Control of MXenes’ electronic properties through termination and intercalation

MXenes are an emerging family of highly-conductive 2D materials which have demonstrated state-of-the-art performance in electromagnetic interference shielding, chemical sensing, and energy storage. To further improve performance, there is a need to increase MXenes’ electronic conductivity. Tailoring the MXene surface chemistry could achieve this goal, as density functional theory predicts that surface terminations strongly influence MXenes' Fermi level density of states and thereby MXenes’ electronic conductivity. Here, we directly correlate MXene surface de-functionalization with increased electronic conductivity through in situ vacuum annealing, electrical biasing, and spectroscopic analysis within the transmission electron microscope. Furthermore, we show that intercalation can induce transitions between metallic and semiconductor-like transport (transitions from a positive to negative temperature-dependence of resistance) through inter-flake effects. These findings lay the groundwork for intercalation- and termination-engineered MXenes, which promise improved electronic conductivity and could lead to the realization of semiconducting, magnetic, and topologically insulating MXenes.


Supplementary Figures
Supplementary Figure 1 MXene stability with electron irradiation and annealing. Low magnification (a) and high magnification (b) TEM images of overlapping Ti3C2Tx flakes after annealing at >500 °C within the TEM and exposure to electron irradiation for imaging, EELS, and diffraction. For a, the scale bar is 500 nm, and for b, the scale bar is 10 nm. While there is some TiO2 formation seen in a, there are no voids present in the MXene flakes, in stark contrast to recent reports by Sang et al 1 . TEM imaging shows that with low current (and low current density) imaging and spectroscopy, there is minimal electron beam induced sample degradation. This claim is further supported by the retention of -O terminations after annealing Ti3C2Tx at 775 °C (see Fig. 3c in the main text).

Supplementary Figure 3
Effect of adsorbed species on resistance. Normalized resistance of the Ti3C2Tx, Ti3CNTx, and Mo2TiC2Tx MXene films upon initial insertion into the TEM. R0 is the resistance at time = 0 s, right before the sample was inserted into the TEM vacuum. The decrease in sample resistance is attributed to the loss of adsorbed species, e.g. H2O and O2. The Ti3CNTx(TBA + ) sample is not shown, since initial resistance values were so high that they could not accurately be determined with the experimental set-up which limited the applied bias to 5 mV. shown. In all cases, a peak in the H2O ion current at ~150 °C indicates the release of H2O intercalants. In most cases, the -OH signal mirrors the H2O signal, indicating that de-protonated H2O dominates the measured -OH ion channel. However, the -OH signal from Ti3CNTx (see panel b) shows a clear peak at ~375 °C which is distinct from the measured H2O ion current, indicating the release of -OH surface groups. In a and b, the -F and HF ion currents indicate the loss of -F surface terminations from Ti3C2Tx and Ti3CNTx, beginning as low as ~400 °C. The broad peaks centered at 150 °C in the -F and HF signals perfectly match that of the H2O channel, indicating that these peaks are related to H2O loss. In c and d, the loss of TBA + from Ti3CNTx(TBA + ) and Mo2TiC2Tx is indicated by the peak in the C2H6 (m/e = 30) ion current. In d, a clear signal in the -O ion channel (m/e = 16) is observed beginning at ~1000 °C, suggesting loss of -O from Mo2TiC2Tx. For the Ti based MXenes, no loss of -O was observed with TGA-MS up to 1500 °C. In all cases, the drop in mass % and the spike in the CO ion current at temperatures above ~800 °C indicate the onset of MXene decomposition. Figure 5 Analysis of concurrent resistance and dR/dT changes with annealing. Plot of η for each annealing step. For a given annealing step, η is the ratio of the proportional change in dR/dT to the proportional change in resistance (Supplementary Equation 21). A positive value of η is consistent with a change in the intra-flake resistance, and a negative value of η is consistent with a change to the inter-flake resistance. For a change in resistance solely due to a change in the intra-flake carrier concentration, the Drude formula predicts that the value of η = 1. The solid markers correspond to the Ti3C2Tx, Ti3CNTx, and Mo2TiC2Tx samples shown in Fig. 1 of the main text. The colored lines are a guide to the eye for these samples. The open markers represent MXene samples not shown in Fig. 1. The Ti3CNTx(TBA + ) sample was annealed at 500 and 700 °C, the Ti3CNTx(2) sample was annealed at 300, 500 and 700 °C, and the Mo2TiC2Tx(2) sample was annealed at 600 and 790 °C (see Supplementary  Tables 1 and 2). Error bars represent the measurement standard deviation accounting for the linear fit to the dR/dT data and assuming a base uncertainty of 1.2% in the resistance measurements. Figure 6 Magnetoresistance (MR) of annealed MXenes. MR of Ti3CNTx (a) and Mo2TiC2Tx (b) measured at 10 K. The MR measurements were performed within the PPMS and were conducted after the samples had been annealed within the TEM to ≥700 °C. The negative MR of Ti3CNTx is similar to that of Ti3C2Tx 3 , and the positive MR of the Mo2TiC2Tx sample is consistent with previous reports 4 .

