Inorganic Chemistry (29) Chen, J.; Hu, L.; Deng, J. X.; Xing, X. R. Negative thermal expansion in functional materials: controllable thermal expansion by chemical modifications. Chem. Soc. Rev. 2015, 44, 3522−3567. (30) Zheng, X. G.; Kubozono, H.; Yamada, H.; Kato, K.; Ishiwata, Y.; Xu, C. N. Giant negative thermal expansion in magnetic nanocrystals. Nat. Nanotechnol. 2008, 3, 724−726. (31) Yamada, I.; Tsuchida, K.; Ohgushi, K.; Hayashi, N.; Kim, J.; Tsuji, N.; Takahashi, R.; Matsushita, M.; Nishiyama, N.; Inoue, T.; Irifune, T.; Kato, K.; Takata, M.; Takano, M. Giant negative thermal expansion in the iron perovskite SrCu3Fe4O12. Angew. Chem., Int. Ed. 2011, 50, 6579−6582. (32) Azuma, M.; Chen, W. T.; Seki, H.; Czapski, M.; Olga, S.; Oka, K.; Mizumaki, M.; Watanuki, T.; Ishimatsu, N.; Kawamura, N.; Ishiwata, S.; Tucker, M. G.; Shimakawa, Y.; Attfield, J. P. Colossal negative thermal expansion in BiNiO3 induced by intermetallic charge transfer. Nat. Commun. 2011, 2, 347. (33) Greve, B. K.; Martin, K. L.; Lee, P. L.; Chupas, P. J.; Chapman, K. W.; Wilkinson, A. P. Pronounced Negative Thermal Expansion from a Simple Structure: Cubic ScF3. J. Am. Chem. Soc. 2010, 132, 15496−15498. (34) Xuan, H. C.; Xie, K. X.; Wang, D. H.; Han, Z. D.; Zhang, C. L.; Gu, B. X.; Du, Y. W. Effect of annealing on the martensitic transformation and magnetocaloric effect in Ni 44.1Mn44.2Sn11.7 ribbons. Appl. Phys. Lett. 2008, 92, 242506. (35) Trung, N. T.; Zhang, L.; Caron, L.; Buschow, K. H. J.; Brück, E. Giant magnetocaloric effects by tailoring the phase transitions. Appl. Phys. Lett. 2010, 96, 172504. (36) Wang, J. T.; Wang, D. S.; Chen, C. F.; Nashima, O.; Kanomata, T.; Mizuseki, H.; Kawazoe, Y. Vacancy induced structural and magnetic transition in MnCo1−xGe. Appl. Phys. Lett. 2006, 89, 262504. (37) Bilovol, V.; Ferrari, S.; Derewnicka, D.; Saccone, F. D. XANES and XPS study of electronic structure of Ti-enriched Nd-Fe-B ribbons. Mater. Chem. Phys. 2014, 146, 269. (38) de Groot, F.; Vankó, G.; Glatzel, P. The 1s X-ray absorption pre-edge structures in transition metal oxides. J. Phys.: Condens. Matter 2009, 21, 104207. (39) Kuzmin, A.; Mironova, N.; Purans, J. The influence of pd mixing and magnetic interactions on the pre-edge peak intensity at the Co (Ni) K absorption edge in Co(Ni)cMg1−cO solid solutions. J. Phys.: Condens. Matter 1997, 9, 5277−5286. (40) Guillou, F.; Ollefs, K.; Wilhelm, F.; Rogalev, A.; Yaresko, A. N.; Yibole, H.; van Dijk, N. H.; Brück, E. Electronic and magnetic properties of phosphorus across the first-order ferromagnetic transition of (Mn,Fe)2(P,Si,B) giant magnetocaloric materials. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 92, 224427. (41) Bazela, W.; Szott, I.; Szytuła, A. Magnetic properties of the CoxNi1‑xMnGe system in the hexagonal phase. Phys. Stat. Sol. (a) 1982, 69, K49−K54.

14207

DOI: 10.1021/acs.inorgchem.8b02195 Inorg. Chem. 2018, 57, 14199−14207