<<

DOI:10.1002/cphc.201600414 Articles

On the Formation of N3H3 in Irradiated Bearing Ices: (H2NNNH) or Triimide (HNHNNH) MarkoFçrstel,[a, d] Yetsedaw A. Tsegaw,[b] Pavlo Maksyutenko,[a, d] Alexander M. Mebel,[c] Wolfram Sander,[b] and Ralf I. Kaiser*[a, d]

The remarkable versatility of triazenesinsynthesis, polymer theoretical studies with our novel detection scheme of photo- chemistry and pharmacology has led to numerousexperimen- ionization-driven reflectron time-of-flight mass spectroscopy tal and theoretical studies.Surprisingly,only very little is we can obtain information on the isomersoftriazene formed known aboutthe most fundamental triazene:the parentmole- in the films. Using isotopically labeled starting material, we can cule with the N3H3.Here we observe molecu- additionally gain insightinthe formation pathways of the iso- lar,isolated N3H3 in the gas phase after it sublimes from ener- mers of N3H3 under investigation and identify the isomers getically processed ammonia and films. Combining formedastriazene (H2NNNH) andpossibly triimide(HNHNNH).

1. Introduction

During the last decades, —a class of organic mole- life time of at least 1mswas also inferred as an intermediate cules carrying the =N N=N moiety—have received substan- in the radiolysis of an aqueous solution of based on À À tial attention both from the theoretical and organic chemistry asingle absorption feature at 230 nm.[6] The cyclic of [1] communities. Derived from cis-and trans-triazene (HN=NNH2 ; triazene, cyclotriazane, was first reported crystallographically in Scheme1), the substituted counterparts have significant appli- zeolite A, where it was stabilized by asilver cation as [1a,c] [1d] + [7] + cations in synthetic chemistry, polymer science, and phar- Ag(N3H3) . Finally,lithium-ion-complexed speciesLi(N3H3) of macology as antitumordrugs such as , Temozolo- unknown structures were generated in amicrowavedischarge mide, and [2] with their biological activity attribut- of hydrazine–helium mixtures.[8] ed to their purported capability to alkylate deoxyribonucleic The aforementioned spuriousindication of “triazene” iso- [3] acid (DNA). Althoughtriazenes have been synthesized for mers (N3H3)triggered significant computational efforts span- over 65 years, their stem compound[4] trans-and cis-triazene ning three decades.[9] An early study by Nguyen et al. identified

(Scheme 1) could not be isolatedsince triazene undergoes a cis-and trans isomer of triazene (HN=NNH2)with all six [5] facile acid-catalyzed decomposition. An unknownisomer of atoms arranged in the same plane (Cs symmetry);the trans 1 N3H3 has been detected mass spectrometrically via signalat isomer is thermodynamically more stable by 27 to 38 kJmolÀ mass-to-charge m/z=45 as atransient speciesbydischarging compared to the cis structure (Scheme 1).[9b,c] Alater study by hydrazine(N2H4); the ionization energy of the unknown isomer Magers et al. identified two additional isomersoftriazene, trii- [5a] was reported to be 9.6 0.1 eV. Triazene (N3H3)with ahalf- mide (azimine) andcyclo- (triaziridine) with triimide Æ 1 about 54 to 130 kJmolÀ and cyclo-triazane about 170 to 1 [a] Dr.M.Fçrstel, Dr.P.Maksyutenko, Prof. Dr.R.I.Kaiser 190 kJmolÀ less stable than trans-triazene. Therein it was Department of Chemistry pointed out that triazeneis“not exactly planar” and that “low- University of Hawaii, 2545 McCarthy Mall temperature isolation of these species would likely succeed”.[9d] 96822 Honolulu HI (USA) [9a] E-mail:[email protected] These findings were refinedbyPye et al. 1-amino-1,1-dia- [b] Y. A. Tsegaw,Prof. Dr.W.Sander zene was first located on the potential energy surface by Salter [9e] Lehrstuhl fürOrganische Chemie II et al., who suggested that this isomer (called isotriazene in 1 Ruhr UniversitätBochum their paper) is 42 to 50 kJmolÀ lower in energy than cyclo-tria- 44780 Bochum(Germany) zane. Combined, these resultsshow that the stabilities of tria- [c] Prof. Dr.A.M.Mebel zenes decrease in the order trans-triazene, cis-triazene, triimide, Department of Chemistry and Biochemistry Florida InternationalUniversity 1-amino-1,1-diazene and cyclo-triazane (Scheme 1). 11200 SW 8thStreet, Miami, FL 33199 (USA) Here, we exploit anovel experimental approach to synthe-

[d] Dr.M.Fçrstel,Dr. P. Maksyutenko, Prof. Dr.R.I.Kaiser size N3H3 in low-temperature matrices via an interaction of ion- W. M. Keck Research Laboratory in Astrochemistry izing radiation with frozenfilms containing ammonia (NH3)and University of Hawaii, 2545 McCarthy Mall nitrogen (N )along with their deuterated (ND )and 15N-labeled 96822HonoluluHI(USA) 2 3 (15N )counterparts. Upon sublimation of the newly formed Supporting Information and the ORCID identification number(s) for the 2 author(s) of this article can be found underhttp://dx.doi.org/10.1002/ molecules, N3H3 is identified for the first time via fragment-free cphc.201600414. single-photon vacuum ultraviolet (VUV) photoionizationcou-

