This is the accepted version of the article: Winter, L.E. , et al. Spin

density wave and superconducting properties of nanoparticle organic

conductor assemblies in Physical Review B (Ed. American Physical

Society), vol. 91, issue 3 (Jan. 2015), art. 35437.

Available at: DOI 10.1103/PhysRevB.91.035437

This version is published under a “All rights reserved” license.

Spin density wave and superconducting properties of nanoparticle organic conductorassemblies

Laurel E. Winter,1, ∗ Eden Steven,1, † James S. Brooks,1 Shermane Benjamin,1

Ju-Hyun Park,2 Dominique de Caro,3, 4 Christophe Faulmann,3, 4 Lydie Valade,3, 4

Kane Jacob,3, 4 Imane Chtioui,3, 4 Belen Ballesteros,5, 6 and Jordi Fraxedas5, 6

1Department of Physics and National High Magnetic Field Laboratory,Florida State University, Tallahassee, FL 32310, USA

2National High Magnetic Field Laboratory, Florida State University, Tallahassee, FL 32310, USA3CNRS, LCC (Laboratoire de Chimie de Coordination),

205 route de Narbonne, BP 44099, 31077 Toulouse Cedex 4, France4Universite de Toulouse, UPS, INPT, 31077 Toulouse Cedex 4, France

5ICN2 - Institut Catala de Nanociencia i Nanotecnologia,Campus UAB, 08193 Bellaterra (Barcelona), Spain

6CSIC - Consejo Superior de Investigaciones Cientificas,ICN2 Building, 08193 Bellaterra (Barcelona), Spain

(Dated: November 3, 2014)

The magnetic susceptibilities of nanoparticle assemblies of two Bechgaard salts, (TMTSF)2PF6

and (TMTSF)2ClO4, have been studied vs. temperature and magnetic field. In the bulk these ma-terials exhibit a spin density wave formation (TSDW = 12 K) and superconductivity (Tc = 1.2 K),respectively. We show from inductive (susceptibility) measurements that the nanoparticle assembliesexhibit ground state phase transitions similar to those of randomly oriented polycrystalline samplesof the parent materials. Resistivity and diamagnetic shielding measurements yield additional infor-mation on the functional nanoparticle structure in terms of stoichiometric and non-stoichiometriccomposition.

I. INTRODUCTION

Restricted geometries have been realized in many ma-terials such as semiconductors and metals, and often thebulk properties are modified when the size of a struc-ture approaches the characteristic extent of the groundstate order parameter. For instance, in semiconductordevices1 or type I elemental metal superconductors likealuminum2,3 the evolution from an insulator to a metalor superconductor is observed with increasing electrondensity or film thickness, and in nanoparticles of tin4 orlead5,6 a modification of the superconducting propertiesis observed for sizes smaller than the coherence lengths.The single molecule magnet material Mn12-acetate hasalso been produced as a thin film7. Although the filmretains aspects of the more dramatic magnetic hystere-sis effects seen in the bulk crystalline materials, modi-fications such as an additional magnetic phase also ap-pear. However, progress in the study of thin film andnanoparticle geometries in the important class of organicsuperconductors, which are predominantly charge trans-fer organic salts with relatively large donor and acceptororganic molecules, has remained elusive. This is due tothe reluctance of these materials to naturally form epi-taxial layers on substrates – rather, the default growthmorphology, for instance via electrocrystallization, is of-ten the nucleation of small crystallites at arbitrary ori-entations. Some progress has been made by evapora-tion/sublimation methods that produce small patches ofordered charge transfer complexes on metallic substratesthat can then be measured by scanning probe methods8.Such sample geometries and measurements present con-

siderable challenges, and are not conducive to other char-acterization methods that can probe the ground statethermodynamics of a nanostructure.

Recently dispersed nanoparticles of donor-acceptor(DA) organic conductors have been synthesized by eitheroxidation or electrocrystallization in the presence of sta-bilizing agents acting as growth inhibitors9. Additionalrefinements of the synthesis conditions have produced anumber of DA organic superconductors where strong cor-relations between the spectroscopic and crystallographicproperties of the bulk materials and the nanoparticleshave been established10. In light of these advances,the main purpose of the present work is to identify thelow temperature thermodynamic ground states of thenanoparticle species. An example of the nanocrystallineand nanoparticle structure of (TMTSF)2ClO4 is shown inFig. 1. High resolution transmission electron microscope(HR-TEM) studies10 show that single crystal constituentnanoparticles of approximately 3 to 5 nm in size (see Sup-plementary Figure 111) make up nanoparticles that TEMimages show are on average about 10 times larger (e.g.average cluster size of 34 nm). These nanoparticles forma fine powder (assembly) with a random crystallographicorientation (see Supplementary Figure 211) that is thenused for the characterization of the physical properties.

