Chirality - American Chemical Society

Chapter 12

Downloaded by UNIV OF MASSACHUSETTS AMHERST on May 31, 2018 | https://pubs.acs.org Publication Date: March 1, 2002 | doi: 10.1021/bk-2002-0810.ch012

X-ray Natural Circular Dichroism: Introduction and First Results Robert D. Peacock Department of Chemistry, University of Glasgow, Glasgow G12 8QQ, Scotland, United Kingdom

We have extended the measurement of Natural Circular Dichroism to the X-ray region for the first time by using the high circular polarization rates available from the helical undulator source Helios II at E S R F (Grenoble, France). X N C D spectra for two enantiomeric pairs of uniaxial single crystals have been measured: Na Nd(dig) .2NaBF .6H O and 2[Co(en) Cl ].NaCl.6H O. The X A N E S parts of the N d L and Co Κ edge X-ray absorptions show circular dichroism corresponding to chiral multiple scattering paths of the photoelectron. In addition both compounds show quadrupole allowed pre-edge features (2p --> 4f for Nd and 1s --> 3d for Co) which have exceptionally large Kuhn dissymmetry factors. 3

3

1

3

2

3

4

2

2,3

Dedicated to the memory of Agnes Louise Davis Peacock (1912-1999)

© 2002 American Chemical Society

Hicks; Chirality: Physical Chemistry ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

157

158


Introduction In 1895, the same year that Rôntgen reported the discovery of X-rays (i), Cotton coined (2) the term circular dichroism to describe the differential absorption of right and left circularly polarized light by a chiral substance He had previously measured the ellipticty induced in plane polarized light after it had passed through basic copper and chromium d-tartarate solutions and subsequently looked at the absorption of left and right circularly polarized sodium light by the same solutions. Since ellipticity and circular dichroism are simply related, and since, until relatively recently, it was much easier to measure ellipticity than circular dichroism, most C D experiments prior to the 1960's were made in this way The adoption of the Pockel's Cell in 1960 (3) and the photoelastic modulator in 1966 (4), together with developments in signal processing and photomultiplier technology, led to commercial C D instruments which operate between 1400 and 180 nm. This energy range expanded in the 1970*8 with the development of C D in the infra-red region (5, 6) and the vacuum ultraviolet region (7). At much the same time the complementary spectroscopies of Raman CID (8) and Circularly Polarized Luminescence (9) were developed. Before we began our excursions into the X-ray region, C D could be measured between roughly 1000 cm" in the infra-red region and 75,500 cm* in the vacuum uv region. Put another way, C D could be measured in vibrational spectra and in electronic transitions within valence states and between valence andRydberg orbitals. Magnetic circular dichroism (MCD) was developed in the 1970's (10) and was first measured in the X-ray region in 1987 (11). It is worth asking the question why M C D was measured twelve years ago and is now considered a routine X-ray spectroscopy, while the measurement of Natural C D had to wait till 1997. There are two main reasons: firstly M C D is generally a large effect the dissymmetry factor (g = ΔΑ/Α) often being between 10" and 10' . Estimates of the likely value of g for X N C D were variously put at between 10" and 10' (see below). Secondly the magnetic experiment has the considerable advantage that the sign changes with the direction of the magnetic field (and is zero in the absence of the field). This enables the experiment to be performed with a single hand of circularly polarized light and field reversal. N C D requires both hands of circularly polarized light, and both enantiomers and a racemate if (at least for the first experiments) a good baseline is to be obtained. The synchrotron radiation obtained from bending magnets varies from horizontal polarization at the centre of the beam to high rates of circular polarization at the top and bottom of the beam. However the beam intensity decreases from the centre to the edges, so it is only possible to obtain relatively high rates of circular polarization with low flux. To measure Natural C D , either 1

1

1

2

3


8


159 the sample must be physically moved up and down relative to the beam centre, or a moveable slit placed in the beam path to enable the top or bottom parts of the beam to be focussed on the sample. Clearly neither approach is suited to making the first measurements of what was considered to be a weak phenomenon. The helical undulator (12) built at ESRF in 1994-6 (13) produces a beam of right or left circularly polarized X-rays which is switched by altering the relative phase of the periodic magnetic arrays. Thus the position and focus of the beam on the sample is the same for both hands of light. This presented a real possibility of measuring X N C D . One previous attempt at measuring natural optical activity in the X-ray region should be mentioned. In 1990 Siddons et al measured (14) the rotation of plane polarized light at the Co /f-edge for powdered samples of Co(en) Br . Enantiomorphous spectra in the pre-edge region were obtained for the enantiomeric complexes but there is a considerable problem with interpreting the data (discussed in more detail below). 3