Supplementary Figure 7
Summary of Ti3CNTx(TBA + ) heating and biasing results. a, Resistance versus temperature for the Ti3CNTx(TBA + ) sample during an anneal at 750 °C. The sample was held at 750 °C for 1 h. Prior to this measurement, the sample was held at 200 °C to achieve adequate contact between the MXene film and Pt electrodes, given the presence of free (i.e. non-intercalated) TBA + , which prevented good electrical contacts. Heating is shown with a solid line and cooling is shown with a dotted line. The large decrease in resistance is mainly attributed to the loss of TBA + , as seen in Supplementary  Figure 4. b, Fitting of the cooling curve shown in a. The sample closely follows the relation ∝ exp ⁄ , with W ~ 80 meV and an R value of 0.999. The physical meaning behind this relation is unclear, but we speculate that it is related to the inter-flake hopping mechanism. Figure 8 Supplementary EELS data. a, In situ EELS measurements of the Ti3CNTx F Kedge demonstrating the reduction in -F concentration with vacuum annealing. b, In situ EELS measurements of the Mo2TiC2Tx F K-edge demonstrating the elimination of -F terminations with annealing at 500 °C. EELS data was acquired at room temperature after annealing.

Supplementary Figure 13
Time resolved resistance measurements of Ti3C2Tx. The data is the same as that presented within Fig. 1 of the main text. Heating and cooling rates were 1 °C/s. Solid (dashed) arrows represent sample heating (cooling). For a, the sample was annealed in atmosphere, and for b-g, the sample was annealed within the TEM. In between heating and cooling, the sample was held at the annealing temperature for 10 s for a-c, and 5 min for d-g. In addition to labeling the initial temperature (25 °C) and the maximum annealing temperature, we also mark the temperature where the previous annealing step was performed (marked with a star). In some cases, not all tick marks on the top x-axis (the temperature axis) could be labeled. The unmarked ticks in b, c, and g, correspond to 125, 200, and 775 °C, respectively.

Supplementary Figure 14
Time resolved resistance measurements of Ti3CNTx. The data is the same as that presented within Fig. 1 of the main text. Heating and cooling rates were 1 °C/s. Solid (dashed) arrows represent sample heating (cooling). For a, the sample was annealed in atmosphere, and for b-f, the sample was annealed within the TEM. In between heating and cooling, the sample was held at the annealing temperature for 10 s for a-c, and 5 min for d-f. In addition to labeling the initial temperature (25 °C) and annealing temperature, we also mark the temperature where the previous annealing step was performed (marked with a star). In some cases, not all tick marks on the top x-axis (the temperature axis) could be labeled. The unmarked ticks in b and c correspond to 125 and 200 °C, respectively.

Supplementary Figure 15
Time resolved resistance measurements of Mo2TiC2Tx. The data is the same as that presented within Fig. 1 of the main text. Heating and cooling rates were 1 °C/s. Solid (dashed) arrows represent sample heating (cooling). For a, the sample was annealed in atmosphere, and for b-g, the sample was annealed within the TEM. In between heating and cooling, the sample was held at the annealing temperature for 10 s for a-c, and 5 min for d-g. In addition to labeling the initial temperature (25 °C) and annealing temperature, we also mark the temperature where the previous annealing step was performed (marked with a star). In some cases, not all tick marks on the top x-axis (the temperature axis) could be labeled. The unmarked ticks in b, c, and g, correspond to 125, 200, and 775 °C, respectively.

Supplementary Figure 16
Effect of molecular adsorption and surface re-functionalization. Normalized resistance of the Ti3C2Tx, Ti3CNTx, and Mo2TiC2Tx MXene films measured as they were removed from the TEM. R0 is the resistance at time = 0 s, right before the sample was removed from the TEM vacuum and exposed to atmosphere. The increase in resistance for all samples is attributed to molecular absorption and surface re-functionalization.