We focus on the low temperature and high magneticfield properties of two new nanoparticle systems in theclass of the Bechgaard salts12. The first is (TMTSF)2PF6

(also, NP-PF6) which undergoes a spin density wave(SDW) and metal-insulator transition at TSDW = 12K, and the second is (TMTSF)2ClO4 (also, NP-ClO4)which, due to the low symmetry of the tetrahedral an-

2

70605040302010

0

ClO

4 C

ou

nt

(#)

605040302010Size (nm)

Average = 34 nm FWHM = 12 nm

NP-ClO4

Gaussian Fit

(a) (b) (c)

FIG. 1. (TMTSF)2ClO4 nanoparticles. (a) Low resolution TEM image of nanoparticles dispersed on a substrate. (b) HR-TEM image of nanoparticle structure: dashed envelope, approximate extent of the overall nanoparticle; circles, positions ofconstituent nanoparticles of 3 to 5 nm in size. (c) Size distribution of nanoparticles in (a). Nanoparticles of (TMTSF)2PF6

exhibit similar morphologies and size distributions

ion ClO4, first undergoes an anion ordering transition atTAO = 24 K, followed by a superconducting transitionat Tc = 1.2 K13. The main goal of the present work isto determine if the nanoparticle morphologies of thesematerials exhibit the same ground states as the bulk sin-gle crystalline materials. We will show that the groundstate properties of the nanoparticles compare favorablywith their bulk counterparts when the highly anisotropicnature14of the Bechgaard salt properties are taken intoaccount. (Bulk crystals form in needle-like morphologieswhere the most conducting axis is the a-axis orientedalong the needle direction, the a-b plane constitutes aconducting layer, and the c-axis is the least conductingdirection, normal to the layers.) However, an unambigu-ous demonstration of a possible alteration of the bulkground state properties when the nanoparticle size is lessthan the ground state coherence lengths will require fur-ther studies.

II. RESULTS AND DISCUSSION

For clarity in terminology in what follows, we notethat the objects on the order of 30 to 60 nm as seenin the lower resolution TEM image (Fig. 1a) are referredto as nanoparticles. These nanoparticles are made up ofsmaller entities, referred to as the constituent nanopar-ticles, with size 3 to 5 nm as seen from the HR-TEMimage (Fig. 1b). For susceptibility and inductive studies,the samples measured were encapsulated powders or as-semblies (see Supplementary Figure 211) comprised ofnanoparticles. For electrical transport measurements,nanoparticle assemblies were studied in a four-terminalconfiguration. In the following sections measurements onthe nanoparticle assemblies are presented in parallel withcomparisons with bulk single crystal results. Since theBechgaard salts are highly anisotropic in terms of theirelectronic structure and magnetic field effects12, a com-parison of the electronic properties and susceptibility of a

randomly oriented assembly of nanoparticles (and there-fore the constituent nanoparticles) with previous bulksingle crystal measurements requires an average of thea, b and c-axis properties. For this purpose, we assumethat the electrical current path or applied magnetic fielddirection lies with equal probability along any of the threeprinciple axes of each nanoparticle. This leads to a sim-ple 1/3 contribution of the property from each direction,averaged to allow a comparison with the correspondingnanoparticle assembly data.

A. Spin density wave transition in (TMTSF)2PF6

nanoparticles

We first discuss the electronic transport of the SDWtransition in the (TMTSF)2PF6 nanoparticle assembly.In powders, the nanoparticles tend to form a dense, inter-connected assembly unlike that observed in crushed sin-gle crystal powders where the crystals are well dispersedand separated (Supplementary Figure 211). This indi-cates a strong interparticle interaction between nanopar-ticles (Supplementary Video 111). The source of this un-usual mutual attraction is not presently understood, butbased on the sample preparation we do not believe thereare residual solvents or stabilizing agents on the surfaceof the nanoparticles. We find the assembly is electricallyconducting, and the temperature dependent resistance ofthe nanoparticle assembly was studied using an adhesivestamp electrode method15 and a lock-in amplifier (SRS830) in a 4-probe configuration down to cryogenic tem-peratures. As observed in Fig. 2, the SDW transition(accompanied by a metal-insulator transition) occurs atTSDW = 12 K, similar to that of the bulk single crystals.We note that in bulk single crystals, the temperature de-pendent resistance above TSDW is metallic, i.e. dR/dT> 0, but the nanoparticle assembly shows very little de-pendence of the resistance above TSDW. However, be-low TSDW the activated component of the resistance rises

3

1

2

3

4

56

10

2

3

4R

No

rm

20181614121086

Temperature (K)

3

2

1

0

Ln

(RN

orm

)0.200.160.120.08

1/Temp (1/K)

