Synthetic pathways to vanadyl phosphates

Solid State Ionics 32/33 ( 1989 ) 57-69 North-Holland. Amsterdam

S Y N T H E T I C PATHWAYS TO VANADYL P H O S P H A T E S Daniel B E L T R A N - P O R T E R , Pedro A M O R 6 S , Rafael IBAIqEZ, Eduardo MARTfNEZ, Aurelio B E L T R A N - P O R T E R UIBCM, Departamento de Quimica Inorgdnica, Universitat de Valencia, C/Doctor Moliner 50, 46100 Burjassot (Valencia), Spain

Armel Le BAIL, Gerard FEREY Laboratoire des Fluorures, U.A. C.N.R.S. 449, Facult~ des Sciences, Universit~ du Maine, Route de Laval, 72017 Le Mans Cedex, France

and Gerard V I L L E N E U V E Laboratoire de Chimie du Solide du C.N.R.S., Universitb de Bordeaux L 351 Cours de la Libbration, 33405 Talence Cedex, France

Received 15 June 1988; accepted for publication 15 September 1988

A general scheme concerning the synthesis and reactivity of oxovanadium phosphates is presented. In particular, the isolation of new materials of composition NH4(VO2)(HPO4), [M(en)2]xVOPO4.(2-2x)H20 (M=Cu or Ni; 0.03~.
1. Introduction In recent years there has been significant interest by solid-state chemists in the V - P - O system. Thus, oxovanadium (V), VOPO4, and derived phases have been widely studied owing to its catalytic relevance [ 1,2]. Otherwise, vanadyl (IV) phosphates offer a large variety of crystal structures suitable to yield low dimensional magnetic interactions [ 3-5 ]. Hence, the theoretical study of magneto-structural correlations becomes a challenging task in this system. Oxovanadium (V) phosphate dihydrate, VOPO4"2H20, is a specially outstanding phase, because it can be considered as a chemical precursor of a great variety o f oxovanadium (V or IV) derivalives [ 5 - 9 ] . Acid-base and redox reactions can lead to very different " v a n a d i u m phosphates". The structure of VOPO4-2H20 is composed of (~-VOPO4)oo layers and water molecules, one H 2 0 group being co-

ordinated to a vanadyl group and the other located in a vacant site [ 10 ]. The layered morphology o f the dihydrate (which makes it a very versatile host for intercalation chemistry [ 11-13 ] ), the variety of ways in which phosphate tetrahedra and vanadyl octahedra can be interlinked, and the accesibility of more than one vanadium oxidation state, are the main factors contributing to such a diverse solid-state chemistry. The present study is intended to offer a unified view o f vanadyl phosphates chemistry. Synthetic pathways are re-examined. New structural data are presented, and the magnetic properties of V ( I V ) containing compounds are analized on the basis of the structural results.

58

D. Behr(zn-Porter et al. I I'anadvl phosphates

V O ( H ~ P O 4 ) , . y H 2 0 ( x = 1, y = 1.5, 2, 3 and 4; x=2, y = 0 ) [6]. Salts of composition AHVPO6 (henceforth, ~-A (VO2)(HPO4); A = N H + , K + , Rb + , Cs + ) have been prepared as described in ref. [ 14].

2. Experimental

2.1. Syntheses The general features concerning the synthetic pathways to vanadyl phosphates have been summarized in table 1. We have previously detailed the phenomenological aspects of the procedures leading to mixed-valence sodium-containing intercalates, Na~(V]VVV_ ,.O ) PO4. ( 2 - x) H20 (0.25~
2.1.1. Preparation of ec-NH4(VO,)(HPOQ An 1.3 M solution of NH4VO2HPO4 was prepared as described in ref. [14]. Then, the solution was evaporated to a smaller volume until incipient cloudiness. A 15 cc. portion of this suspension was introduced in a Teflon bucket, and heated at 200°C for

Table 1 Synthetic pathways to vanadyl phosphates.

VO(HPO4)- 1H20 I / (VO)2P207J

15
BASE

Heterogeneous phase

8
CHEMI$1RY

Totalle-reduction

R*<8,YO(HPO41"4H20

-

R*>27,VO(HP04)"1.5H20 Aqueousmediun,go(H2P04)2 +H3PO4

Modified

+MI2CO3

Ladwig's

Heterogeneousphase

Acetone ~-~[YO(HPO4).O.5H20 ]

procedure [21]

Total le- reduction

medium

1

I AQUEOUS ,SOLUTION

I

.YOP042H20

1

Heterogeneousphase /

[M(en)2lxvoPO4nH20 Partial reduction J [M(en)2]l 2 M: Cu2+, x=O.O~, 0.13,0.20. M=NI2+ , x=0.03,

dan6er's

procedure [ 6]

B-AYO2HPO4

Homogeneousphase

0.20

Partial reduction NBxYOPO4- (2-x) H20

A= NH4*, K+ t

Rb+,cs+.I~

x= 0.02, 0.25, 0.29 0.45, 0.46. x=0.46

-~OUS ~ Homogeneousphase reduction I A=NH4 IZ JTION I [Partial 1:200~ Hydrothermal °(-AVO2HP04]

T=600_oC

'I(,S/

[Nao.46VOP04

P= 2300Bars ltydrothermal

* R is the ratio acetone: water in the reaction medium. ** Aqueous solution containing AHzPO40.770 MAVO~ 0. 134 M and HNO~ 0.745 M.

D. Beltrdn-Porteret al. / Vanadylphosphates 24 h. The apparatus was allowed to cool down to room temperature for 48 h. The resulting yellow crystals were filtered-out and washed with small portions of acetone. (Found: V, 25.8; P, 15.6; N, 7.2; H, 2.7. VPO6NH5 requires V, 25.9; P, 15.7; N, 7.1; H, 2.5%).