ChemPhysChem 2016, 17,2726 –2735 2726  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Scheme1.Structures, ionization energies,and relativeenergies with respect to the most stable isomer of the triazene. Distances are given in Šngstrom, angles in degrees. Plain and italic numbersare calculated at the B3LYP/6-311G** and CCSD(T)/6-311G** levels of theory,respectively. Ionization energiesare adiabatic values. Numbersinparenthesis in italics denote values derived with CCSD(T)/6-311G** optimized geometries. The two bottom rows show the struc- tures of the ionic species used for the calculation of the adiabatic ionization energy.

pled to areflectrontime-of-flightmassspectrometer(PI- Experimental Section ReTOF-MS)[10] through its parention at mass-to-charge m/z= Ice layers with thicknesses of 600 50 nm were prepared from four 45. This observation is substantiated by the detection of its iso- Æ different gases along with their mixtures. These ices were ammonia topically labeled counterparts N3D3 (m/z=48) along with (NH ), D3-ammonia (ND ), ammonia and nitrogen (NH ,N,1:1.0 15 15 15 3 3 3 2 Æ NN2H3(m/z=46), N2NH3 (m/z= 47), and N3H3 (m/z=48), re- 0.2) and ammonia and 15N-nitrogen (NH , 15N ,1:1 0.2) with puri- 3 2 Æ spectively. ties as follows:NH3 (Matheson;99.999%), ND3 (Isotopes Inc; 15 99+%D), N2 (Matheson;99.9999%) and N2 (Cambridge Isotope

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2727  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Inc. 98+% 15N). The gases and/or their mixtures were introduced into the main chamber via aglass capillary array and condensed Table 1. Data applied to calculate the irradiation doseper molecule:* marks values from CASINO simulations. onto a5.5 0.2 Kcold, rhodium-coated silver wafer.The deposition Æ of each gas took about ten minutes. During the deposition, the initial kinetic energy of the electrons, Einit 5keV pressure in the main vacuum chamber increased from (5 2)” 11 8 Æ irradiation current, I 15 2nA 10À torr to (3 1)”10À torr.The ice thickness was monitored Æ Æ total number of electrons (3.4 0.3)”1014 [11] Æ during the deposition via in situ He–Ne laser interferometry. averagekinetic energyofbackscattered electrons, 3.3 0.9 keV Using Equation (1), alaser wavelength of l=632.8 nm and refrac- Æ Ebs*(NH3 ice) [12] [13] tive indices of NH3 and N2 of 1.35 0.05 and 1.2 0.1 and an averagekinetic energyofbackscattered electrons, 3.3 0.9 keV Æ Æ Æ angle of incidence of q=48,the number of observed interference Ebs*(NH3 :N2)ice) fringes (N )can be related to the thickness (d)ofthe ice: fraction of backscattered electrons, f *(NH ice) 0.34 0.1 f bs 3 Æ fraction of backscattered electrons, f *(NH :N ice) 0.35 0.1 bs 3 2 Æ average kinetic energyoftransmitted electrons, 1.5 0.5 keV Nf l Æ 1 d 1 E *, (NH ice) ð Þ ¼ p 2 2 ð Þ trans 3 2 n sin q average kinetic energyoftransmitted electrons, 1.6 0.5 keV À Æ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Etrans*(NH3:N2 ice) The composition of the ice mixtures was determined by relating fraction of transmitted electrons, f *(NH ice) 0.16 0.05 1 trans 3 the NH absorption features at 1092 cmÀ with avalue of 1.7” Æ 3 fraction of transmitted electrons, ftrans*(NH3 :N2 ice) 0.08 0.05 17 [14] Æ 10À cm to the ice thickness determined by interferometry.Each average penetration depth, l*(NH ice) 365 80 nm 3 Æ sample was then irradiated for 60 min with 5keV electrons at acur- average penetration depth, l*(NH3 :N2 ice) 350 80 nm Æ 3 rent of 15 2nAbyscanning the electron beam over the target density of the NH3 ice, 1 0.66 0.05 gcmÀ Æ 3 Æ 2 density of the NH :N ice, 1 0.75 0.08 gcmÀ surface of 0.9 0.1 cm at an angle of 708 with respect to the sur- 3 2 Æ Æ irradiated area, A 0.9 0.1 cm2 face normal of the substrate. The average deposited dose D per ir- Æ total number of molecules processed (7 3)”1017 radiated molecule can be calculated using Equation (2): Æ dose per NH molecule, D (NH ice)1.5 0.4 eV 3 3 Æ dose per NH molecule, D (NH :N ice) 1.3 0.4 eV 3 3 2 Æ Itm dose per N molecule, D (NH :N ice)1.9 0.5 eV 2 D Einit ftransEtrans fbsEbs 2 2 3 2 Æ ð Þ ¼ eNA 1 Alð À À Þ ð Þ

where I, t, m, e, NA, 1, A and Einit are the irradiation current, irradia- generated using the frequency doubled (222.6 nm) output of tion time, molecular mass of the molecule, the electron charge, adye laser (445.1 nm, Coumarin 450, Sirah, Cobra-Stretch) which Avogadro’sconstant, the density of the ice, the irradiated area of was pumped by the third harmonic of an Nd:YAG laser (355 nm). the ice, and the initial kinetic energy of the electrons, respectively. The second laser beam was generated using the direct output of