Supplementary Tables
Supplementary

Supplementary Note 1
MXene dR/dT analysis: To better understand the effects of de-intercalation and surface defunctionalization on MXene electronic properties, we analyze the changes in MXene dR/dT with in situ annealing. We argued in the main text that Ti3C2Tx, Ti3CNTx and Mo2TiC2Tx are all intrinsically metallic in their as-prepared state, and that the semiconductor-like behavior of Ti3CNTx and Mo2TiC2Tx is due to inter-flake effects. This argument implies that the ensemble resistance of each MXene studied here, across all annealing temperatures, can be described by a metallic intra-flake term in series with an insulating inter-flake term The equation is a proportional equation because the resistance is also dependent upon the device geometry. The first term of Supplementary Equation 1 is the metallic (Drude) intra-flake resistance, characterized by the effective electron mass, m, the electron scattering rate, ω, and the carrier density, n. This metallic term contributes a positive component to the ensemble value of dR/dT owing to the linear temperature dependence of ω at high temperatures due to electronphonon coupling 6,7 . The second term represents the insulating inter-flake resistance, where p differentiates between thermally activated transport and variable range hopping formulae; these formulae have previously been used to successfully fit semiconductor-like MXene behavior 4,8,9 .
The inter-flake term contributes a negative component to the ensemble value of dR/dT. Because the intra-flake and inter-flake resistances have opposing temperature dependencies, the ensemble value of dR/dT is determined by the balance of these two terms. Owing to this balance, a decrease in the metallic intra-flake resistance of Supplementary Equation 1 (through, e.g., a change in n with annealing) will decrease the ensemble value of dR/dT. Conversely, a decrease in the insulating inter-flake resistance (through a change in R0 or T0 with annealing) will increase dR/dT. Thus, by measuring the change in dR/dT with annealing, we can differentiate between a decrease in the inter-flake resistance (mediated by de-intercalation) or a decrease in the intraflake resistance (driven by termination loss).
To verify the preceding claims regarding changes in dR/dT and inter-flake versus intra-flake effects, we consider the ensemble value of dR/dT. For the intra-flake Drude term, the total electron scattering rate, ω, is the sum of many scattering processes. In general, the two largest scattering processes are impurity scattering and electron-phonon scattering 6 . The total resistance is then given by where the brackets encompass the metallic intra-flake resistance, ωi is the impurity scattering rate, and ωp is the electron-phonon scattering rate. Impurity scattering is largely temperature independent, but electron-phonon scattering is temperature dependent. Above the Debye temperature, ωp scales linearly in temperature 6,7 . For our in situ TEM measurements from RT up to 775 °C, we are in the regime where ωp is proportional to T. This claim is supported by the approximately constant temperature dependence of resistance measured for Ti3C2Tx, Ti3CNTx, and Mo2TiC2Tx after high temperature annealing (Fig. 1). As such, the ensemble value of dR/dT is given by where Cp is the temperature coefficient of electron-phonon scattering, i.e. ωp = TCp. Our aim is to understanding how the ensemble value of dR/dT changes given a change in either the intraflake or the inter-flake resistance terms. To understand this correlated behavior, we take the derivatives of R and dR/dT with respect to the material parameters which determine the ensemble resistance, i.e. ωi, m, n, Cp, R0, and T0. First we investigate the intra-flake term: The main takeaway from Supplementary Equations 4-11 is that a decrease in R driven by a change in m, Cp, or n will be accompanied by a decrease in dR/dT. For example, dR/dm and ∂ 2 R/∂m∂T are both positive, hence, an increase in m will increase both R and dR/dT. For ωi, ∂ 2 R/∂ωi∂T is zero, thus a change in resistance driven by a change in ωi will not affect the ensemble value of dR/dT.
Next we consider the insulating inter-flake term: The derivative of R with respect to R0 and T0 is positive, but the derivative of dR/dT with respect to R0 and T0 is negative. Hence, a decrease in resistance driven by a change in R0 or T0 will increase the ensemble value of dR/dT. We assume that the inter-flake transport mechanism does not change with annealing, i.e. the exponent p does not change.
Next, we consider the specific case where the insulating inter-flake resistance is negligible, and with annealing, the intra-flake resistance decreases solely due to a change in n We calculate the proportional change in R given a change in n with annealing, where the proportional change in resistance is defined as the resistance after annealing, R2, minus the initial resistance, R1, divided by the initial resistance where n1 is the initial carrier concentration and n2 is the carrier concentration after annealing.
Similarly, we calculate the proportional change in dR/dT given a change in n with annealing 1 (20) Importantly, the proportional change in dR/dT with a change in n is equivalent to the proportional change in R.
To represent how the MXene resistance and dR/dT evolve with annealing, we introduce a parameter η, defined as the ratio of the proportional change in the RT dR/dT to the proportional change in the RT resistance for a given annealing step and 775 °C, and for Ti3CNTx, η ~ 1 for annealing at 700 °C. This data is consistent with an increase in the MXene intra-flake conductivity via surface de-functionalization and an increase in n. The slower transition from negative to positive η for Ti3CNTx relative to Ti3C2Tx is indicative of the increased lattice response of Ti3CNTx to H2O intercalation, as evidenced by XRD measurements (Fig. 2c). For Mo2TiC2Tx, η remains negative for every in situ annealing step. This behavior suggests that at all annealing temperatures, the measured decreases in Mo2TiC2Tx resistance are due (in part) to decreases in the inter-flake resistance. We speculate that residue from the TBA + decomposition continues to affect the inter-flake resistance for annealing at ≥775 °C, causing η < 0. This claim is supported by the Ti3CNTx(TBA + ) sample, where η < 0 for annealing at 500 and 700 °C, in contrast to Ti3CNTx intercalated with only H2O and Li + . The negative value of η for Mo2TiC2Tx does not, however, contradict a correlation between -O termination loss and improved conductivity in Mo2TiC2Tx. It is also possible that the negative value of η for Mo2TiC2Tx arises due to a nontrivial relation between surface termination loss and intra-flake conductivity, which is not captured by the simple Drude formula. We also calculated η as the MXenes were removed from the TEM and exposed to ambient atmosphere (Supplementary Figure 5 and Supplementary Figure 16). For all three MXenes, this process caused an increase in resistance and an increase in dR/dT, giving η ~ 1. As such, we attribute this increase in resistance to a decrease in carrier concentration with surface re-functionalization and molecular adsorption.