3

X-Ray Spectroscopy Figure 1 shows a schematic energy level diagram appropriate to the ϋΓ-edge spectrum of a first row transition metal ion and the L2 -edge spectrum of a lanthanide ion. In the lanthanide case, the principal absorption is the electric dipole allowed transition from the 2p orbital to the unoccupied continuum ed and es states. Some 10 eV to low energy of the electric dipole allowed transitions ("white line") is an electric quadrupole allowed transition from the 2p orbital to the bound 4f orbitals. This weak (pre-edge) transition has not been directly observed in absorption. It has been observed in the X M C D spectrum of G d (15) and in the resonant inelastic X-ray scattering of Y b metal (16) some 5 - 1 0 eV to low energy of the L edge. In the case of 3d transition metal AT-edge spectra, the electric dipole allowed absorption is Is -> ερ and the electric quadrupole Is —> 3d pre-edge feature, although weak, is well resolved from the edge and can be observed directly in the X-ray spectra of transition metal compounds with partially filled d orbitals, eg Co(III). f3

3 +

3

In addition to the atomic type transitions described above, X-ray spectra show considerable fine structure for several hundred eV above the edge. The region extending to around 50 - 100 eV above the edge is called X A N E S (X-ray Absorption Near Edge Structure) while that to higher energy is called E X A F S (Extended X-ray Absorption Fine Structure). In both cases the fine structure is caused by the ejected photo-electron being back-scattered by neighbouring atoms. The difference between E X A F S and X A N E S comes from the relation between the photoelectron de Broglie wavelength and the distance travelled by the photoelectron between scattering events.



160

Figure J: Schematic diagram of the energy levels of a lanthanide ion (left) and a 3d transition metal ion (right). The allowed electric dipole transitions (μ) are shown as bold arrows; the allowed electric quadrupole transitions (Q) as dotted arrows


161

The de Broglie wavelength decreases from around 2Â in the middle of the X A N E S region to less than around 0.5Â by 500 eV after the edge. The result of this is that E X A F S is dominated by single scattering events (ie the photoelectron is scattered by a single neighbour) while X A N E S has a much larger contribution from multiple scattering paths.


The Origins of X-ray C D The Rosenfeld equation relates the circular dichroism of a transition i —> f in a randomly oriented sample to the scalar product of its electric and magnetic dipole moments ( E l - M l mechanism)

R = Im(i\i\f).(f\m\i) In the case of an achiral chromophore in a chiral environment, such as a tris-chelated metal complex, first order perturbation theory may be employed. The zero order electric moment and magnetic moment arise from different transitions and are mixed by the chiral perturbation (V). Considering the case of the N d ion , and taking the electric dipole transition as 2p —» nd, 3 +

R « Im(2ρ |μ| nd)(nd |V| η'

ρ |m| 2p)

Unfortunately there is a considerable problem for the E l - M l mechanism in that the magnetic dipole transition is forbidden . The magnetic dipole selection rule I àé = 0, allows the transition from 2p to the np and continuum ερ states. Unfortunately m is a pure angular operator and cannot connect states which are radially orthogonal. This results in the I Δη| = 0 selection rule for bound states and also clearly forbids 2p —» ερ, except via core-hole relaxation. A n estimate can be made (16) of the contribution of core-hole relaxation from the relevant radial overlap integral, , which can be calculated to be approximately equal to 10' from the relevant hydrogenic wavefunctions. Remembering that typical values for the dissymmetry factor, | Δε/ε|, are 10" for a magnetic dipole allowed transition (for example the A i —» E( Ti) transition of [Co(en) ] or the η —> π* transition of a ketone) or 10' for an electric dipole allowed transition (for example a charge transfer transition) we can estimate that the maximum value of the dissymmetry factor for the 2p —> εd transition from the E l - M l mechanism is likely to be around 10" for a metal complex. This is far too small to be measured using the circular polarization rates available above and below the orbit plane of radiation coming from conventional bending magnets, and indeed is on the limit of what might be measurable using radiation from a helical undulator. 2