The values ftrans, fbs, Ebs, Etrans and l denote the fraction of electrons adye laser (607 nm, Rhodamine 610, Sirah, Precision Scan). This transmitted through the ice, the fraction of electrons which are dye laser was pumped with the second harmonic of an Nd:YAG backscattered, the average kinetic energy of the backscattered laser 532 nm). The respective laser beams where then coupled electrons, the average kinetic energy of the transmitted electrons, using adichroic mirror and led into adifferentially pumped and the average penetration depth of the electrons, respectively. vacuum chamber through amagnesium fluoride (MgF2)window.A These values are determined exploiting the Monte Carlo simulation fused silica bi-convex lens (Thorlabs LB4265, f=150 mm) focused program CASINO[15] averaging over 20000 trajectories. The deposit- the respective beams into asection of apulsed jet of krypton ed energy per ammonia molecule in these experiments is 1.5 (9.67 eV) or xenon (9.1 eV) released by apiezoelectric pulsed valve Æ 0.2 eV.Inthe mixed ices we determined adose of 1.3 0.2 eV per operated at 30 Hz. Alithium fluoride (LiF) lens mounted off-center Æ irradiated ammonia molecule and 1.9 0.2 eV per irradiated nitro- from the beam path of the generated and fundamental laser Æ gen molecule. The simulation parameters are summarized in beams separated the beams according to their refractive indices Table 1. After the irradiation, each ice was kept at 5.5 Kfor one spatially.Apin hole behind this lens was then used to block the hour.During the irradiation and the equilibration phase, infrared fundamentals (w1; w2)from entering the interaction region, and 1 spectra (FTIR, Nicolet6700) were recorded from 6000 to 500 cmÀ only the desired light was introduced into the main chamber.A 1 with aresolution of 4cmÀ .The substrate was then warmed up schematic of the experimental setup is shown in Scheme 2. 1 with arate of 0.5 KminÀ to 300 K.

Molecules subliming into the gas phase were photoionized at Computational Methods 10.49 eV,9.67 eV and 9.1 eV and detected using areflectron time- of-flight mass spectrometer (ReTOF).[16] The 10.49 eV photons were Calculated structures, ionization energies and relative energies generated by frequency tripling of the third harmonic of an with respect to the lowest-energy isomer of N3H3 are shown in

Nd:YAG (354.7 nm) (Spectra Physics, PRO-250, 30 Hz) laser in ajet Scheme 1. Geometries of various N3H3 isomers and transition states [17] of pulsed xenon (Xe) gas. 9.67 eV and 9.1 eV photons were pro- as well as of products of N3H3 dissociation were optimized at the duced using resonant four-wave difference mixing of two frequen- hybrid density functional B3LYP level of theory[19] with the 6– cies (2w w ).[18] 9.67 eV photons were generated from two beams 311G** basis set[20] and vibrational frequencies were computed 1À 2 overlapping in krypton gas, where the first beam was generated using the same B3LYP/6–311G** method to characterize stationary using the frequency-tripled (202 nm) output of adye laser points as local minima or transition states and to obtain zero-point (606 nm, Rhodamine 610, Sirah, Cobra-Stretch) which was pumped energy corrections (ZPE). For selected isomers and their cations we by the second harmonic of an Nd:YAG laser (532 nm). The second additionally performed geometry optimization at the coupled clus- laser beam was generated using the direct output of adye laser ters CCSD(T) level of theory[21] with the same 6–311G** basis set. (478 nm, Coumarin 480, Sirah, Precision Scan). This dye laser was Single-point energies at the optimized geometries were then re- pumped with the third harmonic of an Nd:YAG laser (355 nm). fined employing the explicitly correlated coupled clusters CCSD(T)- 9.1 eV photons were via mixing in xenon gas. The first beam was F12 method[22] with Dunning’scorrelation-consistent cc-pVTZ-f12

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2728  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Scheme2.Schematictop view of the experimental setup.Subliming molecules are ionizedbyVUV light generated via four-wave mixing and detected in are- flectrontime-of-flight mass spectrometer (not to scale). basis set;[23] the CCSD(T)-F12/cc-pVTZ-f12 energy is expected to we performed CCSD(T)/6–311G** geometry optimizationfor closely approach the complete basis set frozen core fc-CCSD(T)/ selected most favorable N3H3 isomers, trans-and cis-H2NNNH CBS limit and hence the relative energies from the present calcula- and trans–trans-NHNHNH, and their cations.The differences 1 tions are anticipated to provide accuracy within 5kJmolÀ or between optimized B3LYP and CCSD(T) geometric parameters better.Vertical and adiabatic ionization energies for all N H iso- 3 3 (Scheme 1) appeared to be rather minor and did not exceed mers were evaluated at the same CCSD(T)-F12/cc-pVTZ-f12//B3LYP/ 6–311G**+ZPE(B3LYP/6–311G**) level and also calculated were ap- 0.03 Š and 48 and in most cases were much lower than these pearance energies of fragmentation pathways. The CCSD(T) and maximal values. The effect of the use of the CCSD(T)/6–311G** CCSD(T)-F12 energies were computed within the restricted closed optimized geometries on the relative energies of the N3H3 iso- (open) shell scheme with restricted HF wavefunction used as aref- mers obtained by single-point CCSD(T)-F12 calculations were 1 erence, RHF-RCCSD(T). All electronic structure calculations were below 1kJmolÀ .For adiabatic ionization energies, the differ- [24] [25] carried out using the Gaussian 09 and MOLPRO 2010 program ences between the values obtained with B3LYP and CCSD(T) packages. The potential energy diagram including N3H3 isomers, geometries were only 0.01 eV for trans-H NNNHand trans– transition states, and dissociation products is illustrated in Figure 1. 2 trans-NHNHNHand 0.05 eV for cis-H NNNH. Energy calculations are summarized in Table S1. Geometric struc- 2 tures of all isomers and transition states are summarized in Triimides can be formed from triazenes by 1,2-H migrations Table S2. Calculated vibrational frequencies are summarized in from the terminal NH2 group to the central nitrogen atom 1 Table S3. Vertical and adiabatic ionization energies are presented in overcoming barriers of 265 kJmolÀ (trans-H2NNNH trans–cis- 1 ! Table S4. The transition states in these tables are labeled according NHNHNH) and 259 kJmolÀ (cis-H NNNH trans–trans- 2 ! to Figure 1. NHNHNH). Alternatively,a1,2-H shift from the terminal NH