Ea = 2 meV

Ea = 0.75 meV

RSC

RNP

RModel

RModel = RS

RSC

RP

FIG. 2. Electrical transport in the (TMTSF)2PF6 nanoparti-cle (NP) assembly compared with single crystal (SC) (c-axis)measurements. Upper inset: Comparison of Arrhenius acti-vation plots for the nanoparticle assembly and single crys-tals showing the lower activation energy for the nanoparticleassembly. Lower insert sketch: Series (RS) & parallel (RP)model for the nanoparticle assembly resistance.

quickly and is clearly observable above the background.Notably, the activation energy (E a) of the NP-PF6 as-

sembly is much lower compared to the case for a bulksingle crystal (Fig. 2, upper inset). In the simplest case,this lower E a can be assigned to the presence of non-stoichiometric contributions to the overall transport sig-nal. If the electrical transport in the NP-PF6 assembly(RNP) is purely stoichiometric, then the activation energyshould be close to that seen in a single crystal. However,if there are non-stoichiometric resistances in series (RS)and parallel (RP) to the stoichiometric resistance (RSC),then the temperature dependence of the transport willbe modified. We have used a simple series-parallel model(Fig. 2, lower inset) to describe the apparent suppres-sion of E a. Our model assumes that each nanoparticlein a compacted assembly is separated from the othersby a thin non-stoichiometric coating. Hence the coatingwill give rise to both series (RS) and parallel (RP) resis-tances with respect to the stoichiometric resistance RSC.We further assume the non-stoichiometric resistances areconstant with the temperature, and the stoichiometricresistance RSC follows the temperature dependence of abulk single crystal (which is small above TSDW, and risesrapidly below TSDW). For the analysis, we have normal-ized the resistances to the data at 20 K and includeda weighting ratio, η, which represents the portion of thetotal resistance that is due to parallel elements in the cir-cuit. The model for the temperature dependence of thetotal resistance of the assembly is RModel = (1 – η)∗RS

+ η∗[RSC∗RP/(RSC + RP)]. It provides an excellent fitto the experimental data (Fig. 2 upper inset), where wefind η = 0.2, RS= 1, and RP= 19.1. This means thatat most the stoichiometric contribution to the overall re-

sistance of the NP-PF6 assembly is 20% of RSC (in thelimit that RP → ∞). The rest is due to contributionsthat could originate from some combination of contactresistances between individual nanoparticles and/or non-stoichiometric material.

The main results of this analysis are that forthe self-adhering powdered nanoparticle assembly of(TMTSF)2PF6, electrical conductivity occurs via a com-bination of transport through stoichiometric and non-stoichiometric material in effectively parallel and seriespathways. This implies that the effective activation en-ergy in the SDW state (of the stoichiometric material)will be reduced by the presence of conduction throughnon-stoichiometric parallel pathways. It also indicatesthe contribution of the stoichiometric resistance becomesmost evident below the SDW transition where its resis-tance rises by orders of magnitude with respect to thenon-stoichiometric material.

Next, we examine the magnetic susceptibility χ of theSDW transition in (TMTSF)2PF6 nanoparticles shownin Fig 3a. To study the magnetic properties, NP-PF6

powders of mass 63.8 mg were placed in a polycarbonatecapsule and measured in a commercial superconductingsusceptometer (MPMS-Quantum Design) vs. tempera-ture and magnetic field. The background signal fromthe polycarbonate capsule was measured independently.The spin susceptibility is obtained by subtracting themolecular diamagnetism16 χmol = 3.5×10−4 emu/mol of(TMTSF)2PF6 from the total susceptibility. For compar-ison (Fig. 3b), the most complete previous study of theanisotropic susceptibility for a bulk single crystal was inthe isostructural material (TMTSF)2AsF6 (with nearlyidentical properties) reported by Mortensen et al.16 forthe a, b, and c directions with respect to an applied fieldof H = 0.3 T. It was found that below TSDW = 12 K,χb dropped rapidly, indicating antiferromagnetic orderdevelops along the b-axis. In contrast, χa and χc remainparamagnetic below TSDW. It was further found (Fig. 3c)that a magnetic field applied along the b-axis induces aspin-flop at Hsf ≈ 0.45 T, where χb then approaches χa

and χc.

From Figs. 2 and 3 it is clear that the (TMTSF)2PF6

nanoparticle assembly exhibits qualitatively the sameSDW ground state of a bulk single crystal, including themetal insulator transition and the spin-flop behavior inmagnetic field compared with a simple averaging argu-ment based on bulk single crystal results. There is no ap-parent shift in the SDW transition temperature betweenthe nano and bulk materials, indicating that the effec-tive coherence length (ξ) of the SDW order parameter iscomparable to or smaller than the effective nanoparticlesize (estimated from a simple BCS type argument to beξ = hνF/2πkBTSDW ≈ 10 nm). Qualitatively, in Fig. 3there may be some functional differences in the field andtemperature dependence of the nanoparticle susceptibil-ity and in the higher spin-flop saturation (1.5 T vs. 0.45T), but at present we cannot make any significant assign-ment of the behavior to size effects.