2.1.2. Preparation of Nao 46 V O P 0 4 A 0.1974 g portion of Nao.46VOPO4.1.58H20 (0.982 mmol) was suspended in 0.40 cc. of water in a gold bucket, and the apparatus was heated at 600 ° C (2300 Bars) for 24 h. The resulting deep-green crystals were filtered-out, and washed with small portions of cold water and acetone (Found: V, 30.0; P, 17.6; Na, 6.1. Nao.46VPO 5 requires V, 29.5; P, 17.9; Na, 6.1%). 2.1.3. Preparation of [M(en) 2]~[V ~v V ~_ 2,:) 01t904 . ( 2 - 2x)H20 (M= Cu, Ni; en=ethylendiamine; O.03 <~x <~0.20) One gram portions (5.05 mmol) of VOPO4'2H20 were added to 30 cc of 0.11 M solutions of [M(en)2)Iz ( M = C u [17], Ni [18]). After prolonged stirring at room temperature the phosphate was partially dissolved and reduced, forming brown suspensions from which microcrystalline solids were separated by filtration. These were washed with small portions of CC14 (to remove I2), water and acetone, and dried in air. The intercalation-reduction degree depends essentially on the reaction time. Thus, when M = Cu it was obtained: t= 1 h, [Cu(en)z]o.osVOPO4'l.88H~O (Found: V, 25.0; P, 15.0; Cu, 1.6; N, 1.5; C, 1.1; H, 2.2; H20, 16.5. VPO6.88Cuo.osNo.zoCo.2oH4.56 requires V, 24.9; P, 15.1 ; Cu, 1.6; N, 1.4; C, 1.2; H, 2.2; H20, 16.5%); t=2 h, [Cu(en)2]o ~3VOPO4-1.78H20 (Found: V, 23.3; P, 14.2; Cu, 3.9; N, 3.4; C, 2.7; H, 5.5; H20, 14.6. VPO6.78Cuo.13No.52Co.52Hs.64requires V, 23.4; P, 14.2; Cu, 3.8; N, 3.3; C, 2.9; H, 5.6; H20, 14.7%); t = 6 h, [Cu(en)2]o.2oVOPO4' 1.57H20 (Found: V, 22.4; P, 13.4; Cu, 5.7; N, 4.7; C, 4.4; H, 2.9; H20, 12.5. VPO5.57Cuo.2oNo.8oCo.8oH6.34 requires V, 22.4; P, 13.6; Cu, 5.6; N, 4.9; C, 4.2; H, 2.8; H20 , 12.5%). When M-Ni t= 1 h[Ni(en)2]o.o3VOPO4.1.96H20 (Found: V, 25.1; P, 15.1; Ni, 0.9; N, 0.9; C, 0.7; H, 2.1; H 2 0 , 17.3. VPO6.96Nio.o3No.12Co.12H4.4o requires V, 25.1; P, 15.3; Ni, 0.9; N, 0.8; C, 0.7; H, 2.2; H20, 17.4%); t = 6 h [Ni(en)z]o.2oVOPO4'l.65H20 (Found: V,

59

22.3; P, 13.4; Ni, 5.2; N, 4.8; C, 4.2; H, 3.0; H 2 0 , 13.2. VPO6.65Nio.2oNo.8oCo.soU6.5o requires V, 22.4; P, 13.6; Ni, 5.2; N, 4.9; C, 4.2; H, 2.9:H20 13.1%). Vanadium content and oxidation states were determined as described in [ 11 ]. Phosphorus and metal contents were determined by atomic absorption (Perkin-Elmer Zeeman 5000), and water thermogravimetrically. C, N and H were determined by elemental analysis. Thermal analysis (under a flowing N 2 atmosphere) were carried out using a Setaram B70 simultaneous TG-DTA thermobalance.

2.2. Physical measurements IR spectra (KBr pellets) were obtained with an FTIR Perkin-Elmer 1750 spectrophotometer. EPR spectra were recorded on a Bruker ER200D spectrometer equipped with a nitrogen cryostat. Magnetic measurements were performed in the temperature range 2-298 K with a magnetometer SQUID S.H.E. (variable magnetic field 10-2-6 Teslas). Xray powder diffraction patterns were obtained by means of a Siemens Kristalloflex 810 diffractometer using CuKct radiation and having the Pt peaks as standard. High power 3~p NMR spectra were recorded at 80.96 MHz on a Bruker MSL 200 spectrometer equipped with a 4.7 Tesla superconducting magnet using solid echo sequence. Diamagnetic K (VO2) (HPO4) was used as reference.

3. Results and discussion

3.1. Synthesis and reactivity of vanadyl phosphates A simplified scheme of reactions, showing the relation between the layered vanadyl phosphate dihydrate, VOPO4'2H20, and many of its derived phases known to date is given in fig. 1. We have adapted Sch611horn's nomenclature [19] to discuss the different reaction pathways observed. We have previously emphasized the acid character of solid VOPO4'2H20, and its capability to act both as Lewis and Bronsted [11] (because of the presence of H 3 0 + ions in the lattice) acid. Lewis' acidity o f VOPO4"2H20 is recognized in the formation of coordination intercalates (a) in which neutral basis such as pyridine, substituted pyridines [13], and

60

D. Be/tran-Porter eta/. / Ianady/ phosphate.s

×B

[ZO]2-[HI*[AI *

[Z]-IH]+[H2Oln ~.

A*~,(d) OH- ~

H* (g)

(h)

exB

n H20 [Z] . . . . . .

=-- [Z]-[HI*[Blx[H20].

~ [Z][H20]n

= [ZI[Blx[H2Oln

(b)

(e)

(c)

n Solv, [Zl[solv.] n

x AY* xye-

~

x [Mkm]Y

(f) ~

xye-

%,

[ZlXY-[AlY* x[H20]n

[Z]xy- [MLm]Y x[H2Oln

Fig. 1. Scheme of solvation, exchange, and redox reactions of vanadyl phosphates. Z, VOPO~: A '+ and [ ML,,]'+, guest cations; B. neutral polar guest molecule.