moiety to the central nitrogen leads from trans-H2NNNH to 1 H2NNHN via a299 kJmolÀ barrier. The latter can ring-close to 2. Theoretical Results 1 c-N3H3,overcoming asignificant barrierof272 kJmolÀ .Anal-

Our computations identified seven N3H3 isomers. In agreement ternative pathway is the decomposition to molecular nitrogen 1 with earlier theoretical calculations, trans-(trans-H2NNNH) and plus ammonia via alower barrier of only 81 kJmolÀ .The cyclic cis-triazene(cis-H2NNNH) are the moststable isomers of N3H3 c-N3H3 isomer can also be formed by ring closure in trans–cis- 1 1 with the trans isomer residing21kJmolÀ below the cis struc- NHNHNH going through abarrier of 238 kJ molÀ .The thermo- ture. Both are connected via atransitionstate lying dynamically mostfavorable dissociation channel of trans-tria- 1 187 kJmolÀ higherinenergy than trans-H2NNNH. The next zene is trans-H2NNNH H2NNHN N2 + NH3,leadingtothe ! 1! group of isomersinthe order of stabilityincludes triimides products exoergic by 193 kJmolÀ but the highest barrieron 1 (HNHNNH), which can exist as trans–trans, trans–cis,and cis– this pathway is at 299 kJmolÀ . Trans-triazenecan also directly 1 cis;78, 81, and 102 kJmolÀ above trans-H2NNNH, respectively. lose molecular (H2)producing hydrogen azide (HN3) 1 The different triimides are connected to one another via rota- with an endoergicity of 62 kJmolÀ and overcoming abarrier 1 tional barriers around the N Nbonds ranging between 107 of 304 kJmolÀ .Both trans-and cis conformations of triazene 1 À and 132 kJmolÀ .The nitrene (H2NNHN) and cyclic isomer (c- can dissociate to amidogen (NH2)plus dinitrogen monohydride 1 N3H3)are the least stable isomersbeing 120 and 178 kJmolÀ (N2H) by acleavage of the N Nsingle bond without an exit À 1 higher in energy than trans-triazene. It should be noted that barrier. The products lie 199 kJmolÀ higher in energy than

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2729  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Figure 1. Singlet potential energy surface (PES) of possible N3H3 isomers togetherwith predicted decomposition pathways. Dashed lines indicatespin-forbid- 3 1 den decomposition pathwaysresulting in the formationoftriplet ground state (NH;XSÀ)and towardsthe lowest-lying singletstate (NH;aD).

1 trans-H2NNNH. Finally,various triimides can in principle decom- ((Z)/(E)-N2H2), 322 and 342 kJmolÀ above trans-triazene. The 3 3 1[26] pose to ground triplet state imidogen (NH, X SÀ)plus triplet ground state of imidogen (NH, C SÀ)lies 150 kJmolÀ

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2730  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles below the first excited state singlet state (the spin allowed 10.49 eV with the isotopically labeled compounds show apeak product), and adecay involvingaN Nbond rupture is likely with the same TPD profile at mass-to-charge ratios of m/z= 36 À to proceed via singlet–triplet intersystem crossing(ISC). Both (ND3 ice) andatm/z= 34, 35 and 36 with intensity decreasing 15 possibilities are shown in Figure 1. in that order in the NH3 : N2 ice. The next highestcontribution