4

1.95

1.90

1.85

1.80

1.75

1.70

1.65

1.60

cs

pin

(10

-4 e

mu

/mo

l)

2016128Temperature (K)

3.0 T 1.5 T 1.0 T 0.7 T 0.5 T 0.3 T 0.1 T

(a)

1.2

0.8

0.4

0.0cs

pin

(10

-4 e

mu

/mo

l)

252015105Temperature (K)

1.5

1.0

0.5

0.0

cS

C,s

pin

(10

-4 e

mu

/mo

l)

3.02.52.01.51.00.5Field (T)

0.3

0.2

0.1

0.0

Dc

NP

,sp

in (

10

-4 e

mu

/mo

l)

DcNP

ca

cb

cc

cave

ca

cb

cc

(b)

(c)

FIG. 3. Magnetic properties of (TMTSF)2PF6. (a) Temperature and magnetic field-dependent susceptibility of a randomlyoriented (TMTSF)2PF6 nanoparticle assembly. The curves have been shifted to coincide with the 3 T data at 13 K to accountfor a systematic field-dependent background. (b) Single crystal results for (TMTSF)2AsF6 from Ref. 16 for a, b and c-axisorientations at 0.3 T. The solid line is a simple average of the susceptibility for all three directions (compare with 0.3 T datain Fig. 3a). (c) Spin-flop behavior at 5 K (left axis: ∆χNP = χNP(5 K, B) – χNP (5 K, 0.1 T)) compared with χb results forsingle crystal (TMTSF)2AsF6 from Ref. 16 (right axis: χa, χb, χc), where we note that the field dependences of χa and χc arenegligible on this scale.

B. Superconductivity and critical magnetic fieldsin (TMTSF)2ClO4 nanoparticles

We next turn our attention to the (TMTSF)2ClO4

nanoparticle system. Due to the low temperature of thesuperconducting transition temperature (Tc = 1.2 K) in(TMTSF)2ClO4, which was below the 1.5 K limit of theMPMS, we used a high sensitivity inductive method ina dilution refrigerator in a 16 T superconducting magnetto study the diamagnetic signal associated with the su-perconducting state. (The superconducting magnet wasswept through zero magnetic field to eliminate remnantfield effects due to trapped flux in the solenoid.) The in-ductive method involves a tunnel diode oscillator (TDO)self-resonating LC circuit (L = inductor, C = capacitor)operating with a frequency f and driven by a tunnel diodebiased in the negative resistance region17,18. Changes inthe resonant frequency are directly proportional to anymechanism that changes the effective inductance of a coilin which the sample is placed. In the case of a supercon-ducting transition, this occurs due to the diamagnetic re-sponse that effectively reduces the volume of the coil dueto flux excluded from the sample, leading to an increasein the resonant frequency. For this work, an assembly ofNP-ClO4 of mass 7.385 mg (corresponding to a sample

volume of 3.1 mm3) was placed inside a non-magneticgelatin capsule with an approximate capsule volume of18.08 mm3, which was then placed inside of a 40 turncoil (L) with an approximate length and volume of 4.6mm and 23.68 mm3, respectively. The rms field (h) dueto the measurement coil is estimated to be 0.04 Oe, wellbelow the lower (Meissner) critical field Hc1 for T < Tc.For comparison with the superconducting diamagneticsignal χd obtained by conventional methods, we presentour results as the temperature dependent change in theresonant frequencies f (Tc) – f (T) ≡ -δf(T) ∼ χd.

We note that in preparation for the low tempera-ture measurements, the nanoparticle assembly was cooledvery slowly (< 0.1 K/min) through the anion orderingtransition TAO = 24 K in order to avoid quenching19,20.The absolute change in the TDO frequency as a func-tion of temperature for the NP-ClO4 assembly, with noexternal applied magnetic field, is shown in Fig. 4. Adecrease in -δf is observed starting at approximately 1.2K due to the diamagnetic onset of the superconductingtransition that approaches a minimum around 0.03 K,the lowest temperature available for this study. The fre-quency shift data in Fig. 4 is characteristic of the responseof a superconductor below Tc for which the decrease inthe susceptibility χ(T), penetration depth λ(T) or theinductive signal -δf are equivalent indications of the ex-

5

1.00.80.60.40.20.0T/Tc

-600

-500

-400

-300

-200

-100

0-d

f (k

Hz)

1.0

0.8

0.6

0.4

0.2

0.0

Df,

Dc (

arb

. u

nit

s)

b''-(ET)2SF5CH2CF2SO3

(Prozorov et al. Df )