amides [20], enter topotactically into the interlaminar host lattice space with displacement of coordinated water. In fact, the formation of V O P O 4 " 2H20 from a-VOPO4 (b) can be considered itself a topotactic Lewis acid-base reaction, like the one that leads to the formation of Ladwig's phases (c) containing included NH3 or ethanol [15]. Otherwise, the formation (d) of V(V) salts of composition AHVPO~, ( A = N H 2 , K + , R b +,Cs +) [9,14] is nicely accountable in terms of the Bronsted acid character of the VOPO4"2H20 (i.e., given the real V O P O 4 . 2 H 2 0 ~ - (HOVO)PO4 (H30 + ) equilibrium [ 11 ] ). Nevertheless, a structural rearrangement occurs in the course of the reaction, and the resulting phases must actually be formulated as A(VO2) (HPO4) (see below). Chemical reduction of V O P O 4 " 2 H 2 0 can lead to mixed-valence [ V ( I V ) - V (V) ] intercalates (e, f) or to V (IV) containing phases (g), depending mainly on the reducing agent. Notwithstanding, the ultimate nature of the resulting solid is strongly influenced by both kinetic factors and preparative procedures. The reactions leading to mixed-valence intercalates occur without changes in the stacking sequence of the host layers [ 11,12,21 ]. As discussed below, all our results indicate that the inserted ions

are very likely replacing coordinated water molecules (i.e., localized t r a n s to the V = O bond of VO 2+ reduced vanadyl groups). On the other hand, the V (IV) phases (g) may formally be considered as resulting from the redox intercalation of"protons" into the VOPO4.2H20 host lattice, but they show a great structural diversity [6-8, and this work] and in no case the layered morphology of the parent VOPO4"2H20 is maintained. According to recent results [ 22 ], vanadyl (IV) phosphates might undergo reversible topotactic acid-base intercalation reactions (h). Our synthetic strategies (see table 1 ) fit in well with this reactivity pattern. While no modification has been introduced in the preparation of the 13A(VO2) (HPO4) salts [14], it is to be noted, however, that a new polymorph, ct-NH4 (VO2) (HPO4), results when the synthesis is carried out hydrothermally (see section 2). More relevant are the results concerning the reduced phases. Thus, sodium containing intercalates are advantageously prepared when working in homogeneous medium [11 ]. In fact, heterogeneous phase reduction can yield mixed solid phases in which the fraction of vanadium (V) reduced is higher than the intercalation degree [ 11,12 ]. Moreover, the sol-

D. Be/tran-Porter et al. / ~naclyl phosphates

ids obtained from an homogeneous medium are more crystalline. This is probably related to the higher chance for the occurrence of an uniform staging process. Besides, kinetic factors do not affect the end product when working in a homogeneous medium, but could become determinant in heterogeneous phase. Actually, the synthesis of metal-complex containing intercalates (see section 2) is governed by the reaction time. In this case, homogeneous phase reduction is precluded by the p H range of stability of the metallic complexes. To finish this discussion, it must be stressed the large variety of unrelated synthetic procedures reported in the literature in order to obtain oxovanadium (IV) hydrogenphosphates. Thus, a diversity of reducing agents (hydrazine, HC1, etc.) and/or reaction media (2-buthanol, isobuthanol, ethanol, water, acetone, etc.) has been proposed as adequate to yield the different VO(HvPO4).,..r/H20 derivatives [5,7,8,23,24]. Our unified synthetic procedure [6 ], which is based on the use ofiodhydric acid both as reducing agent and donor of protons, has allowed us to obtain all the known phases and four new hydrates (involving additional structural types). This procedure shortens to a great extent the reaction times, leading to single phases. Furthermore, the preparative chemistry of the hydrogen phosphates may be systematized as a function of the acetone: water ratio (R) in the reaction medium (see table 1). 3.2. A(VO2)(HP04) ( A = N H ] , K +, Rb +, Cs + salts

As stated above, there are two known polymorphic varieties of these salts, which may be considered as resulting from a methatetical reaction between the VOPO4,2H20 and the respective metallic carbonate. Whereas the [3 phases were studied by Bordes et al. [ 9 ], the synthesis of the ct phase is reported here for the first time. Some inconsistencies between Bordes' results [9] and our own, led us to approach the structural study of these materials. The structure of ctNH4(VO2) (HPO4) (fig. 2a) has been determined from X-ray single-crystal diffraction data [25], and that ofB-A(VO2) (HPO4) (fig. 2b, A = K + ; [3phases are isostructural) has been solved [25 ] from X-ray powder diffraction data using direct methods corn-

61

a D

CI

T

3

b

a

Fig. 2. STRUPLO84 [27] representations showing the unit cell of (a) ct-NHa(VO2)(HPO4) and (b) [3-K(VO2)(HPO4). The structures are viewedalongthe b axis displayingthe directions of lhe V-V chains. Hydrogenatomsof HPO4groupsare omitted for clarity. Spacegroup found are Pb2~a (a) and Pbca (b).

bined with Rietveld profile refinement principles [26]. Both polymorphs are built up of square pyramidal (VOs) units that share one corner to give chains running along the crystallographic a axis. The (VOs) units in each chain are linked two abreast through one (HPO4) group, with no direct link among the chains. The polymorphs differ in the relative arrangement of the (VOs) units. We find the space group of 13 phases to be Pbca [25], and not Pbcm or Pbc2~ as suggested by Bordes [9]. Therefore, there is no relation between this structure and that of VOPO4-2H20, and the suggested presence of hydroxy groups (OH) trans to V = O vanadyl double bonds is also incorrect [9]. Then, some refinements on previous spectroscopic results are necessary. Indeed, the lack of [ O H - V - O ] entities makes unreliable Bordes' IR band assignment. Consequently, this has been revised on the light of the structural results. Listed in table 2 are the new proposed assignments for the bands due to P O - H groups. These results are consistent with those found for oxovanadium (IV) hydrogenphosphates [ 6 ]. According to the structural results, the formulation A(VO2)(HPO4) is more realistic than the previously adopted, A (VOOH) PO4 [ 9 ]. Isostructural [3A (VO2) (HAsO4) phases have been prepared. IR bands due to AsO-H groups appear very close to the above [28 ].