is found in the 10.49 eV probed NH3 ice at amass-to-charge 3. Experimental Results ratio of m/z= 45. This TPD profile is observed in the 150 to 190 Kregion with apeak at around175 K. This profile is also The PI-ReTOF-MS results are shown in Figure 2and Figure 3, observed in the 9.67 eV experiment butnot in the 9.1 eV ex- which include the data of irradiated ammonia measured at periment. The ND3 experiment reveals apeak with similar three different ionization energies (left) and, measured all at an shape at amass-to-charge ratio of m/z= 48, butreduced in in- ionization energy of 10.49 eV,the PI-ReTOF-MS data of irradiat- tensitybyafactor of six comparedtothe ammonia ice. The 15 ed ND3,NH3 :N2 and NH3 : N2 ice (right). Dominating all spectra NH3 :N2 experiment also exposes this feature, but with an inten- measured with 10.49 eV is the signal at mass-to-charge ratio of sity reduced by afactor of 2.5 compared to the ammonia ex- 15 17 from the host matrix molecule ammonia (NH3)and,respec- periment. The NH3: N2 experiment depicts intensity with asim- tively at m/z= 20 from ND3. This signal depicts asublimation ilar TPD profile at mass-to-charge ratios of m/z= 45, 46 and 47 onset of 75 2K,peaks at around 105 2K and reaches zero (Figure 2fand Figure 3f). Again, the intensity decreases as the Æ Æ intensity above 130 K 5K.Asecond TPD peak is observed at mass-to-charge ratio rises with factorsof0.24, 0.12 and 0.06 Æ this m/z ratio between 150 Kand 190 K, with amaximum at compared to the peak at m/z=45 from the pure NH3 ice. At around175 K. The next strongest contribution is observed at amass-to-charge ratio of m/z=30 we also observeaTPD pro- amass-to-chargeratio of m/z= 32, which is observable at all file in the range from 157 to 180 K, peakingat172 K. This peak three ionization energies. This TPD peak is observed in the 145 is observed in the pure ammonia ice only in the 10.49 eV ex- to 200 Krange, peakingataround 155 K. The experiment at periment, where it reveals apeak intensity comparable to that

Figure 2. PI-ReTOF-MS spectra withthe initial ice composition andthe ionization energy given in each panel.

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2731  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

of the TPD profile at amass-to-charge ratio of m/z=45 (Fig- m/z=32 from N2H4 and m/z=45 from N3H3.Molecular nitro- ure 2a and Figure 3a). In the NH3 :N2 ice mixture, this peak is gen (N2)isnot observed because its ionization energy (IE= afactor of two higher in intensity than that of m/z=45 (Fig- 15.58 eV[26])resides well above our highest ionization energy of ure 2e and Figure 3e). We also observe avery smallsignal at 10.49 eV.The N2H2 isomer was identified as cis-and/or trans- [27] amass-to-charge ratio of m/z=29 in the ammonia ice at diimide(m/z=30, IE= 9.58 eV). The N2H4 isomer is hydrazine 10.49 eV (Figure 2a and Figure 3a). The TPD profile of this (m/z=32, IE=8.1 eV). With this low ionization energy it is the signal shows some intensity in the same range as that of m/ only molecule observed in the 9.1 eV experiment. In order to z=45. However,the signal-to-noise ratio of that signal is very identify the isomers of N3H3 we have to use data obtainedat poor.Nosignal is observed at amass-to-charge ratio of m/z= different ionization energies. Figure 3a depictsthe TPD profile 28. of all subliming moleculeswith m/z= 45 and an ionization energy below 10.49 eV.Sublimation starts at 150 Kand the in- 4. Discussion tensity increases slowly up to atemperature of 173 K. After that we observe afurther sharp increase in intensity,which Due to the simplicity of the host ice, consisting only of two dif- reachesits maximum at 180 K. At 185 Kthe intensity is back to ferent atoms, the molecular identification of the observed 20%ofits maximum value and then slowly decreases towards mass-to-charge ratios is straightforward. The signal at m/z=30 zero at around 200 K. must stem from amolecule with the chemical formula N2H2,

Figure 3. TPD profiles extracted from the PI-ReTOF-MS spectrashown in Figure 2. For intensity and shape comparison,the TPD trace of NH3 ice at 10.49eV (a) is includednormalized to the respective spectrum as adotted line in (b), (d) and (e).

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2732  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Figure 3b shows the result when the sublimingmolecules are probed using an ionization energy of 9.67 eV.Itcan be seen that the total observedintensity is afactor of six lower than that measured in the 10.49 eV experiment. The TPD pro- file, however,shows the same shape as that measured at 10.49 eV.For comparison, the normalized 10.49 eV TPD profile is shown in Figure 3b as adotted line. Decreasing the ioniza- tion energy furtherwearrive at Figure 3c.Here, at 9.1 eV ioni- zation energy,weobservenosignal at this mass-to-charge ratio at all. Comparing these observations with the calculated ionization energies we can already concludethat neither iso- triazene nor cyclo-triazaneare detected. Iso-triazene can be ex- cluded because the ionization energy in the 9.1 eV experiment is well above the ionization energy of iso-triazene (IE= 8.62 eV). The fact that the shape of the TPDprofile remains un- changed when changing the ionization below that of cyclo-tri- (Figure 3b)isavery strongindication that cyclo-triazane is not observed in our experiment. Adiscrimination between the remaining isomers, triazene and triimide, based solely on their ionization energies is com- plicated because of the small differences in their ionization en- ergies and the uncertainty of the calculated ionization energies in the order of upto 0.2 eV.[28] However,there is one obser- Æ vation that suggests that the observed isomersare cis-and/or trans-triazene, but not triimide. That is, we observe asmall signal at amass-to-charge ratio of m/z=29, which has aTPD profile that matches that of the trace at m/z= 45. Theappear- + ance energy of the N2H fragment from trans-triazene is 10.26 eV and from cis-triazeneitis10.03 eV.Thissuggeststhat signal at m/z= 29 can only stem from fragmented cis-and/or trans-triazene. It cannot originate from triimidebecauseour computations suggestthat this isomer cannot fragment to Scheme3.Principalreaction pathways towards the formation of N3H3 start-

