Deuterium Isotope Effects in Polymerization of Benzene underPressureWeizhao Cai,† Mihindra Dunuwille,† Jiangang He,‡ Trevor V. Taylor,† Jasmine K. Hinton,†

Mary C. MacLean,† Jamie J. Molaison,§ Antonio M. dos Santos,§ Stanislav Sinogeikin,⊥

and Shanti Deemyad*,†

†Department of Physics and Astronomy, University of Utah, 115S 1400E, Salt Lake City, Utah 84112, United States‡Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208, United States§Quantum Condensed Matter Division, Neutron Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831,United States⊥High Pressure Collaborative Access Team (HPCAT), Geophysical Laboratory, Carnegie Institute of Washington, Argonne, Illinois60439, United States

*S Supporting Information

ABSTRACT: The enormous versatility in the properties ofcarbon materials depends on the content of the sp2 and sp3

covalent bonds. Under compression, if intermoleculardistances cross a critical threshold, then unsaturated hydro-carbons gradually transform to saturated carbon polymers.However, the mechanism of polymerization, even for benzene,the simplest aromatic hydrocarbon, is still not understood. Weused high-pressure synchrotron X-ray, neutron diffraction, andmicro-Raman spectroscopy together with density functionalcalculations to investigate the isotope effects in benzeneisotopologues C6H6 and C6D6 up to 46.0 GPa. Raman spectraof polymeric products recovered from comparable pressuresshow the progression of polymerization exhibiting apronounced kinetic isotope effect. Kinetically retarded reactions in C6D6 shed light on the mechanism of polymerization ofbenzene. We find that C6D6-derived products recovered from P < 35 GPa actively react with moisture, forming polymers withhigher sp3 hydrogen contents. Significant isotopic shift (≥7 GPa) in persistence of Bragg reflections of C6D6 is observed.

Compression-induced bonding change makes pressure avaluable route in the synthesis of functional polymers

with exceptional properties and superhard materials thatpossess diamond-like strength.1−4 Benzene (C6H6), which isthe simplest and the most abundant aromatic hydrocarbon, isonly stable in a limited range of pressures and temperatures,beyond which an irreversible chemical reaction to a polymericphase takes place. The polymers of benzene recovered slowlyfrom 20 GPa have recently been identified as carbonnanothreads with speculated superior mechanical propertiesand have long-range order in two dimensions.5,6 Whileproperties of benzene under extreme conditions have longbeen under investigation, the mechanism of polymerization ofbenzene and the nature and the pressure dependence of itspolymers are not yet fully understood.7−17 External pressurecan change parameters such as the available reaction volume,potential energy, compressibility, and viscosity of materials, allof which influence the rate of reactions and help elucidate themechanism of polymerization.18

For light compounds, an additional knob to control thereaction’s kinetics is by isotopic substitutions. Studying the

kinetic isotope effect (KIE) helps us to identify the rate-determining steps in a reaction and determine the mechanism.Perdeuterobenzene (C6D6) is an isotopologue of benzene(C6H6). Both compounds have significant numbers of C−H/Dcovalent bonds, and KIE can result in variations in theirreaction progression. Even though some high-pressure studieson C6H6 have been reported on C6D6, the properties of C6H6

and C6D6 have only been compared below 25 GPa by Thiery etal. using IR and Raman spectroscopy.19−21 Moreover, thevibrational properties of C6H6 and C6D6 recovered products asa function of pressure have not been reported before.19−21 Thepresent phase diagram of benzene reports seven polymorphs(space group: Pbca, Cmca, P43212, P21/c, C2/c, Pbam, andP21),

12,13 from which only two, Pbca and P21/c, are agreedupon (see the Supporting Information for a summary of priorliterature).14,15

Received: March 5, 2017Accepted: April 10, 2017Published: April 10, 2017

Letter

pubs.acs.org/JPCL

© 2017 American Chemical Society 1856 DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

Our current study is aimed at understanding the structuralchanges leading to the pressure-induced polymerization andhow the reaction can be controlled, by exploring the isotopeeffect at play. In this work, using synchrotron X-ray andneutron diffraction, micro-Raman spectroscopy, and densityfunctional theory (DFT) calculations, we present a detailedstudy of structural and vibrational properties of C6H6 and C6D6under pressure and their recovered products (Table 1). We

study the changes in the vibrational properties of the polymericproducts as a function of maximum compression and presentgradual progression of the polymerization to formulate themechanism of the pressure-induced chemical reaction ofbenzene.The room temperature (RT) structural phase diagrams of

both C6H6 and C6D6 are summarized in Figure 1. (X-ray and

neutron diffraction patterns and diffraction images are shown inFigures S1−S9.) Both polycrystalline samples are colorless upto ∼16 GPa and then turn pale yellow before darkening above20 GPa. This behavior indicates gradual electronic structurechanges of both polycrystalline samples under compression(see calculated band gap changes in Figure S10). As shown inFigure 2a, we find that the phase boundaries of C6H6 and C6D6