2

l

l

l

3

3

5


3+

162 There is an additional problem with the E l - M l mechanism for AT-edge spectra or L edge spectra. Magnetic dipole transitions are forbidden from s orbitals so the only possible source of magnetic dipole intensity involves Is - np orbital mixing. Two attempts at calculating the E l - M l contribution to N X C D have been published. Alagna et al (18) calculated the dissymmetry factors for the carbon £-edge spectra of propylene oxide to be of the order of 10" . This relatively large value can be attributed to the closeness of the Is and 2p orbitals in carbon compared to the Is and 4p orbitals in a transition metal ion. Goulon et al (19) calculated the sulfur Κ and L edge X N C D for the chiral conformation of H S . Dissymmetry factors were found to be around 10" for the Κ edge and 10* for the 1^,3 edges. Fortunately, for oriented systems, there is an additional source of circular dichroism intensity via the electric dipole-electric quadrupole (E1-E2) mechanism. This mechanism was first developed by Chiu (20), elaborated by Buckingham and Dunn (21) and applied by Barron (22) to explain the C D of the A E ( T ) transition in the axial crystal spectrum of [Co(en) ] . This transition, which is magnetic dipole forbidden and electric quadrupole allowed was the only case, until the present work, where the E1-E2 mechanism had been used to account for C D . x


3

2

8

l

{

l

!

2

4

2

3

3+

Applying the E1-E2 mechanism to N d we have, for the pre-edge 2p - » 4f, electric quadrupole allowed, transition, and assuming the principal source of electric dipole intensity is the 2p —> nd transitions: 3 +

R~{2p\Q\4f)(4f\V\nd)(nd^\2p) Conversely the C D of the main, electric dipole allowed transition 2p -> nd is given by

/?-(2ρ|μ|^>(^|ν|4/)(4/|0|2ρ)

Choice of samples Based on the experience of visible/uv C D and the expectation that the E l E2 mechanism, which is only applicable to oriented species, might be significant, we decide to perform our first measurements on oriented single crystals. The choice of our first sample, Na Nd(digly)3.2NaBF .6H 0, was dictated by the energy at which E S R F beamline ID12A had the maximum circular polarization. In fact, as will be seen later, the choice was an excellent one The crystal structure of Na Nd(digly) .2NaBF .6H 0 had not been 3

3

3

4

4

2


2

163 determined (although that of the [C10 ]" analogue had been) and so we determined (23) the crystal structure. This is shown in Figure 2. The complex is a tris-chelate in which the N d ion is coordinated by 6 carboxylate and 3 ether oxygen atoms. Included in the picture are the 6 nearest neighbour water molecules which are connected to the anion by hydrogen bonds. (The chirality of a tris chelate is designated Λ if the three chelate ligands form a left handed screw or Δ if they form a right handed screw.) 4


3 +

Figure 2; View of the A-[Nd(digly) ] ' anion (and associated water molecules) perpendicular to the C$ axis 3

3

Having established the viability of measuring X N C D we chose 2[Co(en) Cl3].NaC1.6H 0 as our second subject of investigation. The [Co(en) ] ion has been of historic importance in the development of transition metal optical activity. A crystal of 2[A-Co(en) Cl ].NaC1.6H 0 provided the first transition metal complex to have its absolute configuration determined by the anomalous scattering of X-rays (24). The same compound was the first transition metal complex to have its electronic C D spectrum measured in both the solid state and in solution (25, 26) and as such has been the complex of choice for testing the various theories of transition metal C D in the visible region (27 - 30). In addition, the presence of a well-resolved pre-edge (Is - » 3d) feature some 18 eV to low energy of the Κ edge absorption and the expectation that the E1-E2 mechanism would be dominant, made [Co(en) ] a natural choice. The structure of 2[A-Co(en) Cl ].NaC1.6H 0 was originally determined by 2-D 3

3

2

3+

3

3

2

3

3

3

3+

2


164 crystallography some 40 years ago. Accordingly we have redetermined (31) the structure.


Results and Discussion

Na Nd(digly) .2NaBF .6H 0 3

3

4

2

Figure 3 shows the Nd L edge absorption spectrum and the X N C D spectra (multiplied by 100) of enantiomorphic single crystals of Na Nd(digly)3.2NaBF4.6H 0 (32). The spectra clearly show that we have measured C D in the X-ray region of the spectrum. A s expected, the enantiomorphic single crystals have mirror image circular dichroism spectra. The intensity of the C D is much larger than that expected from the E l - M l mechanism, the dissymmetry factor in the X A N E S region being - 3 χ 10" , and suggests that the intensity comes from the E1-E2 term. This is confirmed by the disappearance of the C D in the powder spectra as required by this mechanism 3

3

2

3

3 r

-1 I 6190

1

1

1

I

I

i

6210

6230

6250

6270

6290

6310

Energy/eV Figure 3: Axial absorption and CD* 100 spectra for single crystals of A(dark) and 4- (light) Na Nd(digly)s.2NaBF .6H 0 3

4

2

The X N C D spectra are best divided into three regions for discussion: the quadrupole allowed 2p -> 4f pre-edge transition, the white line/ X A N E S region and the E X A F S region.


