2.1. 1H NMR
Hydrogen sites located at the terminals of BETS molecules (
Figure 1a) have negligibly small hyperfine coupling with conduction electrons [
12], therefore the NMR frequency shift should be determined by the dipolar fields from the 3
d Mn
2+ ion electron spin moments (
Sd = 5/2,
g ≈ 2). For the arbitrary orientation of the magnetic field, the
1H NMR spectrum will count 16 peaks: there are 8 inequivalent crystallographic hydrogen sites and two magnetically inequivalent orientations of the BETS dimer (
Figure 1c). The BETS molecules within the dimer are inversion-symmetric to each other, thus magnetically equivalent.
Figure 2 shows
1H NMR spectrum in
κ-(BETS)
2Mn[N(CN)
2]
3 at
T = 74 K in field
H0 = 7 T oriented at
θ = 22°from
![Crystals 02 00224 i017]()
towards
![Crystals 02 00224 i018]()
direction. The spectrum is shown with respect to
1v0 =
1γH0, where
1γ = 42.5759 MHz/T is the proton gyromagnetic ratio. For the chosen geometry
![Crystals 02 00224 i022]()
, the two different orientations of the BETS dimer shown in
Figure 1c become magnetically equivalent which reduces the total number of peaks to 8. Arrows in
Figure 2 indicate the 8 peaks.
Figure 2.
1H NMR spectrum in
κ-(BETS)
2Mn[N(CN)
2]
3 at
T = 74 K in field
H0 = 7 T oriented at
θ = 22°from
![Crystals 02 00224 i017]()
towards
![Crystals 02 00224 i024]()
direction. The spectrum is shown with respect to
1v0 =
1γH0 .
Figure 2.
1H NMR spectrum in
κ-(BETS)
2Mn[N(CN)
2]
3 at
T = 74 K in field
H0 = 7 T oriented at
θ = 22°from
![Crystals 02 00224 i017]()
towards
![Crystals 02 00224 i024]()
direction. The spectrum is shown with respect to
1v0 =
1γH0 .
Figure 3 presents the angular evolution of
1H NMR peak positions in
κ-(BETS)
2Mn[N(CN)
2]
3 measured at 74 K in field
H0=7 T. The field is in the (
a*c)-plane, and the polar angle,
θ, is reckoned from
![Crystals 02 00224 i017]()
direction towards
![Crystals 02 00224 i023]()
.
Figure 3.
Angular evolution of 1H NMR spectrum in the (a*c) plane measured at 74 K in field H0=7 T. Circles are the measured peak positions. Lines are model calculations using Equation 1.
Figure 3.
Angular evolution of 1H NMR spectrum in the (a*c) plane measured at 74 K in field H0=7 T. Circles are the measured peak positions. Lines are model calculations using Equation 1.
To model the observed proton spectrum one needs to sum up the dipolar fields,
hdip, created at the nucleus site by Mn
2+ electronic spins, and take into account the sample geometry resulting in a demagnetizing field,
hD, and the Lorentz field,
hLor, which is the mean field induced at the nucleus site by the bulk of the sample located outside the dipolar summation sphere [
13].
For the magnetic field in the (
a*c)-plane,
![Crystals 02 00224 i031]()
we model the spectrum as
Here, γn = 1γ; μMn is the thermal average of the Mn2+ magnetic moment projection on the field direction, riis the length of the position vector from the proton site to the Mn site i, αi is the angle between this vector and the field direction, VMn is the unit cell volume per Mn2+ ion, and N = N┴ cos2θ + N║sin2θ is the demagnetization factor.
Since the DC magnetization in
κ-(BETS)
2Mn[N(CN)
2]
3 is determined by Mn
2+ magnetic moments [
3], we put in Equation 3
μMn = M/N A, where
M is the measured DC magnetization per mole (3800 emu/mol at
H0= 7 T,
T = 74 K) and
NA is Avogadro’s number. For our very thin-plate sample we assume the demagnetization factors
N┴ = 1,
N║ = 0. Crystallographic positions of Mn and H atoms required to calculate Equation 1b are available online from the CCDC library as mentioned in the Experimental Section. The sum in Equation 1a has been taken over ~ 200 Mn sites within ±20 Å to provide reasonable convergence.