are similar to each other up to 16 GPa, and only two phases,orthorhombic phase I (space group Pbca, Z = 4) andmonoclinic phase II (space group P21/c, Z = 2), are identified.The refined diffraction data reveal a molecular volume Vm (V/Z) collapse of ∼3% at the transition pressure of ∼2.2 GPa forC6H6 and C6D6, which has not been clearly resolved before. (Avolume collapse of ∼4% was suggested by Thiery et al. withdistinct differences in the phase diagram (Figure 2b).19) Severalstudies, including early measurements of compressibility ofC6H6 up to 3.9 GPa by Bridgman, could not resolve the volumecollapse.8 The phase I to II transition is sluggish, with twophases coexisting during isothermal compression at RT.Observation of volume collapse by compressibility measure-ments in the presence of the mixed phases and a slow gradualtransition is difficult. In addition, fast compression can lead to alarger fraction of the benzene crystalline in metastable phase Ito persist in a wider pressure range.8 Our results on C6H6 areconsistent with reports by Ciabini et al.; however, we find ournonannealed samples only exist in mixture of phase I and phaseII under isothermal compression at RT (Figure 2c and FiguresS3−S9).14 Moreover, the mixed crystalline phases of C6H6 andC6D6 coexist with newly formed polymers even after thechemical reaction is initiated (>16 GPa). This coexistence isevidenced by (100) peak splitting together with visualobservation of liquid and solid material upon recovery (FiguresS5 and S6). Comparing the X-ray diffraction (XRD) peaks ofC6H6 and C6D6 suggests that in the mixed-phase region therelative ratio of phase II to I is higher in C6H6 than C6D6 atsimilar pressures (e.g., see phase II (100) peak changes at 12.5GPa in C6H6 and 12.8 GPa in C6D6, Figures S6 and S7). Thisobservation is further supported by Raman analysis, asdiscussed next.The main difference between the structural properties of

C6H6 and C6D6 is the upper limit of pressure where the Braggreflections can be observed, which is more than ∼7 GPa higherin C6D6. At 31.2 GPa, almost no reflections of C6D6 crystallinephase can be detected except for a broad (100) peak, whichpersists up to the experimental limit of 46.0 GPa (Figure S6).The presence of this peak may arise from the reflections ofremaining unreacted C6D6 or 1D or 2D long-range order ofpolymers.5 However, in C6H6, most of the Bragg reflections arelost above 24.1 GPa, and less residual liquid is observed atrecovery (Figure S7). The difference suggests that the C6D6polymerizes slower than C6H6 under compression. Thisconsiderable pressure shift may indicate significant vibrationalcontributions to the chemical reactivity of C6H6 and C6D6 anda strong KIE, which will be discussed later from analysis of thevibrational properties of the recovered products.22,23 Thepresence of the (100) peak is consistent with the change in theRaman signature of the sample which, as we will discuss next,shows a significant drop in the intensity of skeletal breathingmode (ν1) above 30 GPa.To extend the phase diagram to low temperatures, we also

studied the temperature dependence of transitions of C6D6using neutron diffraction up to 6.2 GPa in a Paris−Edinburghpress (Figures S8 and S9 and SI). We observe that thestructural phase I−II transition in C6D6 is suppressed at lowertemperatures, which extends the pure phase I stability regime innonannealed samples below RT. This result can haveimplications in retarding benzenes polymerization, which maydefine a low-temperature compression path to metallization.13

Furthermore, we collected the Raman spectra of C6H6 andC6D6 at RT upon compression to 35.2 and 40.2 GPa,

Table 1. Symmetry and the Type of the Vibrational ModesObserved from Phase I of C6H6 and C6D6 and TheirFrequencies under Ambient Conditions (X = H/D in C6H6/C6D6)

frequency (cm−1)

vibrationalmode symmetry type of vibration C6H6 C6D6

ν1 a1g skeletal ring breathing 991 949ν2 a1g C−X stretching mode 3061 2302ν5 b2g out-of-plane trigonal motion 1006 825ν6 e2g planar C−C−C angle bend 606.5 585ν7 e2g planar C−X bond stretch 3044 2274ν8 e2g planar C−C bond stretch 1585 1556ν9 e2g planar C−C−X angle bend 1169 857ν10 e1g out of plane C−C−X angle

bend852 657

Figure 1. Schematic phase diagrams and onset of chemical reactions inC6H6 and C6D6 at RT. The microphotographs show the appearance ofC6H6 and C6D6 polycrystalline samples at different pressures.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1857

respectively (Figures S11−S36, more discussions in SI). Thepressure-induced Raman shifts of lattice modes and fourselected vibrational modes in C6H6 and C6D6 showdiscontinuities above 4.2 GPa (Figures S11 and S12). Theobserved shifts match well with previous studies below 4.2GPa.19,24 The differences observed for the pressure-inducedshifts for all modes, above 4.2 GPa, indicate that the phasetransition we observed is different from that previouslyreported.13,19,24 These spectral changes above 4.2 GPa can beattributed to the structural transition to a mixture of phases Iand II in both isotopologues, which is concluded from the X-ray and neutron diffraction studies, albeit at lower pressure(∼2.2 GPa). It is plausible that variation in the frequenciesobserved for phase I modes (when extrapolated to higherpressures) is due to the coexistence of two different crystallattices, influencing the molecules in the lattices differently thanin their pure forms. We find that both ν1 (skeletal ringbreathing) and ν2 (C−X stretching mode) in phase I are stiffer

than in mixed phase for both isotopologues. For the C6H6

phase I the slopes of ν1 and ν2 modes against pressure dν1/dPand dν2/dP are 4.32 cm−1/GPa and 7.69 cm−1/GPa, where forC6D6 they are 4.55 cm−1/GPa and 4.97 cm−1/GPa,respectively. In the mixed phase of C6H6, both ν1 and ν2modes split into two modes separately: 2.61 cm−1/GPa and3.25 cm−1/GPa for dν1/dP, 5.96 cm−1/GPa and 6.82 cm−1/GPa for dν2/dP. In comparison, C6D6 only shows slopes of3.71 cm−1/GPa and 3.6 cm−1/GPa. The absence of splitting ofν1 and ν2 modes in the mixed phase of C6D6 could be aconsequence of differences in the vibrational frequencies of thetwo phases in the mixed phase, which may not get pronouncedin its spectral features. On the basis of the above analysis, it isevident that the C−H bonds in C6H6 are becoming significantlyshorter than the C−C bonds for phase I and the mixed phase.However, for C6D6, the bond distances of C−C and C−D inthe same phase seem comparable. The shortening of C−Hbonds upon compression and the higher compressibility in

Figure 2. (a) Proposed phase diagram of C6D6. The gray and red dashed lines indicate suggested phase boundaries of C6H6 and P−T paths of C6D6taken in this study.16 Half circles show X-ray and IR data of C6H6.