+ ing with NH3 and with N2H2 as an intermediate. The primaryirradiation prod- N2H +NH2 upon ionization at 10.49 eV.Itneeds to be noted ucts are shown in blue. Also included are two principal reaction pathways that the observed intensity is barely above the noise level and towards N2H4. that even if it were higher,itwould not exclude that triimideis formed. However,itdoes imply that at least some of the ob- served signal stems from cis-and/or trans-triazene. Another ar- and not the main final irradiation product hydrazine (N2H4)or gument why the observed isomer could be trans-triazene is other intermediates like HN3 or HN2 are intermediates in the the fact that only hydrogen tunneling is necessary to arrive N3H3 formation. They also show that aformation of N3H3 in- from any of the triimides or the cis-triazene to the energetically volvingmolecular nitrogen (N2)isunlikely. lowest lying trans-triazene. Processes like this were shown to First, the N3H3 yield decreases by afactor of 2.5 when com- have half-life times of two hours at temperatures as low as paringthe NH3 to the NH3:N2 ices. The yield of N2H2 decreases 1 [29] 11 Kand abarrier of 126 kJmolÀ . We reach the sublimation by afactor of 1.6 in these ices. These values are only in agree- temperature of around 170 K6h40min after irradiation ment with each other if we assume that N2H2 is aprecursor in stoppedand the calculated barriers are in the same order of N3H3 formation.This first point is furtherbacked by the obser- magnitude as in the systems discussed in Ref. [29] (Figure 1). vation that the ratio of diimide molecules with no 15Natoms After having established that mostly triazene and also possi- to diimidemolecules with one 15Natom is, with 0.53 0.05, Æ bly triimideare forming in the ice, let us now concentrate on the same as that of N H to 15NN H (0.52 0.05). 3 3 2 3 Æ the potentialformation pathways of these molecules.Wesug- Second, using the same argumentwecan exclude hydrazine gest that the most likely formation pathway towards N3H3 (N2H4)asaprecursor because the yield of N2H4 decreases by occurs via the diimide(N2H2)intermediate after reactionwith afactor of 4.3 when comparing its yields in the NH3 and the either imidogen via N H + :NH N H or with amidogen via NH :N ice. This means that if N H were aprecursor in N H 2 2 ! 3 3 3 2 2 4 3 3 N H +·NH N H +Hasshown in the last step in Scheme 3. production,the N H yield should decrease at least as much as 2 2 2 ! 3 3 3 3 Both, imidogen and amidogen, are primary formation products the N2H4 yield. of the irradiation process. Besides the computationally predict- Third,inthe NH3 :N2 ices, we observe adecrease in the yield ed decomposition pathways andenergetics, the four following of N3H3 compared to pure ammonia ice of about 2.5. The experimental observations strongly suggest that diimide (N2H2) shape of the TPD curve is, however,unchanged. This indicates

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2733  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