NP - (TMTSF)2ClO4

(This work -df ) SC - (TMTSF)2ClO4

(Gubser et al. Dc)

FIG. 4. A comparison between the change in frequency (-δf) for (TMTSF)2ClO4 nanoparticles (Tc ∼ 1.2 K), the nor-malized average relative susceptibility (∆χ) of a bundle of(TMTSF)2ClO4 single crystals (Tc ∼ 1.1 K) by Gubser etal.21, and the averaged penetration depth change (∆f) of β”-(ET)2SF5CH2CF2SO3 (Tc ∼ 5.5 K)22 as a function of reducedtemperature T/Tc.

cluded sample volume due to diamagnetic shielding cur-rents. The results for -δf are in qualitative agreementwith previous susceptibility measurements performed byGubser et al.21 for a bundle of aligned crystals (Fig. 4).The weak, sigmoidal nature of the temperature depen-dence of -δf in Fig. 4 does not correspond to the muchstronger temperature dependence of a BCS-type modelλ(T) = λ(0)∗(1 – (T/Tc)γ)−1/2 where the exponent (typ-ically between 1.5 and 4) depends on the nature of thesuperconducting order parameter. This is made evidentby comparison with another organic superconductor, β”-(ET)2SF5CH2CF2SO3

22 where the rise in λ(T) (∆f inthe Fig. 4) near Tc (∼ 5.5 K) is rapid. Due to the lay-ered, anisotropic structure of organic superconductors,the orientation of the sample in the inductor coil plays asignificant role. Defining the layers to lie in the a-b plane,for the oscillating magnetic field h perpendicular to thec-axis, λ⊥(T) can be much larger and less temperaturedependent than λ‖(T) for h parallel to the c-axis. A sim-ple averaging to find the contributions for the a, b, and cdirections in β”-(ET)2SF5H2CF2SO3 merely reduces therange, but not the temperature dependence of ∆f .

The shielding fraction of the NP-ClO4 assembly maybe estimated from the experimental parameters. Tofacilitate the discussion, a model of the NP-ClO4, tobe discussed below, is presented in Fig. 5. Followingon the analysis for NP-PF6, we assume that there isa stoichiometric nanoparticle core surrounded by non-stoichiometric material; their sum makes up the totalmass (and volume) of the nanoparticle.

The total nanoparticle assembly volume, based on themass of the assembly, is V assem. = 3.1 mm3, and thevolume of the inductor coil is V coil = 23.68 mm3. TheTDO resonant frequency is f0 = AV coil

−1/2 (where A

is a constant dependent on physical constants and geo-metrical factors). If a superconducting sample is placedin the coil, then the altered frequency due to an effec-tive shielded volume V s due to diamagnetic currents willbe fs = A(V coil – V s)

−1/2. Using Figs. 4 and 5a wemay estimate V s at 0.03 K from the simple relationship(V coil – V s)/V coil = (f0/fs)

2, where fs – f0 = -δf(0.03K). We find that V s = 0.075∗V coil = 1.79 mm3, whichimplies that at 0.03 K, about 58% of nanoparticle mate-rial is superconducting and shields the field. Therefore42% of the nanoparticle assembly volume is either pen-etrated by field or is non-stoichiometric material. Thereare two different scenarios of how the nanoparticle assem-bly can act: a) the nanoparticles are either electricallyconnected or Josephson coupled, acting as a monolithicsolid body where the shielding currents appear at the pe-riphery (Fig. 5a); b) as an assembly of weakly interactingnanoparticles (Fig. 5b) where the field easily penetrates,and the diamagnetic shielding arises from the sum of in-dividual superconducting nanoparticles23.

Scenario (a): If we assume that there is no non-stoichiometric material (i.e. r stoich. = rassem.), we mayestimate how far the magnetic field penetrates into thenanoparticle assembly volume at low temperature, i.e.the effective change in penetration depth ∆λeff = λ(Tc)– λ(T = 0) by assuming λ(Tc) ∼ rassem. for a sphericalsample shape of radius rassem.(≡ [ 3

4πV assem.]1/3), and

shielded sample radius r s(≡ [ 34πV s]

1/3). Using the rela-

tion 43π(rassem. – ∆λeff)3 = V s we find ∆λeff ∼ 150 µm.

If there is non-stoichiometric material, then ∆λeff will besmaller. We note that ∆λeff ∼ 150 µm is in the rangeof values for the penetration depth reported for other or-ganic superconductors, (λ⊥ ∼ 800 µm and λ‖ ∼ 1 µm

for anisotropic β”-(ET)2SF5CH2CF2SO322), and that it

is significantly larger than the range of coherence lengthsin (TMTSF)2ClO4: 80 nm < ξ < 2 nm.