14 The half-filled triangles indicate ambient-pressure data of C6D6.35 The label IV

indicates Pbam phase, as discussed in refs 13 and 14. The inset shows the selected neutron diffraction patterns of C6D6. (b) Third- and second-orderBirch−Murnaghan EoS fits to the molecular volume Vm (V/Z) data. The decompression neutron data are represented in unfilled circles. Hexagonsshow points on C6H6 measured at the same beamline for comparison. Red asterisks and stars show the literature data on C6H6 at RT.

15,24,38,39 Thephase transition at ∼2.2 GPa is indicated by vertical red dashed lines. (c) La Bail fit of the diffraction patterns of C6D6 at 0.5 and 9.4 GPa. The blackcircles and red solid lines are the measured scattering intensity and the Le Bail fit to the data, respectively. The insets magnify the fitting at lowangles. The vertical bars indicate Bragg reflections of the phases I (black) and II (olive), and the blue lines indicate difference profiles. (d) Relativeenthalpy difference (ΔH) between phases I and II as a function of pressure at 0 K. The insets show the crystal structure of phases I and II.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1858

C6H6 phase I may lead to efficient packing, meaning that therelative ratios of phase II/I in C6H6 are higher than in C6D6 atcomparable pressures. This leads to differences in theirpolymeric product compositions.To understand the phase stabilities of C6H6, we performed

DFT calculations. The calculated equilibrium volumes ofphases I and II at 0 K are 110.6 and 109.2 Å3/C6H6,respectively. This is consistent with the experimentalobservation that the denser phase II is stabilized at elevatedpressure. The results show that phase I−II transition takes placeat 0.9 GPa at 0 K (Figure 2d). The transition pressuredifference between calculations and experiments is from thesubtle energy difference between phase I and II, which couldnot be captured by DFT.25 As shown in Figure S38, theinfluence of vibrational entropy is significant in the I−II phasetransition: The critical pressure of I−II phase transitiondecreases with temperature increasing, which is in line withthe experimental observation. This is because phase I has largervibrational entropy, which originates from its lower density andsofter low-frequency phonon modes. Herein, the residual phase

I observed in our study exists in the metastable state at RT. Onthe basis of on DFT calculations the intrinsic origin of thecoexistence of phases I and II is probably a consequence ofquasi-hydrostatic conditions, deviatoric stresses between thepolycrystalline samples, or simply a sluggish phase transition.Raman measurements were also performed on both samples

in several compression−decompression cycles to determinetheir thresholds of irreversible chemical reaction at RT. Ramanspectra of C−C ring breathing modes ν1 (C6H6: ∼990 cm−1;C6D6: ∼950 cm−1) were used to track the progression ofchemical reactions.19 No appreciable change in the backgroundis observed when C6H6/C6D6 is released from 15.5 GPa(Figure 3 and Figure S39). However, when pressure wasincreased above ∼16 GPa, a strong photoluminescence (PL)background appeared, which most likely arises from thehydrogenated carbon clusters formed from the pressure-induced chemical reactions (Figure 3a,b).26 The backgroundincreases continuously during decompression, indicating theacceleration of the chemical reaction, owing to the increasedavailable volume.27 Another distinct signature of this chemical

Figure 3. Raman spectra of (a) C6H6 and (b) C6D6 upon compression to ∼16 GPa and slow decompression to ambient pressure. The aromatic C−H/C−D sp2 stretching modes (ν2 and ν7) are shown in olive dashed squares. Thin polymer layer resides in the original boundary of gasket chamberafter recovery marked by arrows. Raman spectra in the C−H/C−D stretching region of the recovered product are shown in the upper insets. Thestrong background in the inset of panel b from second-order Raman spectrum of type Ia diamond is close to the C−D stretching band. (c)Microphotographs of unreacted C6D6 liquid exposed to the atmosphere after release of pressure from 26.6 GPa. After the liquid sample evaporates,white polymers were left around the gasket chamber (green arrows).

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1859

reaction is the formation of a thin film in the original boundaryof the gasket chamber after the evaporation of unreacted C6H6/C6D6 (insets in Figure 3a,b). The emerging sp3 C−H/C−Dstretching bands (H: ∼2900−3000 cm−1; D: ∼2170 cm−1)demonstrate that the recovered solids lose aromatic character.28

Hence the onset of the chemical reaction at RT for bothsamples is ∼7 GPa lower than previous studies of C6H6.

16,27

Annealing of C6H6 and C6D6 samples at 503 K for 24 h,followed by slow cooling to RT, increased the homogeneity ofthe samples, but the onsets of their chemical reaction remained

around 16 GPa (Figure S40). Our findings are quite differentfrom the previous observations, where polymerization wasobserved at considerably higher pressures.27,29,30 As discussedpreviously, the structural phase transitions of C6H6 exhibitsignificant dependence on the thermal path of the sample. Itappears that isothermal compression of C6H6 at RT does notcompletely remove the low-pressure phase I. Together withprevious results, one possibility that arises is that the presenceof metastable phase I aids to the polymerization. In this case thediscrepancy in the onset of polymerization can be attributed to

