How optical excitation controls the structure and properties of vanadium dioxide

Otto, Martin R.; René de Cotret, Laurent P.; Valverde-Chavez, David A.; Tiwari, Kunal L.; Émond, Nicolas; Chaker, Mohamed; Cooke, David G.; Siwick, Bradley J.

doi:10.1073/pnas.1808414115

How optical excitation controls the structure and properties of vanadium dioxide

Martin R. Otto https://orcid.org/0000-0002-3884-8424 [email protected], Laurent P. René de Cotret, David A. Valverde-Chavez, +4 , Kunal L. Tiwari, Nicolas Émond, Mohamed Chaker, David G. Cooke, and Bradley J. SiwickAuthors Info & Affiliations

Edited by Richard Averitt, University of California, San Diego, Del Mar, CA, and accepted by Editorial Board Member Zachary Fisk November 12, 2018 (received for review May 15, 2018)

December 26, 2018

116 (2) 450-455

https://doi.org/10.1073/pnas.1808414115

PDF/EPUB

Significance

Revealing how the complex interplay between charge, spin, orbital, and lattice-structural degrees of freedom gives rise to emergent properties is central to materials research. Using external perturbations to modify this interplay and control materials is at the forefront of these efforts. Here we use femtosecond laser excitation to stimulate two independent insulator-to-metal transformations in a strongly correlated material:

{V O}_{2}

. One transformation involves a change in lattice structure familiar from the equilibrium phase diagram; the other one involves a change in electronic structure that has no equilibrium analog. We use a multimodal approach to directly watch the

{V O}_{2}

unit cell reorganization that accompanies these optically induced transformations and directly determine the impact that these structural transformations have on electronic transport properties.

Abstract

We combine ultrafast electron diffraction and time-resolved terahertz spectroscopy measurements to link structure and electronic transport properties during the photoinduced insulator–metal transitions in vanadium dioxide. We determine the structure of the metastable monoclinic metal phase, which exhibits antiferroelectric charge order arising from a thermally activated, orbital-selective phase transition in the electron system. The relative contribution of the photoinduced monoclinic and rutile metals to the time-dependent and pump-fluence–dependent multiphase character of the film is established, as is the respective impact of these two distinct phase transitions on the observed changes in terahertz conductivity. Our results represent an important example of how light can control the properties of strongly correlated materials and demonstrate that multimodal experiments are essential when seeking a detailed connection between ultrafast changes in optical-electronic properties and lattice structure.

The insulator–metal transition (IMT) in vanadium dioxide (

{V O}_{2}

) is a benchmark problem in condensed-matter physics (1–4), as it provides a rich playground on which lattice-structural distortions and strong electron correlations conspire to determine emergent material properties. The equilibrium phase diagram of pure

{V O}_{2}

involves a high-temperature tetragonal (rutile,

R

) metal that is separated from several structurally distinct low-temperature insulating phases (monoclinic

M_{1}

,

M_{2}

, and triclinic

T

), depending on pressure or lattice strain. The transition to lower-symmetry insulating phases occurs in the vicinity of room temperature and is sensitive to doping (Cr and W), making

{V O}_{2}

interesting for a range of technological applications (5–7). There has been a lively discussion in the literature about the driving force responsible for the IMT in

{V O}_{2}

and the nature of the insulating and metallic phases that has revolved around the role and relative importance of electron–lattice and electron–electron interactions. The stark dichotomy between Peierls (2) and Mott (8) pictures characterizing the earliest explanations has recently given way to a nuanced view that the insulating phases of

{V O}_{2}

are nonstandard Mott–Hubbard systems where both electron–lattice and electron–electron interactions play important roles in determining the electronic properties of all of the equilibrium phases (9–14).

Photoexcitation using ultrafast laser pulses has provided another route to initiate the transition between the insulating and metallic phases of

{V O}_{2}

since it was discovered that the IMT occurs very rapidly following femtosecond laser excitation with sufficient fluence (15). Since this discovery,

{V O}_{2}

has been the focus of many time-resolved experiments including X-ray (16, 17) and electron (18–20) diffraction, X-ray absorption (21, 22), photoemission (23), and optical spectroscopies (24–32) from terahertz to UV aimed at uncovering the connection between the photoinduced IMT and changes in lattice structure.

Recently, ultrafast electron diffraction (UED) and midinfrared spectroscopy were combined to show that there are two distinct photoinduced IMTs in

M_{1} {V O}_{2}

(19). The first one, accessible at relatively high pump fluence, is an analog of the equilibrium IMT and is associated with the lattice-structural transition between

M_{1}

and

R

crystallography expected from the equilibrium phase diagram. The second one, accessible at lower pump fluence, has no equilibrium analog and yields a metastable, structurally distinct monoclinic metal phase (

M

) that retains the crystallographic symmetry of its parent equilibrium monoclinic phase. Here, UED and time-resolved terahertz spectroscopy (TRTS) are combined (Fig. 1) to perform a detailed structure–property correlative study of the photoinduced phase transitions in

{V O}_{2}

. UED measurements are used to determine the real-space structure of the transient metastable

M

phase (Structure of the Monoclinic Metallic Phase). The fluence dependence of the

M

- and

R

-phase fractions that form the heterogeneous multiphase specimen following photoexcitation is measured and shown to directly correspond with the TRTS measurements (Fluence Dependence). From this correspondence we determine the low-frequency terahertz conductivities of both photoinduced

M

and

R

phases. The kinetics of the

M_{1} \to M

phase transition are consistent with thermal activation driven by electron temperature with an activation energy of

304 \pm 109

meV (Activation Energy and Kinetics). Information on the free energy landscape in

{V O}_{2}

and its dependence on structural distortions and orbital occupancy is also discussed. Our results provide a unified view of the photoinduced structural phase transitions in

{V O}_{2}

and their relationship to changes in the low-frequency terahertz conductivity.

Fig. 1.