Scenario (b): The nanoparticles act individually. Not-ing that the rms ac field h = 0.04 Oe << Hc1, no vortexstate enters an individual nanoparticle, and since λ (mi-crons) >> ξ ∼ nanoparticle size (nanometers), the pres-ence of the field would result solely in a reduced super-conducting order parameter within the nanoparticle24,and the diamagnetic response would come from the sumof the nanoparticles in the assembly, mimicking a bulksolid body response. Previous work on Pb nanoparticlesindicates that even in a compacted state, the nanoparti-cles act as individual entities5 and give rise to a bulk-likediamagnetic signal. Moreover, there is no suppression inTc until the nanoparticle size is only about 10% of thebulk coherence length.

We next turn to the magnetic field dependence, deter-mined systematically by isothermal magnetic field sweepsfor temperatures between 1.5 K and 33 mK. Shown inFig. 6 are the normalized changes in the TDO frequencyδfnorm, which is normalized with respect to -δf at zerofield for the lowest temperature sweep (T = 33 mK,see Supplementary Figure 311), for increasing magneticfield for two representative temperatures below Tc ∼

6

(a) (b)

FIG. 5. Two scenarios for superconductivity in the nanoparticle assembly of (TMTSF)2ClO4: (a) Model of the nanoparticleassembly acting as a monolithic solid body in an inductive coil. (b) Visualization of magnetic field penetration in a weaklyinteracting assembly of superconducting nanoparticles.

1.2 K, where a quadratic field-dependent instrumentalbackground present at all temperatures has been sub-tracted from the data. For T < Tc, δfnorm increasesrapidly with field. For comparison, we show the nor-malized field dependent changes in susceptibility data,χave = (2χ‖ + χ⊥)/3, averaged to estimate a randomorientation from Ref. 21 for two similar temperatures(expanded view in lower inset). We find again a goodqualitative similarity between the magnetic field depen-dent susceptibility of the nanoparticle assembly and bulksingle crystal data. Quantitatively, the single crystal sus-ceptibility data shows a slightly faster suppression of thediamagnetic shielding with field.

For higher fields δfnorm asymptotically approachesunity at different fields for different temperatures, whichwe interpret as the onset of the upper critical field H∗c2.Given the uncertainties in estimating asymptotic inter-cepts, we have determined the H∗c2 onset vs. temperaturefor our complete set of field sweep data from - 2 T to 2T (see Supplementary Figure 311). The results are plot-ted in Fig. 7, along with previous determinations of Hc2

by Murata et al.25 for the three crystal directions, andtheir averaged (solid line) contribution. Clearly, H∗c2 ofthe NP-ClO4 assembly follows closely the expectationsfor Hc2 of the bulk crystal. We note that up to 16 T at33 mK, there was no evidence for orbital features asso-ciated with the field induced SDW phases, nor the rapidoscillations26. Since these effects are periodic in inversefield for B‖c, they may average out due to the randomnanoparticle orientations, although size effects may alsoplay a role.

Summarizing, to the best of our knowledge the(TMTSF)2PF6 and (TMTSF)2ClO4 nanoparticle sys-tems are the first assemblies of organic charge transfermetal nanoparticles to exhibit, respectively, a SDW for-mation and superconductivity. Our characterizations of

1.0

0.8

0.6

0.4

0.2

0.0

df n

orm

(a

rb.

un

its

)

1.41.21.00.80.60.40.20.0

Field (T)

1.0

0.8

0.6

0.4

0.2

0.0df n

orm

0.100.00 Field (T)

Relative SusceptibilityGubser et al.

650 mK 35 mK

TDO dfnorm

630 mK 33 mK

H*c2 (630 mK) H*c2 (33 mK)

FIG. 6. A comparison of the magnetic field dependence of thenormalized changes in the TDO frequency δfnorm (normalizedwith respect to -δf at zero field for the lowest temperaturesweep, 33 mK) compared with the normalized average relativesusceptibility χave = (2χ‖ + χ⊥)/3 obtained by Gubser et

al.21 (inset: expanded low field view). Estimates of the uppercritical field H∗c2 (arrows) are obtained from the field where thesignal asymptotically merges with the constant backgroundat higher field. The randomly aligned nanoparticles exhibita field dependence that is in qualitative agreement with theaverage response of aligned signal crystals.

their ground state properties indicate a close correspon-dence to the bulk single crystal parent materials, andalso give new information about the functional descrip-tion of the nanoparticle structures beyond spectroscopyand electron microscopy. Specifically, the temperaturedependent resistance measurements on (TMTSF)2PF6

indicate the presence of non-stoichiometric material that

7

2.0

1.5

1.0

0.5

0.0

m0H

c2

(T

esla

)

1.21.00.80.60.40.20.0Temperature (K)