165 A major feature of the spectra is a strong dichroism (g is at least 7xl0" , and is probably considerably larger, since the transition cannot be independently detected in the absorption spectrum) exactly where the predicted electric quadrupole-allowed 2p —> 4f transitions are expected. The position of the transition some 9 eV to low energy of the white line is similar to that of equivalent transitions in other rare earths (15, 16). Enantiomeric features with g ~ 10" are also seen under the white line and in the X A N E S region of the spectrum with weaker features evident in the E X A F S . The X A N E S region of the spectrum has a significant contribution from multiple scattering. Single scattering paths cannot be chiral by definition and therefore cannot contribute to the dichroism (the theory of chiral multiple scattering is described in detail in the following chapter). The C D spectrum, therefore, singles out multiple scattering pathways, which explains why the dichroism is relatively stronger in X A N E S than in E X A F S . We have calculated the C D spectrum using the 3


3

g

-0.015 |-0.02 -40

-20

0

20

40

60

80

100

Energy /eV Figure 4: Comparison between the experimental and calculated XNCD signals, normalised to the respective atomic absorptions. The experimental curve has been multiplied by a factor of 4.5.

multiple scattering formalism. A comparison of experiment and calculation is shown in Figure 4. The result is an excellent agreement. A Fourier transform of the X A N E S C D gives distances of 7 and 8.2 Â while the equivalent analysis of the absorption X A N E S gives peaks at about 2.5 and 3.5À . Thus while the absorption has a major contribution from the single scattering paths involving the Nd and the ligating oxygen atoms (2.433 , 2.535Â) and the Nd and the first carbon neighbours (3.51Â) the C D is entirely due to multiple scattering. The


166 two chiral paths which best fit the data are N d - O(carboxyl) - C(carboxyl) (7.05 Â ) and Nd - O(carboxyl) - 0(carboxyl adjacent ligand) (8.25 Â ) .

2[Co(en) Cl ].NaCL6H 0 3

3

2

Figure 5 shows the absorption and X N C D spectra of axial single crystals of 2[Λ- and A-Co(en) Cl ].NaC1.6H 0 (33) . The most noticeable feature of the spectra is the spectacular size of the Is - » 3d C D observed below the C o Κ edge. The Kuhn dissymmetry factor, g = (ΔΑ/Α), is 12.5% i f the raw C D and absorption are used. A s expected, a racemic crystal gave no X N C D signal. To attempt to explain this dissymmetry factor, which is comparable to that of the magnetic dipole allowed A i - » A ( T ) and E(T ) d d transitions, we


3

3

2

]

]

2

l

lg

lg

1.50 r

7690

7700

7710

7720

7730

7740

7750

7760

7770

Energy/eV Figure 5: Axial absorption and XNCD spectra of 2[Co(en) Cl ].NaCL6H 0 - the dark curve is the CD of the Δ crystal, the light curve that of the A crystal 3

3

2

extended the ab initio approach to the calculation of core-valence C D for the first time. In order to investigate the efficacy of the E1-E2 mechanism for the Is - » 3d transition, we have performed both frozen core and relaxed core H F calculations in a Gaussian orbital basis (33). Such calculations of the absorption spectra of transition metal complexes by ab initio methods are rare (34) and there is only one example of an ab initio calculation of transition metal d *Α (Τι ) and E(Ti ) valence transitions were reproduced satisfactorily, with similar agreement to experiment to that found in references 30 and 35. The results confirm the importance of the quadrupole-dipole interference term in the mechanism of the X N C D in this oriented crystal. The sign of the Is —> 3d preedge signal is correctly reproduced as is the order of magnitude of the Kuhn dissymmetry factor ( g i = 0.11, g = 0.125; = ^ . S x l O ^ c g s u , R bs=8.7X10* cgsu). The source of electric dipole transition moment for the pre-edge C D is approximately 97% Co-based, as expected for a transition emanating from the Is core orbital. 3

3

3


l

3

2

2

l

β

g

ca

c

obs

0

44

The large magnitude of the C D in the pre-edge excitation is satisfactorily explained by the E1-E2 mechanism which is particularly efficient in this system. The g factor for Is —» 3d is the same order of magnitude as that of the d

Chirality - American Chemical Society

Recommend Documents