Figure 4. Relative percentages of (a) sp2 and (b) sp3 to the total character for C6H6 and C6D6 recovered products. X is H for C6H6 and D for C6D6.(c,d) Deconvoluted spectra for C6H6 and C6D6 sample recovered from 25.1 and 25.7 GPa, respectively. (e,f) Raman spectra of the recoveredproducts of C6H6 and C6D6, right after opening the cell (black) and after ∼24 h of exposure to air (blue). The pressures indicated are the highestpressures each sample was reached to prior to their slow recovery. Inset in panel e shows increase in the C−H sp3 mode of C6D6 polymers after 2weeks of exposure to moisture.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1860

the differences in the progression of polymerization in theabsence of phase I.31,32 Because the polymerization can dependon the packing of molecules, it is plausible to find differences inthe nature of the polymers recovered from different isothermalcompression paths.We further compressed C6H6 and C6D6 samples above 16

GPa. The Raman spectra collected upon decompression onlyshowed the characteristic vibrons of crystalline C6H6 (C6D6).The amount of reacted products formed depends on thehighest pressure that the sample was compressed to, prior toslow unloading. The ν1 mode remains observable even upondecompression from 35.4 GPa (41.1 GPa) for C6H6 (C6D6),indicating that the chemical transformation is not complete.20

This unreacted material quickly evaporated once the diamondanvil cell (DAC) was opened, and at the highest pressures, thesynthesized polymers seem to expand and overfill the gaskethole (35.4 GPa/41.1 GPa for C6H6/C6D6; see Figure 3c). Onthe contrary, the Raman spectra obtained upon decompression(and before opening the cell) did not show vibrational modescorresponding to the polymeric product at any pressure.Because of increased PL background, even the vibronscorresponding to the crystalline phases of C6H6 (or C6D6)were difficult to distinguish.The Raman spectra of the polymers were only detectable

after recovery under ambient conditions (Figures S41 and S42).For C6H6, we observed vibrons with frequencies correspondingto both C−C breathing modes and C−H stretching modeswith both sp2 and sp3 character,5 the C−H bending mode withsp3 character,28 and the G mode arising from the bondstretching of sp2 C atoms in rings.33 The C−H stretchingmodes show three bands at ∼2880, 2925, and 2970 cm−1 ofsp3-bonded C−H characteristics and a weak peak at ∼3050cm−1 resembling sp2-bonded C−H. The triplet characterindicates that the recovered product has three different typesof rings. The sp3 C−H mode becomes less resolved when thesample is recovered from higher pressures due to the increaseof PL background. The sp2 C−H stretching mode decreases inintensity and becomes hardly visible when recovered frompressure above 28.0 GPa, indicating change in hybridizationfrom sp2 to sp3 (Figure 4e,f).We observed similar modes for C6D6 with frequencies

corresponding to the D-isotope, a strong band at ∼2170 cm−1

(sp3 C site), and a shoulder at ∼2265 cm−1 (sp2 C site).34

Similar to C6H6, Raman spectra of sp2 C−D bands showappreciable changes under pressurization, indicating reductionin sp2 character (Figure 4f). The spectra obtained right afteropening the cell from low pressures (<35 GPa) show relativelyweak C−H stretching modes of sp3 character.20 Even thoughthese peaks are weaker than the other peaks, this broad featurecan still be deconvoluted to two to three peaks. As evident fromthe spectra obtained after 24 h of exposure to air, we observedthat the amount of sp3 C−H component had increased and thepeaks were well resolved (Figure 4d and Figure S41b). Thisimplies that polymerized products of C6D6 are unstable inatmospheric pressure. This phenomenon has been previouslyreported in IR studies, and it was suggested that a reaction withwater may be taking place.20 To verify the nature of the newpeaks and exclude other possible reactions in C6D6-derivedpolymers, we opened the cells, which were compressed tovarious maximum pressures, inside a high-purity argonglovebox and exposed them to a water saturated environment.The Raman spectra of the recovered samples show similarspectra to the samples exposed to air. Hence, this newly

emerged Raman signature indeed is coming from the reactionwith moisture. Moreover, we did not observe any significantdifferences over time in water-deficient environments (i.e.,enhancement of the C−H stretching and bending modes).Extended exposure to the atmosphere not only causes anincrease in intensity of sp3 C−H modes but also leads to anincrease in ratio of sp3 C−D modes to sp2 C−D modes. Inaddition, it is evident that the relative intensities of the Gmodes observed for both isotopologues increase as a functionof the recovery pressure. However, exposure of the recoveredC6H6 product to atmosphere and D2O-rich environment doesnot lead to distinguishable changes in Raman spectra due to (a)overlap of old and new sp3 C−H stretching bands and (b) thepresence of substantially lower fractions of polymericintermediates from comparable pressures (Figure S41a).To qualitatively understand these features, we calculated the

relative content of sp2 and sp3 C−X modes from integrated areaof their stretching bands to estimate the sp2 and sp3 percentages(Figure 4a,b and Figure S42). For C6H6, the percentage of sp

2

area to both sp2 and sp3 characteristics was ∼10%, whereas forC6D6 it is ∼26%. The sp2 percentage decreases as the recoverypressure is increased for both samples. However, even whenC6D6 was recovered from 41.1 GPa, it showed ∼13% of sp2