NP Assembly:

SC (Murata et al.): Hc2 a-axis

Hc2 c-axis

Hc2 b-axis

Hc2 average

H*c2 Average

H*c2 Fit

FIG. 7. A comparison between the upper critical field Hc2

reported on a single crystal (SC) of (TMTSF)2ClO4 for eachaxis (broken lines) and their average (solid line)25 and our de-termination of H∗c2 for the (TMTSF)2ClO4 nanoparticle (NP)assembly (solid circles). The uncertainty-bars represent therange of H∗c2 values based on determinations from all fieldsweep data at each temperature. The gray-scale dashed lineis a polynomial guide-to-the-eye fit, and indicates that H∗c2for the (TMTSF)2ClO4 nanoparticle assembly is similar tothe of Hc2 for single crystals averaged over the three crystaldirections.

we have modeled as a coating that in a nanoparticle as-sembly acts as both a series and parallel conducting pathwith the stoichiometric material. It is likely that the(TMTSF)2ClO4 functional structure is similar. Hence inthe superconducting state, there are normal state regionsbetween any two superconducting centers, and this willreduce the Josephson coupling in the assembly. The rel-atively large effective penetration depth ∆λeff ∼ 150 µmmay be a manifestation of this since the entire mass ofthe assembly is taken into account when estimating thenanoparticle effective volume, but only part of it actu-ally goes superconducting and contributes to the changein diamagnetic shielding below Tc.

Based on the observations above, it is probable thatthe NP-ClO4 act as isolated superconducting nano-sizedregions below Tc. In this respect, previous work on Pbnanoparticles is instructive. Bulk Pb is a type I super-conductor with a critical temperature of Tc = 7.2 K, acoherence length ξ(0) ∼ 90 nm, and a penetration depthλ(0) ∼ 40 nm. Li et al.5 investigated both the susceptibil-ity and resistivity of nanoparticle powders of Pb (NP-Pb)for a variety of nanoparticle sizes between 86 and 2 nmin capsules with a packing mass density of ∼ 7% of thebulk (less than half of what we determined for our ClO4

capsule). They found the Tc of NP-Pb did not decrease

until the particle size was less than 10% of the bulk coher-ence length. Two important implications of this work arethat the NP-Pb act individually and not as a connectedsystem, and that the nanoparticles must be significantlysmaller in size that the coherence length before size ef-fects cause a significant change in the superconductingorder parameter.

III. CONCLUSIONS

The above experiments on nanoparticle Bechgaardsalts, and the implications provided by comparisons withprevious single crystal studies, are subject to a numberof uncertainties, namely the distribution of nanoparticlesizes, the random orientation of the nanoparticles andexperimental parameters including the relatively highfrequency used (15 MHz – the quasiparticle skin depthabove Tc is δ = 15 µm in the TDO probe). In spite ofthese complications, we find a remarkably strong correla-tion between the nano and bulk ground state properties,and the nanoparticle materials do indeed reflect the in-trinsic materials with nearly unperturbed transition tem-peratures. Less clear is evidence for the nano-size effectson the temperature and magnetic field dependences of theground states. Based on previous results on Pb nanopar-ticles, it may be necessary to significantly reduce the sizeof the (TMTSF)2ClO4 nanoparticles by an order of mag-nitude from 30 nm to 3 nm. Indeed, this is the size of theconstituent nanoparticles that comprise a single nanopar-ticle (Fig. 1b). Future work will be needed to specificallyaddress the effects of size on these constituent nanopar-ticles with, for instance, low temperature scanning probespectroscopy.

ACKNOWLEDGMENTS

This work was supported by NSF-DMR 1005293 and1309146, and the NHMFL is supported by NSF Co-operative Agreement No. DMR-1157490, the State ofFlorida and the US Department of Energy. Funding forthe dilution refrigerator used for the 16 T experimentcame from DOE NNSA SSAA DE-NA0001979. Col-laboration within the European Associated LaboratoryTrans-Pyreneen (LEA) (program: “de la Molecule auxMateriaux”) is appreciated. Work done at the Labora-toire de Chimie de Coordination (LCC) was supportedby the French Centre National de la Recherche Scien-tifique (CNRS). I. C. thanks the French Ministere del’Enseignement Superieur et de la Recherche (MESR) fora Ph.D. grant.

∗ Corresponding Author: laurelewinter@gmail.com† Corresponding Author: esteven@magnet.fsu.edu

1 E. Abrahams, S. V. Kravchenko, and M. P. Sarachik, Re-views of Modern Physics 73, 251 (2001).

8

2 D. B. Haviland, Y. Liu, and A. M. Goldman, PhysicalReview Letters 62, 2180 (1989).

3 K. Ohshima, T. Kuroishi, and T. Fujita, J. Phys. Soc.Jpn. 41, 1234 (1976).

4 S. Bose, A. M. Garcia-Garcia, M. M. Ugeda, J. D. Urbina,C. H. Michaelis, I. Brihuega, and K. Kern, Nat Mater 9,550 (2010).

