Академический Документы
Профессиональный Документы
Культура Документы
North-Holland
The following binary metal oxides have yielded high-resolution “0 nuclear magnetic resonance (NMR) spectra, obtained by
magic-angle spinning (MAS) without isotopic enrichment: a-Al,O,, TiOs (anatase and rutile), TirOr, monoclinic VOr, CurO,
ZnO, tetragonal and monoclinic Zr02, AgrO, tetragonal SnO, hexagonal Laz03, Ce02, HQ, a-Pm, also Na &41203 and cubic
MgO-Zr02. Detection sensitivity limits observation to cases where the nuclear quadrupole coupling is less than 4 MHz. Different
ceramic phases can readily be distinguished by their chemical shift. As empirical chemical shift correlation is sug~~estedbetween
“0 in the oxide MrO, and 19Fin the corresponding fluoride MF,,,.
phase, viz. the standard NazO* 11Alz03 structure. ing typically 4000-8000 responses with the repeti-
When necessary, samples were calcined to remove tion period of 20 s. The x/2 pulse length for liquid
hydroxide or carbonate contaminant. The magnesia- water was 9 and 6 us in two different probes. The
fully-stabilized cubic zirconia was of eutectic com- spinners supplied by Bruker for use in the former
position ( 13 mol% MgO ) , fired at 1970 K for 4 h and probe were of alumina, and for the latter MgPSZ.
rapidly cooled, and then characterized by electron Each of them contributed a characteristic “0 NMR
diffraction. The magnesia-partially-stabilized zir- signal which served as a frequency marker and inten-
conia (MgPSZ) was the material of spinners sup- sity calibrator. Spinner and probes were selected to
plied by Bruker. The composition of the nominal reduce spectral overlap with the specimen. The 170
T&O3 was analysed as Ti203.04 by thermogravimetry resonance in distilled water was taken as the fre-
during oxidation to the dioxide. quency standard; its linewidth was 1.5 ppm ( fwhm ).
NMR spectra were obtained with a Bruker MSL The spectra are shown in figs. t-3, relative fre-
400 spectrometer operating in a nominal field of 9.4 quency (6) being expressed in parts per million
T and at the frequency of 54.242 MHz. All experi- (ppm ). For display purposes additional line broad-
ments were carried out at room temperature, 296 K. ening has been imposed on the raw spectra, as indi-
Magic-angle spinning was done in a Bruker double cated_ Nevertheless the figures demonstrate the sig-
bearing gas turbine and a 7 mm outer diameter spin- nal-to-noise ratio that is available in these oxides
ner of 0.3 1 ml capacity. The MAS rotation frequency when the low natural abundance is compensated by
was variously set between 3.5 and 4.9 kHz. The spec- moderate patience. (Eleven other compounds that
tra were obtained by Fourier transforming the free were examined in the same way without any reso-
induction decay following a 4 us rf pulse, accumulat- nance being detected are noted in section 4 below. )
ah
(a)
Ceo2
1
15OHz
* *
4
Ag,O
6.9
3ooHr
Illlll’lLllf~‘l IIIIIII(I’~‘~~~
1000 800 600 400 200 0 -200 -400 ,000 8W BOO 400 200 0 -200 -400
Ppm ppm
Fig. I. MAS “0 absorption speeua of polyerystalline metal oxides: (a) CeO,, (b) Cu10, (c) Ag20, (d) a-PbO, (e) black SnO, ( f ) Na
P_A1203. Peaks due to (A) alumina spinner, (2) zirconia spinner, ( l ) spinning sidebands; additional line broadening (fwhm). Fre-
quency reference is H2”0.
T.J. &stow, S.N. Stuart I “0 NMR in simple oxides 461
(4
h-&&i 2oot-k
* l
A
tb%
160Hz
Z A
(d)
TiO, (h)
rutile 200Hz Mg c-Zr@ 2OOHz
I I II I I I 01 I 81 13 I
1000 800 600 400 200 0 -200 -400 moo 800 600 400 200 0 -200-400
fwm pm
Fig. 2. h4.4S I70 absorption spectra of polycrystalline metal oxides: (a) hexagonal La203, (b) TizOs, (c) TiOl anatase, (d) TiOz rutile,
(e) HfOz, (f) monoclinic ZrOz, (g) tetragonal ZrOl, (h) MgO-fully-stabilized cubic ZrO2. Peaks due to (A) alumina spinner, (Z)
zirconia spinner, ( s ) spinning sidebands; additional line broadening (fwhm). Frequency reference is Hz”O.