(C6H6 recovered from 35.4 GPa shows ∼5% of sp2). Thismeans that the extent of reaction progression is affected by theisotope as well as the recovery pressure.The structures of the recovered polymers are unknown,

which makes the interpretation of the observed phenomenadifficult. We compared our results with previously proposedpolymers with different saturation degrees.5,13,35,36 In thecrystalline phase II, the nearly hexagonally packed columns oftilted C6H6/C6D6 molecules go through a Diels−Alder type[4π + 2π] cycloaddition reaction provided the mechanicalenergy available through PdV upon compression allowsovercoming the energy barrier for this reaction. The sp2

carbon-dominant intermediates (e.g., 4 + 2 polymer) thusformed go through various kinds of zipping process to formpolymers entirely sp3 in character, for example, tube (3,0),5

polymer I.13 Moreover, slow decompression facilitates thesereactions due to their negative activation volumes.31 Thispolymerization mechanism will have a secondary KIE, which,based on Streitwieser’s rehybridization model, should lead to areverse KIE.37 Therefore, while this proposed pathway canexplain the reduction of C-X sp2 bonding fraction underpressure, as discussed above, we find that based on the aboveproposed polymerization reaction mechanisms a reverseisotope effect should be observed, which is in contradictionwith our results (normal KIE). The normal KIE observed, here,however can be explained based on the model above if thecrystal packing at high pressure increases the out-of-plane C-Xbending to an extent that is substantially larger than that of sp3

C−X bending, in which case a normal KIE will be expected.We find that the percentage of sp2-dominant polymers (e.g.,

4 + 2 polymer) in C6H6 is considerably lower than that in C6D6at comparable pressures. The polymers resulting from zippingreactions are also present in both recovered samples even at 20GPa, suggesting that some of the sp2-dominant polymers getconverted to the zipper polymers. The sp3 C−H vibrationalmodes contain three different bands in C6H6-derived polymers,which indicates that three types sp3 sites are present in thesepolymers, for example, polymer I, which has five- and six-membered rings,13 and tube (3,0), which has six-memberedrings in the entire structure.5 However, such information

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1861

cannot be deduced from C6D6 simply because the vibrationalfrequency variation corresponding to similar structuresresulting from the isotopic substitution is small. On the basisof these observations, we propose that in C6D6 the chemicalreactions are retarded, and thus substantial amount of unstablesp2 intermediates (e.g., 4 + 2 polymer) is present in recoveredproducts. For example, the 4 + 2 polymer has a significantnumber of sp2 sites, which easily react with moisture. Duringthis process some sp2-hybridized C−D bonds convert to sp3

C−D and C−H bonds. The presence of multiple C−H modesimplies that once the sp2-dominant polymer reacts withmoisture, the resulting product will induce a chain reactionpromoting the formation of a new type of polymers withadditional hydrogen contents.In summary, the isotopic substitution of C6H6 (to C6D6)

crystalline samples shows negligible effect in the first-orderstructural phase transition (∼2.2 GPa) and onset of chemicalreaction (∼16 GPa) at RT. The subtle free-energy differencebetween phases I and II revealed by DFT calculations suggeststhe residual phase I is a metastable phase after phase I−IItransition. Isothermal compression of benzene at low temper-atures may sufficiently affect the transition from I → II,allowing a metallization to take place. The crystalline molecularphase in C6D6 at RT persists for pressures as high as 31.2 GPa,which is significantly higher than the maximum pressure inwhich Bragg reflections are observed in C6H6 (24.1 GPa).The character of polymers changes gradually as a function of

maximum pressure, and a distinct isotope effect upon initiationof chemical reaction is observed. The variations in polymer-ization progression lead to different fractions of sp3 character atcomparable pressures in both C6H6 and C6D6. A completetransformation of sp2- to sp3-type polymers in bothisotopologues may occur if the reaction pressure is extendedat RT. Our work provides a route to synthesize carbon-basedpolymers from C6H6 and C6D6 with different sp2/sp3 fractionsusing a high-pressure solid-state reaction method. In addition,the destabilization of recovered C6D6 products in atmospherepresents an approach to obtain more saturated sp3-bondedmaterials with higher hydrogen contents than the parentcompounds, with possible improvement in their mechanicalstrength. The isotope effect in polymerization reactionobserved in our studies also suggests the possibility ofcontrolling the reaction rate with different isotopic reactantsunder high pressure and sheds light on developing amicroscopic picture of pressure-induced polymerization inhydrocarbons.

■ EXPERIMENTAL DETAILSSample Preparation and Data Analysis. Liquid samples of C6H6and C6D6 from Sigma-Aldrich (≥99.9% purity) were usedwithout further purification. The samples were loaded in theform of either solid or liquid, in a gasket hole (rhenium orstainless steel) placed between diamond anvils of 500 μm culetflats. The details of the low-temperature solid loading in a DACcan be found elsewhere.40 The internal pressure of the sampleswas determined using the R1 shift of ruby luminescence using aquasi-hydrostatic ruby pressure scale as well as the equation ofstate (EoS) of Re where applicable.41,42 Because the phasetransformations of C6H6 and C6D6 are sluggish, we compressedthe samples slowly (2 GPa/h) using membrane DACs andsimultaneously collected XRD data at RT. Neutron scatteringstudies were only performed on C6D6 in a Paris−Edinburghpress fitted with cubic boron nitride single toroid gaskets that,

owing to the natural boron absorption characteristics, yield noextra peaks in the diffraction data. The pressure was controlledvia a computer controlled syringe pump. Prior to loading intothe press, the C6D6 sample was frozen into a solid, ground, andloaded into a SME gasket (soft metal encapsulating) fabricatedof a TiZr null-scattering alloy. Once the gasket was closed, thecell was quickly sealed, and pressure rose to 10 tons as quicklyas the oil-pressure driving syringe pump, keeping the sample inthe solid state even at RT. Such procedure aims to preventmelting, ensuring that there is no sample loss to evaporationand that it remains a good powder average. The pressure wasestimated by monitoring predominant Bragg peaks of thediffraction data and their corresponding shift in d spacing withpressure and comparing these with the data obtained from theX-ray experiments.The X-ray and neutron diffraction data were analyzed by Le

Bail fitting method using GSAS-EXPGUI package.43 Theprevious low-temperature phase I (Pbca) was used as thestarting model to carry out the refinements.44,45 The structuralmodel of high-pressure monoclinic phase II (P21/c) wasadopted from C6H6 at 2.5 GPa.