Multimodal pump–probe experiments performed on the photoinduced phase transitions in ${V O}_{2}$ . (A) Ultrafast electron diffraction. The 50-nm-thick polycrystalline ${V O}_{2}$ sample is photoexcited by a 35-fs 800-nm optical pump pulse. The probe electron pulse has a temporal full width at half maximum of $\sim$ 100 fs and arrives after a precisely controlled time delay. Diffraction patterns showing Debye–Scherrer rings are collected by a charge-coupled device (CCD) camera for each time-delay value. A typical azimuthally averaged diffraction pattern is shown at *Right*, identifying the Bragg peaks for ${V O}_{2}$ most discussed in this work: (200), (220), and (30 $\bar{2}$ ). (B) The pump-induced difference in diffracted intensity up to 10 ps that follows photoexcitation at 23 mJ/ ${c m}^{2}$ . Blue (dashed) and red (solid) arrows indicate important aspects of the slow and fast processes, respectively. (C) Time-resolved terahertz (THz) spectroscopy. The electric-field waveform of the transmitted, single-cycle multi-THz pulse with spectral components spanning 1–30 THz is measured in the absence, $E_{ref} (t)$ , and presence, $E_{pump} (t, τ)$ of a photoexcitation at pump–probe delay time $τ$ . The differential $Δ E (t, τ) = E_{ref} (t) - E_{pump} (t, τ)$ and $E_{ref} (t)$ are Fourier transformed and the resulting amplitude and phase spectra are used to extract the dynamically changing complex optical conductivity $\tilde{σ} (ω, τ) = σ_{1} (ω, τ) + i σ_{2} (ω, τ)$ . (D) Change in the real part of the optical conductivity ( $σ_{1} (ω, τ)$ ) from 1 THz to 16 THz computed from the differential THz waveforms.

Experimental Results

UED measurements of pulsed laser-deposited 50-nm

{V O}_{2}

films (optical depth is

\sim 130

nm at 800-nm wavelength) reveal rich pump-fluence–dependent dynamics up to the damage threshold of

\sim

40 mJ/

{c m}^{2}

(35 fs, 800 nm,

f_{rep} = 50 - 200

Hz). Fig. 1A shows a typical baseline-corrected, 1D powder diffraction pattern for equilibrium

{V O}_{2}

in the

M_{1}

phase and identifies the (200), (220), and (30

\bar{2}

) peaks (33). The (30

\bar{2}

) peak acts as an order parameter for the

M_{1} \to R

transition, since it is forbidden by the symmetry of the

R

phase, while (200) and (220) peaks are present in all equilibrium phases. Consistent with previous work (19), the pump-induced changes to diffracted intensity (Fig. 1B, 23 mJ/

{c m}^{2}

) indicate two distinct and independent photoinduced structural transformations. The first one is a rapid (

τ_{(30 \bar{2})} \approx 350

fs) nonthermal melting of the periodic lattice distortion (dimerized V–V pairs) present in

M_{1}

, evident in Figs. 1B and 2A as a suppression of the (30

\bar{2}

) and related peak intensities. The second one is a slower (

τ_{(200), (220)} \approx 2

ps) transformation associated with a significant increase in the intensity of the (200), (220), and other low-index peaks whose time dependence is also shown in Figs. 1B and 2A. As we will show, at low pump fluences (

\sim

4–8 mJ/

{c m}^{2}

) the slow process is exclusively observed, while at high pump fluences (

> 35

mJ/

{c m}^{2}

) the fast process dominates. Note that these structural transitions are independent; the slow process does not follow the fast process.

Fig. 2.

Comparison of UED and TRTS measurements. (A) The red triangles show the suppression of the (30 $\bar{2}$ ) peak ( $\sim 350$ fs, red line) associated with the $M_{1} - R$ transition (Fig. 1). The blue circles show the intensity increase of the (200) and (220) peaks associated with the slow reorganization ( $\sim 2$ ps) and the formation of the $M$ phase. (B) The transient change in THz conductivity (spectrally integrated from 2 THz to 6 THz, shown as gray circles) is well described by a biexponential function composed of both fast ( $\sim 350$ fs, red line) and slow ( $\sim 2.5$ ps, blue line) time constants, in quantitative agreement with those of the structural transitions observed with UED.

Complementary TRTS measurements were performed on the same samples under identical excitation conditions to determine the associated changes in the time-dependent complex conductivity,

\tilde{σ} (ω, τ)

(Fig. 1 C and D). The pump-induced changes in real conductivity,

σ_{1} (ω, τ)

, over the

\sim

2- to 20-THz frequency range (Fig. 1D) also exhibit fast (

Δ σ_{1}^{fast}

) and slow (

Δ σ_{1}^{slow}

) dynamics, consistent in terms of timescales and fluence dependence with those described above for the UED measurements and similar measurements performed on sputtered

{V O}_{2}

films in the 0.5- to 2-THz window (28). Additional structure at higher frequencies is due to optically active phonons associated with O-cage vibrations around V atoms (25). We connect the observed THz response to the two structural transformations by focusing on the integrated spectral region from 2 THz to 6 THz, which includes exclusively electronic contributions to the conductivity (Drude-like) and omits phonon resonances (25–27). Fig. 2B shows an example of the transient real conductivity measured at 22 mJ/

{c m}^{2}

along with the fast and slow exponential components plotted individually. These time constants are in excellent agreement with those of the fast and slow processes determined from the UED measurements (Fig. 2A), over the entire range of fluences investigated.

Structure of the Monoclinic Metallic Phase.

Since its discovery (19), the structure of the photoinduced

M

phase and its relationship to the parent

M_{1}

phase have remained unclear. Here we use measured UED intensities to determine the changes in the electrostatic crystal potential,

Φ (x)

, associated with the transformation between

M_{1}

and

M

phases. The centrosymmetry of the monoclinic and rutile phases provides a solution to the phase problem (34) and allows for the reconstruction of the full 3D real-space electrostatic potential from each 1D diffraction pattern obtained using UED. Our analysis procedure is described in detail in SI Appendix.