Table 1
Cubic oxide structures
given for spin I= 5/2 by in terms of Cartesian tensor components, and where
dQ is defined by eq. (2 ) . A lower bound to M2 can be
s=s,- &(VQ/V#(l+$$). (1) found from the amplitude of the sidebands visible
above the noise: with an upper bound for dQ, eq. (4)
Here, as usual, =&e2qQ/h is the nuclear quadru-
vQ
gives a lower bound for 11 Aa]].
pole resonance frequency, and q the asymmetry pa- If the static broadening interactions are small
rameter of the quadrupole interaction. Thus the der-
enough, then rotational sidebands of the satellite
ivation of isotropic chemical shift & from centroid transitions will be visible [ 8 1. Sidebands of the first
frequency requires a second-order quadrupolar cor- pair of satellites, + (3/2, i/2), are 7/24 as wide as
rection. The standard deviation (rms half-width) of the centre band of the central transition (not count-
the centre-band absorption, relative to vL, has the ing residual broadening) and are slightly shifted in
quadrupolar contribution [ 91 frequency. Sidebands of the second pair of satellites,
+ (5/2, 3/2), being 20 times weaker than the first,
~Q=&vQ/vL)2(1+~tzz), (2)
are not important. A lower bound to the first side-
according to the adiabatic approximation [ 6 1. The band amplitude provides an upper bound to if the
VQ
observed standard deviation d of the central line sideband envelope is smooth. If, on the other hand,
therefore provides an upper bound to the quadrupole vQ is much greater than about (1+1/2)/7=0.75
coupling, thus, MHz, 7 being the rf pulse length, then the satellite
transitions are no longer excited.
KQ= ]e2qQ/h((1+fq2)‘/’ Nonetheless, simple observation of rotational side-
bands does not permit direct inference about the static
< +&(+t)“2, (3)
broadening mechanisms unless the responsible tran-
The quadrupolar splitting of the centre band, which sitions can positively be identified. The sidebands
is approximately $(vQ/~~)~(l-_) [6], has not might be those of the central transition broadened by
been seen in the present study. dipole interaction, those of the first satellite transi-
Under certain conditions distinct rotational side- tions broadened by quadrupole interaction or some
bands of the central transition will be visible. The combination of the two. In most cases where the sig-
pattern of centre band and sidebands has a variance nal-to-noise ratio is low we are unable to identify the
or second moment that is independent of the rota- sidebands.
tion frequency, and to which the static broadening The results of our measurements are presented in
mechanisms contribute independently. In the pres- table 2. Both raw chemical shift 6’ and corrected
ence of shielding anisotropy Aa and quadrupole in- chemical shift & are given. The width of the central
teraction, the second moment, relative to I$ is [ 10,9] line, fwhm = 2.3 55d (for equivalent Gaussian ) , de-
termines both the chemical shift correction and the
M 2=&]lAal]2+~~&, (4) stated error in So. In the cubic structures the error is
where taken as id; in cases of lower site symmetry, where
the lineshape has some quadrupolar contribution, the
llW2= z C&d2, C %a=O, correction 2dQ is estimated conservatively as d, the
(Y
T.J. Bastow, S.N. Stuart I “0 NMR in simple oxides 463
Table 2
“0 NMR at 54.242 MHz with magic-angle spinning of about 4 kHz. Centroid frequency and corrected chemical shift (ppm with respect
to water); linewidth (fwhm, ppm) of centr&and absorption; upperbound of quadrupole coupling (MHz) derived from linewidth,
relative intensity of sidebands; extent of visible sideband pattern (standard deviation, ppm)
al See text for informed correction. b, Derived from positions of satellite sidebands.
to a distribution of chemical shift. In the case of CutO, [ 3 ] and an order-of-magnitude calculation of 0.16
it must be in spite of the coherent Cu environment, MHz [ 141.
