Nature | Vol 577 | 2 January 2020 | 67obtain elemental maps and determine the extent of MX 2 penetration.
These experiments reveal that the M and X spatial distributions of each
variant match well with that of Zr (Extended Data Figs. 1–4), suggesting
uniform dispersion of sheets throughout the crystal, rather than their
accumulation at regions near the crystal surface. Importantly, these
results also indicate that an equilibrium exists between the clusters and
dissolved metal species under the reaction conditions, enabling migra-
tion of MX 2 species to the crystal interior and reversible sheet formation.
The formation of partially filled sheet fragments in 1 (NiBr 2 )9.9 sug-
gested that snapshots of sheet growth could be monitored as a function
of metal halide loading, and towards this end single-crystal structures
were determined at 100 K for samples of 1 treated with increasing equiv-
alents of NiBr 2 or Ni(DME)Cl 2 relative to bipyridine (Fig. 3 ). Reaction
with one equivalent of either metal source exclusively resulted in met-
allation of the bipyridine linkers^17 , confirming that cluster nucleation
occurs at these sites. For nickel(ii) bromide, additional equivalents
populate the rest of the sites, preferring edge sites over those at the
interior (Extended Data Fig. 5a). This trend implies that nickel(ii) bro-
mide sheet formation initiates at the bipyridine sites, followed by a
progressive inward growth towards the centre. In contrast, the remain-
ing sites in the nickel(ii) chloride sheets fill uniformly with increasing
NiCl 2 loading (Extended Data Fig. 5b), further indicating the tendency
of nickel(ii) chloride to form completely filled sheets.
Because the sheets represent fragments of the corresponding metal
halide monolayers, we anticipated that their magnetic behaviour might
be related to that of the bulk material. In the latter, ferromagnetic cou-
pling is dominant within monolayers, and antiferromagnetic coupling
occurs between adjacent layers^21 ,^22. Accordingly, the product of the
molar magnetic susceptibility times temperature (χMT) for each bulk
material initially increases with decreasing temperature as the spins
within each monolayer align ferromagnetically. Below the Néel tem-
perature (TN) for each solid, a sharp decrease in χMT is then observed
as alternating monolayers adopt opposite spin orientations to form an
antiferromagnetic ground state^21 ,^22. Notably, the isolation of individual
layer fragments within 1 presents an opportunity to eliminate the anti-
ferromagnetic interlayer interactions and simultaneously confine the
magnetic domains to the nanoscale.
To compare the magnetic properties of the framework-confined clus-
ters to those of the bulk solids, d.c. magnetic susceptibility data were
collected on microcrystalline powder samples of 1 (NiCl 2 ) 15 , 1 (NiBr 2 ) 12 ,
1 (FeCl 2 ) 19 and 1 (CoCl 2 ) 18 between 300 and 2 K under applied fields of
0.01, 0.1 and 1 T. For each framework, the per-metal susceptibilities
measured at room temperature are slightly lower than those observed
in the corresponding bulk material, with the exception of 1 (NiBr 2 ) 12 ,
which exhibits a large temperature-independent paramagnetic con-
tribution (Fig. 4a–d). In the case of 1 (FeCl 2 ) 19 and 1 (CoCl 2 ) 18 , it is likely
that the tetrahedral metal sites contribute a lower magnetic moment
than do the octahedral sites, thereby suppressing the per-metal sus-
ceptibility. Analogous to the bulk metal halides, χMT increases for all
cluster-containing materials upon cooling below 300 K, indicative of
ferromagnetic coupling of individual spins within each sheet fragment
to form a total spin S. Notably, χMT continues to increase well below the
Néel temperature for each corresponding bulk material, consistent with
the behaviour expected for isolated monolayers. A steep decrease in
χMT is eventually observed below 10 K for all confined sheets, which we
attribute primarily to Zeeman splitting of the high-spin ground state,
and, in the case of 1 (FeCl 2 ) 19 , to magnetic blocking (as discussed below).
To characterize further the static magnetic behaviour of the frame-
work-confined sheets, we collected zero-field-cooled (ZFC) and field-
cooled (FC) magnetization data for temperatures ranging from 2 to
300 K. Whereas these data are completely superimposable for the
cobalt(ii) and nickel(ii) materials and are indicative of simple para-
magnetism at low temperatures, a divergence is observed for 1 (FeCl 2 ) 19
at around 3 K (Fig. 4e), suggesting the immobilization of the total spin050 100 150 200 250 300024681012T (K) T (K)T (K) T (K) T (K)H (T)FMT per Fe (emu K mol–1)Zr 6 O 4 (OH) 4 (bpydc) 6 (FeCl 2 ) 19
Bulk FeCl 20.01 T
2468 10 12 14 16 18 2000.20.40.60.81.0 ZFC
FCM(PB)M(PB)0.01 T
–1.0 –0.5 00 .5 1.0–30–20–1001020302 K
9 mT s–1050 100 150 200 250 300012345FMT per Ni (emu K mol–1)FMT per Ni (emu K mol–1)FMT per Co (emu K mol–1)Zr 6 O 4 (OH) 4 (bpydc) 6 (NiBr 2 ) 12
Bulk NiBr 21 T
050 100 150 200 250 3000123456780.01 TZr 6 O 4 (OH) 4 (bpydc) 6 (NiCl 2 ) 15
Bulk NiCl 2ab050 100 150 200 250 3000246810Zr 6 O 4 (OH) 4 (bpydc) 6 (CoCl 2 ) 18
Bulk CoCl 20.01 Tcde fFig. 4 | Magnetic data. a–d, d.c. magnetic susceptibility data collected for
1 (NiBr 2 ) 12 (a, red symbols), 1 (NiCl 2 ) 15 (b, green symbols), 1 (CoCl 2 ) 18 (c, purple
sy mbols), 1 (FeCl 2 ) 19 (d, orange symbols), and their corresponding bulk metal(ii)
halides (grey symbols) under a 1 T or 0.01 T applied field. emu, electromagnetic
unit. e, Zero-field-cooled (filled orange circles) and field-cooled (empty orange
circles) magnetization per mole of 1 (FeCl 2 ) 19 (M) versus temperature (T) data
taken under a 0.01 T applied field for 1 (FeCl 2 ) 19. f, Magnetization per mole of
1 (FeCl 2 ) 19 (M) versus applied d.c. magnetic field (H) data for 1 (FeCl 2 ) 19 (empty
orange circles) collected at 2 K using a sweep rate of 9 mT s−1. Solid lines are
guides for the eye. μB, Bohr magneton.