that the same isomers are forming in both systems. The iso- culations of the possible decomposition pathways of N3H3. topically labeled ices revealthe reason for the strong decrease Summarizing we can conclude that our study gives adetailed 15 in N3H3 yield:even thoughthe N2 part of the ice contributes insight in the formation of N3H3 in energetically processed two thirds of all the nitrogen atoms available in the ice, 56 films and hope that it helps in future attempts to isolate mo- Æ 3% of the observed triazene molecules did not incorporate lecular triazene to make it availableasadirect component in any 15Natom. 29 2% of the molecules built in only one 15N, the synthesis of more complexmolecules. Æ and merely 14 2% hold two 15Natoms. Apossible reaction Æ pathway leadingtotriazene including isotopically labelledmo- 15 lecular nitrogen without initial N2 cleavage would be N2 + Acknowledgements 15 15 15 ·H · N2Hfollowed by· N2H+ ·NH2 N2NH3. Only 14%ofthe ! ! The authors thank the W. M. Keck Foundation and the U. S. Army observed N3H3 molecules have that signature despite the abundance of 15Natoms. This means that hypothetical forma- Research Office (W911NF-14–1–0167) for support. M.F.acknowl- edges fundingfrom the DFG (FO 941/1). The Bochum team (YAT, tions of N3H3 via N2H(from N2 +H) or via N3H(from N2 + NH) are not important pathways. WS) thanks the Cluster of Excellence RESOLV (EXC 1069), funded by the Deutsche Forschungsgemeinschaft, for support. Fourth,the remainingpossibility to form N3H3 starts with acleavageofmolecular nitrogen (N2)totwo nitrogen atoms, which could then form imidogen (:NH) after recombination Keywords: ammonia · matrix isolation · nitrogen chain with ahydrogenatom;imidogencan react with diimide(N2H2) molecules · triazene · triimides to N3H3.Alsothese pathways can only play aminor role in the 15 formationofN3H3 since they would result in NN2H3.All these [1] a) D. B. Kimball, M. M. Haley, Angew.Chem. Int. Ed. 2002, 41,3338 –3351; experimental considerations together with our energetic calcu- Angew.Chem. 2002, 114,3484 –3498;b)A.Schmiedekamp,R.H. Smith, Jr., C. J. Michejda, J. Org.Chem. 1988, 53,3433 –3436;c)M.L. lations are pointingtoN2H2 as the reaction intermediate in the N H formation. Gross, D. H. Blank, W. M. Welch, J. Org.Chem. 1993, 58,2104 –2109;d)J. 3 3 Stebani, O. Nuyken,T.Lippert, A. Wokaun, Die Makromolekulare Chemie, As depicted in the reaction scheme (Scheme3), there are Rapid Commun. 1993, 14,365–369;e)F.Marchesi, M. Turriziani, G. Tor- two possibilities to arrive from N2H2 to N3H3.Using our tech- torelli, G. Avvisati, F. Torino,L.DeVecchis, Pharmacol. Res. 2007, 56, nique we cannot differentiatebetween these two.The same 275–287;f)Triazanes:Chemical, Biologicaland Clinical Aspects,Springer holds for the formation of N H .Here, four differentformation Science +Business Media, LLC, 1989;g)Y.A.Jeilani, T. M. Orlando,A. 2 2 Pope, C. Pirim,M.T.Nguyen, RSC Adv. 2014, 4,32375–32382. pathways are feasible. One starting with the primary irradiation [2] a) D. E. V. Wilman, The Chemistry of AntitumorAgents,Glasgow,Blackie; product ·NH2 and three possibilities startingwith the primary New York :Chapman and Hall, 1990;b)E.L.Carvalho, A. P. Francisco, J. irradiation product :NH. Iley,E.Rosa, Bioorg. Med. Chem. 2000, 8,1719 –1725;c)M.Ludwig, M. Kabícˇkovµ, Collection of Czechoslovak Chemical Communications 1996, 61,355–363. 5. Conclusions [3] R. B. Livingston, Single Agents in Cancer Chemotherapy,Springer Science &Business Media, 2012. [4] K. Saunders, Practical Organic Chemistry,2nd ed.,Longmans, Green and We have processed homogeneous ammonia (NH3)and deuter- ated ammonia (ND )films as well as mixed NH :N and Co.,New York, 1949,pp. 157–179. 3 3 2 [5] a) S. Foner,R.L.Hudson, J. Chem.Phys. 1958, 29,442 –443;b)S.Foner, 15 NH3 : N2 films with energetic electrons at low temperatures to R. Hudson, Adv.Chem. Ser. 1962, 36,34. study the formation of the hitherto poorly described N3H3 iso- [6] a) E. Hayon, M. Simic, J. Am. Chem. Soc. 1972, 94,42–47;b)J.W.Suther- land, J. Phys. Chem. 1979, 83,789 –795. mers. N3H3 as well as its isotopically labeled counterparts N3D3, 15N H , 15N H and 15NN H are observed via their mass-to- [7] a) Y. Kim,J.W.Gilje, K. Seff, J. Am. Chem. Soc. 1977, 99,7057 –7059; 3 3 2 3 2 3 b) N. H. Heo,Y.Kim, J. J. Kim, K. Seff, J. Phys. Chem. C 2016, 120,5277 – charge ratios m/z= 45, 48, 48, 47 and 46, respectively,after 5287. they are photoionized upon sublimation.Using our newly es- [8] T. Fujii, C. P. Selvin,M.Sablier,K.Iwase, J. Phys. Chem. A 2002, 106, tablishedmethod of photoionization driven reflectrontime-of- 3102 –3105. flight mass spectroscopy (PI-ReTOF-MS), asensitive detection [9] a) C. C. Pye, K. Vaughan, J. F. Glister, Can. J. Chem. 2002, 80,447 –454; b) M. T. Nguyen, L. Hoesch, Helv.Chim.Acta 1986, 69,1627 –1637; scheme utilizing tunable photonenergies, in combination with c) M. T. Nguyen, J. Kaneti,L.Hoesch, A. S. Dreiding, Helv.Chim. Acta the calculation of ionization energies of the possible N3H3 iso- 1984, 67,1918–1929;d)D.H.Magers,E.A.Salter,R.J.Bartlett, C. Salter, B. A. Hess, Jr., L. J. Schaad, J. Am. Chem. Soc. 1988, 110,3435 –3446; mers, we show that the observed N3H3 signal can only stem from the triazene and/or the triimideisomers. Both the iso-tria- e) E. A. Salter,R.Z.Hinrichs, C. Salter, J. Am. Chem. Soc. 1996, 118,227 – 228. zene and the previously described cyclo-triazaneare not ob- [10] a) S. Maity,R.I.Kaiser, B. M. Jones, Phys. Chem.Chem.Phys. 2015, 17, served. Further,bycalculating the appearance energy of the 3081 –3114; b) R. I. Kaiser,S.Maity,B.M.Jones, Phys. Chem. Chem. Phys. fragment ions of triazeneand triimide andcomparing those to 2014, 16,3399 –3424. our experimental observations we present evidence that at [11] a) O. S. Heavens, OpticalProperties of Thin Solid Films,Butterworhts Sci- entific Publications, London, 1955;b)M.S.Westley,G.A.Baratta, R. A. least parts of the observed N3H3 signalmust originate from Baragiola, J. Chem. Phys. 1998, 108,3321 –3326. trans-and/or cis-triazene. To elucidatethe formation mecha- [12] M. .Satorre,J.Leliwa-Kopystynski, C. Santonja,R.Luna, Icarus 2013, 225,703–708. nism of N3H3 we have then conducted isotopic substitution studies and suggest that the N H formation occurs preferably [13] M. .Satorre, M. Domingo, C. Millµn, R. Luna, R. Vilaplana, C. Santonja, 3 3 P&SS 2008, 56,1748–1752. with the diimide (N2H2)asaprecursor via the reaction with imi- [14] L. B. d’Hendecourt, L. J. Allamandola, Astron. Astrophys. Sup. 1986, 64, dogen (NH). These considerations are backed by energetic cal- 453–467.