62

D. Beltran-Porter el al. / ~anac!vl phosphate.s

Table 2 IR absorption wavenumbers (cm ~) for the A( VO~) ( HPO4 ) salts. Only POH bands are included. A=NH4~

A=K +

A=Rb+

A=Cs +

Assignment

3425 w 2940 vw 2835 vw

3410 vw 2905 vw 2830 vw 2700 w 2385 m 1670 m 1250 m 725 s

3400 vw 2900 vw

3400 vw 2880 vw

v(PO-H)

2600 w 2285 m 1680 m 1256 m 728 m

2600 w 2285 m 1680 m 1254 m 725 s

2350 m 1660 m 1225 m 731 s

3.3. M i x e d valence p h a s e s

Partial reduction of V ( V ) in the lamellar VOPO4'2H20 can lead to intercalates in which the induced defect of charge is neutralized by inserted cations. Although the phases having uni- [ 11,12 ] or multivalent [12,29] simple ions have been considered with some detail, only preliminary results concerning metal-complexes-containingintercalates have been advanced [ 30 ]. In a recent publication [11], we have discussed extensively the main features concerning the synthesis and characterization of sodium intercalates with formula Na,(Vl, vv} ' , ) P O 4 . ( 2 - x ) H e O . Some of these results deserve to be emphasized in the present context. Thus, in contrast with the stated for the V ( V ) salts, the structure of the partially reduced phases is very similar to that of the parent VOPO4"2H20, but Na + insertion and H20 evolution induce an orthorhombic distortion of the original tetragonal cell [10,11]. Sodium ions in the intercalates are very likely localized trans to the V = 0 bond of VO 2+ reduced vanadyl groups (i.e., replacing coordination water molecules). This way, the charge defect induced by the reduction should be neutralized locally, and Na + ions may be considered as "'pillars" fixing adjacent layers by introducing ionic interactions in the Van der Waals gap. In fact, the interlaminar distance decreases as the r e d u c t i o n - i n tercalation degree increases. A similar effect has been observed for the intercalation of some di- and trivalem cations [11,291. A previously overlooked aspect concerns the presence of an exothermic effect at relatively high temperature (ca. 4 0 0 - 5 0 0 ° C , i.e. clearly above the

~5(POH)+ u(P()~) 8( POH ),n~,~...... 8( POH ).,.1., pl

......

temperature of total dehydration) in the course of the thermal decomposition of sodium containing intercalates. This effect is associated to an irreversible phase transformation leading to a lower symmetry anhydrous NA,VOPO4 polymorph. Whereas at temperatures below this transition the d e h y d r a t i o n - r e hydration process is reversible, anhydrous phases obtained at higher temperatures do not rehydrate. On this basis, we have been able to isolate single crystals of anhydrous Nao 4,,VOPO4 (see section 2 ) and the solution of its structure is in progress. Some new results concerning the metal-complexes-containing intercalates [M ( e n ) : ] ,V~, V~'~ ~ , ~ O ] P O 4 ( 2 - 2 x ) H : O , may contribute to progress in the knowledge of the parlially reduced derivatives as a whole. Thus, the thermal study ( T G - D T A ) of water evolution from these intercalates show that, as occurs with sodium phases [ 11 ]. water molecules are always lost in two stages (table 3 ). The first one corresponds to the evolution of one water molecule per v a n a d i u m atom, regardless of the intercalation degree, and the a m o u n t of water eliminated in the second stage corresponds now to practically ( 1 - 2 x ) molecules per vanadium atom. This result, as well as the intensity lowering of the v ( O H ) bands due to coordinated water as the red u c t i o n - i n t e r c a l a t i o n degree increases, suggests that also the intercalated metal-complex would be replacing stoichiometrically coordinated water molecules. Again, the interlaminar distance decreases steadily as the a m o u n t of intercalated metal increases (table 3 ). EPR results on mierocrystalline samples fairly complement the above information. When working with phases with a low content of V ( I V ) (,v=0.03.

63

D. Beltrdn-Porter et al. / Vanadyl phosphates Table 3 Analytical results and interlaminar distance in [M(en )2] ,VOPO4' (2 - 2 x ) H 2 0 compounds. [ M (en) 2]/V~o,a,

H20/V,o,a~

molar ratio

M = Cu-' * M = Cu 2 * M = C u 2+ M = N i -'+ M = N i 2~ :') Water molecules evolved,

0.05 0.13 0.20 0.03 0.20

1.88 1.78 1.57 1.96 1.65

d(A) "'

1st step

2nd step

H~ a)

H~ al

1.07 1.00 0.98 1.08 0.97

0.8I 0.78 0.59 0.87 0.67

7.21 7.15 6.95 7.39 7.18

t,) lntcrlaminar distance in ~.

0.05) the hyperfine structure due to vanadium nucleus is clearly resolved. This hyperfine structure "disappears" by increasing the V(IV) concentration, because the dipolar and exchange interactions. Copper derivatives spectra show an axial signal, but nickel (I1) remains EPR inactive even at 4.2 K. Therefore, it does not seem that the planar environment of the nickel undergoes significant distortions as consequence of the intercalation (see below). More details on the geometry around the copper (II) ion and on the V(IV) centres can be inferred from the [Cu(en)2]2+-"doped '' phase (i.e., x ca. 0.001) EPR spectrum (fig. 3b). The observed spectrum results from the superimposition of the signals due to copper (H) (fig. 3a) and those corresponding to two different types of vanadium (IV) ions. The main EPR results are reported in table 4. It is evident that the coordination polyhedron around copper (II) rea

25'00

Dehydration

35'00

Fig. 3. X-Band EPR spectra o f (a) [ Cu (en) 2 ] z+_doped VO£O4.

i

H (gauss) [Cu(en)2]12 and (b)

Table 4 EPR parameters

[Cu(en)2]12 [Cu(en)2]2+-doped [VO2+](1) [VO2+](I1) Na+-doped VOPO4.2H20

gll

gt

2.195 2.202 1.935 1.935 1.934

2.053 2.056 1.985 1.985 1.985

A, (Gauss)

AL (Gauss)