38 The previously reported high-pressure data were used for refinement of C6H6.

15,39 The bulkmoduli were calculated using EOSFIT 5.2c software.46

High-Pressure Synchrotron X-ray Measurements. The XRDmeasurements were performed at both High PressureCollaborative Access Team (HPCAT) at the Advance PhotonSource (APS), Argonne National Laboratory using 16-ID-Bbeamline (λ = 0.4066 Å) and Advanced Light Source (ALS) atLawrence Berkeley National Laboratory using 12.2.2 beamline(λ = 0.4966 Å). All X-ray data were collected duringcompression at RT.Neutron Dif f raction Measurements. Time-of-flight neutron

diffraction measurements at variable temperatures andpressures were performed at the Spallation Neutrons andPressure (SNAP) beamline of the Spallation Neutron Source atOak Ridge National Laboratory. Polycrystalline sample of C6D6was loaded in a TiZr null scattering alloy gasket, and pressurewas applied to the sample in a Paris−Edinburgh press fittedwith cubic boron nitride anvils. The sample was cooled with theaid of a custom-made liquid-nitrogen cooling device connecteddirectly to the anvils, allowing measurements in the range of103 and 298 K. Pressure was estimated based on calibratedload. The data were collected using a neutron wavelength bandcentered at 2.1 Å, allowing access to Bragg reflections between0.5 and 5.0 Å.Micro-Raman Spectroscopy. Raman measurements were

performed on a confocal micro-Raman system (WITECAlpha SNOM) with backscattering geometry using both 1800lines/mm and 600 lines/mm gratings. Both 488 nm (blue) and532 nm (green) laser lines were used for excitation. Because ofthe high cutoff from the Notch filter, the measurable Ramanspectrum is above 150 cm−1. Attention was paid to possiblephotoinduced effects of excitation laser used on the samples.Minimum laser power and lowest exposure times were usedunless intentional laser damage was desired. To study thereactivity of the polymer products with air moisture, therecovered samples were exposed to the following conditions forextended periods of time: normal atmospheric conditions,H2O-saturated environments, and moisture-free environment(inside a high-purity argon glovebox where both H2O and O2levels were below 0.1 ppm), and the results were compared inseveral experimental runs.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1862

Density Functional Theory Calculations. All the calculations arebased on DFT as implemented in Vienna ab intio SimulationPackage (VASP).47,48 The van der Waals exchange-correlationfunctional optB86b49 and the plane-wave basis set with energycutoff of 600 eV were used. The gamma-centered Monkhorst−Pack k-meshes of 7 × 7 × 6 and 9 × 9 × 7 were used for phasesI and II, respectively. The finite displacement method asimplemented in Phononphy package50 was adopted for phononand thermal properties calculations. The 2 × 2 × 2 supercellswere used in second-order force constants calculations.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jpclett.7b00536.

Specific discussions on X-ray and neutron diffraction andRaman measurements of C6H6 and C6D6 samples.Synchrotron X-ray and neutron diffraction data andRaman spectra and DFT calculations of C6H6 and C6D6under pressure are shown in Figures S1−S43. Thecalculated bulk moduli of C6H6 and C6D6 from X-ray andneutron diffraction data at RT are summarized in TableS1. (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: deemyad@physics.utah.edu. Tel: +1 801 585 5955.ORCIDWeizhao Cai: 0000-0001-7805-2108Jiangang He: 0000-0001-9643-3617Shanti Deemyad: 0000-0001-5661-8801NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe acknowledge Dr. K. Dziubek for helpful comments on themanuscript and Prof. R. Hoffmann and Dr. B. Chen for fruitfuldiscussions. We are grateful to Dr. E. Olejnik for her assistanceduring synchrotron X-ray and Raman spectra data collection,Drs. X.-D. Wen and P. Raiteri for providing theoretical modelsof benzene structures, Dr. R. Hrubiak for collecting essentialcomplementary data in HPCAT, and Dr. M. Kunz forexperimental support in ALS. We are grateful for J. Blantonfor editing the manuscript. The experimental works wereperformed at HPCAT (Sector 16), Advanced Photon Source(APS), Argonne National Laboratory. Beam time for theseexperiments was provided by the Carnegie-DOE AllianceCenter, which is supported by DOE-NNSA under grantnumber DE-NA-0002006. HPCAT operations are supportedby DOE-NNSA under Award No. DE-NA0001974 and DOE-BES under Award No. DE-FG02-99ER45775, with partialinstrumentation funding by NSF. The APS is a U.S.Department of Energy (DOE) Office of Science User Facilityoperated for the DOE Office of Science by Argonne NationalLaboratory under Contract No. DE-AC02-06CH11357. Meas-urements were performed at the high-pressure beamline 12.2.2,ALS, which is supported by the Director, Office of Science,Office of Basic Energy Sciences, of the U.S. Department ofEnergy under Contract No. DE-AC02-05CH11231. Researchconducted at ORNL’s Spallation Neutron Source wassponsored by the Scientific User Facilities Division, Office of

Basic Energy Sciences, U.S. Department of Energy. Research inthe University of Utah is supported by American ChemicalSociety grant PRF no. 55176-ND10. J.K.H and M.C.M researchis supported by NSF-REU grant number 1263394. Travelsupport for J.K.H. is funded by the Undergraduate ResearchOffice of the University of Utah.

