Arxiv:0807.1907V1 [Physics.Atom-Ph] 11 Jul 2008 † ∗ Lrcl Oeuefrainhsfcse Nakl Metal on Alkali Work on Dimers

arXiv:0807.1907v1 [physics.atom-ph] 11 Jul 2008 † ∗ lrcl oeuefrainhsfcse nakl metal on alkali work on dimers. most focussed has species, formation other metal molecule than alkali ultracold cool Feshbach Since to zero-energy easier 3]. are through [1, atoms fields tuning be magnetic by photoassoci- with can or by resonances 2] either Molecules gases [1, atomic ation ultracold degeneracy. in Fermi formed con- and Bose-Einstein molecular densation achieving in and molecules ‡ on ttsb ae-ae ehd uha stimulated as such deeply methods to laser-based major by molecules states been Feshbach bound have transferring there in significant recently, have advances Very molecules moments. heteronuclear dipole which occur effort cannot for losses abso- current inelastic and their which intense for in is states, vibrational molecules ground there ultracold lute bound producing this, weakly at of directed in Because are essentially states. they are Fur- when molecules 12]. fast nonpolar 11, heteronuclear [10, undergo loss even molecules trap thermore, to the scattering lead cases that very large other collisions in to in inelastic For dimers tuned only [9]. fermion level, long-lived lengths homonuclear vibrational are highest as they be the such but can cases, 8], molecules specific such 7, of [6, vibra- gases excited formed Quantum highly in states. initially tional are that molecules pro- form and storage information quantum might [5]. in molecules cessing used dipolar have be Ultracold to also predicted [4]. are gases properties quantum novel molecules. dipolar result, nonpolar a and between As exist quadrupole-quadrupole that forces the stronger dispersion than both longer-range are and interactions Dipole-dipole molecules. lcrncades -al [email protected]. E-mail: address: Electronic lcrncades -al [email protected] E-mail: address: Electronic lcrncades -al [email protected]. E-mail: address: Electronic hr spriua neeti omn ultracold forming in interest particular is There ultracold of formation the in interest great is There ohpoosoito n ehahrsnnetuning resonance Feshbach and photoassociation Both ASnmes 71.q 11.j 33.15.Pw 31.15.aj, isotopologs. 37.10.Pq, RbCs for numbers: PACS kHz of tens spin-spi few nuclear st a the explore nonrotating of and and For part isotopologs electron-mediated RbCs fields. the and magnetic by nated and RbK electric for of constants presence coupling density-functio magnetic use and We molecules. ultracold ground-state h oetrttoa ee ftegon lcrncstate, electronic ground the of level rotational lowest the eivsiaeteeeg eeso eeoula laime alkali heteronuclear of levels energy the investigate We .INTRODUCTION I. eateto hmsr,Dra nvriy ot od DH Road, South University, Durham Chemistry, of Department .Aldegunde, J. rudsaeplrmlclsi lcrcadmgei field magnetic and electric in molecules polar ground-state h yefieeeg eeso laimtldimers: metal alkali of levels energy hyperfine The ∗ e .Rvntn it S. Piotr Rivington, A. Ben Dtd a 0 2018) 30, May (Dated: uk uk polar eeo b n bs hc r oia o current for topical focus are We which RbCs, fields. and experiments. magnetic RbK and how on electric explore here levels to in the energy and behave lowest of dimers they metal the purpose alkali investigate The heteronuclear to of con- is gases. to paper quantum methods present developing resulting in the crucial specific trol be in will molecules and produce states to design- methods in laser-based essential an is ing Nevertheless, levels resolu- energy rota- these techniques. the lowest of beyond spectroscopic understanding their are most in splittings of en- tiny tion dimers the The metal of states. alkali structure tional of hyperfine remarkably levels the However, about ergy 19]. known 18, is 17, little Feshbach [16, by espe- formed tuning states dimers, resonance near-dissociation metal the alkali in homonuclear cially of levels energy now is molecules ground-state of reach. within gases Forma- quantum 15]. of 14, tion [13, (STIRAP) passage adiabatic Raman irtoa,rttoa,hprn,SakadZeeman to and analysis our Stark restricting By hyperfine, de- terms. be electronic, rotational, can the fields contributions: vibrational, different electric six and into magnetic composed external of ence oincnb rte 2,2,22] 21, rota- Hamil- [20, the the written of and be parts value can Zeeman constant tonian and level, a Stark vibrational take hyperfine, fixed terms tional, a two in first and the state electronic ground Zuchowski, ˙ hr a enacnieal muto oko the on work of amount considerable a been has There h aitna fadaoi oeuei h pres- the in molecule diatomic a of Hamiltonian The a hoyt aclt ula quadrupole nuclear calculate to theory nal hc r motn neot oproduce to efforts in important are which opig hyaeafwkzfrRbK for kHz few a are They coupling. n ts h eofil pitnsaedomi- are splittings zero-field the ates, a iesi eescreaigwith correlating levels in dimers tal † .MlclrHamiltonian Molecular A. H n eeyM Hutson M. Jeremy and h yefiesrcuei the in structure hyperfine the = L,Uie Kingdom United 3LE, 1 H I THEORY II. rot + H hf + H S + 1 s oeue nthe in molecules Σ ‡ H Z , (1) 2 where III. EVALUATION OF THE COUPLING CONSTANTS

2 2 2 Hrot = BvN − DvN · N ; (2) Nuclear quadrupole coupling constants have been mea- 2 sured for several alkali metal dimers as shown in Table Hhf = X Vi : Qi I. However, the only such species for which the mag- i=1 netic coupling constants have been measured is Na2 [26], 2 and even there the experiments did not resolve hyper- + X ci N · Ii + c3 I1 · T · I2 + c4 I1 · I2; (3) fine splittings for the N = 0 state. To the best of our i=1 knowledge, no experimental data are available for the hy- HS = −µ · E; (4) perfine structure of the molecules we consider here, KRb 2 and RbCs, in their ground electronic state. We therefore carry out electronic structure calculations to estimate HZ = −grµN N · B − X giµN Ii · B(1 − σi). (5) i=1 them. The electric quadrupole coupling constants (eqQ)1 and (eqQ)2, the nuclear shielding, the spin-rotation constants c1 and c2 and the spin-spin coupling constants c3 The three different sources of angular momentum in a 1Σ and c4 are evaluated by density-functional theory (DFT) diatomic molecule are the rotational angular momentum using the ADF package [27, 28], which uses Slater func- N and the spins I1 and I2 of nuclei 1 and 2. The rota- tions and allows the inclusion of relativistic corrections. tional and centrifugal distortion constants of the molecule The rotational g-factor (not implemented in the ADF are Bv and Dv (the centrifugal distortion contribution code) is evaluated with the DALTON package [29]. will not be considered in the calculations). The hyper- The objective of the present paper is to explore the fine Hamiltonian of equation 3 consists of four terms. The behaviour of the molecular energy levels in the presence first is the electric quadrupole interaction with coupling of external fields. A detailed discussion of the features constants (eqQ)1 and (eqQ)2, where qi is the electric field and effectiveness of the many different methods and basis gradient at nucleus i and eQi is its nuclear quadrupole sets available for the calculation of the coupling constants moment. The second is the interaction between the nu- is beyond the scope of the work. However, to estimate clear magnetic moments and the magnetic field created the reliability of the functionals and basis sets employed by the rotation of the molecule, with spin-rotation cou- here we compare the coupling constants obtained for a pling constants c1 and c2. The two remaining terms rep- group of molecules containing alkali metal atoms with resent the tensor and scalar interactions between the nu- experimental results in tables I, II, III and IV. For sim- clear dipole moments, with spin-spin coupling constants plicity we have omitted experimental uncertainties and c3 and c4 respectively. The tensor T describes the angle- vibrational state dependences. It may be seen that the dependence of the direct spin-spin interaction and the calculated coupling constants are generally within 30% of anisotropic part of the indirect spin-spin interaction [22]. the experimental values, except in occasional cases where the experimental values are unusually small (such as c4 The Stark and Zeeman Hamiltonians, equations 4 and for 85Rb35Cl). 5, describe the interaction of the molecule with an exter- Evaluation of hyperfine coupling constants requires a E B µ nal electric field and magnetic field , where is the basis set that properly describes the electron density near molecular dipole moment. The Zeeman Hamiltonian con- the nuclei. Because of this, we employ all-electron basis sists of two terms representing the rotational and nuclear sets rather than valence basis sets with effective core po- Zeeman effects. The former arises because the molecular tentials. However, for core orbitals of heavy elements N rotation produces a magnetic moment grµN , where gr such as those considered here, relativistic effects can be is the rotational g-factor of the molecule, which interacts important. In the present work, relativistic corrections with the external magnetic field. The latter arises from were included by means of ZORA, the two-component I the interaction of the nuclear magnetic moments giµN i zero-order regular approximation [41, 42, 43], including with the magnetic field, where gi is the nuclear g-factor spin-orbit coupling as well as scalar effects (which are I for nucleus i and i is its nuclear spin. The nuclear shield- the equivalent of Darwin and mass-velocity terms in the σ ing tensor i is approximated here by its isotropic part Breit-Pauli Hamiltonian). σ ; terms involving the anisotropy of σ are extremely i i DFT generally performs well for calculations of electric small for the states considered here. The diamagnetic quadrupole coupling constants for main-group elements Zeeman effect is not included in the Hamiltonian as it [44, 45, 46, 47, 48, 49, 50, 51, 52]. Following most of causes level splittings less than 1 Hz for the range of these examples, we use the B3LYP functional [53, 54] in magnetic fields considered in this work. our calculations with the QZ4P basis set (a quadruple-ζ The nuclear g-factors and quadrupole moments are all-electron basis set with four polarization functions). well known [23]. The dipole moments of KRb and RbCs Shielding tensors were evaluated using the KT2 func- have been calculated from relativistic electronic structure tional [55] with the same basis set and relativistic cor- calculations [24, 25]. rection as for the quadrupole coupling constants. For 3

TABLE I: Comparison of electric quadrupole coupling con- TABLE IV: Comparison between rotational g-factors calcu- stants for alkali metals atoms calculated as described in the lated as described in the text and experimentally measured. text with experimental values. The units are MHz. Calc Exp Molecule gr gr Ref. Molecule (eQq)Calc (eQq)Exp Ref. 23 Na2 0.0324 0.0386 [39] 23 − − 39 Na2 0.456 0.459 [26] K2 0.0247 0.0212 [39] 39 K2 −0.279 −0.158 [30] 23Na39K 0.0253 0.0253 [40] 39 19 − − 85 K F 7.87 7.93 [31] Rb2 0.0082 0.0095 [39] 39 7 − − 133 K Li 0.830 1.03 [32] Cs2 0.0051 0.0054 [39] 39K23Na −0.671 −0.718 [32] (for K) a 39K23Na −0.216 0.171 [32] (for Na) 85 Rb2 −2.283 −1.1 [30] distribution). The coupling constant for the direct inter- 85Rb19F −73.1 −70.7 [33] action is [22, 58] 85Rb35Cl −53.5 −52.8 [34] 2 85Rb79Br −46.8 −47.2 [35] µ0 µN −3 RDD = g1g2hR i, (6) 85Rb127I −39.6 −58.9 [36] 4π h 85 7 Rb Li −8.04 −9.12 [32] where R is the internuclear distance. The indirect inter- 133Cs19F 1.30 1.25 [37] action is represented by a tensor J [22, 58] with isotropic 133 35 a Cs Cl 1.05 ≤ 1.1 [38] part Jiso and anisotropy ∆J = Jk − J⊥. The coupling a constants c and c are related to the direct and indirect Only the absolute value was determined experimentally. 3 4 components by [22, 58]

∆J TABLE II: Comparison between spin-rotation coupling con- c3 = RDD − . (7) stants calculated as described in the text and experimentally 3 measured. The label 1 refers to the less electronegative atom and (K, Rb or Cs) and the label 2 to the more electronegative one. The units are kHz. c4 = Jiso (8) Calc Exp Calc Exp Molecule c1 c1 c2 c2 Ref.

23 In the present work, c3 and c4 were evaluated from equa- Na2 0.299 0.243 0.299 0.243 [26] −3 −3 39 19 tions 6 to 8 with hR i ≃ Re , where Re is the equi- K F 0.235 0.270 17.5 10.7 [31] librium distance. The components of J were calculated 85 19 Rb F 0.598 0.498 16.1 10.6 [33] using the same methods as for the quadrupole coupling 85Rb35Cl 0.457 0.395 0.569 0.394 [34] constants, except that the PBE [59] functional was used. 133Cs19F 1.05 0.662 21.9 15.1 [37] This functional produced results slightly closer to the experimental measurements than KT2 for the molecules considered in table III (although the diﬀerences were calculation of shielding tensors of main-group atoms (H, small). BLYP performed well for all except Na2, for C, N, O and F), the performance of this functional is which it gave the wrong sign and order of magnitude; excellent, and is better [56] than that of more popular it also gave qualitatively diﬀerent results from PBE and functionals such as BLYP [53, 57] and B3LYP. KT2 for KRb and RbCs. Two nuclear magnetic moments can interact both di- ADF does not calculate spin-rotation constants directly (through space) and indirectly (via the electron rectly. However, the spin-rotation constants are given approximately by [60, 61, 62]

2meBvgi ci ≈ (σik − σi⊥) for i =1, 2, (9) TABLE III: Comparison between spin-spin coupling constants mp calculated as described in the text and experimentally measured. The units are kHz. where mp and me are the proton and electron masses, Calc Exp Calc Exp Bv is the rotational constant, gi is the nuclear g-factor Molecule c3 c3 c4 c4 Ref. and σik − σi⊥ is the anisotropy of the nuclear shielding 23 Na2 0.298 0.303 1.358 1.067 [26] tensor σi. Two approximations underlie this expression. 39K19F 0.470 0.540 0.032 0.030 [31] First, a quadrupole term has been neglected. Secondly, 85Rb19F 0.751 0.797 0.151 0.237 [33] it was obtained in the frame of the non-relativistic theory 85Rb35Cl 0.032 0.033 0.010 0.026 [34] developed by Flygare [60]. However, previous studies [63] 133 19 and our own results (see table II) suggest that it can be Cs F 0.875 0.927 0.471 0.627 [37] applied reliably in the relativistic case. 4

TABLE V: Nuclear properties and coupling constants for the diﬀerent isotopic species of the KRb molecule. 39K85Rb 39K87Rb 40K85Rb 40K87Rb 41K85Rb 41K87Rb

IK 3/2 3/2 4 4 3/2 3/2

IRb 5/2 3/2 5/2 3/2 5/2 3/2

gK 0.261 0.261 −0.324 −0.324 0.143 0.143

gRb 0.541 1.834 0.541 1.834 0.541 1.834 Bv/GHz 1.142 1.134 1.123 1.114 1.104 1.096

(eQq)K/MHz −0.245 −0.245 0.306 0.306 −0.298 −0.298

(eQq)Rb/MHz −3.142 −1.520 −3.142 −1.520 −3.142 −1.520

σK(ppm) 1321 1321 1321 1321 1321 1321

σRb(ppm) 3469 3469 3469 3469 3469 3469

cK/Hz 19.9 19.8 −24.2 −24.1 10.5 10.4

cRb/Hz 127.0 427.5 124.8 420.1 122.8 413.1

c3/Hz 11.5 38.9 −14.2 −48.2 6.3 21.3

c4/Hz 482.5 1635.7 −599.0 −2030.4 264.3 896.2

gr 0.0144 0.0142 0.0141 0.0140 0.0139 0.0138 µ/D 0.76 0.76 0.76 0.76 0.76 0.76

Lastly, the rotational g-factors were evaluated with the TABLE VI: Nuclear properties and coupling constants for the DALTON program using the KT2 functional and the all- diﬀerent isotopic species of the RbCs molecule. electron basis sets of Huzinaga and coworkers [64, 65]. Again, the choice of the functional is based on its reli- 85Rb133Cs 87Rb133Cs ability for this molecular property [66]. No relativistic corrections were included in this case. Previous calcu- IRb 5/2 3/2 lations [67] for hydrogen halides and noble gas hydride ICs 7/2 7/2 cations including atoms as heavy as I and Xe suggest that gRb 0.541 1.834 relativistic corrections are relatively small for rotational gCs 0.738 0.738 g-factors (less than 5% of the non-relativistic value). Bv/GHz 0.511 0.504 The coupling constants obtained for KRb and RbCs (eQq)Rb/MHz −1.803 −0.872 are given in tables V and VI. All the calculations were (eQq)Cs/MHz 0.051 0.051 carried out at the equilibrium geometries, Re = 4.07 A˚ σRb(ppm) 3531 3531 ˚ for KRb [68] and Re = 4.37 A for RbCs [69]. This ne- σCs(ppm) 6367 6367 glects small corrections due to vibrational averaging even cRb/Hz 29.4 98.4 for v = 0, but nevertheless gives results that are quali- cCs/Hz 196.8 194.1 tatively valid for any low-lying vibrational state. ADF generally gives coupling constants for only one isotopic c3/Hz 56.8 192.4 species, but the others may be obtained by simple scaling. c4/Hz 5116.6 17345.4 The nuclear quadrupole coupling constants scale with the gr 0.0063 0.0062 nuclear quadrupoles Qi, the spin-spin coupling constants µ/D 1.25 1.25 with the product of nuclear g-factors gigj, and the spin- rotation coupling constant with the product of gi and the rotational constant Bv. The rotational g-factor scales in basis sets, a more complicated way that depends on Bv and the shift of the center of mass [70]. |I1M1I2M2NMN i (uncoupled basis); (10)

|(I1I2)IMI NMN i (spin-coupled basis); (11)

|(I1I2)INFMF i (fully coupled basis). (12)

IV. HYPERFINE ENERGY LEVELS Here I and F are quantum numbers for the total nuclear spin and total angular momentum and MI and MF rep- resent their projections onto the Z axis defined by the We calculate the hyperfine levels by diagonalizing the external field. We consider here only cases in which only complete Hamiltonian of equations 2 to 5 in a basis set of one field, electric or magnetic, in present. The matrix ele- angular momentum functions. We employ three different ments corresponding to the different terms of the Hamil- 5

I tonian in each of the basis sets are calculated through 2 4 standard angular momentum techniques [71]. 4 c4

0 M M 3 Rb K

30 3 c4 The use of three basis sets rather than one helps in as- -5/2 -3/2 2 c -2 2 4 39K85Rb signing quantum numbers to the energy levels. Although 1 -5/2 -1/2

0 1

20 -3/2 -3/2 the Hamiltonian matrix is not diagonal in any of the ba- -5/2 1/2 -3/2 -1/2 sis sets employed, it is usually closer to diagonal for one -5/2 3/2 -3/2 1/2 basis than for the others. When one coeﬃcient of an 10 -1/2 -3/2 -1/2 -1/2 eigenvector is much larger than the others, it is possi- -3/2 3/2 -1/2 1/2 1/2 -3/2 ble to assign approximate quantum numbers to the state 0 -1/2 3/2 concerned. However, diﬀerent basis sets achieve this in 1/2 -1/2

Energy (kHz) 1/2 1/2 3/2 -3/2 diﬀerent ﬁeld regimes. 1/2 3/2 -10 3/2 -1/2 3/2 1/2 5/2 -3/2 3/2 3/2 5/2 -1/2 -20 5/2 1/2 5/2 3/2

0 5 10 15 20 Magnetic Field (G) A. Zeeman splitting for rotational ground-state molecules (N = 0) FIG. 1: Zeeman levels for 39K85Rb(v = 0, N = 0).

Figure 1 shows the Zeeman splittings for energy levels 15 39 85 of K Rb with N = 0. The splittings are dominated by 39 85 K Rb(MI = −1) the scalar nuclear spin-spin interaction and the nuclear Zeeman effect, which are the only terms in the Hamilto- nian with matrix elements diagonal in N for N = 0. It 10 should be noted that the scalar spin-spin coupling is en- tirely mediated by the electron distribution, and has no contribution from the direct dipolar interaction. In the 5 absence of external fields, the energy levels are split into groups labeled by the total nuclear spin I. For small mag- Energy (kHz) netic fields B, I remains a nearly good quantum number and the levels split according to the value of its projec- 0 tion MI (which in this case coincides with the projection of the total angular momentum, which is always a good quantum number). Energy levels corresponding to the -5 same value of M display avoided crossings as a function 0 5 10 15 20 I Magnetic Field (G) of the field as shown in figure 2. For fields well above the crossings (which are at 2 to 10 G in this case), I FIG. 2: Zeeman splitting and avoided crossings (indicated is destroyed and the good quantum numbers are MRb 39 85 with arrows) for the MI = −1 levels of K Rb(v = 0, N = and MK. Since both nuclear g-factors are positive for 0). 39K85Rb, states where both projections are positive are high-field-seeking and those where both are negative are low-field-seeking. in the spin-coupled and fully coupled basis sets, Although the splittings at low fields are dominated by the scalar spin-spin coupling, there are several terms in hN = 0(I1I2)IMI |c4 I1 · I2|N = 0(I1I2)IMI i = the Hamiltonian that are off-diagonal in N. The energies hN = 0(I1I2)IFMF |c4 I1 · I2|N = 0(I1I2)IFMF i = are therefore obtained by diagonalizing a full matrix that 1 c4[I(I + 1) − I1(I1 + 1) − I2(I2 + 1)]. (13) includes enough rotational levels for convergence. For 2 the Zeeman effect, the only off-diagonal terms involving N = 0 are the electric quadrupole coupling and the tensor The nuclear Zeeman Hamiltonian is diagonal in the un- spin-spin coupling, both of which are small. Convergence coupled basis set, with nonzero elements given by for N = 0 is achieved with Nmax = 2 and the splittings obtained differ from those calculated with only N = 0 − [g1M1(1 − σ1)+ g2M2(1 − σ2)]µNB. (14) by less than 1%. For the Stark effect, however, the Stark The splitting pattern is therefore determined by the al- term itself mixes N = 0 states with N > 0. Terms off- lowed values of the total nuclear spin quantum number I diagonal in N are then very important and much larger and by the magnitudes and signs of the scalar spin-spin basis sets are needed. coupling constant c4 and the rotational g-factors. The The scalar spin-spin interaction for N = 0 is diagonal nuclear shielding constants σi are only a few parts per 6

0.0 85Rb133Cs 39 85 80 K Rb I

6 6 -4.0x10

40 6 c4

6 5 I = 4

-8.0x10 0 I = 3 5 4

c Energy (kHz) I 0 = 2 I = 1 4

-5 Energy (kHz) 4 c4

3 -1.2x10 3 -10

4 0.00 0.01 0.02 -40 c

2 2 c4 1

0 10 20 30 40 50 −1 -80 Electric Field (kV cm )

-5 0 5 10 15 20 Magnetic Field (G) FIG. 4: Stark eﬀect on energy levels of 39K85Rb correlating with (v = 0, N = 0) for electric ﬁelds up to 50 kV/cm. FIG. 3: Zeeman levels for 85Rb133Cs(v = 0, N = 0).

MI =0 ±2 39K85Rb thousand. For large values of the magnetic field, where MI = ±1 the nuclear Zeeman effect is the dominant term in the 0 Hamiltonian, the magnetic moment (gradient of the en- 3 ±1 I =4 ergy with respect to B) is close to −(g1M1 + g2M2)µN. ±3 The Zeeman splittings for 85Rb133 Cs are shown in ±2 figure 3. They are qualitatively similar to those for ±1

39 85 K Rb, except that the range of I is different and the I =3 0 0 spin-spin coupling constant c4 is significantly larger. Be- I =2 Energy (kHz) cause of this, I remains a good quantum number up to I =1 significantly higher magnetic fields. At high fields, once

-3 the magnitude of the scalar spin-spin interaction can be ±4 MI = ±1 MI = ±2 ±3 neglected compared to the Zeeman effect, MRb and MCs become good quantum numbers. The splitting patterns for other KRb and RbCs iso- 0.0 0.5 1.0 Electric Field (kV cm−1) topologs are qualitatively similar to those discussed above and the corresponding figures are available as FIG. 5: Stark splitting for energy levels of 39K85Rb correlat- supplementary online material. The spin-spin coupling ing with (v = 0, N = 0) for electric fields up to 1 kV/cm. constant and the potassium g-factor are negative for 40 85 40 87 The levels are shown relative to their field-dependent average K Rb and K Rb. The sign of c4 determines energy. whether the lowest zero-field energy corresponds to the highest or lowest value of I. In general the fields where the avoided crossings occur and above which M1 and M2 els: while in the Zeeman case this mixing is very weak become good quantum numbers scale with |c4/(g1 − g2)|. and is exclusively due to hyperfine terms, in the Stark When g1 and g2 are equal, as in homonuclear dimers, case it is strong and is caused directly by the electric there are no avoided crossings for N = 0 and the I quan- field. At low fields the mixing is weak and can be treated tum number is conserved even at high fields. by second-order perturbation theory, giving rise to a quadratic Stark effect. However, as the field increases the mixing becomes increasingly important: the N = 1 B. Stark splitting for rotational ground-state basis functions contribute around 25% at 10 kV/cm and molecules (N = 0) 40% at 20 kV/cm. Eventually the molecule becomes fully oriented by the field and the linear Stark effect overcomes The Stark effect for levels of 39K85Rb correlating with the quadratic effect. The mixing also has numerical con- N = 0 is shown in figure 4 to 6. Corresponding fig- sequences as the number of rotational levels required for ures for the remaining isotopologs of KRb and RbCs are convergence increases with field: for example, calcula- available as additional online material. The Stark ef- tions at 50 kV/cm require Nmax = 6. fect is quadratic at low fields but becomes linear at high The magnitude of the Stark shift in figure 4 obscures fields, as is usual for diatomic molecules in Σ states [72]. the splittings between hyperfine levels. Figure 5 therefore This arises from mixing between different rotational lev- shows the levels correlating with N = 0 relative to their 7

MI MRb MK 0 ±1/2 ∓1/2 oriented along the ﬁeld direction. In this limit the split- 1 1 2 1 2 39 85 ± ± / ± / K Rb 0 1 2 1 2 300 ∓ / ± / tings are mostly determined by the nuclear quadrupole ±2 ±1/2 ±3/2 ±1 ∓1/2 ±3/2 coupling constants, with relatively small contributions 2 ±3 2 ±1 2 150 ± / / from the magnetic hyperﬁne terms. ±1 ±3/2 ∓1/2 ±3 ±3/2 ±3/2 0 3 2 3 2 0 ± / ∓ / 0 ∓3/2 ±3/2

-150 Energy (kHz) V. CONCLUSION

-300 ±3 ±5/2 ±1/2 ±2 ±5/2 ∓1/2 4 5 2 3 2 -450 ± ± / ± / We have investigated the hyperﬁne level splittings ex- ±1 ±5/2 ∓3/2

0 10 20 30 40 50 pected for alkali metal dimers in their rotational ground −1 Electric Field (kV cm ) state in the presence of electric and magnetic fields. We have carried out density-functional calculations of the 39 85 FIG. 6: Stark splitting for energy levels of K Rb correlat- electronic structure of RbK and RbCs at the equilibrium ing with (v = 0, N = 0) for electric fields up to 50 kV/cm. geometry of the ground 1Σ state and evaluated all the The levels are shown relative to their field-dependent average hyperfine coupling constants necessary to calculate en- energy. ergy level patterns. For nonrotating states, the zero-field splittings between hyperfine states range from a few kHz for isotopologs of KRb to a few tens of kHz for isotopologs average energy, for fields up to 1 kV/cm. As expected, of RbCs. They are dominated by the electron-mediated each zero-field level splits into I + 1 components labeled contribution to the nuclear spin-spin coupling. The re- by the different possible values of |M |. For |M | > 0 the I I sults will be valuable in designing laser-based schemes levels exist in degenerate pairs corresponding to chang- to produce ultracold molecules in their absolute ground ing the sign of M and M . However, changing the sign 1 2 states in applied fields. of one of M1 and M2 produces a different state with a Acknowledgments different value of |MI |. For MI = 0 there is an extra sym- metry corresponding to reflection in a plane containing the electric field vector. The authors are grateful to EPSRC for funding of the At higher field, as shown in figure 6, the projections of collaborative project QuDipMol under the ESF EURO- the individual nuclear spins become well-defined as well CORES Programme EuroQUAM and to the UK National as their sum. At sufficiently large fields the splittings ap- Centre for Computational Chemistry Software for com- proach a limiting value as the molecules become strongly puter facilities.

[1] J. M. Hutson and P. Soldán, Int. Rev. Phys. Chem. 25, 1 (2007). 497 (2006). [13] K. Winkler, F. Lang, G. Thalhammer, P. van der Straten, [2] K. M. Jones, E. Tiesinga, P. D. Lett, and P. S. Julienne, R. Grimm, J. H. Denschlag, A. J. Daley, A. Kantian, Rev. Mod. Phys. 78, 483 (2006). H. P. Büchler, and P. Zoller, Phys. Rev. Lett. 98, 043201 [3] T. Köhler, K. Goral, and P. S. Julienne, Rev. Mod. Phys. (2007). 78, 1311 (2006). [14] S. Ospelkaus, A. Pe’er, K.-K. Ni, J. J. Zirbel, B. Neyen- [4] M. Baranov,L. Dobrek, K. Góral, L. Santos, and huis, S. Kotochigova, P. S. Julienne, J. Ye, and D. S. Jin, M. Lewenstein, Phys. Scr. T102, 74 (2002). arXiv:physics/0802.1093 (2008). [5] D. DeMille, Phys. Rev. Lett. 88, 067901 (2002). [15] J. G. Danzl, E. Haller, M. Gustavsson, M. J. Mark, [6] S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, R. Hart, N. Bouloufa, O. Dulieu, H. Ritsch, and H.-C. S. Riedl, C. Chin, J. H. Denschlag, and R. Grimm, Sci- Nägerl, arXiv:physics/0806.2284 (2008). ence 302, 2101 (2003). [16] M. Mark, T. Kraemer, P. Waldburger, J. Herbig, C. Chin, [7] M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. H.-C. Nägerl, and R. Grimm, Phys. Rev. Lett. 99, 113201 Raupach, S. Gupta, Z. Hadzibabic, and W. Ketterle, (2007). Phys. Rev. Lett. 91, 250401 (2003). [17] M. Mark, F. Ferlaino, S. Knoop, T. Kraemer, C. Chin, [8] M. Greiner, C. A. Regal, and D. S. Jin, Nature 426, 537 H.-C. Nägerl, and R. Grimm, Phys. Rev. A 76, 042514 (2003). (2007). [9] D. S. Petrov, C. Salomon, and G. V. Shlyapnikov, Phys. [18] C. Chin, V. Vuletić, A. J. Kerman, S. Chu, E. Tiesinga, Rev. Lett. 93, 090404 (2004). P. J. Leo, and C. J. Williams, Phys. Rev. A 70, 032701 [10] J. Herbig, T. Kraemer, M. Mark, T. Weber, C. Chin, (2004). H. C. Nägerl, and R. Grimm, Science 301, 1510 (2003). [19] J. M. Hutson, E. Tiesinga, and P. S. Julienne, [11] P. Soldán, M. T. Cvitaˇs, J. M. Hutson, P. Honvault, and arXiv:physics/0806.2583 (2008). J. M. Launay, Phys. Rev. Lett. 89, 153201 (2002). [20] N. F. Ramsey, Phys. Rev. 85, 60 (1952). [12] J. M. Hutson and P. Soldán, Int. Rev. Phys. Chem. 26, [21] J. M. Brown and A. Carrington, Rotational Spectroscopy 8

of Diatomic Molecules (Cambridge University Press, Phys. Lett. 258, 330 (1996). Cambridge, 2003). [45] W. C. Bailey, J. Mol. Spectrosc. 190, 318 (1998). [22] D. L. Bryce and R. E. Wasylishen, Acc. Chem. Res. 36, [46] W. C. Bailey, Chem. Phys. Lett. 292, 71 (1998). 327 (2003). [47] W. C. Bailey, Chem. Phys. 252, 57 (2000). [23] I. Mills, T. Cvitaˇs, K. Homann, N. Kallay, and K. Ku- [48] E. van Lenthe and E. J. Baerends, J. Chem. Phys. 112, chitsu, Quantities, Units and Symbols in Physical Chem- 8279 (2000). istry (Blackwell, Oxford, 1988). [49] I. Hung and R. W. Schurko, Solid State Nucl. Magn. [24] S. Kotochigova, P. S. Julienne, and E. Tiesinga, Phys. Reson. 24, 78 (2003). Rev. A 68, 022501 (2003). [50] M. H. Palmer and A. D. Nelson, J. Mol. Struct. 828, 91 [25] S. Kotochigova and E. Tiesinga, J. Chem. Phys. 123, (2007). 174304 (2005). [51] F. A. Bischoff, O. Hübner, W. Klopper, L. Schnelzer, [26] P. E. Van Esbroeck, R. A. McLean, T. D. Gaily, R. A. B. Pilawa, M. Horvatić, and C. Berthier, Eur. Phys. J. B Holt, and S. D. Rosner, Phys. Rev. A 32, 2595 (1985). 55, 229 (2007). [27] G. te Velde, F. M. Bickelhaupt, S. J. A. van Gisbergen, [52] H. Behzadi, N. L. Hadipour, and M. Mirzaei, Biophysical C. Fonseca Guerra, E. J. Baerends, J. G. Snijders, and Chemistry 125, 179 (2007). T. Ziegler, J. Comput. Chem. 22, 931 (2001). [53] C. Lee, W. Yang, and P. R. G., Phys. Rev. B 37, 785 [28] ADF2007.01, http://www.scm.com (2007), SCM, The- (1988). oretical Chemistry, Vrije Universiteit, Amsterdam, The [54] A. D. Becke, J. Chem. Phys. 98, 5648 (1993). Netherlands. [55] T. W. Keal and D. J. Tozer, J. Chem. Phys. 119, 3015 [29] DALTON, a molecular electronic (2003). structure program, Release 2.0, [56] T. W. Keal, D. J. Tozer, and T. Helgaker, Chem. Phys. http://www.kjemi.uio.no/software/dalton/dalton.html Lett. 391, 374 (2004). (2005). [57] A. D. Becke, Phys. Rev. A 38, 3098 (1988). [30] R. A. Logan, R. E. Coté, and P. Kusch, Phys. Rev. 86, [58] J. Vaara, J. Jokisaari, R. E. Wasylishen, and D. L. Bryce, 280 (1952). Prog. Nucl. Magn. Reson. Spectrosc. 41, 233 (2002). [31] P. A. Bonczyk and V. W. Hughes, Phys. Rev. 161, 15 [59] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. (1967). Lett. 77, 3865 (1996). [32] P. J. Dagdigian and L. Wharton, J. Chem. Phys. 57, [60] W. H. Flygare, J. Chem. Phys. 41, 793 (1964). 1487 (1972). [61] T. D. Gierke and W. H. Flygare, J. Am. Chem. Soc. 94, [33] J. Cederberg, E. Frodermann, H. Tollerud, K. Huber, 7277 (1972). M. Bongard, J. Randolph, and D. Nitz, J. Chem. Phys. [62] R. E. Wasylishen, D. L. Bryce, C. J. Evans, and M. C. L. 124, 244304 (2006). Gerry, J. Mol. Spectrosc. 204, 184 (2000). [34] J. Cederberg, S. Fortman, B. Porter, M. Etten, M. Feig, [63] S. A. Cooke and M. C. L. Gerry, Phys. Chem. Chem. M. Bongard, and L. Langer, J. Chem. Phys. 124, 244305 Phys. 6, 4579 (2004). (2006). [64] S. Huzinaga and B. Miguel, Chem. Phys. Lett. 175, 289 [35] E. Tiemann, B. Holzer, and J. Hoeft, Z. Naturforsch. A (1990). 32, 123 (1977). [65] S. Huzinaga and M. Klobukowski, Chem. Phys. Lett. [36] E. Tiemann, B. Holzer, and J. Hoeft, Z. Naturforsch. A 212, 260 (1993). 31, 236 (1976). [66] D. J. D. Wilson, C. E. Mohn, and T. Helgaker, J. Chem. [37] J. Cederberg, J. Ward, G. McAlister, G. Hilk, E. Beall, Theory Comput. 1, 877 (2005). and D. Olson, J. Chem. Phys. 111, 8396 (1999). [67] T. Enevoldsen, T. Rasmussen, and S. P. A. Sauer, J. [38] J. Hoeft, E. Tiemann, and T. Torring, Z. Naturforsch. A Chem. Phys. 114, 84 (2001). 27, 1516 (1972). [68] A. J. Ross, C. Effantin, P. Crozet, and E. Boursey, J. [39] R. A. Brooks, C. H. Anderson, and N. F. Ramsey, Phys. Phys. B: At. Mol. Opt. Phys. 23, L247 (1990). Rev. Letters 10, 441 (1963). [69] H. Katôand H. Kobayashi, J. Chem. Phys. 79, 123 [40] R. A. Brooks, C. H. Anderson, and N. F. Ramsey, J. (1983). Chem. Phys. 56, 5193 (1972). [70] T. R. Lawrence, C. H. Anderson, and N. F. Ramsey, [41] E. van Lenthe, E. J. Baerends, and J. G. Snijders, J. Phys. Rev. 130, 1865 (1963). Chem. Phys. 99, 4597 (1993). [71] R. N. Zare, Angular Momentum (John Wiley & Sons, [42] E. van Lenthe, E. J. Baerends, and J. G. Snijders, J. 1987). Chem. Phys. 101, 9783 (1994). [72] R. N. Townes and A. L. Schawlow, Microwave Spec- [43] E. van Lenthe, E. J. Baerends, and J. G. Snijders, J. troscopy (Dover Publications, New York, 1975). Chem. Phys. 110, 8943 (1999). [44] O. L. Fedotov, M. A. Malkina and V. G. Malkin, Chem.