Fig. 3 shows slices of

Φ (x)

for

{V O}_{2}

in the

R

(Fig. 3B) and

M_{1}

(Fig. 3C) phases obtained using this procedure. The autocorrelation of

Φ (x)

is in quantitative agreement with the Patterson function computed directly from the UED data (SI Appendix, Fig. S9). The slices shown are aligned vertically along the rutile

c_{R}

axis and horizontally cut the unit cell along

a_{R} + b_{R}

as indicated in the 3D structural model of

M_{1} {V O}_{2}

(Fig. 3F). In this plane, adjacent V chains are rotated by

90 °

, with dimers tilting either in or orthogonal to the plane of the page as indicated in Fig. 3 C and F. The lattice parameters obtained from these reconstructions are in excellent agreement with published values for the two equilibrium phases (Fig. 3A).

Fig. 3.

Real-space view of the photoinduced changes to $Φ (x)$ in ${V O}_{2}$ in a plane spanned by $c_{R}$ and $a_{R} + b_{R}$ . (A) Line cuts of $Φ (x)$ along the red and blue dashed lines shown in B and C, respectively. (B) $Φ (x)$ for equilibrium $R$ -phase ${V O}_{2}$ showing the absence of V–V dimerization and tilting. (C) $Φ (x)$ for equilibrium $M_{1}$ -phase ${V O}_{2}$ showing V–V dimerization and tilting illustrated by solid black (in-plane tilt) and red (out-of-plane tilt) lines which are also illustrated in the 3D structure (F). The line cuts in A show the expected undimerized/dimerized V–V lengths of 2.9 Å/2.6 Å, respectively. (D) $Δ Φ (x)$ for $M$ -phase ${V O}_{2}$ computed 10 ps after photoexcitation at a fluence of 6 mJ/ ${c m}^{2}$ . V–V dimerization from C is preserved and O atoms nearest to V atoms display an increase in $Φ (x)$ resulting in antiferroelectric order up the $c_{R}$ axis indicated by arrows. (E) Line cut of $Δ Φ (x)$ shown in D along the dashed black line which intersects a chain of O atoms. The antiferroelectric order is seen as an increase in $Φ (x)$ on alternating O atoms. (F) Three-dimensional structure of ${V O}_{2}$ showing V/O atoms as red/gray spheres. In-plane (out-of-plane) V–V dimers are connected by black (red) lines. (G) Three-dimensional structure of ${V O}_{2}$ looking down the $c_{R}$ axis.

In Fig. 3D the changes in

Φ (x)

associated with the

M_{1} - M

transition are revealed. This map is computed from the measured

Δ I_{G}

between the

M

and

M_{1}

phases 10 ps after photoexcitation at 6 mJ/

{c m}^{2}

. The preservation of

M_{1}

crystallography is clear, i.e., V–V dimerization and tilting along the

c_{R}

axis. Also evident is the transition to a 1D antiferroelectric charge order along

c_{R}

. In the equilibrium phases all O atoms are equivalent, but the

M

phase exhibits a periodic modulation in

Φ (x)

at the O sites along the

c_{R}

axis indicated by arrows in Fig. 3D. This modulation is commensurate with the lattice constant (Fig. 3 C–E). The O atoms exhibiting the largest changes are those associated with the minimum V–O distance in the octahedra and, therefore, the V–V dimer tilt. This emphasizes the importance of the lattice distortion present in the parent

M_{1}

phase to the emergence of the

M

phase. The antiferroelectric lattice distortion in

M_{1}

was already emphasized by Goodenough (2) in his seminal work on

{V O}_{2}

. Significant changes in electrostatic potential are also visible between V atoms in the octahedrally coordinated chains along

c_{R}

that are consistent with a delocalization or transfer of charge from the V–V dimers to the region between dimers. All of these observations suggest that the

M

phase emerges from a collective reorganization in the electron system that preserves the monoclinic lattice distortion.

Fluence Dependence.

We have established that there are two qualitatively distinct ultrafast photoinduced phase transitions in

{V O}_{2}

. The pump-fluence dependence of the sample response, specifically the heterogeneous character of the film following photoexcitation (due to both

M_{1} \to M

and

M_{1} \to R

transformations) and the corresponding changes in conductivity, is addressed in this section. UED intensities report on the fluence dependence of both structural phase transitions (Fig. 4A). The change in the (

30 \bar{2}

) peak intensity provides an order parameter exclusively for the

M_{1} \to R

transition, while the (200) and (220) peak intensities report on both

M_{1} \to M

and

M_{1} \to R

transformations (SI Appendix). Measurements of the (30

\bar{2}

) peak intensity (Fig. 4A, red triangles) clearly demonstrate a fluence threshold of

\sim 8

mJ/

{c m}^{2}

for the

M_{1} \to R

transformation that is consistent with previous work (16, 18, 19). Above this threshold the suppression of the (30

\bar{2}

) peak increases approximately linearly with fluence up to a magnitude greater than

75 %

at

\sim 30

mJ/

{c m}^{2}

, a result that is inconsistent with a “two-step” model that involves fast V–V dimer dilation followed by slow dimer rotation. Complete V–V dimer dilation yields a maximum (30

\bar{2}

) peak suppression of

50 %

. Instead, these data are consistent with a picture where the periodic lattice distortion of the

M_{1}

phase simply melts in

\sim 350

fs. The photoinduced fraction of

R

-phase

{V O}_{2}

reaches

\sim 75 %

of the film on this timescale at the highest pump fluences reported. The (200) and (220) peaks (Fig. 4A and SI Appendix, Fig. S3, blue and green circles) show a more complicated fluence dependence, reaching a maximum change in intensity in the 20 mJ/

{c m}^{2}

range. At the highest excitation fluences reported the intensity changes in the (200) and (220) peaks correspond to the relative increase expected for the

R

phase compared with the

M_{1}

of

{V O}_{2}

. The maximum at

\sim 20

mJ/

{c m}^{2}

is entirely due to the presence of the

M

phase as we demonstrate by converting the changes in UED intensities to phase volume fractions (SI Appendix).

Fig. 4.