■ REFERENCES(1) Chelazzi, D.; Ceppatelli, M.; Santoro, M.; Bini, R.; Schettino, V.High-Pressure Synthesis of Crystalline Polyethylene Using OpticalCatalysis. Nat. Mater. 2004, 3, 470−475.(2) Mao, W. L.; Mao, H.-k.; Eng, P. J.; Trainor, T. P.; Newville, M.;Kao, C.-C.; Heinz, D. L.; Shu, J.; Meng, Y.; Hemley, R. J. BondingChanges in Compressed Superhard Graphite. Science 2003, 302, 425−427.(3) Citroni, M.; Ceppatelli, M.; Bini, R.; Schettino, V. Laser-InducedSelectivity for Dimerization Versus Polymerization of Butadiene underPressure. Science 2002, 295, 2058−2060.(4) Meng, X.; Qi, G.; Zhang, C.; Wang, K.; Zou, B.; Ma, Y. VisibleMechanochromic Responses of Spiropyrans in Crystals Via Pressure-Induced Isomerization. Chem. Commun. 2015, 51, 9320−9323.(5) Fitzgibbons, T. C.; Guthrie, M.; Xu, E.-S.; Crespi, V. H.;Davidowski, S. K.; Cody, G. D.; Alem, N.; Badding, J. V. Benzene-Derived Carbon Nanothreads. Nat. Mater. 2014, 14, 43−47.(6) Silveira, J. F. R. V.; Muniz, A. R. First-Principles Calculation ofthe Mechanical Properties of Diamond Nanothreads. Carbon 2017,113, 260−265.(7) Bridgman, P. W. Change of Phase under Pressure. I. The PhaseDiagram of Eleven Substances with Especial Reference to the MeltingCurve. Phys. Rev. 1914, 3, 126−141.(8) Bridgman, P. W. Further Rough Compressions to 40,000 kg/cm2,Especially Certain Liquids. Proc. Am. Acad. Arts Sci. 1949, 77, 129−146.(9) Schmidt, S. C.; Moore, D. S.; Schiferl, D.; Shaner, J. W. BackwardStimulated Raman Scattering in Shock-Compressed Benzene. Phys.Rev. Lett. 1983, 50, 661−664.(10) Kobayashi, T.; Sekine, T. Raman Spectroscopy of Shock-Compressed Benzene and Its Derivatives. Phys. Rev. B: Condens. MatterMater. Phys. 2000, 62, 5281−5284.(11) Ciabini, L.; Santoro, M.; Bini, R.; Schettino, V. High PressurePhotoinduced Ring Opening of Benzene. Phys. Rev. Lett. 2002, 88,085505.(12) Raiteri, P.; Martonak, R.; Parrinello, M. Exploring Poly-morphism: The Case of Benzene. Angew. Chem., Int. Ed. 2005, 44,3769−3773.(13) Wen, X.-D.; Hoffmann, R.; Ashcroft, N. W. Benzene under HighPressure: A Story of Molecular Crystals Transforming to SaturatedNetworks, with a Possible Intermediate Metallic Phase. J. Am. Chem.Soc. 2011, 133, 9023−9035.(14) Ciabini, L.; Gorelli, F. A.; Santoro, M.; Bini, R.; Schettino, V.;Mezouar, M. High-Pressure and High-Temperature Equation of Stateand Phase Diagram of Solid Benzene. Phys. Rev. B: Condens. MatterMater. Phys. 2005, 72, 094108.(15) Katrusiak, A.; Podsiadło, M.; Budzianowski, A. AssociationCH•••π and No Van Der Waals Contacts at the Lowest Limits ofCrystalline Benzene I and II Stability Regions. Cryst. Growth Des. 2010,10, 3461−3465.(16) Cansell, F.; Fabre, D.; Petitet, J. P. Phase Transitions andChemical Transformations of Benzene up to 550 °C and 30 GPa. J.Chem. Phys. 1993, 99, 7300−7304.(17) Wang, Y.; Wang, L.; Zheng, H.; Li, K.; Andrzejewski, M.;Hattori, T.; Sano-Furukawa, A.; Katrusiak, A.; Meng, Y.; Liao, F.; et al.Phase Transitions and Polymerization of C6H6−C6F6 Cocrystal underExtreme Conditions. J. Phys. Chem. C 2016, 120, 29510−29519.(18) Trakhtenberg, L. I.; Klochikhin, V. L. Effect of Pressure andTemperature on the H-Atom Tunneling in Solid Phase ChemicalReactions. The Acridine/Fluorene System. Chem. Phys. 1998, 232,175−187.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1863