since a clearly defined 63Cu nuclear quadrupole res-
onance signal was measured previously in the same 4.3. Sesquioxides
material [ 12 1. The chemical shift in Cu,O is lower
than that reported by Oldfield et al. [ 5 1. In hexagonal Laz03 the oxygen atoms occupy two
There is appreciable complexity in magnesia cubic distinct sites, one with octahedral coordination, one
zirconia, which is a solid solution having the average tetrahedral, both of axial symmetry. The two ob-
structure of fluorite with oxygen vacancies (Zr, served (centre-band) resonances (fig. 2a) can be as-
Mg) (0, 0 )2_ The fully stabilized material ( 13 mol% signed on the basis of their integrated intensity. The
MgO) gives a broad band (fig. 2h) indicating the site of higher 6 has the lower coordination. This line
range of oxygen chemical shift. In MgPSZ (9 mol% shows evidence of low-level satellite sidebands, rather
MgO, in typical commercial grades ) such a spectrum like ZnO; from the shape of the envelope the quad-
is apparently swamped by the signal from the accom- rupole coupling constant may be about 0.37 MHz; the
panying tetragonal phase (fig. 3, discussed in section prominence of the first sidebands is presumably due
4.4 below). In neither material is there any trace of to the central transition_ An observation of lower res-
the MgO line at 47 ppm [ 31. olution has been reported [ 5 1.
The solid alumina spinners give the same spectrum
4.2. Uniaxial monoxides (marked A in figs. 1 and 2 ) as a powder specimen of
a-Al,O, (in a zirconia spinner). The centre-band
In the tetragonal structure of litharge, oxygen at- lineshape is approximately Gaussian. First side-
oms occupy a tetrahedral site of axial symmetry. bands are visible with about the same width as the
Strong sideband patterns are observed in the spectra centre band: they may be attributed to the central
of black SnO and a-PbO (figs. 1d, 1e ). Static dipole- transition. Now the I70 quadrupole interaction has
dipole broadening is expected to be minor. If the been measured in a single crystal of ruby as
sidebands should be attributed to the central transi- le2qQ/hl =2.167(5) MHz, ~=0.517(5) [15]. Us-
tion, it would follow from eq. (4) that the shielding ing those parameters, the chemical shift correction in
anisotropy ]]Aa]] is at least 300 and 270 ppm, respec- eq. ( 1) amounts to 10.4 ppm, giving the corrected
tively. The chemical shift of SnO agrees with a recent shift S,= 7 1 ppm. At the frequency of 67.8 MHz em-
report, but our value for PbO is higher [ 5 1. ployed by Oldfield et al., the reported position (6’ )
Our measurement of ZnO (spectrum not shown) [ 1 ] is subject to a correction of 6.7 ppm, which leads
gives a sharp central line at S’ = - 18.4 ppm, which to &= 73 (2 ) ppm; and another report [ 4 ] gives the
agrees with Oldfield et al., value [ 3 1. The centre band corrected value 75 (2) ppm. From eq. (2) the quad-
is sharp enough to ensure that the satellites are ex- rupolar contribution to the width of the centre band
cited also. The strong sidebands have a symmetrical is 12.3 ppm (fwhm of equivalent Gaussian); the ob-
envelope, which indicates a first-order quadrupolar served linewidth of 14 ppm is just enough to blur the
powder pattern with q=O (in accordance with site quadrupolar splitting of about 5.6 ppm. The second
symmetry) and small shielding anisotropy; the stan- moment A42 of the static central transition is esti-
dard deviation (excluding the centre band) is at least mated from the sidebands to be at least 1460 ppm2.
10 kHz. On that interpretation, the observation of Of this, 358 ppm2 is due to the above quadrupole in-
sidebands equidistant from the centre places an up- teraction, 490 ppm* to dipole-dipole interaction from
per bound on the quadrupole coupling constant [ 81 27Al (according to Walter and Oldfield [ 41) , leaving
of about 0.2 MHz. Other such bounds derived from at least 6 10 ppm* which may be attributed to shield-
eq. (3) and from the first sideband amplitude [8] ing anisotropy. This corresponds via eq. (4) to
are less stringent. Further, if the static broadening I]Aa]l> 68 ppm. At the frequency employed by Old-
were purely quadrupolar, the coupling constant would field et al. the three interactions would combine
be at least 0.15 MHz [ 13 1. These estimates are rea- slightly differently, and the static spectrum would be
sonably consistent with a reported value of 0.13 MHz about 5.2 kHz wide (fwhm of equivalent Gaussian ),
T.J. Bastow, S.N. Stuart / “0 NMR in simple oxides 465
which is in agreement with the reported observation The centre bands have been identified from several
[ 41. Sidebands of the satellite transitions are too weak long runs at different rotational frequencies. Signal-
to be visible [ 8 1. to-noise ratio is inadequate for close analysis (spec-
The chemical shift in Tiz03 shows more shielding trum not shown).
than in TiOz (see section 4.4 below), which is possi- Pure tetragonal zirconia shows the same spectrum
bly due to the greater coordination at the oxygen, but (fig. 2g) as the Bruker spinners (marked Z in figs. 1
no very pronounced, paramagnetic shift. This sup- and 2) made of MgPSZ, which contains precipitates
ports the Goodenough view that the titanium elec- of tetragonal Zr02. The sidebands, being mostly nar-
tronic spins (3d’ ) are strongly paired in the ses- rower than the centre band (fig. 3)) are attributed to
quioxide [ 16,17 1. The resonance in Ti203 has about the first pair of satellites [ 81. Their uniform ampli-
the same width as the centre band in u-A&O,. The tude is indicative of a first-order quadrupolar pow-
isomorphic V203 showed no resonance. der pattern with an asymmetry parameter that is not
No resonances were observed in the bixbyite struc- too large. Their position is not significantly displaced
tures, Scz03, Y203, In203 (in which the oxygen at- from the centre band, which implies an upper bound
oms are not close packed). None of the three sites in to the quadrupole coupling constant of about 1.4 MHz
&Ga203 was detected. [ 8 ] ; other bounds, as in the case of ZnO above, are
less stringent. The full extent of the sideband pattern
4.4. Tetragonal and monoclinic dioxides has not been determined.
There are two oxygen sites in the monoclinic bad-
Two forms of TiOa show distinct chemical shifts, deleyite structure, one of tetrahedral and one of tri-
the oxygen being trigonally coordinated in both. Ru- gonal coordination. These show up clearly both in
tile, the denser of the two, produces the higher value monoclinic ZrOz (fig. 2f) and in HfQ, (fig. 2e). The
of 6. Our value for rutile is similar to that recently site of higher 6 may be the one of lower coordination
attributed to TiOz [ 5 1. The “0 quadrupole interac- and shorter bondlength, according to a well-known
tion in x-utile has been measured as e2qQ/h= principle [ 19,201. These lines are not found in the
1.497(4) MHz, ~=0.868(5) [ 18 J. Using those pa- MgPSZ (fig. 3).
rameters, the chemical shift correction in eq. ( 1) Oxygen shielding in the solid oxides of zirconium
amounts to 5.7 ppm, giving the corrected shift and hafnium is greater than has been found in their
&,= 594 ppm; the quadrupolar linewidth predicted by p-0x0 complexes in solution, where the coordination
eq. (2) is marginally greater than that observed. The is two fold [ 2 11. This also may be a matter of ionic-
sidebands (fig. 2d) are due to the central transition, ity, coordination and bondlength.
as in alumina, and their observed extent implies via
eq. (4)) if static dipole-dipole broadening is negligi- 4.5. Other oxides
ble, that llAall> 120 ppm. It would be surprising if
the shielding anisotropy were greater in r-utile than in In sodium p-alumina the strength of the observed
alumina, considering that the quadrupole interaction line (fig. 1f ) indicates that most of the “0 nuclei in
is smaller. The narrower line of anatase implies a the sample are contributing to the signal. The bulk of
smaller quadrupole coupling than in rutile; there are the material has the hexagonal structure of the ca-
no obvious sidebands, presumably because of an in- nonical Na20: 1 1A1203, which involves five distinct
sufficient magnetic dipole interaction. No reso- oxygen sites (with non-cubic point symmetry). We
nances were detected in a third form of TiOz, natural conclude that possibly two of the most populated sites
brookite, or in other oxides with the rutile structure, have (a) nearly equal chemical shift and (b ) small
Ge02, SnOz and Pb02. quadrupole coupling. The shift is not far from that of
In monoclinic VOz ( V204), a distorted rutile a-A&OS and other OAld sites [ 41.
structure, the oxygen atoms occupy two sites of three- V,O, (which has three oxygen sites) and CuO were
fold coordination. The spectrum exhibits two series examined without success.
of spinning sidebands without intense centre bands,
the stronger series extending from 400 to 1300 ppm.
466 T.J. Bastow, S. N. Stuart / “0 NMR in simple oxides
In fig. 4 the chemical shift of I70 in binary oxides The NMR of I70 can successfully be detected by
(M,O,) is compared with that of 19F in the corre- the MAS technique in well-ordered oxides, without
sponding fluorides (MF,). In this comparison, the special isotopic enrichment, provided the nuclear
valence of the electropositive element is the same in quadrupole coupling is less than about 4 MHz. A
a pair of compounds, while coordination about the promising range of industrially useful oxides are
electronegative atom is in general different. The col- found in this category. The resolution is quite suffl-
lection of 38 diamagnetic compounds ranges across cient to reveal the inequivalent sites. To obtain such
the periodic table from ionic solids (BaO, BaF,) to resolution, in cases where the quadrupole coupling is
associated liquids (H20, HF). A broadly positive much greater than that, would seem to call for the
correlation between the two shifts may be discerned, technique of specimen double rotation [ 71 or some
especially with regard to the oxidation states I to III. multiple-pulse equivalent. In a few cases the spectral
The effect may be compared with the chemical shift sidebands, which carry information about static
correlation between I70 in ethers ORR’ and 13Cin broadening mechanisms, can be interpreted.
the corresponding hydrocarbons CH,RR’ [ 22 1. The There is a broad correlation between the chemical
primary variation of both oxygen and fluorine shifts shift of I70 in binary oxides MzO, and that of 19Fin
is probably related to variation in the size of the an- the corresponding fluorides MF,, as might be ex-
ion; and it is possible that relative electronegativity pected of isoelectronic anions. The span of oxygen
has an additional but secondary effect. shifts is about three times larger than fluorine.
Interesting chemical shift effects are seen where the
same system exhibits different phases or inequiva-
lent sites. In zirconia-based ceramics “0 NMR MAS
clearly distinguishes the monoclinic, tetragonal and
cubic solution phases. The latter two are difficult to
separate by routine X-ray diffraction powder meth-
ods [ 231. For La203, Ti,03, Ti02, ZrOz and HQ
the results appear to be consistent with the general
rule [20] that 6 varies inversely with coordination.
l C But it is too early to say whether the shifts are fully
accounted for by the mechanism of overlap.
Acknowledgement
-100 0 100 200 300 We wish to thank I. Grey, R.H.J. Hannink, D. Hay,
IF
R.J. Hill, M. Hoang, H.J. Rossell, S.L. Segel and H.J.
Fig. 4. NMR chemical shift (ppm) of “0 in binary oxides MzO, Whitfield for generously providing us with well-char-
compared with that of “F in the corresponding fluorides MF,,,, acterized materials. We are further grateful to Henry
taking reference frequencies from Hz0 and C6FS, respectively. Rossell for enlightening discussions on the structural
Valence m is indicated by symbol, thus: I ( 0 ) , II ( 0 ), III ( A ), chemistry of zirconia ceramics.
IV (*) or VI (0). Values for Ag20, ZnO, PbO, A1203, La203,
ZrQ, HfO, are present measurements. Other values are taken
from published reports as follows: BeO, MgO, CaO, SrO, CdO,
BaO, HgO [3]; CO2 [24]; SiO2 [2]; SO1, SOa [25]; HF [26];
References
BeFz [27]; MgFz, ZnFz, HgF2 [28]; CaF2 [29-311; SrF2, CdFt,
BaF,[30,31];AgF,AlF3,HfF4[11];LaFj[32];ZrF.,[33];PbF2 [ 1] S. Schramm, R.J. Kirkpatrick and E. Oldfield, J. Am. Chem.
[31,34]; CF4, SiF,, SF,, SFs [35]. Conversions: &(F,)- Sot. 105 (1983) 2483.
&(C,F,)=589 ppm [11,32]; G,(CF,CO,H)-$(C,F,)= [ 21 S. Schramm and E. Oldfield, J. Am. Chem. Sot. 106 ( 1984)
84.4 ppm [29]. 2502.
T.J. Bastow, S.N. Stuart I “0 NMR in simple oxides 467
[ 31 G.L. Turner, SE. Chung and E. Oldfield, J. Magn. Reson. [ 191 R.G. Kidd, Can. J. Chem. 45 (1967) 605.
64 (1985) 316. [20] W.G. Klemperer,Angew.Chem. Intern. Ed. 17 (1978) 246.
I41 T.H. Walter and E. Oldfield, J. Phys. Chem. 93 ( 1989) 6744. [21] G.L. Hillhoise and J.E. Bercaw, J. Am. Chem. Sot. 106
{ 51E. Oldfield, C. Coretsopoulos, S. Yang, L. Reven, H.C. Lee, (1984) 5472.
J. Shore, O.H. Han, E. Ramli and D. Hinks, Phys. Rev. B [ 221 C. Delseth and J.P. Kintzinger, Helv. Chim. Acta 61 ( 1978)
40 (1989) 6832. 1327.
[ 61 E. Kundla, A. Samoson and E. Lippmaa, Chem. Phys. [ 231 A. Paterson and R. Stevens, J. Mater. Res. 1 ( 1986) 295.
Letters 83 ( 1981) 229. [ 241 J. Reuben, cited by A. Velenik and R.M. Lynden-Bell, Mol.
[ 71 B.F. Chmelka, KT. Mueller, A. Pines, J. Stebbins, Y. Wu Phys. 19 (1970) 371.
and J.W. Zwanziger, Nature 339 (1989) 42. [25] B.N. F&is, R.G. Kidd and R.S. Nyholm, Proc. Roy. Sot.
[ 81 A. Samoson, Chem. Phys. Letters I 19 ( 1985) 29. A 269 ( 1962) 469.
[ 91 D. Freude, J. Haase, J. Klinowski, T.A. Carpenter and G. [26] D-K. Hindermann and C.D. Comwell, J. Chem. Phys. 48
Ronikier, Chem. Phys. Letters 119 ( 1985) 365. (1968) 2017.
[ IO] J.S. Waugh, M.M. Maricq and R. Cantor, J. Magn. Reson. [27] R.E.J. Sears, J. Chem. Phys. 59 (1973) 5212.
29 (1978) 183. [28] R.W. Vaughan, D.D. Elleman, W.K Rhim and L.M. Stacey,
[ 111L.M. Avkhutsky, Yu.V. Gagarinsky, S.A. Polyshchuk and J. Chem. Phys. 57 (1972) 5383.
S.P. Gabuda, Spectrosc. Letters 2 ( 1969) 75. [29] T. Terao and T. Hashi, J. Phys. Sot. Japan 36 ( 1974) 989.
[ 121 T.J. Bastow, I.D. Campbell and H.J. Whitfield, Solid State [ 301 D.P. Burum. D.D. Eleman and W.K. Rhim, J. Chem. Phys.
Commun. 33 ( 1980) 399. 68(1978) 1164.
[ 131 N. Bloembergen and T.J. Rowland, Acta Metallurg. 1 [ 3 1] N. Boden, P.K. Kahol, A. Mee, M. Mortimer and G.N.
(1953) 731. Peterson, J. Magn. Reson. 54 (1983) 419.
[ 141 T.J. Bastow and S.N. Stuart, Phys. Stat. Sol. 145b (1988) [32] Yu.V. Gagarinskii and S.P. Gabuda, J. Struct. Chem. 11
719. ( 1970) 897.
[ 151 E. Brun, B. Derighetti, E.E. Hundt and H.H. Niebuhr, Phys. [ 331 S.P. Gabuda, Yu.V. Gagarinskii and A.G. Lundin, J. Struct.
Letters A 31 ( 1970) 416. Chem. 7 (1966) 192.
[ 161 R.M. Moon, T. Riste, W.C. Koehler and S.C. Abrahams, J. [34] R.E.J. Sears, Q.Zh. Guo and H.J. Mackey, J. Chem. Phys.
Appl. Phys. 40 ( 1969) 1445. 80 (1984) 5448.
[ 17 ] M.G. Vincent, K. Yvon, A. Griitter and J. Ashkenazi, Acta [ 351 E.L. Muetterties and W.D. Phillips, J. Am. Chem. Sot. 81
Cryst. A 36 ( 1980) 803. (1959) 1084.
[ 181 C. Gabathuler, E.E. Hundt and E. Brun, Magnetic
Resonance and Related Phenomena, Proceedings of the 17th
Congress Ampere, Turku, ed. V. Hovi (North-Holland,
Amsterdam, 1973) pp. 499-50 I.