(A) Phase volume fractions for the $M$ (blue line) and $R$ (red line) phases calculated using the UED data shown via the volume phase fraction model (*SI Appendix*). Red triangles are the ( $30 \bar{2}$ ) data points and blue circles are the (200) data points (scaled for clarity) with the $M_{1} \to R$ phase transition contribution subtracted. (B) Fluence dependence of the fast $Δ σ_{1}^{fast}$ and slow $Δ σ_{1}^{slow}$ components of the transient terahertz optical conductivity (red and blue, respectively). Solid lines serve as a guide to the eye. $Δ σ_{1}^{fast}$ increased steadily and corresponds to the formation of $R$ crystallites and $Δ σ_{1}^{slow}$ attains a maximum at 25 mJ/ ${c m}^{2}$ and corresponds to the formation of $M$ crystallites. Error bars represent the error given by the fitting routine (*SI Appendix*).

We denote

F_{R}

as the phase volume fraction for the

R

phase and

F_{M}

that for the

M

phase. The results of the model are shown in Fig. 4A. For fluences below the structural IMT fluence threshold of

\sim 8

mJ/

{c m}^{2}

, we observe clearly that only a small percentage (

\sim 10 %

) of

M

crystallites have been formed by photoexcitation. As the fluence increases, the photoexcitation of

R

crystallites begins at the threshold and increases roughly linearly afterward. The

M

phase achieves a maximum in the vicinity of 20 mJ/

{c m}^{2}

(consistent with Fig. 4A), where we determine that

F_{M} = 45 \pm 13 %

. At greater fluences,

F_{M}

decreases as the material becomes increasingly

R

phase due to stronger photoexcitation. The data points shown in Fig. 4A as blue circles are an average of the (200) and (220) data points from SI Appendix, Fig. S3 with the contribution from the

M_{1} \to R

phase transition subtracted (SI Appendix).

We obtain quantitatively consistent results for the fluence dependence of the transient conductivity obtained by TRTS, firmly establishing a link between the differential structural and differential electronic responses. Fig. 4B shows the fluence dependence of the fast (

Δ σ_{1}^{fast}

) and slow (

Δ σ_{1}^{slow}

) conductivity terms. The

Δ σ_{1}^{fast}

component corresponds to the conductivity response associated with the transition from

M_{1} \to R

as it increases steadily with fluence in accordance with

F_{R}

shown in Fig. 4B. Furthermore, we clearly observe that

Δ σ_{1}^{slow}

achieves a maximum at a fluence of

\sim

20 mJ/

{c m}^{2}

beyond which it decreases, consistent with the behavior of the (200) and (220) diffraction peaks and

F_{M}

shown in Fig. 4A. By analyzing the conductivity terms in an effective medium model, we find good agreement for

F_{M}

in the metallic limit (SI Appendix).

Activation Energy and Kinetics.

The

M_{1} \to R

and

M_{1} \to M

transitions exhibit qualitatively different kinetic behavior as evidenced by the fluence dependence of the time constants

τ_{fast}

and

τ_{slow}

obtained from both UED and TRTS (Fig. 5). The time constant for the

M_{1} \to R

transition (

τ_{fast}

), captured by the dynamics of the (30

\bar{2}

) peak in the UED measurements and by

τ_{fast}

in the THz measurements, is

350 \pm 100

fs independent of fluence. This demonstrates that the photoinduced

M_{1} \to R

transition—the melting of the periodic lattice distortion—is nonthermal and barrier-free in the kinetics sense (the transition rate is independent of the drive). The results for the

R

-phase volume fraction (Fig. 4), however, also show that the excitation threshold for this nonthermal phase transition is heterogeneous in the films. This last fact was also previously observed using ultrafast electron microscopy (35) and by others using nanoscopy (36, 37). The

M_{1} \to M

time constant, conversely, decreases significantly with pump fluence as seen in the (200) and (220) peak dynamics from UED and the

τ_{slow}

from TRTS. The exponential increase in the

M_{1} \to M

rate with excitation energy deposited in the electron system strongly suggests that the

M_{1} \to M

is an activated process. We can extract the activation energy

E_{A}

from these data by determining the electronic excitation energy per unit cell

E

which is deposited in the sample as a function of pump fluence (SI Appendix). Invoking the Eyring–Polanyi equation from transition state theory (38)

\ln (\frac{h τ_{slow}^{- 1}}{E}) = - \frac{E_{A}}{E} + \frac{Δ S^{‡}}{k_{B}},

[1]

where

Δ S^{‡}

is the entropy of activation. We take the values of

τ_{slow}

for the (200) peak shown in Fig. 5 and plot

\ln (h τ_{slow}^{- 1} E)

vs.

1 E

, which is shown in Fig. 5, Inset. By fitting to Eq. 1, we determine a lower limit value of

E_{A} = 304 \pm 109

meV. The deposited energy

E

may be related to the electron temperature

T_{e}

through an electronic heat capacity model (SI Appendix), which yields activation energies

E_{A}

ranging from

\sim 300

meV to 800 meV for photoexcited carrier densities of 0.1–

2 e^{-} /

unit cell. This describes a fundamental property of the photoinduced

M_{1} \to M

transition. Furthermore, the fluence required to deposit

E_{A}

per unit cell is

F \approx 3.7

mJ/

{c m}^{2}

, which is the value previously attributed to the

M_{1} \to R

IMT threshold (15, 16, 25). This is also in agreement with the fluence threshold extracted from the low-fluence data points in Fig. 4A.

Fig. 5.

Time constants for the photoinduced phase transitions in ${V O}_{2}$ . Shown are time constants for the slow [(200), blue; (220), green] and fast (30 $\bar{2}$ , dark red) peak dynamics as determined from the UED data. Not all fluences were carried out with sufficient time resolution to resolve the fast dynamics, and thus certain data points are not shown. TRTS-measured time constants pertaining to the slow/fast components of $σ_{1}$ are depicted by gray/red diamonds, respectively. The red shaded area represents the temporal region ( $350 \pm 150$ fs) associated with the photoinduced structural phase transition from $M_{1} \to R$ which dominates at high pump fluence. The solid blue line is an exponential fit to the (200) peak time constants $τ_{slow}$ . *Inset* shows the plot of $\ln (h τ_{slow}^{- 1} / E)$ vs. inverse energy deposited per unit cell $1 / E$ using values of $τ_{slow}$ . The solid line is a fit to Eq. 1, from which the activation energy $E_{A}$ and entropy $Δ S^{‡}$ are determined. The error bars represent the SE determined from the fitting described in *SI Appendix*.