5 W.-H. Li, C. C. Yang, F. C. Tsao, and K. C. Lee, Phys.Rev. B 68, 184507 (2003).

6 S. Reich, G. Leitus, R. Popovitz-Biro, and M. Schechter,Phys. Rev. Lett. 91, 147001 (2003).

7 J. Means, V. Meenakshi, R. V. A. Srivastava, W. Teizer,A. A. Kolomenskii, H. A. Schuessler, H. Zhao, and K. R.Dunbar, J. Magn. Magn. Mater. 284, 215 (2004).

8 K. Clark, A. Hassanien, S. Khan, K. F. Braun, and S. W.Hla, Nat. Nano 5, 261 (2010).

9 D. de Caro, L. Valade, C. Faulmann, K. Jacob,D. Van Dorsselaer, I. Chtioui, L. Salmon, A. Sabbar,S. El Hajjaji, E. Perez, S. Franceschi, and J. Fraxedas,New J. Chem. 37, 3331 (2013).

10 D. de Caro, C. Faulmann, L. Valade, K. Jacob, I. Chtioui,S. Foulal, P. de Caro, M. Bergez-Lacoste, J. Fraxedas,B. Ballesteros, J. S. Brooks, E. Steven, and L. E. Win-ter, European Journal of Inorganic Chemistry 2014, 4010(2014).

11 See Supplementary Material at [ ] for additional nanopar-ticle photos, videos, and data.

12 T. Ishiguro, K. Yamaji, and G. Saito, Organic Supercon-ductors, 2nd ed. (Berlin, Heidelberg, New York:Springer-Verlag, 1998).

13 P. Alemany, J.-P. Pouget, and E. Canadell, Phys. Rev. B89, 155124 (2014).

14 The anisotropic properties of (TMTSF)2X are as follows12

for the a, b and c-axis respectively. (TMTSF)2ClO4 - thelattice constants are 7.266 A, 7.678 A, and 13.275 A; thetransfer integrals are 250 meV, 25 meV, and 1.5 meV; theapproximate coherence lengths at zero temperature are 800

A, 300 A, and 20 A; the critical fields Hc1(50 mK) are 0.2Oe, 1 Oe, and 10 Oe; and the upper critical fields Hc2(0K) are 28 kOe, 21 kOe, and 1.6kOe. (TMTSF)2PF6 - thelattice constants are 7.297 A, 7.711 A, and 13.522 A, andthe transfer integrals are 0.51 meV, 0.52 meV and 1.73meV.

15 E. Steven, E. Jobiliong, P. M. Eugenio, and J. S. Brooks,Review of Scientific Instruments 83, 046106 (2012).

16 K. Mortensen, Y. Tomkiewicz, and K. Bechgaard, Phys.Rev. B 25, 3319 (1982).

17 W. A. Coniglio, L. E. Winter, K. Cho, C. C. Agosta,B. Fravel, and L. K. Montgomery, Phys. Rev. B 83, 224507(2011).

18 G. J. Athas, J. S. Brooks, S. J. Klepper, S. Uji, andM. Tokumoto, Review of Scientific Instruments 64, 3248(1993).

19 J.-P. Savy, D. de Caro, L. Valade, J.-C. Coiffic, E. S. Choi,J. S. Brooks, and J. Fraxedas, Synthetic Metals 160, 855(2010).

20 J. P. Pouget, G. Shirane, K. Bechgaard, and J. M. Fabre,Phys. Rev. B 27, 5203 (1983).

21 D. U. Gubser, W. W. Fuller, T. O. Poehler, D. O. Cowan,M. Lee, R. S. Potember, L. Y. Chiang, and A. N. Bloch,Phys. Rev. B 24, 478 (1981).

22 R. Prozorov, R. W. Giannetta, J. Schlueter, A. M. Kini,J. Mohtasham, R. W. Winter, and G. L. Gard, Phys. Rev.B 63, 052506 (2001).

23 S. Haddad, S. Charfi-Kaddour, and J.-P. Pouget, Journalof Physics: Condensed Matter 23, 464205 (2011).

24 C. Kittel, Introduction to Solid State Physics, edited by7th (New York: Wiley, 1996).

25 K. Murata, M. Tokumoto, H. Anzai, K. Kajimura, andT. Ishiguro, Japanese Journal of Applied Physics 26, 1367(1987).

26 S. Uji, T. Terashima, H. Aoki, J. S. Brooks, M. Tokumoto,S. Takasaki, J. Yamada, and H. Anzai, Phys. Rev. B 53,14399 (1996).