195.0 201.2 196.0

73.1 80.1 72.9

mains basically unaltered. Otherwise, the observation of two component spectra with different A, and A• values has allowed us to conclude the existence of two different environments for the V(IV) ions. Given that, in both cases, g~ < g i , it can be stated that the vanadium orbital ground state is essentially d~,.. The spectra of both (VO 2+) groups show the same fine structure (g values), which in turn is very similar (see table 4) to that corresponding to sodiumdoped VOPO4"2H20 (i.e. Nao o2Vo.o2 iv Vo.98 v POs' 1.98H20). Nevertheless, the hyperfine parameters of one of them, [VO 2+ ] (II), are clearly divergent. While it seems reasonable to assume that the environment of [VO 2+ ] (I) groups is very similar to that found in the dihydrate (i.e. they belong to O=VO4 chromophores), the raised hyperfine parameters of [VO 2+ ] (II) are indicative of a more ionic bonding system, this meaning that the vanadium to oxygen double bond character in this chromophore is lower than that corresponding to the [VO 2+ ] (I) site. All these results allow us to suggest that the planar [ M (en)2 ] 2+ entities enter the interlayer space sandwiched between two [VO 2+ ] groups (fig. 4), giving

64

D. Behrdn-Porler et a//[anadylphosphates ~c b

,

® 0

~

2+

2+

• 74

Fig. 4. Schematic view of the proposed intercalation site of the [M(en ):]~+ species in the VOPO4 matrix. rise to formally trimeric -[[VO 2 + ] ( l l ) - M [ VO 2+ ] ( I ) } units along the crystallographic c axis. A bonding interaction between the intercalated-metal and one oxygen atom (from [VO 2+ ] ( I I ) ) would be consistent with the reduction of the V - O double bond character in the { V O 2 + ] ( I I ) site. Moreover, the maintenance of the planar geometry of the complexes once intercalated, with van der Waals forces partly substituted by ionic interactions, would agree with the observed shortening of the interlaminar distance (table 3). These conclusions are similar to the obtained by Antonio et al. [29], when dealing with simple multivalent ions, through EXAFS data. Magnetic measurements can provide more insight on the nature of these intercalates. Indeed, the phases containing ca. 40% of V (IV) (i.e. [Cu(en)2]o.2VOPO4'l.57H20 and [Ni(en)2lo.2VOPO4" 1.65H20 ) behave similarly. From 300 down to 35 K the XTvalues stay practically constant, being 0.210 and 0.164 cm 3 mol -~ K for Cu and Ni derivatives, respectively. Obviously, these values include the parametric contribution of V ( I V ) (0.146 cm 3 mol -L K). Below 35 K, these values decrease to become 0.165 (Cu) and 0.125 (Ni) cm 3 mol -~ K at 4.2K. The above proposal o f quasi-square planar environments (the local symmetry, including the rela-

tively close oxygen atom, actually would be C4~) around the intercalated metal (with 2B~-Cu- and ~A~ or ~B2-Ni- ground states) is compatible with the values of X T at high temperatures. Likewise, the N i ( I I ) singlet ground state would explain its EPR inactivity. The diamagnetism of the nickel derivative, in particular, suggests that the interactions between the intercalated metal and the oxygen atom of the [ VO 2+ ] (II) group are weak. Nevertheless, given the interlaminar distance values (6.95 A. and 7.1 g for M = C u and Ni, respectively), and assuming a linear [ V ( IV ) - M - V ( IV ) ] array, a metal (Cu or Ni ) to oxygen (from [VO 2+ ] ( I t ) ) distance shorter than ca. 2.7-2.8 .~. can be predicted (particularly if the electrostatic repulsion between the metal and the neighbouring V (IV) ion taken into account ). Hence. in linewidth that predicted from the EPR results, the intercalated metal to oxygen interaction cannot be neglected. Under C4v symmetry, the diamagnetism of nickel ( II ) might be rationalized assuming that. in such a system ( N i ~ O = V O 4 ) , the "additional" c~ oxygen to nickel donation would require some synergistic n electron density transference from the nickel d,-, d,._- doublet. Thus, the oxygen request for electron density would be satisfied through )z orbitals which have antibonding character in the vanadiumoxygen internuclear region and, in accordance with previous considerations, the vanadium to oxygen bond would result weakened. Parallely~ the nonbonding character of the nickel d,.,. and d:2 orbitals should be reinforced, and their energetic separation from the d,: ,.: level would remain high. Therefore, a n IA t o r IB 2 ground state would be predicted. On the other hand, it is evident that the decrease of X T below 35 K must be due to an antiferromagnetic coupling among paramagnetic centres. However, the occurrence of this coupling across the { [VO 2 + ] ( I I ) - M - [ V O 2+] (I)} entities seems very unlikely, owing to the relative orientation of the spin containing orbitals from each metal ion. In fact, the variation of X T p e r vanadium (IV) mol along the 4 30 K range is the same for both copper (once substracted the paramagnetic contribution of copper (II) ions) and nickel derivatives (see fig. 5 ). Even though weak antiferromagnetic coupling among vanadium atoms might occur through the diamagnetic [ Ni (en) 2 ] 2 + system, a similar mechanism is unreasonable for the analogous copper compound. The

D. Beltran-Porter et al. /Vanadylphosphates

(cm3 mol

k) 0

o ~ . ° o . . e ~ , . * 8 ** o . .

0.15

O 0

0

0 4

oo

g 0.10

0.05

20i

4J°

e~°

al°

T~KJ

Fig. 5. Temperature dependence of XTper V(IV) mol along the 4-30 K range. Values corresponding to the copper derivative have been correcled for the copper(II) contribution ( O ) [Ni(en)2]o.2oVOPOa'l.65H20; (*) [Cu (en)2 ]o.2oVOPO4' 1.57H20.

observation of two independent V (IV) EPR signals with hyperfine structure in low concentration copper (x~<0.05) and nickel (x=~<0.02) phases is the strongest argument against the existence of intratrirneric magnetic interactions. On the other hand, the EPR signals collapse into a single broad one for systems containing ca. 25% o f V ( I V ) ( x ~ 0 . 1 3 ) . Above this x value the probability of finding pairs of V (IV) ions in each sheet becomes significant. All these considerations lead us to conclude that the antiferromagnetic interactions must occur in the [VOPO4]~ layer themselves, being the phosphate groups which act as magnetic connection between vanadium atoms. Effectively, preliminary 3~p NMR experiments on the Nao.46VOPO4 derivative show a line shift of ca. 240 ppm.

65

tetrahedra [7] ) and that of the parent V O P O 4 " 2 H 2 0 . The tetrahydrate (type III) can be viewed as double chains of alternating VO6 octahedra and PO4 tetrahedra [ 7 ], and directly gives the sesquihydrate upon heating [6]. Concerning the structural type II, we have approached the resolution of the structure of V O ( H P O 4 ) ' 2 H 2 0 from X-ray powder diffraction data [26]. The crystal structure (fig. 6) consists of vanadyl hydrogenphosphate layers stacked along the a axis and held together by hydrogen bonding via water molecules [31 ]. Fig. 6 shows the layer organization: each adjacent pair of VO6 octahedra shares one corner to give infinite chains running along the b axis. These chains are linked together by H P O 4 groups which share three oxygen atoms with three different VO6 octahedra. The X-ray powder diffraction pattern of the trihydrate, VO (HPO4). 3H30, is identical to that registered for the dihydrate. It can be thought that there is one "zeolite-type" water molecule occupying the (1 0 0) direction channels. The monohydrate, VO(HPO4)'H2O, retains all the features of the dihydrate, except for the elimination of the interlayer water molecule. This dehydration results in a shortening of the interlaminar distance [4,6 ]. The available structural results enabled us to reconsider the IR spectroscopic data and propose a new assignment of the bands [6] on the basis of the site group analysis and the factor group splitting effect [34,35 ]. As stated above, we have benifited from ¢

hi"

3.4. VO(HvPO4)x'xH20 phases The use of H1 to carry out the complete l e - reduction of oxovanadium (V) phosphates yield a diversity ofvanadyl (IV) hydrogenphosphates showing a striking structural variety. Indeed, it has been possible to classify the monohydrogenphosphate series ( x = 1 ) into three structural types, namely: type I, y=0.5; type II, y = l, 2, 3 and type III, y = 1.5, 4 [6]. The hemihydrate (type I) is lamellar, but there is no relation between its structure (which contains pairs of face shared VO6 octahedra linked by PO4

Fig. 6. STRUPLO 84 [27] representation showing the unit cell of VO(HPO4).2H20.

66

D. Beltrdn-Porter el a/. / I ~nadyl phosphates

these results in the study of the A ( V O 2 ) ( H P O 4 ) phases. The analysis of the known crystal structures of oxovanadium (IV) derivatives reveals a great variety of bridging modes between vanadium atoms suitable to yield magnetic interactions. However, such a diversity really makes difficult to establish unambiguous predictions about the magnetic behaviour of a given compound. Even more, as we shall see, the structural connectivity might lead to initially "'evident" previsions which disagree with the experimental results. Clearly, the most exciting aspect from the theoretical point of view would be the attainment of magneto-structural correlations which could explain the observed magnetic behaviour. In this context, our present approach is intended to set criteria on the superexchange pathways which result effective in these materials. The main features concerning the experimental magnetic behaviour of these compounds have been summerized in table 5. A result to be emphasized is the observation, when measure& of significant line shifts in the '~P NMR spectra. Moreover, these shifts are nicely correlated with the calculated J values. Hence. some spin density, related to the strength of the interaction, is detected on the P nucleus. Thus, as predicted for the complex-containing intercalates, and verified on the sodium derivative, phosphate groups must be involved in the exchange pathways. The structural moieties which a priori might yield exchange pathways among V (IV) ions are shown in fig. 7. The coordination polyhedra around vanadium atoms are (VOw,) octahedra strongly distorted axially. Therefore, the d,, vanadium orbitals, v mixed with in plane oxygen p orbitals, would be the "'magnetic orbitals". Consequently, significant coupling through l,-oxo bridges (type O, fig. 7) is precluded because of the orientation of the magnetic orbitals. As stated above, it is very likely the exchange pathway involving phosphate groups the one responsible for the observed magnetic behaviours. At this point, to make predictions concerning both the type and the strength of the interactions on the only basis of geometrical considerations is not an easy task. Nevertheless, some arguments encompassing the principle of optimum matching between the metallic part of the magnetic orbitals can be advanced. Thus, coplanarity and an adequate relative orientation of

.E ',.,.2

,-5 --: E

-2

,'5

e'll

=

2"

+

~..=_

~2

C

.. ~

Z

.

~-

~

:z: ;-7

:ao

.:

0 h

d

2, I E b

.~'~o

E

'rJ

~

II

II

II

II

II ,J'

II

II

II

II

II

D. Beltr(~n-Porter eta/. / Vanadvl phosphates

67

DI

D II

0 III

© D IV

D III

?

MII

MI

•

0

DV

D Vl

M III

~ D

Vll :'

Fig. 7. Bridging modes observed in the known structures of vanadyl (IV) derivatives which might act as exchange pathways. (a) O: /loxo bridges (O III includes also one lL-phosphato (O, O ' ) bridge) and M: iL-phosphato (O, O ' ) bridges: (b) D: di-iL-phosphato (O, O' ) bridges.

the local coordinate system of adjacent (VO6) octahedra are reasonable requirements for an efficient coupling. Furthermore, double bridging modes (type D, fig, 7) would lead to interactions stronger than the single ones (type M, fig. 7), provided a similar relative orientation of the (VO6) units. Likewise, di/l-phosphato-(O) bridges (D I, and D II, fig. 7), minimizing vanadium-vanadium distances, would be more efficient than the di-/~-phosphato- (O, O ' ) ( D I I I to D VII, fig. 7) ones. All these considerations allow us to propose the following interaction strength sequences: (a) D>M>~O; (b) O I>~O II; (c) M I ~ M II>>M III; (d) D I > D I I > D V ~ D IV>>D VI ~ D VII ~ D I I I . Accordingly, dealing with each one of the studied compounds, and taking into ac-

count the experimental results in table 5, we propose: (a) VO(H2PO4)2. It is built up of O It entities, leading to chains along the (0 0 1 ) direction, which are linked among them through M It units in the ab plane. In both moieties, the mismatch of the magnetic orbitals would result in a poor V-V magnetic interaction. In fact, the temperature of the susceptibility maximum is very low (T(Xma~) :-2.7 K). X versus T data fit in very well with that expected for an isotropic simple 1-D chain, this indicating that O II>>M Ill. Actually, the predominance of M I I I would make operative a 2-D behaviour, and no fit to an I-D model would work. Some spin density is present in the V-bond, indicating a certain contribution of the d:~ orbital to the mainly d,,. ground

68

D. Bellrdn-Porler el a[. / l anaclrl l~hosphate.~

state. An additional 31p N M R experiment could confirm the inactivity of the phosphate groups in the exchange mechanism for this derivative. (b) (VO)eP,O~. It may be considered that the basic entities in the p y r o p o s p h a t e structure [ 33 ] are of the type D I. These are connected two abreast by means of two oxo bridges (like in O I) in the ( 1 0 0 ) direction. Moreover, they are also linked to two adjacent D I units through phosphate groups (like those in D IV) in the (0 1 0) direction, and to two more D 1 units in the (0 1 1) direction. To explain the observed magnetic properties, such a complex structural pattern was, however, simplified by neglecting initially all the possible pathways involving phosphate groups, and described, consequently, as isolated double "'ladder" chains ( D 1-O I) in the a direction [36]. The susceptibility data showing an antiferromagnetic b e h a v i o u r with 7"(X. ..... ) = 79.6 K (inconsistent with an isolropic dimer m o d e l ) were nevertheless "accurately described" on the basis of the Heisenberg alternating S = 1/2 chain model, even though the authors noted the incongruity of their results [36]. Whereas intradimeric interactions in D I are expected to be significant, it is unlikely that O I moieties be able to account for the observed l-d behaviour. As noted above (see ( a ) ) , the O I entities are inadequate as exchange pathway. On the contrary, when the availability for the exchange o f the previously ignored D IV and D VI entities is considered, and given their relative topologies, it is reasonable to conclude the magnetic activity of the D IV unit. Hence, the pyrophosphate can truly be described as alternating magnetic chains formed by consecutive D I and D IV entities running along the h axis. The "striking" results reported in ref. [ 36 ] are then fully reasonable. (c) V O ( H P O 4 ) . 4 H , O . The structure of the tetrahydrate [ 7 ] can be adequately described as a zigzag chain in which two kinds of very similar D VII units are linked through M II entities. Such a framework leads to isolated "ladder-like" double chains [7]. Actually, all the above structural moieties do not have a topology suitable to transmit magnetic interactions, what is in good agreement with both the observed low value of T ( X ...... ) ~ 6 K and the chemical shift in the ~ P N M R experiment. While only the use of a Heisenberg linear chain model provides a

good fit o f the X data, a "'forbidden A n l , = 2'" signal is clearly observed in the EPR spectrum. Although, this last result is generally understood as evidence of dimeric unities, it has also been observed in some ld systems [ 37 ]. In fact, for any alternated chain there is an energy gap between the singlet ground state and the lowest triplet excited state [36] which can result in half-field EPR transitions. Indeed, we found again this signal when performing low temperature EPR experiments on the pyrophosphate. The experimental results could be rationalized assuming thai the interactions transmitted by D III entities overcome those involving M II unit. The fact that the observed b e h a v i o u r be accountable in terms of an alternating chain may be related with the structural differences between both D VII units involved, but very low t e m p e r a t u r e experiments are necessary to confirm this hypothesis. ( d ) V O ( H P O 4 ) . 2 H 2 0 . A view of this structure has been shown in fig. 6. In terms of the structural schemes in fig. 7, it can be described as formed by O !11 unit chains (lying in the (0 1 0) direction) interlinked by means of bridges similar to D V. Following the above considerations, O II1 entities would be a poor support of magnetic interactions. In contrasl, D V phosphate bridges may provide good exchange pathways. The experimental behaviour, which reflects a moderate antiferromagnetic coupling ( T ( X ...... ) = 2 5 K ) , as well as the X data, are nicely accountable on the basis of the isotropic diiner model. There is no possible fit using 1-D models. Room t e m p e r a t u r e EPR spectrum shows the halt-field signal. and the '~P N M R chemical shift indicates the participation of the phosphate groups in the exchange phenomena. It can be considered that, flom the magnetic point of view, the dihydrate behaves as isolated magnetic D V pairs antiferromagnetically coupled. (e) V O ( H P O 4 ) ' 0 . 5 H 2 0 . The hemihydrate has previously been analyzed in terms o f d i m e r i c entities [3]. Its structure can be described as built up of D II units linked among them (in the x)' plane) through D 1II and D VI bridges, which play the same role that D IV and D V1 do in the ),z plane in the pyrophosphate structure. Neither D Ill nor D VI (see above) have an adequate topology to yield good magnetic exchange pathways. On the other hand, some remarks about the magnetic activity of D II bridges can

D. Beltran-Porter et al. / Vanaa~vl phosphaws

be d o n e . T h u s , w h e n g o i n g f r o m D I to D II t o p o l o g y t h e angle d e f i n e d by t h e m a g n e t i c o r b i t a l s c h a n g e s f r o m 180 to 60 °. T h i s m u s t result in a l o w e r i n g o f t h e o v e r l a p integral o f t h e m a g n e t i c orbitals, w h i c h in t u r n will b e r e f l e c t e d in t h e s t r e n g t h o f t h e magn e t i c i n t e r a c t i o n . In fact, t h e c a l c u l a t e d J v a l u e f r o m the d i m e r m o d e l (43 K ) is o f t h e s a m e o r d e r b u t l o w e r t h a n t h a t f o u n d for t h e p y r o p h o s p h a t e . A l t h o u g h t h i s first a p p r o a c h to t h e m a g n e t i c beh a v i o u r o f the V ( I V ) - c o n t a i n i n g p h a s e s is b a s e d m a i n l y o n t o p o l o g i c a l c o n s i d e r a t i o n s , w e are able to o f f e r a c o n s i s t e n t q u a l i t a t i v e i n t e r p r e t a t i o n o f t h e exp e r i m e n t a l results. H o w e v e r , a m o r e e l a b o r a t e d s e m i q u a n t i t a t i v e s t u d y u s i n g e x t e n d e d H u c k e l calculat i o n s o f t h e o v e r l a p i n t e g r a l s o n e a c h c o m p o u n d is n o w in progress.