Discussion and Conclusion

We have demonstrated that photoexcitation of

M_{1} {V O}_{2}

yields a complex, heterogeneous, multiphase film whose structure and properties are both time and fluence dependent. The character of the fluence-dependent transformation is summarized in Fig. 6. At pump fluences below

\sim 4

mJ/

{c m}^{2}

there is no long-lived (

> 1

ps) transformation of the

M_{1}

structure, and

{V O}_{2}

behaves like other Mott insulators insofar as optical excitation induces a relatively small, impulsive increase in conductivity followed by a complete recovery of the insulating state (19, 23, 26, 28). Above

\sim 4

mJ/

{c m}^{2}

, however, photoexcitation stimulates a phase transition in the electron system that stabilizes metallic properties through an orbitally selective charge reorganization: the

M

phase. Between

\sim

4 mJ/

{c m}^{2}

and 8 mJ/

{c m}^{2}

photoexcitation exclusively yields the

M

phase, which populates 15–20

%

of the film by 8 mJ/

{c m}^{2}

. In this fluence range, time-resolved photoemission experiments show a complete collapse of the band gap (23), TRTS experiments show a dramatic increase in conductivity (26, 28), and optical studies show large changes in the dielectric function (16, 27, 31), all of which are persistent, long lived, and characteristic of a phase transition. Given the nature of the equilibrium phase diagram, these observations were interpreted mostly as evidence of the

M_{1} \to R

transition. The

M_{1} \to R

transition, however, exhibits a minimum fluence threshold of

\sim

8–9 mJ/

{c m}^{2}

consistent with surface-sensitive experiments (18, 35) and coherent phonon investigations (27). Above 8 mJ/

{c m}^{2}

photoexcitation yields a heterogeneous response with both

M

- and

R

-phase fractions increasing with fluence to approximately 20 mJ/

{c m}^{2}

where each phase occupies

\sim 50 %

of the film. At higher fluences

M_{1} \to R

dominates and the

M

phase occupies a decreasing proportion of the film. Our measurements are ensemble averaged over the probe volume and are thus not sensitive to the spatial detail of this heterogeneity. Spatially resolved studies have indicated

M_{1}

- and

R

-phase separation on length scales of 50–100 nm (35–37), which is on the order of the average crystallite size in our samples. Given this, and the fact that it would be energetically unfavorable to have a mixture of

M_{1}

and

R

phases for reasons related to strain, we believe

M_{1}

- and

R

-phase coexistence within a single crystallite to be unlikely. Phase coexistence between

M_{1}

and

M

, however, since they share the same crystal structure, could be occurring and would be interesting to investigate with sensitive spatially resolved techniques.

Fig. 6.

Illustration of the multiphase heterogeneity of photoexcited ${V O}_{2}$ . (*Top*, *Left* to *Right*) Scanning electron microscope image of polycrystalline ${V O}_{2}$ grown by pulsed laser deposition. Shown is a schematic representation of crystallite phases with increasing photoexcitation strength. The possibility of phase coexistence within crystallites is not excluded since our measurement is spatially averaged over the sample volume. (*Bottom Right*) Below the fluence threshold of $\sim 4$ mJ/ ${c m}^{2}$ the response of the material is Mott–Hubbard-like: instantaneous transient metallization from excited carriers followed by a recovery to the insulating phase within $\sim 100$ fs. When the fluence threshold for the $M_{1} \to M$ IMT is reached in a particular crystallite (orange arrow), the gap collapses and the crystallite transitions to the $M$ phase for $\sim 100$ ps (black arrow). Eventually the crystallographic structural IMT fluence threshold ( $\sim 9$ mJ/ ${c m}^{2}$ ) is reached and crystallites transform from $M_{1} \to R$ (gray arrow). Formation of the $M$ and $R$ phases is the result of two independent transitions occurring in specific proportions of the sample and may be dependent on crystallite size, grain boundaries, and strain. (*Bottom Left*) Free energy landscape. The monoclinic metal phase is described by a free energy minimum and the associated IMT is driven by electron temperature with an activation energy of $E_{A} = 304 \pm 109$ meV (depicted by an orange arrow). The occupancy of the $d_{x z}$ orbital changes from partial to full filling during this process. The monoclinic-rutile structural IMT has a barrier which is removed when the threshold fluence of $\sim 9$ mJ/ ${c m}^{2}$ is reached, leading to an increase in the V–V distance (removal of dimerization).

Nonthermal melting as a route to the control of material structure and properties with femtosecond laser excitation has been known for some time and there are examples in several material classes. Much more interesting is the (

M_{1} \to M

) which has no equilibrium analog and represents another direction for using optical excitation to control the properties of strongly correlated materials. The

M_{1} \to M

is thermally activated and does not involve a significant lattice structural component, representing a phase transition in the electron system alone. Our results are consistent with the recent computations of He and Millis (39) which indicate that an orbital selective transition can be driven in

M_{1} {V O}_{2}

through the increase in electron temperature following femtosecond laser excitation. This transition depletes the occupancy of the V-

3 d_{x^{2} - y^{2}}

band that is split by the V–V dimerization in favor of the V-

3 d_{x z}

band that mixes strongly with the O-

2 p

orbitals due to the antiferroelectric tilting of the V–V dimers (Fig. 3F). The

M_{1}

band gap collapses along with this transition, yielding a metallic phase. The three salient features of this picture are in agreement with our observations: thermal activation on the order of

100

meV, orbital selection, and band-gap collapse to a metallic phase. Of interest is the fact that depletion of the V-

3 d_{x^{2} - y^{2}}

band where states are expected to be localized on the V–V dimers does not seem to significantly lengthen the V–V dimer bond. The question arises whether this phenomenon can be entirely understood inside a picture that treats

M_{1} {V O}_{2}

as a

d^{1}

system or whether more than a single V-

3 d

electron is involved as dynamical mean-field theory (DMFT) calculations suggest (10). Such DMFT results from Weber et al. (10) suggest that

M_{1} {V O}_{2}

is a paramagnetic metal with antiferroelectric character, like that shown in Fig. 3D, when intradimer correlations are not included.

The combination of UED and TRTS measurements also makes it possible to address the question of a structural bottleneck associated with the photoinduced IMT in

M_{1} {V O}_{2}

. Here we definitively show that the timescale associated with the IMT, i.e., the timescale associated with the emergence of metallic conductivity similar to that of the equilibrium metallic phase, is determined by that of the structural phase transitions (Figs. 2 and 5). Following photoexcitation at sufficient fluence, there is overwhelming evidence that there is an impulsive collapse of the band gap in

M_{1} {V O}_{2}

with equally rapid changes in optical properties. However, the emergence of metallic transport properties occurs on the same timescale as the structural phase transitions. Clearly the localization–delocalization transition that leads to a 5 order-of-magnitude increase in conductivity is inseparable from the structural phase transitions.

In conclusion, we have combined UED and TRTS measurements of

{V O}_{2}

and decoupled the concurrent structural phase transitions along with their contribution to the multiphase heterogeneity of the sample following photoexcitation. We have shown that the monoclinic metal phase is the product of a thermally activated transition in the electron system, which provides another avenue for the optical control of strongly correlated material properties (40).

Materials and Methods

Growth of Vanadium Dioxide Films.

The 50-nm

{V O}_{2}

samples were deposited by pulsed laser deposition on a 40-nm

{S i N}_{x}

substrate. The sample area is formed by a 250-

μ

m × 250-

μ

m silicon window. Details of the deposition process can be found in ref. 5.

RF-Compressed UED.

The UED measurements are carried out at

\sim 300

K in transmission mode with 90 keV radio-frequency–compressed electron pulses which have a bunch charge of

\sim 0.1

pC (41, 42). The sample and the beam line are under high vacuum (

\sim 1 0^{- 7}

mbar). The laser setup is based on a commercial Ti:Sapphire amplified system (Spectra-Physics Spitfire XP-Pro) operated at a repetition rate of 50–200 Hz (depending on excitation fluence) to allow for sufficient sample recovery. The duration of the optical pump pulse is 35 fs (FWHM) with a spot size of 350

μ

m (FWHM).

TRTS.

Time-resolved multiterahertz spectroscopy experiments are based on two-color laser plasma generation of single-cycle, broadband terahertz pulses and air-biased coherent detection providing a spectral range from 0.5 THz to 30 THz and temporal resolution of

\sim 40

fs (43). The terahertz spectrometer is driven by 35-fs, 795-nm pulses from an amplified Ti:sapphire femtosecond laser operating at 250 Hz repetition rate to allow for sample recovery between shots. The pump spot size was

\sim 700 μ

m which was at least four times the size of the terahertz pulse for the lowest frequencies analyzed here (2 THz).

Note Added in Proof.

After the final revision of this article was submitted, a new report on the photo-induced

M_{1} \to R

transition in vanadium dioxide appeared in Science (44). This new work is in agreement with both the results presented here and in our previous work on the

M_{1} \to R

part of the photo-induced phase transition (19); however, the opposite impression is given in the text. We agree that the

M_{1} \to R

transition is best understood as a melting or disordering transition, not a directed motion along an optical phonon coordinate. The results presented here (and in our earlier article) also clearly indicate that there are no structural intermediate along the

M_{1} \to R

phase transition pathway. We emphasize that the

M_{1} \to M

metal transition also observed in our work [but not discussed in this new report (44)] is an independent phase transition of a very different character than the

M_{1} \to R

transition.

Data Availability

Data deposition: The data and scripts reported in this paper have been deposited in the following location: ftp://132.206.175.129/.

Acknowledgments

The authors acknowledge support from the Natural Sciences and Engineering Research Council of Canada, the Fonds de Recherche du Québec–Nature et Technologies, the Canada Foundation for Innovation, and the Canada Research Chairs program.

Supporting Information

Appendix (PDF)

Download
2.03 MB

References

1

FJ Morin, Oxides which show a metal-to-insulator transition at the Néel temperature. Phys Rev Lett 3, 34–36 (1959).

Crossref

Google Scholar

2

JB Goodenough, The two components of the crystallographic transition in

{V O}_{2}

. J Solid State Chem 3, 490–500 (1971).

Crossref

Google Scholar

3

V Eyert, The metal-insulator transitions of VO₂: A band theoretical approach. Ann Phys 11, 650–704 (2002).

Crossref

Google Scholar

4

DN Basov, RD Averitt, D van der Marel, M Dressel, K Haule, Electrodynamics of correlated electron materials. Rev Mod Phys 83, 471–541 (2011).

Crossref

Google Scholar

5

A Hendaoui, N Émond, M Chaker, É Haddad, Highly tunable-emittance radiator based on semiconductor-metal transition of VO₂ thin films. Appl Phys Lett 102, 061107 (2013).

Crossref

Google Scholar

6

JD Ryckman, et al., Photothermal optical modulation of ultra-compact hybrid Si-

{V O}_{2}

ring resonators. Opt Exp 20, 13215–13225 (2012).

Crossref

PubMed

Google Scholar

7

KJ Miller, KA Hallman, RF Haglund, SM Weiss, Silicon waveguide optical switch with embedded phase change material. Opt Exp 25, 26527–26536 (2017).

Crossref

PubMed

Google Scholar

8

NF Mott, L Friedman, Metal-insulator transitions in

{V O}_{2}

,

{T i}_{2} O_{3}

and

{Ti}_{2 - x} V_{x} O_{3}

. Philos Mag J Theor Exp Appl Phys 30, 389–402 (1974).

Google Scholar

9

V Eyert, VO₂: A novel view from band theory. Phys Rev Lett 107, 016401 (2011).

Crossref

PubMed

Google Scholar

10

C Weber, et al., Vanadium dioxide: A Peierls-Mott insulator stable against disorder. Phys Rev Lett 108, 256402 (2012).

Crossref

PubMed

Google Scholar

11

S Biermann, A Poteryaev, AI Lichtenstein, A Georges, Dynamical singlets and correlation-assisted Peierls transition in VO₂. Phys Rev Lett 94, 026404 (2005).

Crossref

PubMed

Google Scholar

12

WH Brito, MCO Aguiar, K Haule, G Kotliar, Metal-insulator transition in VO₂: A

D F T + D M F T

perspective. Phys Rev Lett 117, 056402 (2016).

Crossref

PubMed

Google Scholar

13

TJ Huffman, et al., Insulating phases of vanadium dioxide are Mott-Hubbard insulators. Phys Rev B 95, 075125 (2017).

Crossref

Google Scholar

14

O Nájera, M Civelli, V Dobrosavljević, MJ Rozenberg, Multiple crossovers and coherent states in a Mott-Peierls insulator. Phys Rev B 97, 045108 (2018).

Crossref

Google Scholar

15

MF Becker, et al., Femtosecond laser excitation of the semiconductor-metal phase transition in VO₂. Appl Phys Lett 65, 1507–1509 (1994).

Crossref

Google Scholar

16

A Cavalleri, et al., Femtosecond structural dynamics in VO₂ during an ultrafast solid-solid phase transition. Phys Rev Lett 87, 237401 (2001).

Crossref

PubMed

Google Scholar

17

M Hada, K Okimura, J Matsuo, Photo-induced lattice softening of excited-state VO₂. Appl Phys Lett 99, 051903 (2011).

Crossref

Google Scholar

18

P Baum, DS Yang, AH Zewail, 4D visualization of transitional structures in phase transformations by electron diffraction. Science 318, 788–792 (2007).

Crossref

PubMed

Google Scholar

19

VR Morrison, et al., A photoinduced metal-like phase of monoclinic VO₂ revealed by ultrafast electron diffraction. Science 346, 445–448 (2014).

Crossref

PubMed

Google Scholar

20

Z Tao, et al., The nature of photoinduced phase transition and metastable states in vanadium dioxide. Sci Rep 6, 38514 (2016).

Crossref

PubMed

Google Scholar

21

A Cavalleri, et al., Band-selective measurements of electron dynamics in VO₂ using femtosecond near-edge x-ray absorption. Phys Rev Lett 95, 067405 (2005).

Crossref

PubMed

Google Scholar

22

MW Haverkort, et al., Orbital-assisted metal-insulator transition in VO₂. Phys Rev Lett 95, 196404 (2005).

Crossref

PubMed

Google Scholar

23

D Wegkamp, et al., Instantaneous band gap collapse in photoexcited monoclinic VO₂ due to photocarrier doping. Phys Rev Lett 113, 216401 (2014).

Crossref

PubMed

Google Scholar

24

A Cavalleri, T Dekorsy, HHW Chong, JC Kieffer, RW Schoenlein, Evidence for a structurally-driven insulator-to-metal transition in VO₂: A view from the ultrafast timescale. Phys Rev B 70, 161102 (2004).

Crossref

Google Scholar

25

C Kübler, et al., Coherent structural dynamics and electronic correlations during an ultrafast insulator-to-metal phase transition in VO₂. Phys Rev Lett 99, 116401 (2007).

Crossref

PubMed

Google Scholar

26

A Pashkin, et al., Ultrafast insulator-metal phase transition in VO₂ studied by multiterahertz spectroscopy. Phys Rev B 83, 195120 (2011).

Crossref

Google Scholar

27

S Wall, et al., Ultrafast changes in lattice symmetry probed by coherent phonons. Nat Commun 3, 721 (2012).

Crossref

PubMed

Google Scholar

28

TL Cocker, et al., Phase diagram of the ultrafast photoinduced insulator-metal transition in vanadium dioxide. Phys Rev B 85, 155120 (2012).

Crossref

Google Scholar

29

S Wall, et al., Tracking the evolution of electronic and structural properties of VO₂ during the ultrafast photoinduced insulator-metal transition. Phys Rev B 87, 115126 (2013).

Crossref

Google Scholar

30

B Mayer, et al., Tunneling breakdown of a strongly correlated insulating state in VO₂ induced by intense multiterahertz excitation. Phys Rev B 91, 235113 (2015).

Crossref

Google Scholar

31

MF Jager, et al., Tracking the insulator-to-metal phase transition in VO₂ with few-femtosecond extreme UV transient absorption spectroscopy. Proc Natl Acad Sci USA 114, 9558–9563 (2017).

Crossref

PubMed

Google Scholar

32

E Abreu, et al., Ultrafast electron-lattice coupling dynamics in VO₂ and V₂O₃ thin films. Phys Rev B 96, 094309 (2017).

Crossref

Google Scholar

33

LP René de Cotret, BJ Siwick, A general method for baseline-removal in ultrafast electron powder diffraction data using the dual-tree complex wavelet transform. Struct Dyn 4, 044004 (2017).

Crossref

PubMed

Google Scholar

34

T Elsaesser, M Woerner, Photoinduced structural dynamics of polar solids studied by femtosecond X-ray diffraction. Acta Crystallogr Sect A Found Crystallogr 66, 168–178 (2010).

Crossref

PubMed

Google Scholar

35

VA Lobastov, J Weissenrieder, J Tang, AH Zewail, Ultrafast electron microscopy (UEM): Four-dimensional imaging and diffraction of nanostructures during phase transitions. Nano Lett 7, 2552–2558 (2007).

Crossref

PubMed

Google Scholar

36

BT O’Callahan, et al., Inhomogeneity of the ultrafast insulator-to-metal transition dynamics of VO₂. Nat Commun 6, 6849 (2015).

Crossref

PubMed

Google Scholar

37

SA Dönges, et al., Ultrafast nanoimaging of the photoinduced phase transition dynamics in VO2. Nano Lett 16, 3029–3035 (2016).

