Conclusions

[Summary] [Deintercalation Processes in Hg_1.24TiS₂] [Structures of Hg_xTiS₂] [Future research]

The results presented in the dissertation have served two purposes. First, dynamic high resolution TEM has been employed to provide the first detailed dynamic atomic-level observations of deintercalation processes in lamellar intercalation compounds. Hg_xTiS₂ (1.24 x 0.00) was chosen as the model weak-charge-transfer intercalation system for these comprehensive investigations. This system is of particular interest, as it allows the study of elastic intercalate interactions in the virtual absence of electrostatic interlayer guest-guest repulsions. Complete deintercalation of stage-1 Hg_1.24TiS₂ has been induced in-situ by controlled resistive heating (from -170C to above ambient temperature). At all stages of this study special care was taken to avoid (i) intercalate modification due to exposure to the ambient atmosphere, and (ii) vacuum deintercalation of the material prior to DHRTEM observations.

Second, a combination of the complementary techniques of TEM and XPD has been used to thoroughly investigate the unusual structural properties of this model weak-charge-transfer intercalation system, and the structural effects the deintercalation process imposes on the material. The results of this study have demonstrated that Hg-intercalated TMDICs (Transition Metal Dichalcogenide Intercalation Compounds) are very unusual compounds with novel (3+1)-dimensional misfit structures that exhibit extraordinary structural properties. The main findings of this work are summarized below.

Deintercalation Processes in Hg_1.24TiS₂

Our observations provided atomic-level details of the onset and general progression of deintercalation processes in the neutral intercalation system Hg_1.24TiS₂.

The onset of deintercalation is most often observed to occur at the outermost guest layers adjacent to the most flexible external host layers. However, the onset of deintercalation has been frequently observed at internal layers. In this case, the internal layers exhibiting enhanced reactivity can be correlated with the strain induced by defects such as host-layer edge dislocations, interlayer and surface defects. Thus, defects can play a prominent role in deintercalation reaction processes by reducing the deintercalation activation energy of individual guest layers, which can lead to internal onset.

Once started in either outermost or internal guest layers, deintercalation generally progresses away from the onset layers forming primarily randomly-staged regions. Local stage-ordered regions are occasionally formed during deintercalation. The absence of substantial stage ordering, in contrast to DHRTEM studies of ionic Ag⁺_0.17TiS₂^0.17- [68], indicates electrostatic repulsions play a prominent role in stage ordering.

The formation and deintercalation of guest edge dislocations and islands is observed to be an integral part of the deintercalation process. Both asymmetric and symmetric terminations of guest edge dislocations are observed. Their symmetry/asymmetry is strongly associated with the minimization of local host-layer strain energy and host-layer restacking. The formation of kinetically stable staggered domain walls has been routinely observed, providing direct experimental support for theoretical prediction of such walls [15].

Guest and host layer behavior during deintercalation has provided strong atomic-level support for the applicability of the Daumas-Hrold domain model of staging to intercalation/deintercalation processes [28].

The main features of the deintercalation process are summarized in the table below:

Main Features of the Hg_xTiS₂ Deintercalation Process

Feature	Observation	Reason	Comment
Initial state	Stage-1 material with ~8.7 layer repeat distance	Low temperature prevents vacuum deintercalation	Consistent with X-ray observations of the completely intercalated material [71-74]
Onset of deintercalation	At external-most guest layers	High external host-layer flexibility	Most typical observation
	At internal guest layers	Host lattice defects, surface steps, interlayer defects	Frequent observation
Progression of deintercalation	Away from the onset layers forming randomly staged regions	Nearby guest layers are less stable to deintercalation	Most typical observation
	Local stage ordering	Weak elastic and possibly electrostatic interactions	Occasionally observed
Layer behavior	Guest-layer separation and island formation	Deintercalation	Frequent observation
	Staggered domain walls formed during deintercalation	Energetically stable configuration, predicted theoretically [15]	Typical observation
Guest layer terminations	Symmetric and asymmetric	Minimization of local strain energy and host-layer restacking	Typical observation

Structures of Hg_xTiS₂

TEM/HRTEM investigations have revealed a novel in-plane guest-layer structure for Hg_xTiS₂ that is composition independent. Hg guest layers and islands are comprised of infinite one-dimensional chains that are incommensurate with the host layers along their chain-axis direction. Starting with this intralayer model, two different models for the Hg_xTiS₂ crystal structure have been proposed. Both structures are monoclinic (b = 96 and b = 102) and very similar, differing only by a minor shift of the host layers. The two-fold intralayer symmetry of the Hg layers combined with the three-fold intralayer symmetry of the host results in the formation of three equivalent Hg orientational variants with intralayer rotations of 120.

It has been shown that Hg_xTiS₂ forms an unusual four-dimensional layered misfit compound consisting of interpenetrating three-dimensional TiS₂ and Hg sublattices. These sublattices share commensurate a and c axes, but are incommensurate along their b axes. The Hg chains in Hg_xTiS₂ reside either in trigonal-prismatically or trigonal-antiprismatically coordinated sulfur channels (for the b = 102 and b = 96 structures, respectively).

Deintercalation induces a phase transformation from the b = 102 to the b = 96 phase via a 1/2b glide of the host layers. Due to their structural similarity, these phases can coherently coexist, which has been confirmed by HRTEM observations of lamellar fragments of the 102 phase in the 96 matrix. The Hg chains in Hg_xTiS₂ exhibit correlated, thermally activated disorder along their chain-axis direction, which suggests that deintercalation is primarily associated with the motion of the Hg chains along their axes. Deintercalation reaction processes in Hg_xTiS₂ are primarily one-dimensional in character, unlike other M-TMDICs whose reaction chemistry is generally a two-dimensional process [1,3,4].

These compounds exhibit significant metallic guest-guest bonding, with their intrachain Hg-Hg distance (2.76 ) being much shorter than their interchain Hg-Hg distance (3.27 ). These compounds also have significant guest-host interactions between the Hg chains and the sulfur channels. This is evidenced by an unusual intralayer off-chain-axis modulation of the Hg positions (0.2 ) to optimize the Hg-S interactions locally.

The primary structural features of Hg_xTiS₂ are summarized in the table below:

Structural Properties of Hg_xTiS₂

Feature			Description	Comment
In-plane structure			Novel partially incommensurate modulated structure
	Host		Intercalation has a negligible effect on the host in-plane structure	Host in-plane structure remains essentially unchanged for all values of x
	Guest		Hg occupies host galleries in monolayer form. Each Hg layer is comprised of infinite one-dimensional chains incommensurate with the host lattice along chain-axis direction	The in-plane structure of Hg_xTiS₂ is composition independent. During deintercalation, Hg condenses in islands
		Variants	Hg layers can have three equivalent rotational variants with respect to the host layers	Consequence of different in-plane symmetry of Hg and TiS₂ layers
Crystal structure			Four-dimensional compound which can be approximated as interpenetrating three-dimensional monoclinic TiS₂ and Hg sublattices	Guest and host sublattices share commensurate a and c axes, but are incommensurate along their b axes
	Modifications		Two similar structural modifications (b = 102 and b = 96 phases).	Transition from one modification to another can be achieved by a b shift of host layers
		Hg_1.24TiS₂	a= 5.9223 , b= 3.4076 , c= 8.867 , b = 102.33. Mercury chains reside in trigonal-prismatically coordinated sulfur channels	X-ray data show that completely intercalated material has the b = 102 structure
		Hg_xTiS₂	a= 5.92 , b= 3.41 , c= 8.70 , b = 96.5. Mercury chains reside in trigonal-antiprismatically coordinated sulfur channels	TEM data show that partially deintercalated material has the b = 96 structure with occasional b = 102 phase fragments in the form of stacking faults
Effect of intercalation			Expansion of TiS₂ host lattice by 2.95 together with 1/3a rigid glide of the host layers producing a monoclinic distortion
Effect of deintercalation			D-H island/domain formation is accompanied by the b = 102 to b = 96 phase transformation via a 1/2b shift of the host layers to minimize elastic strain energy
Hg chains and deintercalation			Hg chains exhibit correlated, thermally activated disorder. Deintercalation is primarily associated with the motion of Hg chains along the chain axes	Deintercalation processes in Hg_xTiS₂ are primarily one-dimensional in character
Bonding
		guest-guest	Stronger metallic Hg-Hg intrachain bonding than Hg-Hg interchain interactions.
		guest-host	Relatively weak guest-host interactions	Results in an unusual off-chain-axis modulation of Hg chains to optimize Hg-S interactions locally

Future research

Although, these investigations have provided the first detailed atomic-level insight into intercalation/deintercalation dynamic processes, they are only applicable to systems with very weak charge transfer, including the virtual absence of electrostatic interlayer guest-guest repulsions. These repulsions are generally considered to be the primary force associated with stage ordering. A natural extension of this work would be to use DHRTEM to study strong-charge-transfer deintercalation processes, such as thermal deintercalation of alkali-metal GICs [9,16].

Hydrazine (N₂H₄) TMD intercalation is also well suited for DHRTEM of strong-charge-transfer intercalation processes [14] and is currently under study in our laboratories. Hydrazine is a particularly suitable guest for intercalation/deintercalation DHRTEM studies, since it has a convenient vapor pressure near ambient temperature and intercalation can be achieved by exposure of the host material to the hydrazine vapor. Furthermore, deintercalation can be achieved at temperatures close to ambient temperature at reduced pressures. Therefore, this system provides the possibility of in-situ investigations of not only deintercalation processes, but intercalation processes as well (using environmental-cell DHRTEM [68]).

Ammoniated TMDICs are also attractive intercalates for DHRTEM. Very recently, the first in-situ observations of weak-charge-transfer intercalation reactions were performed with ammonia as the intercalant using environmental-cell DHRTEM [68].

Such studies should provide a detailed atomic-level picture of intercalation and deintercalation, including staging phenomena as a function of charge transfer. The outstanding questions to be addressed include:

How do guest islands form/transform/coalesce?

What is the sequence of guest-layer intercalation/deintercalation?

Do the external basal planes exhibit unique chemical reactivity for all systems?

What is the relative importance of electrostatic and elastic interlayer repulsions to staging processes?

How do guest-guest attractive forces affect intercalation processes and staging?

What roles do thermodynamics and kinetics play in intercalation dynamics?

In addition to probing the above key questions, further research will provide a unique opportunity to image the structures of intermediate phases along the intercalation/deintercalation pathway. The identification of these structures may be an important key to the synthesis of new materials.

Last updated: September 27, 2001