The peak positions calculated using Equation 1 for each of the 8 inequivalent hydrogen sites are shown in
Figure 3 by solid lines. The agreement between the calculated and the measured peak positions is clearly reasonable despite the absence of any fit parameters used in the calculations. This indicates that
1H NMR frequency shift is determined by the dipolar fields from Mn
2+ at the hydrogen site, at least at this field and temperature.
To check if this is true for all temperatures, we compare the temperature dependences of the DC magnetization and the
1H NMR peak positions.
Figure 4a,b show temperature dependences of, respectively, the molar DC magnetization normalized to the applied field,
M/N, and the normalized frequency shift of the lowest-frequency peak in the spectrum measured in
H ║
a* geometry (
θ = 0 in
Figure 3), for the magnetic fields 1.4 and 7 T.
Figure 4.
(
a) Molar magnetization
vs. temperature in
H = 1.4 and 7 T; (
b) Position of the lowest-frequency peak in
1H NMR spectrum in
![Crystals 02 00224 i053]()
geometry
vs. temperature in
H = 1.4 and 7 T; (
c) Position of the peak for the temperature range 4-150 K, in function of the DC magnetization per Mn ion,
μMn = M/NA, measured at the same temperatures and fields and expressed in terms of the Bohr magneton,
μB. Open red circles and closed black circles correspond to the fields 1.4 and 7 T, respectively.
Figure 4.
(
a) Molar magnetization
vs. temperature in
H = 1.4 and 7 T; (
b) Position of the lowest-frequency peak in
1H NMR spectrum in
![Crystals 02 00224 i053]()
geometry
vs. temperature in
H = 1.4 and 7 T; (
c) Position of the peak for the temperature range 4-150 K, in function of the DC magnetization per Mn ion,
μMn = M/NA, measured at the same temperatures and fields and expressed in terms of the Bohr magneton,
μB. Open red circles and closed black circles correspond to the fields 1.4 and 7 T, respectively.
The right panel of
Figure 4 depicts the plot of the proton frequency shift,
v -
1v0, in function of the magnetization measured at the same temperatures and fields. The linearity of the data in
Figure 4 demonstrates that for the whole temperature and field ranges covered in the experiment, the positions of
1H NMR spectrum peaks are determined by the magnetic subsystem associated with Mn
2+ moments, and can be utilized as Hall probes of the Mn
2+ dipolar fields at hydrogen sites.
Figure 5a shows the evolution of
1H NMR spectrum (
H0 = 1.4T parallel to
![Crystals 02 00224 i017]()
) with temperature. As can be seen, the frequency span of the spectrum increases with decreasing temperature while its shape is maintained down to ~20 K. At lower temperature the peaks broaden rapidly, which is more pronounced on the right-hand side of the spectrum.
Figure 5b depicts the temperature dependence of the half width at half hight,
![Crystals 02 00224 i059]()
, for the leftmost and the rightmost peaks in the spectrum. As one can see in
Figure 5, the linewidth is relatively flat above
TMI ≈ 23 K (especially for the leftmost peak) and increases sharply below this temperature.
Figure 5c demonstrates the plot of the linewidths in function of the DC magnetization. One can see in
Figure 5c a crossover in the linewidth behavior at
TMI which is observed as an upturn from the linear
M-dependence (shown by solid lines) obeyed at higher temperatures.
Figure 5.
(
a) Temperature evolution of
1H NMR spectrum measured at
H0 = 1.4T parallel to
![Crystals 02 00224 i017]()
; (
b) Temperature dependence of the half-linewidth of the leftmost (circles) and the rightmost (squares) peaks; (
c) The leftmost and the rightmost peak half-linewidths in function of the DC magnetization. Solid lines extrapolate the linewidth behavior above 25 K to low temperatures. Dashed lines with numbers on top mark measurement temperatures.
Figure 5.