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2734  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

[15] D. Drouin, A. R. Couture, D. Joly,X.Tastet, V. Aimez, R. Gauvin, Scanning K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. 2007, 29,92–101. Kitao,H.Nakai, T. Vreven,J.A.Montgomery Jr., J. E. Peralta, F. Ogliaro, [16] a) B. M. Jones, R. I. Kaiser, J. Phys. Chem. Lett. 2013, 4,1965 –1971; b) M. M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin,V.N.Staroverov,R.Ko- Fçrstel, P. Maksyutenko, B. M. Jones,B.-J. Sun, S.-H. Chen,A.H.H. bayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyen- Chang, R. I. Kaiser, ChemPhysChem 2015, 16,3139–3142. gar,J.Tomasi, M. Cossi,N.Rega, J. M. Millam,M.Klene, J. E. Knox, J. B. [17] a) S. Maity,R.I.Kaiser,B.M.Jones, Faraday Discuss. 2014, 168,485 –516; Cross, V. Bakken, C. Adamo, J. Jaramillo,R.Gomperts, R. E. Stratmann, b) S. Maity,R.I.Kaiser,B.M.Jones, Astrophys. J. 2014, 789,36. O. Yazyev,A.J.Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, [18] a) J. W. Hepburn, Laser Techniques in Chemistry,Wiley,New York, NY, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, 1994;b)R.Hilbig, R. Wallenstein, Appl. Opt. 1982, 21,913 –917;c)R. S. Dapprich, A. D. Daniels, Ö.Farkas, J. B. Foresman, J. V. Ortiz, J. Cio- Hilbig, G. Hilber, A. Lago, B. Wolff, R. Wallenstein, Nonlinear Optics and slowski, D. J. Fox, Gaussian,Inc.,Wallingford, CT, 2013. Applications 1986, 613. [25] H.-J. Werner,P.J.Knowles, G. Kinizia, F. R. Manby,M.Schutz,GAUSSIAN [19] a) A. D. Becke, J. Chem. Phys. 1993, 98,5648 –5652;b)C.Lee, W. Yang, 09 version 2010.1 ed., 2010,p.apackage of ab initio programs. R. G. Parr, Phys. Rev.B1988, 37,785–789. [26] NIST,inNIST Standard Reference Database Number 69 (Ed.:P.J.Linstrom, [20] R. Krishnan, M. Frisch,J.Pople, J. Chem. Phys. 1980, 72,4244 –4245. Mallard, W. G.), 2011. [21] a) G. D. Purvis, R. J. Bartlett, J. Chem.Phys. 1982, 76,1910 –1918;b)G.E. [27] M. Fçrstel, P. Maksyutenko, B. M. Jones, B. J. Sun, A. H. H. Chang, R. I. Scuseria,H.F.Schaefer III, J. Chem.Phys. 1989, 90,3700 –3703;c)G.E. Kaiser, ChemPhysChem 2015, 16,3139 –3142. Scuseria,C.L.Janssen,H.F.Schaefer III, J. Chem. Phys. 1988, 89,7382 – [28] a) R. I. Kaiser,S.P.Krishtal, A. M. Mebel, O. Kostko, M. Ahmed, Astrophys. 7387;d)J.A.Pople, M. Head-Gordon, K. Raghavachari, J. Chem. Phys. J. 2012, 761,178;b)O.Kostko, J. Zhou,B.J.Sun, J. S. Lie, A. H. Chang, 1987, 87,5968 –5975. R. I. Kaiser,M.Ahmed, Astrophys. J. 2010, 717,674. [22] a) G. Knizia, T. B. Adler,H.-J.Werner, J. Chem. Phys. 2009, 130,054104; [29] a) P. R. Schreiner,H.P.Reisenauer,D.Ley,D.Gerbig, C.-H. Wu,W.D. b) T. B. Adler,G.Knizia, H.-J. Werner, J. Chem. Phys. 2007, 127,221106 – Allen, Science 2011, 332,1300 –1303;b)P.R.Schreiner,H.P.Reisenauer, 224100. F. C. Pickard IV,A.C.Simmonett,W.D.Allen,E.Mµtyus, A. G. Csµszµr, [23] K. A. Peterson, T. B. Adler,H.-J.Werner, J. Chem. Phys. 2008, 128,084102. Nature 2008, 453,906 –909. [24] Gaussian09(Revision D.01),M.J.Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,M.A.Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Men- nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, Manuscript received:April 22, 2016 A. F. Izmaylov,J.Bloino,G.Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, Final Article published:June 27, 2016

ChemPhysChem 2016, 17,2726 –2735 www.chemphyschem.org 2735  2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim