Table Of Content

VUV-synchrotron absorption studies of N and CO at 900 K 2 5 1 M. L. Niua, A. N. Heaysb, S. Jonesb,c, E. J. Salumbidesa,d, E. F. van Dishoeckb, 0 N. De Oliveirae, L. Nahone, W. Ubachsa,∗ 2 n aDepartment of Physics and Astronomy, LaserLaB, VU University,De Boelelaan 1081, 1081 HV Amsterdam, The a Netherlands J bLeiden Observatory, LeidenUniversity,PO Box 9513, 2300 RA Leiden, The Netherlands cSchoolofPhysicsandAstronomy, TheUniversityofNottingham, UniversityPark,Nottingham,NG72RD,UnitedKingdom 3 dDepartment of Physics, Universityof San Carlos, CebuCity 6000, Philippines 1 eSynchrotron Soleil, Orme des Merisiers, St. AubinBP 48, 91192, Gif sur Yvettecedex, France ] h p - m Abstract o t Photoabsorption spectra of N2 and CO were recorded at 900K, using the vacuum-ultraviolet Fourier- a transform spectrometer at the DESIRS beamline of synchrotron SOLEIL. These high-temperature and . s high-resolution measurements allow for precise determination of line wavelengths, oscillator strengths, and c i predissociativelinebroadeningofhighly-excitedrotationalstateswithJ uptoabout50,andalsovibrational s hot bands. In CO, the perturbation of A1Π−X1Σ+ vibrational bands (0,0) and (1,0) were studied, as y h well as the transitions to perturbing optically-forbidden states e3Σ−, d3∆, D1∆ and a′3Σ+. In N2, we p observedlineshiftsandbroadeninginseveralb1Π −X1Σ+ bandsduetounobservedforbiddenstatesof3Π u g u [ symmetry. The observedstateinteractionsaredeperturbedand,forN ,usedtovalidateacoupled-channels 2 1 modeloftheinteractingelectronicstates. Thisdataisappropriateforuseinastrophysicalor(exo-)planetary v atmosphericapplicationswherehightemperaturesareimportantandinfuturespectroscopicmodelsofthese 5 molecules. 9 1 Keywords: Synchrotron radiation, Fourier-transformspectroscopy, Carbon monoxide, Molecular nitrogen 3 0 1. 1. Introduction wavelengthsintherange40−200nm[2,3]. Inrecent 0 years, this unique instrument has been used to per- 5 ThetechniqueofFourier-transformspectroscopyis form highresolutionspectroscopicstudies on a num- 1 typically applied to the infrared and optical wave- ber of small molecules in the gas phase that exhibit : v length domains. The interferometric principle re- strongly-structuredmulti-linespectra,suchasH2 [4], Xi quires a beam-splitter for which no materials exist HD [5], N2 [6], and CO [7]; as well as for molecules r inthe farvacuumultraviolet(VUV) partofthe elec- with more continuum-like spectra, such as CO2 [8]. a tromagnetic spectrum. At the DESIRS beamline of These studies amply demonstrate the multiplex ad- the SOLEIL synchrotron[1] this problem was solved vantageoftheFourier-transformtechniquebyreveal- by developing a VUV Fourier-transform spectrome- ingmanyhundredsofabsorptionlinesinasingle-scan ter (FTS) based on beam-splitting by wave-front di- window of some 5nm, determined by the bandwidth vision, thus enabling high-resolution spectroscopy at ofthe beamline undulator source. Alternatively,the setup was used to determine photo-absorption cross sections[9]andpredissociationlinewidths(andhence ∗Correspondingauthor rates) of excited states of small molecules [6, 10]. Email address: [email protected] (W.Ubachs) Preprint submitted to Journal of Molecular Spectroscopy January 15, 2015 The FTS-VUV setup has been used for gas-phase absorption spectroscopy under varied measurement conditions. Most studies have been performed in a quasi-staticgasenvironmentwherethegassampleef- fusively flows through a narrow capillary-shaped absorption cell, with the absorption path aligned with the VUV beam emanating from the undulator. This cell was not equipped with windows, to permit pas- sage of the VUV beam through the sample gas into the FTS-VUV instrument for spectroscopic analysis. For this geometry,differentialpumping maintainsan ultrahigh vacuum in the FTS and DESIRS beam line. The column density of absorbing gas is lim- ited by the pumping conditions and vacuum require- ments of the beam line. In any case, there is a pres- sure gradient over the cell length falling off toward bothends,complicatinganyabsolutecolumndensity calibration. In further studies dedicated to cross- section measurements, a movable gas cell was used of ∼19mm length and sealedby either MgF or LiF 2 wedged windows. This allowed for somewhat-higher pressures and controlled gas column densities [11]. Studiesusingthiscellarelimitedinwavelengthrange to λ > 105 nm [12] by the short-wavelength opac- ity of the windows. In some experiments requiring the simplification of congested spectra, a molecular jet expansion was employed as well as cooling of the quasi-static gas cell with liquid-nitrogen or liquid- helium. A comparison between these techniques was performed in a study of the D spectrum [13] which 2 also demonstrated improved spectral resolution and Figure 1: (a) Cross-section drawing of the gas sample cham- accuracythroughreductionoftheDopplerwidthun- ber mounted in the FTS-branch of the DESIRS beam line at SOLEIL. The high temperature windowless cell is located in der these conditions. thecenterofthechamberandisseparatedfromtheultra-high Forthepresentstudyaheatedcellisimplemented, vacuumofthebeamlinebytwostagesofdifferentialpumping. allowingforthe recordingofspectra attemperatures (b)Theinsetshowsdetailsofthecellandtheshieldingwhere of ∼ 1000K. The high-resolution FTS allows for the half ofthe cylindrical shellhas been removed forclarity. The heating element is wrapped all over the cylindrical cell inside measurementandanalysisofseverelycongestedspec- agrooveinordertoincreasesurfacecontactwiththecell. The tra at these elevated temperatures. Such spectra gasisflowingthrougha7.5×4.5mmtubeintotheheatedcell. bear significance for the modeling of astrophysical Thecopperbasecanbecooleddownwithathermalizedwater circulationsystem. shock-wave regions [14], or other high-temperature astrophysicalregionswherethespectroscopyofsmall molecules is key to understanding the phenomena, suchase.g.,inthephotospheresofwhitedwarfs[15]. Another goalofperformingspectroscopyofhot sam- ples is to follow rotational progressions to high J- quantum numbers, where perturbations due to non- 2 Born-Oppenheimer effects are abundantly present. temperature (for the cell) and never went beyond a Some pertinent perturbation features specifically oc- maximum of 200oC with no visible consequence or curring at high rotational quantum numbers will be damage. A thermocouple is connected at one end of shownhereinVUV-absorptionspectraofCOandN the cell to obtain an indication of the gas tempera- 2 recorded at 900K. ture. The quasi-static gas density inside the heated cell was monitored from outside the vacuum by a 1 mbar range capacitive gauge. The gas column den- 2. Experimental sitywasadjustedbyaneedlevalveinordertohavea The VUV Fourier-Transform spectrometer at the constant continuous flow through the cell during the DESIRS beamline is a permanent end station ded- photoabsorption measurements. The effective col- icated to high-resolution photoabsorption studies in umndensityalongtheabsorptionpathinsidethecell the range 4−30 eV [1]. The instrument has been varied from 4×1014 to 1.2×1017cm−2 and was ad- described in detail previously [2, 3]. In short, the justed accordingto the crosssectionsofthe recorded spectrometerisbasedonwave-frontdivisioninterfer- bands. ometryusingreflectivesurfaces,thusallowingtheex- Inthepresentstudy,theFTS-VUVwassettopro- tension ofthe FTS technique into the far VUV spec- vide an instrumental linewidth of 0.27cm−1. The tral range. The undulator white beam is used as the Doppler broadening corresponds to 0.28 cm−1 at a background,feedingtheFTSbranchandpermitsthe frequency of 65000cm−1, temperature of 900K, and recording of a spectral bandwidth ∆E/E =7%, cor- molecularmassof28amu. Afterconvolutionwiththe responding to 5 nm, on each scan. The typical inte- instrumentwidth,aspectrallinewidthof0.39cm−1is grationtime for a single scan is less than 30 minutes anticipated for unsaturated and non-predissociation- to obtain a signal-to-noise ratio for the background broadened N and CO lines. The FTS spectra are 2 continuum level of ∼400. intrinsicallywavelengthcalibratedbymonitoringthe The windowless absorption cell is a 40 cm long movement of the travel arm in the interferometer T-shaped tube with a rectangular cross section, in- which is controlled by a HeNe-laser [2, 3]. Addi- stalled under vacuum inside the multi-purpose gas tional and improved calibration can be derived from sample chamber of the FTS branch (Fig. 1). The co-recording special calibration lines, e.g. resonance cross section of the tube (7.5 × 4.5 mm) is adapted linesofnoblegases[16,17]. Inthepresentcaseofthe to the astigmatic shape and dimensions of the un- CO spectra the very accurate laser-calibration data dulator source in this section of the beam line. An of the low-J rotationallines in the A−X bands, ac- Inconel heating element (thermocoax) is wrapped curate to ∆λ/λ = 3×10−8, were used [18]. For the aroundthe tube sitting in a groovedesignedto max- present high-temperature measurements with larger imize the contact surface between the heating wire Doppler-broadeningtheFTSwasnotusedinitsvery andthecell,andensuringthegasflowinginthe tube highest resolution mode and the spectral accuracy is uniformly heated. Two semi-cylindrical shells are typically reached is estimated at 0.02 cm−1. The pressed aroundthe cell in order to improve the ther- accuracy is somewhat lower for weaker and blended mal contact during the heating operation. An extra lines. stainlesssteel boxis alsoinstalledto shield radiation originating from the cell. Inconel allows for heating thecellupto1000K,although,thepresentmeasure- 3. Absorption spectra of N2 ments were done at a maximum temperature of 900 K. The cell is mounted on a copper base plate that FiveN vibrationalbandswereanalysedappearing 2 can be cooled by water circulation, although during in our spectrum between 100400 and 108500cm−1 the experiments the setup was operated without the (99.6 and 92.2nm). These bands are spectroscopi- cooling system. The temperature of the base-mount cally denoted b1Π −X1Σ+(v′,v′′ = 0) for v′ = 0, u g was carefully monitored within the covered range of 1, 2, and 10, and c′1Σ+ − X1Σ+(v′ = 0,v′′ = 1) 4 u g 3 b−X(2,0) 0.4 34 30 26 22 18 14 10 6 2P(J′′) 37 33 29 25 21 17 13 9 51Q(J′′) ts) 36 32 28 24 20 16 12 8 0R(J′′) i rb.un 0.3 2118 1714 1310 96 521PQ((JJ′′′)′) b−X(1,0) 29 25 211713929510P25(J18′′R)(cJ′4′′−) X(0,1) (a 24 20 16 12 80R(J′′) e c 0.2 n a t t i m s n a 0.1 r T 0.0 101200 101400 101600 101800 102000 102200 Transitionwavenumber(cm−1) Figure2: Photoabsorptionspectrumshowingthebandsb−X(2,0),c′−X(0,1),andpartofb−X(1,0);andfurtherabsorption 4 linesarisingfromH2contamination,high-J′linesofb−X(3,0),andofunknownorigin(circles). Thelowertraceindicatesthe residualerroraftersubtractingamodelspectrum. ′ ′′ (where v and v are upper- and lower-state vibra- In many cases a more useful measurement of the tional quantum numbers, respectively; and hereafter strength of a line than the integratedcross section is we neglect electronic-state term symbols); and have a derived band f-value, calculated by factoring the been previously observed in room-temperature or ground-state rotational thermal population as well expansion-cooled synchrotron- or laser-based experi- as rotation-dependentHo¨nl-Londonlinestrengthfac- ments [19, 20, 21, 22, 23, 24]. Part of our photoab- tors. Band f-values are only weakly dependent on ′ sorption spectrum showing three of these bands is upper-state J for unperturbed bands. plotted in Fig. 2. A listing of the deduced term val- The main difficulties encountered while analysing ′ ues for the e and f components of the observedb(v ) the hot-cell N spectrum were the significant con- 2 levels is given in Table 1. tamination from highly-excited rotational structure The analysis of DESIRS FTS spectra of molecular of nearby bands and obtaining a correct calibration nitrogen has been discussed previously [6]. This in- of the temperature in the cell. Groups of lines from volves simulating each observed absorptionline with the same vibrational band were sometimes analysed a Voigt profile defined by a GaussianDoppler width, while assuming correlated wavenumbers, widths and Lorentzian natural line width, transition wavenum- strengthstofacilitatetheanalysisofblendedspectral ber, and integrated cross section. A summed cross regions. Thatis,P(J′′−1)andR(J′′+1)transitions section is then transformed into an absorption spec- are connected to a common excited state so the dif- trum by the Beer-Lambertlaw and convolvedwith a ference in their transition wavenumbers was fixed to sinc function simulating the instrumental resolution knownground-stateenergylevels[34]andacommon ′ ′ of the FTS. All parameters defining the model ab- linewidth assumed. A weak J -dependence (or J - sorption spectrum are then automatically optimised independence) was also assumed for some linewidths to best agree with the experimental scan. or f-values. 4 Table1: Experimentaluppertermvaluesa forobservedlinesinN indexedbyexcited stateangular-momentum,J′. 2 b(0) b(1) b(2) b(10) J′ e-parity f-parity e-parity f-parity e-parity f-parity e/f-parityb 1 100819.84(2) 100819.91(4) 101454.460(5) 101454.455(8) 102154.82(1) 102154.79(3) 108374.115(7) 2 100825.54(3) 100825.61(1) 101460.090(7) 101460.097(3) 102160.33(1) 102160.359(7) 108378.968(4) 3 100834.25(1) 100834.27(2) 101468.560(2) 101468.541(4) 102168.691(5) 102168.658(9) 108386.239(4) 4 100845.85(2) 100845.821(9) 101479.798(4) 101479.803(3) 102179.754(8) 102179.731(8) 108395.955(4) 5 100860.303(7) 100860.32(1) 101493.874(2) 101493.877(3) 102193.637(4) 102193.70(1) 108408.050(2) 6 100877.67(1) 100877.660(8) 101510.764(3) 101510.75(1) 102210.295(7) 102210.267(6) 108422.591(3) 7 100897.887(6) 100897.895(9) 101530.443(2) 101530.456(4) 102229.655(4) 102229.675(8) 108439.530(2) 8 100921.01(1) 100920.987(7) 101552.946(2) 101552.940(3) 102251.874(9) 102251.846(8) 108458.875(2) 9 100946.968(6) 100946.96(1) 101578.238(2) 101578.24(1) 102276.746(6) 102276.770(9) 108480.618(2) 10 100975.83(1) 100975.815(7) 101606.338(3) 101606.348(6) 102304.416(6) 102304.426(7) 108504.758(2) 11 101007.503(6) 101007.52(3) 101637.230(2) 101637.230(3) 102334.845(3) 102334.856(7) 108531.283(2) 12 101042.036(9) 101042.04(1) 101670.911(2) 101670.912(3) 102368.018(4) 102368.018(7) 108560.191(2) 13 101079.458(7) 101079.44(1) 101707.379(2) 101707.378(3) 102403.933(3) 102403.926(8) 108591.479(2) 14 101119.70(1) 101119.690(8) 101746.629(2) 101746.626(3) 102442.572(3) 102442.635(8) 108625.157(2) 15 101162.74(1) 101162.74(1) 101788.652(2) 101788.654(3) 102483.948(2) 102483.965(6) 108661.209(2) 16 101208.58(2) 101208.600(9) 101833.455(2) 101833.456(3) 102528.044(3) 102528.086(5) 108699.644(2) 17 101257.250(9) 101257.29(2) 101881.006(2) 101881.013(3) 102574.875(2) 102574.887(4) 108740.459(3) 18 101308.77(2) 101308.71(1) 101931.329(3) 101931.332(3) 102624.387(3) 102624.415(4) 108783.684(5) 19 101362.95(1) 101362.91(3) 101984.400(2) 101984.400(4) 102676.610(2) 102676.647(4) 108829.02(1)c 20 101419.98(3) 101419.92(1) 102040.210(3) 102040.215(3) 102731.545(3) 102731.574(3) 108876.398(7) 21 101479.69(2) 101479.66(4) 102098.767(2) 102098.764(5) 102789.153(2) 102789.168(4) 108926.565(6) 22 101542.27(6) 101542.13(3) 102160.044(4) 102160.042(3) 102849.447(2) 102849.489(3) 108979.148(5) 23 101607.38(2) 101607.27(6) 102224.043(3) 102224.045(6) 102912.413(2) 102912.457(4) 109034.005(5) 24 – 101675.24(3) 102290.766(6) 102290.761(4) 102978.050(3) 102978.097(3) 109091.321(4) 25 101745.90(5) 101745.66(7) 102360.177(4) 102360.171(8) 103046.349(2) 103046.384(4) 109150.906(5) 26 – 101819.01(6) 102432.300(8) 102432.275(6) 103117.309(4) 103117.327(3) 109212.899(4) 27 101895.03(5) – 102507.081(5) 102507.08(1) 103190.873(2) 103190.911(5) 109277.088(7) 28 – 101973.59(9) 102584.52(1) 102584.532(8) 103267.083(4) 103267.124(4) 109343.734(9) 29 – – 102664.686(8) 102664.67(2) 103345.914(3) 103345.956(6) 109412.79(1) 30 – 102138.44(6) 102747.45(4) 102747.45(1) 103427.360(6) 103427.403(5) 109484.23(4) 31 – – 102832.91(1) 102832.83(3) 103511.408(4) 103511.446(8) – 32 – – – 102920.94(2) 103598.05(1) 103598.083(6) – 33 – – 103011.62(3) – 103687.265(6) 103687.32(1) – 34 – – – – 103779.06(1) 103779.09(2) – 35 – – – – 103873.41(1) 103873.46(2) – 36 – – – 103299.14(5) 103970.27(3) 103970.33(2) – 37 – – – – 104069.79(4) 104069.80(5) – 38 – – – – – 104171.73(4) – 39 – – – – – 104276.17(9) – 40 – – – – – 104383.05(5) – aWithunitsofcm−1andparenthetical1σfittinguncertaintiesintermsoftheleast-significantdigit. Theestimatedabsolute calibrationuncertaintyis0.04cm−1. bNosplittingofe-andf-paritylevelswasobserved(apartfromforJ′=18)andthesewereassumedidentical. cAsplittingofe-andf-paritylevelswasobserved,withtermvalues 108829.02(1) and108828.82(3)cm−1,respectively. 5 Lines with natural widths below about 0.05cm−1 0.0035 full-width half-maximum (FWHM) are not reliably measured in our experiment due to concurrent in- 0.0030 bbb−−−XXX(((000,,,000))) strumentandDoppler broadeningby 0.27andabout 0.0025 0.4cm−1FWHM, respectively. No linewidths are 0.0020 then measurable from our spectrum for transitions 0.0015 ′ to the weakly-predissociatedc4(0) level [25]. 0.0010 Wecompareourmeasuredf-valuesandlinewidths 0.0005 with those calculated from an existing model of 0.0000 N2 photoabsorption and dissociation, including pho- 0 5 10 15 20 25 30 35 40 toabsorbing 1Π and 1Σ+ excited states and spin- 0.012 u u forbidden but dissociative 3Π states [6, 26, 27, 0.010 bbb−−−XXX(((111,,,000))) u 28, 29]. This model solves a coupled-Schrödinger 0.008 equation (CSE) for the nuclear motion of the excited molecule, where the necessary potential-energy s) 0.006 s curves and state interactions have been optimised le 0.004 P branch n with respect to a large body of room-temperature o Qbranch si 0.002 experimental data. This model has been success- n Rbranch fully employed previously in applications of atmo- me 0.000 0 5 10 15 20 25 30 35 40 spheric [30] and astronomical photochemistry [31, di 0.040 ( 32], including temperatures as high as 1000K. Here, ue 0.035 bbb−−−XXX(((222,,,000))) we seek to validate the extrapolation of the CSE al 0.030 v modeltohightemperaturebycomparisontoournew - 0.025 f measurements. d 0.020 n a 0.015 B 3.1. Temperature calibration 0.010 0.005 The f-values of b−X(v′,0) transitions were used 0.000 to calibrate the ground-state rotational temperature 0 5 10 15 20 25 30 35 40 0.014 and N column density in the hot cell by compar- 2 0.012 bbb−−−XXX(((111000,,,000))) ison with previously-measured absolutely-calibrated ′ 0.010 f-valuesforv =0,1,2,and10[23,24]. Theresultant valuesare(6.35±0.64)×1015cm−2and901±26K,re- 0.008 spectively. The reference data was recorded at room 0.006 temperature, included rotational levels as high as 0.004 J = 23, and themselves have an absolute column 0.002 density uncertainty of 10% which is also the dom- 0.000 inant systematic uncertainty of our f-values. The 0 5 10 15 20 25 30 35 40 final agreement between the present measurements J′ and the reference data, shownin Fig. 3, is very good despitethefactor-of-5differenceingroundstatepop- Figure 3: Band f-values of all transitions observed in our ulations, for example, at J′ =20. experiment as a function of excited-state angular-momentum The c′ − X(0,1) band appears quite weakly in quantumnumber,J′,andwith1σrandomfittinguncertainties 4 (circles with error bars). A10% systematic erroralsoapplies our spectrum and was analysed in order to esti- and some f-values were analysed assuming J′-independent mate the vibrational temperature in the hot cell. ranges (horizontal error bars). Also shown are previously- For this, constant band f-values were assumed over measured f-values [23, 24] (crosses), and calculated by the CSEmodel(solid black curves). 6 ) 0.040 a well-known localised perturbation by the crossing nless 0.035 cc′4′4−−XX((00,,11)) rotTahtieonnaelwtelyrm-mseearsiuesreodf bc′′1Σ−+u(Xv′(0=,11)) [f2-0v]a.lues are o 0.030 4 si somewhat smaller than the simulated values and an n 0.025 e alternative model adopting an 800K distribution of m 0.020 i ground-state levels leads to the better agreement in- d ( 0.015 dicatedinFig.4. Thismayindicateincompletether- e u 0.010 malisation of the N in our experiment leading to a l 2 a v 0.005 lesserdegreeofvibrationalexcitationthanrotational. - f 0.000 A similar result is found in Sec. 4 for the CO rota- nd 0 5 10 15 20 25 30 35 40 tional and vibrational temperatures. a B J′ As a final check on the temperature of our sample of N , the Doppler broadening in our experi- 2 Figure 4: Band f-values of all c′4 −X(0,1) transitions ob- ment was measured by reference to lower-J′ levels servedinourexperimentasafunctionofexcited-stateangular- ofthe b−X(1,0)absorptionband, whosepredissoci- momentum quantum number,J′,andwith1σ randomfitting ationbroadeningisknowntobebelowourresolution uncertainties (circles with error bars). A 10% systematic error also applies and some f-values were analysed assuming limit[21,26]. Wefindakinetictemperaturefromthis J′-independentranges(horizontalerrorbars). Alsoshownare of about 930K, with an uncertainty estimated to be alternative experimental f-values assuming an 800K ground significantly greater than for our deduced rotational stateexcitation(dashed lines, open circles) andreferenceval- temperature. ues calculated from a combination of CSE and experimental data(solid curve). 3.2. Results Transition wavenumbers for all observed b − ′′ ′′ ′ small ranges of most P(J ) and R(J ) lines as in- X(v ,0) bands were reduced to term values using dicated piecewise in Fig. 4. Simulated c′ −X(0,1) accurate N ground-state molecular constants [34]. 4 2 f-values are also shown, with magnitude calculated Term values for these bands have been deduced pre- from the ratio of c′ − X(0,0) and c′ − X(0,1) f- viouslyforrotationallevelswithJ′ ashighas36and 4 4 values deduced by electron-excited fluorescence [33], with about 0.1cm−1 uncertainty. Our term values f /f =6.3±0.4,and an absolute c′ −X(0,0) are listed in Table 1 and have statistical uncertain- (0,0) (0,1) 4 absorptionf-valuemeasurement[23]. Thestatedun- ties of around 0.01cm−1. The absolute calibration certaintiesofthetwoexperimentalvaluesusedinthis of our experiment was made by comparison of ar- comparisonare6%[33]and10%[23],respectively,al- gonresonancelinesappearinginourspectrawiththe thoughthelattershouldbeneglectedbecauseourex- NIST database and has an estimated uncertainty of perimental column-density is calibrated to the same 0.04cm−1. reference. We used the CSE model to simulate the Our deduced band f-values are plotted in Fig. 3. significant rotational dependence of c′ −X(0,1) f- The decrease of b−X(v′,0) f-value with J′ contin- 4 values and assumed a 900K distribution of ground- ues to the highest-excitation lines that we observe state rovibrational levels. This simulation then cor- and is in perfect agreement with values predicted by rectly reproduced the observed splitting of P- and the CSE model. This decrease is effectively due to a R-branch f-values for c′ −X(0,1) transitions with decreasing Franck-Condon overlap of b(v′) and X(0) 4 ′ increasing J . This splitting is also known to occur vibrationalwavefunctionswithincreasingcentrifugal for the c′ − X(0,0) fundamental band [23] and is distortion. 4 ′ ′ the result of a rotational-perturbation of c (v = 0) Measured natural linewidths and comparable val- 4 by nearby 1Π levels [6]. Additionally, c′ −X(0,1) ues from previous photoabsorption and resonantly- u 4 ′ transitions to J = 11, 12, and 13 levels are signifi- enhancedphotoionisationexperiments[19,22,23]are cantly weakened relative to their neighbours due to shown in Fig. 5. The widths of b(0), b(1), and b(2) 7 averaged over their J ≤ 5 levels have been previ- 1.2 ously deduced from laser-based lifetime or linewidth 1.0 bb((00)) measurements [19, 21, 22]. The rotationally-resolved 0.8 J-dependent broadening of b(2) and b(10) levels 0.6 have been measured in synchrotron-based experiments [23, 24], and an increasing b(1) predissocia- 0.4 ′ tion width with J has also been experimentally de- 0.2 duced [35, 36]. Our newly-measured widths show 0.0 goodagreementwithallreferencedatabutwithgen- 0.50 5 10 15 20 25 30 35 40 erally reduced scatter. Two interesting new pieces of information are discussed below. 0.4 bb((11)) First, the decreasing b(2) widths are now shown M) 0.3 to pass through a minimum at J ≃ 28. This com- H e-parity plex behaviour is well-reproducedby the CSE model W 0.2 which includes a mechanism for predissociative line f-parity F broadening by including unbound electronic states 1 0.1 previousresults − amongst its coupled channels [26, 27]. The critical m c 0.0 interactions in this case are the spin-orbit coupling ( 0 5 10 15 20 25 30 35 40 hs 1.0 of b(2) with vibrationally-bound levels of the C3Πu widt 0.8 bb((22)) sthtaeteuannbdoutnhdeircsounbtisneuquumentoefletchteroCni′c3iΠntersatcattieo.nwTihthe e u n 0.6 dominant perturber of b(2) is the C(8) level which i ll lies only 100cm−1 lower in energy and has been pre- ra 0.4 viously identifiedin aphotoabsorptionspectrum [37] u at 0.2 and found to have a linewidth of 18cm−1 for J′ less N than about 10, despite the nominally spin-forbidden 0.0 nature ofthis transition. All other bound 3Π states 0 5 10 15 20 25 30 35 40 u 1.6 are too remote in energy to contribute significantly 1.4 bb((1100)) to the predissociation of b(2) [28] and the observed 1.2 J′-dependence of its widths must then closely scale 1.0 with the broader widths of C(8). 0.8 Second, there is a sharp peak in the linewidths of 0.6 b(10) shown in Fig. 5. Increasing widths beginning 0.4 ′ around J = 15 were known from a poorer signal- 0.2 to-noise-ratio room-temperature spectrum [24], but 0.0 ′ are now better resolved and to higher-J . There is 0 5 10 15 20 25 30 35 40 also a perturbation of b(10) rotational energy lev- J els near J′ = 19, as shown in Fig. 6 as a 0.8cm−1 deflection of its reduced term values. The localised Figure 5: Natural linewidths of e- and f-parity excited-state perturbationofb(10)energiesandwidths indicatesa levelsaccessedinourexperimentasafunctionoftheirangular- momentum quantum number, J′, and with 1σ random fit- level crossing with a predissociation-broadened level ting uncertainties (circles with error bars). Some linewidths of 3Π symmetry, as is known to occur elsewhere in wereanalysedassumingJ′-independentranges(horizontal er- u the N spectrum [37]. rorbars). Alsoshownarepreviously-measuredlinewidths[19, 2 One candidate for the role of b(10) perturber is 21, 22, 23, 24] (yellow lines and crosses), and linewidths cal- culatedbytheCSEmodel(solidblackcurves)andatwo-level the v′ = 16 level of the C3Πu state, which has localinteractionmodel(dashed black curve). been observed for J′ ≤ 10 [37] and has a band 8 ) 1 0.8 CO A(v) - X(v=0) − 0 1 2 3 m b(10) c 0.6 ( 300 K s 900 K e 0.4 u l a v 0.2 CO A(v) - X(v=0) m r 0.0 CO A(v) - X(v=1) 0 1 2 3 e t 0 1 2 3 4 d −0.2 e c u d −0.4 e 0 5 10 15 20 25 30 R J′ 64000 66000 68000 70000 Frequency (cm-1) Figure6: Experimental f-parityterm values ofb(10) reduced bythesubtractionofacubicpolynomialofbestfitintermsof J′(J′+1)(circles). Alsoshownarereducedtermvaluesfrom Figure7: OverviewspectrumoftheCOA1Π−X1Σ+ system theb(10)/3Πu interactionmodel(curve). including (0,0), (1,0), and (2,0) bands, and some hot bands. Thetoppanelshowsthespectrumwhichismeasuredatroom temperature(300K).Inthebottompanelisthehotspectrum (900 K). The sharp absorption line inthe upper spectrum at origin only 80cm−1 below that of b(10). However, 68045.156 cm−1 isaxenonresonanceline. a rotational constant calculated from the observed C(16) levels, 1.153cm−1 [28], is too low to cross model parameters are the strength of the b(10) and the b(10) term series where the observed perturba- 3Π spin-orbit interaction, 7cm−1, and the deper- tion peaks at J ≃ 18. Alternatively, the v = 2 u turbed predissociation widths of 3Π levels. Good level of the G3Π state has been observed [38] to u u lie nearby, 340cm−1 below b(10), and undoubtedly agreement could only be found when assuming the ′ ′ latter increases linearly in term of J (J + 1) from has a larger rotational constant more characteristic ofN Rydberglevels,about1.9cm−1. Acrossingbe- 20cm−1 at J′ = 18, to 60cm−1 at J′ = 29. All of 2 these deduced values are intermediate between those tween G(2) and b(10) is then conceivable. The observed widths of C(16) for J′ ≤ 10 are less than known or predicted for C(16) and G(2), [28, 37, 38], 0.5cm−1FWHM, whereas G(2) is predicted to be indicatingthatthe perturberofb(10)isanelectronic much broader, about 90cm−1FWHM, by the CSE admixture of the C3Πu and G3Πu states. Indeed, the coupled-channels model of Lewis et al. [28] pre- model of Lewis et al. [28]. dicts this, as well as a further significant admixture To analyse the width and term value perturba- of the F3Π Rydberg state into the nominal C(16) tion of b(10) further we defined a two-level model u and G(2) levels. of b(10) interacting by the spin-orbit operator with a 3Π level (including all triplet sublevels) and op- u timised its various parameters to match our experi- 4. Absorption spectra of CO A−X(0,0) and mentaldata. Thiswasdoneinanidenticalfashionto (1,0) bands similar deperturbations of N 3Π /1Π interactions 2 u u by Lewis et al. [37]. Comparisons of experimental The novel hot cell configuration was employed for widths and reduced term values with this model are the further investigation of the A1Π−X1Σ+ system shown in Figs. 5 and 6 and find overall good agree- of CO for the lowestvibrationalbands in the excited mentwhenadoptinga3Π statewithatermoriginof state. Figure 7 shows an overview spectrum of some u approximately108150cm−1,arotationalconstantof bands recordedat300Kand900K.With thehigher 1.6cm−1,andaspin-orbitsplittingof30cm−1(where gastemperature,therotationalenvelopeofeachband the sign of the latter is unconstrained). Two further includes higher J-quantum numbers and hot bands 9 Table2: Observedhigh-J transitionfrequencies(invacuumcm−1)oftheCOA1Π−X1Σ+(0,0)and(1,0)bandsobtainedwith thehotcell. Lower-J transitionsarelistedinRef.[7]. Theestimateduncertainty(1σ)is0.02cm−1 exceptforweakorblended lines. A1Π−X1Σ+(0,0) A1Π−X1Σ+(1,0) J′′ R(J) Q(J) P(J) J′′ R(J) Q(J) P(J) 17 64589.64 21 66140.33 66005.05 18 64583.37 22 66128.24 65986.78 19 64689.77 64566.32 23 66115.44 65967.83 20 64679.86 64549.58 24 66101.66 66023.95 65948.17 21 64669.38 64532.53 25 66091.32 66006.31 65927.76 22 64658.26 64584.85 64515.00 26 66074.03 65989.04 65906.37 23 64646.44 64570.02 64496.90 27 66058.34 65970.28 65888.43 24 64634.07 64554.44 64478.17 28 66042.09 65950.94 65863.57 25 64620.76 64538.20 64458.80 29 66025.18 65929.54 65840.31 26 64604.44 64521.15 64438.67 30 66007.60 65910.53 65816.49 27 64593.32 64501.07 64417.88 31 65989.33 65889.04 65792.02 28 64577.54 64486.13 64393.98 32 65970.28 65866.89 65766.89 29 64559.78 64466.58 64375.28 33 65950.64 65843.98 65741.06 30 64551.43 64445.05 64351.94 34 65930.24 65819.92 65714.55 31 64531.64 64432.91 64326.63 35 65909.14 65798.34 65687.33 32 64512.56 64409.35 64310.71 36 65887.33 65772.41 65659.43 33 64490.76 64386.49 64283.40 37 65864.79 65746.65 65630.83 34 64481.53 64360.95 64256.77 38 65841.53 65720.29 65601.52 35 64459.85 64347.94 64227.48 39 65817.50 65693.27 65571.51 36 64438.85 64322.50 64210.71 40 65793.18 65665.55 65540.79 37 64297.87 64181.54 41 65767.61 65637.11 65509.30 38 64396.16 64272.94 64153.19 42 65741.50 65607.95 65477.54 39 64373.87 64248.16 64124.55 43 65714.55 65577.99 65444.54 40 64350.93 64221.59 64095.36 44 65686.94 65548.84 65410.97 41 64327.65 64194.95 64065.55 45 65658.53 65516.48 65376.69 42 64303.14 64167.66 64035.39 46 65628.96 65484.32 65341.67 43 64278.62 64139.56 64004.55 47 65600.90 65451.21 65305.83 44 64253.03 64111.24 63972.67 48 65569.47 65418.96 65268.92 45 64226.85 64081.88 63940.64 49 65537.24 65383.97 46 64199.92 64052.09 63907.71 50 65348.14 47 64172.37 51 65318.08 48 63989.48 52 65277.96 51 63892.98 53 65402.39 10