(19) Thiery, M. M.; Besson, J. M.; Bribes, J. L. High Pressure SolidPhases of Benzene. II. Calculations of the Vibration Frequencies andEvolution of the Bonds in C6H6 and C6D6 up to 20 GPa. J. Chem. Phys.1992, 96, 2633−2654.(20) Pruzan, P.; Chervin, J. C.; Thiery, M. M.; Itie, J. P.; Besson, J.M.; Forgerit, J. P.; Revault, M. Transformation of Benzene to aPolymer after Static Pressurization to 30 GPa. J. Chem. Phys. 1990, 92,6910−6915.(21) Thibry, M. M.; Besson, J. M.; Bribes, J. L. Pressure Dependenceof Vibrational Modes in Solid Benzenes C6H6 and C6D6. High PressureRes. 1991, 7, 159−161.(22) Wiberg, K. B. The Deuterium Isotope Effect. Chem. Rev. 1955,55, 713−743.(23) Westheimer, F. H. The Magnitude of the Primary KineticIsotope Effect for Compounds of Hydrogen and Deuterium. Chem.Rev. 1961, 61, 265−273.(24) Thiery, M. M.; Leger, J. M. High Pressure Solid Phases ofBenzene. I. Raman and X-Ray Studies of C6H6 at 294 K up to 25 GPa.J. Chem. Phys. 1988, 89, 4255−4271.(25) Azadi, S.; Cohen, R. E. Low-Pressure Phase Diagram ofCrystalline Benzene from Quantum Monte Carlo. J. Chem. Phys. 2016,145, 064501.(26) Rusli; Robertson, J.; Amaratunga, G. A. J. PhotoluminescenceBehavior of Hydrogenated Amorphous Carbon. J. Appl. Phys. 1996, 80,2998−3003.(27) Ciabini, L.; Santoro, M.; Bini, R.; Schettino, V. High PressureReactivity of Solid Benzene Probed by Infrared Spectroscopy. J. Chem.Phys. 2002, 116, 2928−2935.(28) Ristein, J.; Stief, R. T.; Ley, L.; Beyer, W. A ComparativeAnalysis of a-C:H by Infrared Spectroscopy and Mass SelectedThermal Effusion. J. Appl. Phys. 1998, 84, 3836−3847.(29) Ciabini, L.; Santoro, M.; Bini, R.; Schettino, V. High PressureCrystal Phases of Benzene Probed by Infrared Spectroscopy. J. Chem.Phys. 2001, 115, 3742−3749.(30) Ciabini, L.; Santoro, M.; Gorelli, F. A.; Bini, R.; Schettino, V.;Raugei, S. Triggering Dynamics of the High-Pressure BenzeneAmorphization. Nat. Mater. 2007, 6, 39−43.(31) Schettino, V.; Bini, R. Molecules under Extreme Conditions:Chemical Reactions at High Pressure. Phys. Chem. Chem. Phys. 2003, 5,1951−1965.(32) Gaylord, N. G. Critical Factors Affecting Chemical Reactions onPolymers. J. Polym. Sci., Part C: Polym. Symp. 1968, 24, 1−5.(33) Ferrari, A. C.; Robertson, J. Interpretation of Raman Spectra ofDisordered and Amorphous Carbon. Phys. Rev. B: Condens. MatterMater. Phys. 2000, 61, 14095−14107.(34) Ferrari, A. C.; Robertson, J. Resonant Raman Spectroscopy ofDisordered, Amorphous, and Diamondlike Carbon. Phys. Rev. B:Condens. Matter Mater. Phys. 2001, 64, 075414.(35) Chen, B.; Hoffmann, R.; Ashcroft, N. W.; Badding, J.; Xu, E.;Crespi, V. Linearly Polymerized Benzene Arrays as Intermediates,Tracing Pathways to Carbon Nanothreads. J. Am. Chem. Soc. 2015,137, 14373−14386.(36) Xu, E.-S.; Lammert, P. E.; Crespi, V. H. Systematic Enumerationof sp3 Nanothreads. Nano Lett. 2015, 15, 5124−5130.(37) Anslyn, E. V.; Dougherty, D. A. Modern Physical OrganicChemistry; University Science Books: Sausalito, CA, 2006.(38) Piermarini, G. J.; Mighell, A. D.; Weir, C. E.; Block, S. CrystalStructure of Benzene II at 25 Kilobars. Science 1969, 165, 1250−1255.(39) Budzianowski, A.; Katrusiak, A. Pressure-Frozen Benzene IRevisited. Acta Crystallogr., Sect. B: Struct. Sci. 2006, 62, 94−101.(40) Olejnik, E.; Deemyad, S. Note: Simple and Portable Setup forLoading High Purity Liquids in Diamond Anvil Cell. Rev. Sci. Instrum.2016, 87, 036103.(41) Mao, H. K.; Bell, P. M.; Shaner, J. W.; Steinberg, D. J. SpecificVolume Measurements of Cu, Mo, Pd, and Ag and Calibration of theRuby R1 Fluorescence Pressure Gauge from 0.06 to 1 Mbar. J. Appl.Phys. 1978, 49, 3276−3283.

(42) Vohra, Y. K.; Duclos, S. J.; Ruoff, A. L. High-Pressure X-RayDiffraction Studies on Rhenium up to 216 GPa (2.16 Mbar). Phys. Rev.B: Condens. Matter Mater. Phys. 1987, 36, 9790−9792.(43) Toby, B. EXPGUI, a Graphical User Interface for GSAS. J. Appl.Crystallogr. 2001, 34, 210−213.(44) Jeffrey, G. A.; Ruble, J. R.; McMullan, R. K.; Pople, J. A. TheCrystal Structure of Deuterated Benzene. Proc. R. Soc. London, Ser. A1987, 414, 47−57.(45) Craven, C. J.; Hatton, P. D.; Howard, C. J.; Pawley, G. S. TheStructure and Dynamics of Solid Benzene. I. A Neutron PowderDiffraction Study of Deuterated Benzene from 4 K to the MeltingPoint. J. Chem. Phys. 1993, 98, 8236−8243.(46) Angel, R. J.; Alvaro, M.; Gonzalez-Platas, J. Eosfit-7c and aFortran Module (Library) for Equation of State Calculations. Z.Kristallogr. - Cryst. Mater. 2014, 229, 405−419.(47) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Open-Shell Transition Metals. Phys. Rev. B: Condens. Matter Mater. Phys.1993, 48, 13115−13118.(48) Kresse, G.; Furthmuller, J. Efficiency of Ab-Initio Total EnergyCalculations for Metals and Semiconductors Using a Plane-Wave BasisSet. Comput. Mater. Sci. 1996, 6, 15−50.(49) Klimes, J.; Bowler, D. R.; Michaelides, A. Van Der WaalsDensity Functionals Applied to Solids. Phys. Rev. B: Condens. MatterMater. Phys. 2011, 83, 195131.(50) Togo, A.; Oba, F.; Tanaka, I. First-Principles Calculations of theFerroelastic Transition between Rutile-Type and CaCl2-Type SiO2 atHigh Pressures. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 78,134106.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.7b00536J. Phys. Chem. Lett. 2017, 8, 1856−1864

1864