(
a) Temperature evolution of
1H NMR spectrum measured at
H0 = 1.4T parallel to
![Crystals 02 00224 i017]()
; (
b) Temperature dependence of the half-linewidth of the leftmost (circles) and the rightmost (squares) peaks; (
c) The leftmost and the rightmost peak half-linewidths in function of the DC magnetization. Solid lines extrapolate the linewidth behavior above 25 K to low temperatures. Dashed lines with numbers on top mark measurement temperatures.
Since the
1H NMR peak positions are highly anisotropic as can be seen in
Figure 3, there is a number of trivial reasons for the peaks to be broadened, including sample imperfections on the macroscopic and local levels, as well as minor misalignment of the magnetic field from
![Crystals 02 00224 i017]()
direction, as discussed in [
14]. However, all of them should make the linewidth as linear in the magnetization as the peak position itself is (
Figure 4c). The upturn of the linewidth
M-dependence below
TMI ≈ 23 K (
Figure 5c) indicates the onset of yet another broadening mechanism. Provided that the crystal lattice is intact, this should be related with local-level scatter of the dipolar fields from Mn
2+ at the hydrogen site, for example, a variable from site to site tilt of the Mn
2+ static moments. This site variation of the tilt is apparently random or incommensurate with the crystal lattice, because otherwise a splitting of
1H NMR peaks would take place instead of the featureless broadening observed in the experiment (
Figure 5a). The tilt of the Mn
2+ moments emerging below
TMI, together with the deviation of the magnetization from the Curie-Weiss behavior [
3], indicates apparently the tendency of the Mn
2+ spin system towards the AF order. Frustrated geometrically by triangular arrangement of Mn in the anion layer (
Figure 1e), Mn
2+ system is resolved into a disordered (like spin-glass) or an incommensurate spin structure.
2.2. 13C NMR
The
13C NMR spectra were measured in
κ-(BETS)
2Mn[N(CN)
2]
3 in the external field
H = 7 T aligned perpendicular to
![Crystals 02 00224 i001]()
at 45°to
![Crystals 02 00224 i017]()
. For this experimental geometry, the metallic-state spectrum in theory is represented by four peaks arising from the two magnetically different orientations of the BETS dimers (
Figure 1c) and nonequivalent(“inner” and “outer”) carbon sites of the central C=C bond within the dimer (
Figure 1d) [
15]. The dipolar interaction between
13C spins in the C=C bond, which in general provides another factor of two to the number of peaks, is nearly zero for this field orientation.
Figure 6.
13C NMR spectra measured in the external field
H = 7 T aligned perpendicular to
![Crystals 02 00224 i001]()
at 45° to
![Crystals 02 00224 i017]()
. The spectra are shown with respect to
13v0 = 74.946 MHz. (
a) The spectrum at
T = 50 K. Blue solid lines indicate positions of the resonance peaks calculated with the shift tensor of
κ-(ET)
2Cu[N(CN)
2]Br [
15]; (
b) The evolution of the
13C NMR spectrum with temperature.
Figure 6.
13C NMR spectra measured in the external field
H = 7 T aligned perpendicular to
![Crystals 02 00224 i001]()
at 45° to
![Crystals 02 00224 i017]()
. The spectra are shown with respect to
13v0 = 74.946 MHz. (
a) The spectrum at
T = 50 K. Blue solid lines indicate positions of the resonance peaks calculated with the shift tensor of
κ-(ET)
2Cu[N(CN)
2]Br [
15]; (
b) The evolution of the
13C NMR spectrum with temperature.
Unlike the magnetic ion-free quasi-two-dimensional ET-based conductors where the NMR peaks from magnetically nonequivalent
13C sites are usually well resolved,
13C spectrum in the metallic state of
κ-(BETS)
2Mn[N(CN)
2]
3 for this field orientation is represented by a single featureless Gaussian-shaped line. The left panel in
Figure 6 represents the spectrum taken at
T = 50 K. Using the
13C NMR shift tensors obtained for
κ-(ET)
2Cu[N(CN)
2]Br [
15] (which is nearly the same as for
κ-(ET)
2Cu[N(CN)
2]Cl [
16]), for the given experimental geometry one expects a
κ-(BETS)
2Mn[N(CN)
2]
3 sample to produce the resonance peaks at 2.7, 10.7, 22.7, and 40.8 kHz (with respect to
13v0 = 74.946 MHz). Dipolar fields from Mn
2+ moments provide additional shifts to the resonance frequencies, which can be calculated following Equations 3 using the values of Mn
2+ moments known from the bulk magnetization measurements [
3]. At
T = 50 K the calculated dipolar fields from Mn
2+ move the listed peaks to positions at 0.3, 7.7, 19, and 38.2 kHz. The calculated peak positions are shown by vertical lines in
Figure 6a. One can see that the spectrum measured at 50 K covers fairly well the range of the calculated peak positions indicating that the
13C shift tensor in the title compound is not much different from that in
κ-(ET)
2Cu[N(CN)
2]Br.
Evidently, the peaks from individual carbon sites in the metallic state of
κ-(BETS)
2Mn[N(CN)
2]
3 are vastly broadened that merges them into a single line. Some broadening of
13C NMR peaks has been noticed in
κ-(ET)
2Cu[N(CN)
2]Br and a number of reasons has been recruited to explain it [
15], including spacial variation of the
π-electron spin density due to precursors to Anderson localization or spin-density wave. For the sample reported here, the linewidth is expected broader than in non-magnetic compounds due to the presence of the anisotropic dipolar fields from Mn
2+. For example, the same mechanism that creates the
1H NMR linewidth (
Figure 5b) will be responsible for ~10 kHz linewidth of the
13C peaks at carbon sites at 50 K. Anyway, more detailed analysis of the possible broadening mechanisms in the metallic state is beyond the scope of this communication.
The right panel in
Figure 6 demonstrates the evolution of the
13C NMR spectrum with temperature. The single peak characteristic of the spectrum in the metallic state above
T = 23K ≈
TMI develops below this temperature into a broad symmetric pattern counting 5 pronounced peaks. Below
T = 15 K the spectrum spans the range of nearly ± 1 MHz, which is huge compared to the spectrum width in the metallic state. This cannot result from the dipolar fields created by Mn
2+: calculations show that fully polarized Mn
2+ can provide a dipolar shift ranging from -12.5 to -19 kHz (depending on the carbon site) in 7 T field. Therefore the spectrum in the insulating state evidences enhancement of the electron spin density at carbon sites due to electron spins localized on the dimers of the BETS molecules. Moreover, several pronounced peaks are visible in the low-temperature spectrum, which infers a commensurate order of the localized spins. Finally, the symmetric shape of the spectrum indicates the staggered order, because antiparallel components of the staggered electron spins (
![Crystals 02 00224 i069]()
) produce opposite local fields at carbon sites
i and
j:
![Crystals 02 00224 i070]()
, where
![Crystals 02 00224 i071]()
and
![Crystals 02 00224 i072]()
are the nuclear and electron spin operators, respectively, and
A is the hyperfine tensor. In turn, this signifies the AF exchange interaction between the localized spins. The staggered component of the spin magnetization lies apparently somewhere in the plane perpendicular to the magnetic field, since the anisotropic AF exchange term is usually much smaller than 7 T, the external field of this experiment.
The frequency range of the
13C spectrum in
κ-(BETS)
2Mn[N(CN)
2]
3 at T = 5 K (
Figure 6b) is of the same order of magnitude (~±1 MHz) as observed in the AF state of
κ-(ET)
2Cu[N(CN)
2]Cl [
16] and
β΄-(ET)
2ICl
2 [
17]. The magnitude of the electron spin magnetization of 0.5
μB and 1
μB per dimer, respectively, has been reported for these two compounds. Therefore in
κ-(BETS)
2Mn[N(CN)
2]
3 this value should be within the same range since the hyperfine tensor here is expected to be similar, as the measurements in the metallic phase suggest.
More detailed and quantitative information about the spin structure of the localized πelectrons can hardly be derived at the moment. To do this one needs to know more or less exactly the hyperfine tensors for the central carbons in κ-(BETS)2 Mn[N(CN)2]3, which in turn requires the value of the πspin susceptibility never reported so far (but probably accessible via ESR).