Acknowledgement We t h a n k v e w m u c h t h e E.E.C. for p a r t i a l f i n a n cial s u p p o r t o f this w o r k , u n d e r g r a n t ST 2 J - 0 1 6 4 - 4 E ( C D ) , a n d Dr. J.V. F o l g a d o for h e l p f u l d i s c u s s i o n . P.A. t h a n k s t h e S p a n i s h M i n i s t e r i o d e E d u c a c i 6 n y C i e n c i a for a F P I f e l l o w s h i p .

References [ 1] B.L. Hodnett, Catal. Rev.-Sci. Eng. 27 ( 1985 ) 373. [ 2 ] O. Centi, F. Trifiro, J.R. Ebner and V.M. Francheni, Chem. Rev. 88 (1988) 55. [3] G. Villeneuve, P. Amor6s, D. Beltr~in and M. Drillon, Organic and inorganic low dimenisonal crystalline materials, eds. P. Delhaes and M. Drillon, NATO ASI Series, Vol. B 168 (Plenum, New York, 1987). [4] P. Amor6s, R. Ibfifiez, D. Beltrfin, A. Le Bail, G. Ferey and G. Villeneuve, in: Proc. First European Inorganic Chemistry Seminar (EUCHEM), Dourdan, France, April 5-9 ( 1988 ). [5] J.T. Wrobleski, lnorg. Chem. 27 (1988) 946. [6] P. Amor6s, R. Ib~ifiez, E. Martinez, A. Beltrfin, D. Beltrfin and G. Villeuneuve, J.Chem. Soc. Dalton Trans., to be published. [7] M.E. Leonowicz, J.W. Johnson, J.F. Brody, H.F. Shanson and J.M. Newsam, J. Solid State Chem. 56 ( 1985 ) 370; C.G. Torardi and J.C. Calabrese, Inorg. Chem. 23 (1984) 1308. [8]S.A. Linde, Y.E. Gorbunova, A.V. Lavrov and Y.G. Kustnetsov, Dokl. Akad. Nauk SSSR 224 (1979) 1411. [9] S. Pulvin, E. Bordes, M. Ronis and P. Courtine, J. Chem. Res. (S) 29 (1981);

69

S. Pulvin, E. Bordes, M. Ronis and P. Courtine, J. Chem. Res.(M) (1981) 362. [ 10] H.R. Titze, Aust. J. Chem. 34 ( 1981 ) 2035. [ I 1] N. Casafi, P. Amoros, R. Ibafiez, E. Martinez, A. Behr~in and D. Beltran, J. Incl. Phen. 6 (1988) 193. [ 12 ] A.J. Jacobson, J.W. Johnson, J.F. Brody, J.C. Scanlon and J.T. Lewandowski, lnorg. Chem. 24 ( 1985} 1782. [13] J.W. Johnson, A.J. Jacobson, J.F. Brody and S.M. Rich, Inorg. Chem. 21 (1982) 3820. [ 14] F. Preuss and H. Schug, Z. Naturforsch. Tell B, 30 ( 1975 ) 334. [15] G. Ladwig, Z. Anorg. Allg. Chem. 388 (1965) 2. [16] V.G. Jander, K.F. Jahr and H. Witzmann, Z. Anorg. Allg. Chem. 217 (1934) 65. [ 17 ] M. Procter, B.J. Hathaway and P Nicholls, J. Chem. Soc. A (1968) 1678. [18] F.P. Dowyer, Inorg. Syntheses 6 (1962) 198. [19JR. Sch611horn, in: Intercalation chemistry, eds. M.S. Whiningham and A.J. Jacobson (Academic Press, New York, 1982 ) p. 315. [20] M. Martinez Lara, L. Moreno Real, A. Jimenez L6pez, S. Bruque Gfimez and A. Rodriquez Cargia, Mater. Res. Bull. 21 (1986) 13. [21 ] M. Martinez Lara, A. Jimenez Lopez,L. Moreno Real, S. Bruque Gfimez, B. Casal and E. Ruiz-Hitzky, Mater. Res. Bull. 20 (1985) 549. [22] L. Alagna, T. Prosperi and A.A.G. Tomlinson, Mater. Res. Bull. 22 (1987) 691. [23] J.W. Johnson, D.C. Johnston, A.J. Jacobson and J.F. Brody, J. Am. Chem. Soc. 106 (1984) 8123. [24] F. Garbassi, J.C.J. Bart, F. Montino and G. Petrini, Appl. Calal. 16 (1985) 271. [25]P. Amor6s, D. Beltr~in, A. Le Bail, G. Ferey and G. Vitleneuve, Eur. J. Solid State lnorg. Chem., to be published. [26] H.M. Rietveld, J. Appl. Crystallogr. 2 (1969) 65. [27 ] R.X. Fischer, J. Appl. Crystallogr. 18 ( 1985 ) 258. [28] P. Amoros and D. Beltr~in. unpublished results. [29] M.R. Antonio, R.U Barbour and P.R. Blum, Inorg. Chem. 26(1987) 1235. [30] N. Cassafi, E. Martinez, P. Amor6s, N. Castell6, G. Llorente, A. Beltrfl and D. Beltrfin, in: XXIV Internat. Cone Coordination Chemistry, Athens (Greece), August 1986. [31]A. Le Bail, G. Ferey, P. Amor6s, D. Beltrfin and G. Villeneuve, J. Solid State Chem., to be submitted for publication. [23] G. Villeneuve, A. Erragh, D. Behrfin, M. Drillon and P. Hagenmuller, Mater. Res. Bull. 21 (1986) 621. [33] Y.E. Gorbunova and S.A. Linde, Dolk. Akad. Nauk. SSSR 245 (1979) 584. [34] J.R. Ferraro and J.S. Ziomek, Introductory group theory, 2rid. Ed. ( Plenum Press, New York, 1975 ). [35] E. Berry and C.B. Baddiel, Spectrochim. Acta 23A ( 1967 ) 2089. [36]D.C. Johnston, J.W. Johnson, D.P. Goshorn and A.J. Jacobson, Phys. Rev. 25B (1987) 219. [ 37 ] T.T.P. Cheung, Z.G. Soos. R.E. Dietz and F.R. Merritt, Phys. Rev. 17B (1978) 1266.