Crossref

PubMed

Google Scholar

38

MG Evans, M Polanyi, Some applications of the transition state method to the calculation of reaction velocities, especially in solution. Trans Faraday Soc 31, 875–894 (1935).

Crossref

Google Scholar

39

Z He, AJ Millis, Photoinduced phase transitions in narrow-gap Mott insulators: The case of VO₂. Phys Rev B 93, 115126 (2016).

Crossref

Google Scholar

40

D Basov, R Averitt, D Hsieh, Towards properties on demand in quantum materials. Nat Mater 16, 1077–1088 (2017).

Crossref

PubMed

Google Scholar

41

MR Otto, LP René de Cotret, MJ Stern, BJ Siwick, Solving the jitter problem in microwave compressed ultrafast electron diffraction instruments: Robust sub-50 fs cavity-laser phase stabilization. Struct Dyn 4, 051101 (2017).

Crossref

PubMed

Google Scholar

42

RP Chatelain, VR Morrison, C Godbout, BJ Siwick, Ultrafast electron diffraction with radio-frequency compressed electron pulses. Appl Phys Lett 101, 081901 (2012).

Crossref

Google Scholar

43

DA Valverde-Chavez, et al., Intrinsic femtosecond charge generation dynamics in single crystal CH₃NH₃PbI₃. Energy Environ Sci 8, 3700–3707 (2015).

Crossref

Google Scholar

44

S Wall, et al., Ultrafast disordering of vanadium dimers in photoexcited VO₂. Science 362, 572–576 (2018).

Crossref

PubMed

Google Scholar

Information & Authors

Information

Published in

Proceedings of the National Academy of Sciences

Vol. 116 | No. 2
January 8, 2019

PubMed: 30587594

Classifications

Copyright

Data Availability

Data deposition: The data and scripts reported in this paper have been deposited in the following location: ftp://132.206.175.129/.

Submission history

Published online: December 26, 2018

Published in issue: January 8, 2019

Keywords

Acknowledgments

The authors acknowledge support from the Natural Sciences and Engineering Research Council of Canada, the Fonds de Recherche du Québec–Nature et Technologies, the Canada Foundation for Innovation, and the Canada Research Chairs program.

Notes

This article is a PNAS Direct Submission. R.A. is a guest editor invited by the Editorial Board.

Authors

Affiliations

Martin R. Otto¹ https://orcid.org/0000-0002-3884-8424 [email protected]

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

View all articles by this author

Laurent P. René de Cotret

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

View all articles by this author

David A. Valverde-Chavez

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

View all articles by this author

Kunal L. Tiwari

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

View all articles by this author

Nicolas Émond

Institut National de la Recherche Scientifique, Centre Énergie Matériaux et Télécommunications, Université du Québec, Varennes, QC, Canada J3X 1S2;

View all articles by this author

Mohamed Chaker

Institut National de la Recherche Scientifique, Centre Énergie Matériaux et Télécommunications, Université du Québec, Varennes, QC, Canada J3X 1S2;

View all articles by this author

David G. Cooke

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

View all articles by this author

Bradley J. Siwick

Department of Physics, Center for the Physics of Materials, McGill University, Montreal, QC, Canada H3A 2T8;

Department of Chemistry, McGill University, Montreal, QC, Canada H3A 0B8

View all articles by this author

Notes

1

To whom correspondence should be addressed. Email: [email protected].

Author contributions: M.C., D.G.C., and B.J.S. designed research; M.R.O., L.P.R.d.C., D.A.V.-C., K.L.T., and N.É. performed research; M.R.O., L.P.R.d.C., and D.G.C. analyzed data; M.R.O. and B.J.S. wrote the paper; M.R.O. and L.P.R.d.C. wrote SI Appendix; and N.É. and M.C. fabricated samples.

Competing Interests

The authors declare no conflict of interest.

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.

Altmetrics

Citations

Cite this article

M.R. Otto,
L.P. René de Cotret,
D.A. Valverde-Chavez,
K.L. Tiwari,
N. Émond,
M. Chaker,
D.G. Cooke,
& B.J. Siwick,

How optical excitation controls the structure and properties of vanadium dioxide, Proc. Natl. Acad. Sci. U.S.A. 116 (2) 450-455, https://doi.org/10.1073/pnas.1808414115 (2019).

Copied!

Copying failed.

Export the article citation data by selecting a format from the list below and clicking Export.

Cited by

View Options

View options

PDF format

Download this article as a PDF file

DOWNLOAD PDF

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Personal login Institutional Login

Recommend to a librarian

Recommend PNAS to a Librarian

Save for later

Purchase options

Purchase this article to access the full text.

Single Article Purchase

How optical excitation controls the structure and properties of vanadium dioxide

Proceedings of the National Academy of Sciences

Vol. 116
No. 2
pp. 339-700

Restore content access

Restore content access for purchases made as a guest

Featured Topics

Articles By Topic

Featured Topics

Articles By Topic

Featured Topic

Articles By Topic

Significance

Abstract

Sign up for PNAS alerts.

Experimental Results

Structure of the Monoclinic Metallic Phase.

Fluence Dependence.

Activation Energy and Kinetics.

Discussion and Conclusion

Materials and Methods

Growth of Vanadium Dioxide Films.

RF-Compressed UED.

TRTS.

Note Added in Proof.

Data Availability

Acknowledgments

Supporting Information

References

Information

Published in

Classifications

Copyright

Data Availability

Submission history

Keywords

Acknowledgments

Notes

Authors

Affiliations

Notes

Competing Interests

Metrics

Altmetrics

Citations

Cite this article

Cited by

View options

PDF format

Login options

Recommend to a librarian

Purchase options

Restore content access

Share

Share article link

Share on social media

Further reading in this issue

A salivary effector enables whitefly to feed on host plants by eliciting salicylic acid-signaling pathway

Geometry and arithmetic of crystallographic sphere packings

Modeling stochastic processes in disease spread across a heterogeneous social system

Republicans are flagged more often than Democrats for sharing misinformation on X’s Community Notes

Bodily maps of emotions

Neural network retuning and neural predictors of learning success associated with cello training

Sign up for thePNAS Highlights newsletter