Arxiv:1702.08265V2 [Cond-Mat.Quant-Gas] 26 May 2017 Also Invariant Under Spin Rotations

Stepwise Bose-Einstein condensation in a spinor gas

C. Frapolli,1 T. Zibold,1, ∗ A. Invernizzi,1 K. Jiménez-Garc´ıa,1 J. Dalibard,1 and F. Gerbier1 1Laboratoire Kastler Brossel, Collègede France, CNRS, ENS-PSL Research University, UPMC-Sorbonne Universités,11 place Marcelin Berthelot, 75005 Paris, France (Dated: August 27, 2018) We observe multi-step condensation of sodium atoms with spin F = 1, where the different Zeeman components mF = 0, ±1 condense sequentially as the temperature decreases. The precise sequence changes drastically depending on the magnetization mz and on the quadratic Zeeman energy q (QZE) in an applied magnetic field. For large QZE, the overall structure of the phase diagram is the same as for an ideal spin 1 gas, although the precise locations of the phase boundaries are significantly shifted by interactions. For small QZE, antiferromagnetic interactions qualitatively change the phase diagram with respect to the ideal case, leading for instance to condensation in mF = ±1, a phenomenon that cannot occur for an ideal gas with q > 0.

Multi-component quantum fluids described by a vec- on the thermodynamic phase diagram. tor or tensor order parameter are often richer than their The thermodynamics of spinor gases with conserved scalar counterparts. Examples in condensed matter are magnetization has been extensively studied theoretically superfluid 3He [1] or some unconventional superconduc- using various assumptions and methods [16–22]. A tors with spin-triplet Cooper pairing [2]. In atomic generic conclusion is that Bose-Einstein condensation oc- physics, spinor Bose-Einstein condensates (BEC) with curs in steps, where BEC occurs first in one specific com- several Zeeman components mF inside a given hyperfine ponent and magnetic order appears at lower tempera- spin F manifold can display non-trivial spin order at low tures when two or more components condense. Natural temperatures [3–6]. The macroscopic population of the questions are the number of steps that can be expected, condensate enhances the role of small energy scales that and the nature of the magnetic phases realized at differ- are negligible for normal gases. This mechanism (some- ent temperatures. times termed Bose-enhanced magnetism [6]) highlights In this Letter, we report on the observation of multi- the deep connection between Bose-Einstein condensation step condensation in an antiferromagnetic F = 1 conden- and magnetism in bosonic gases, and raises the question sate of sodium atoms. Fig.1 illustrates four situations of the stability of spin order against temperature. that occur when lowering the temperature starting from In simple cases, magnetic order appears as soon as a normal Bose gas. Without loss of generality, we focus a BEC forms. Siggia and Ruckenstein [7] pointed out in this work on the case of positive magnetization, given for two-component BECs [7] that a well-defined relative that the case of mz < 0 can be deduced by symmetry. In phase between the two components implies a macroscopic all cases with mz 6= 0, we find a sequence of transitions transverse spin. BEC and ferromagnetism then occur si- where different Zeeman components condense at differ- multaneously, provided the relative populations can ad- ent temperatures. Depending on the applied magnetic just freely. A recent experiment confirmed this scenario field B and on the magnetization, we find either two or for bosons with spin-orbit coupling [8]. This conclusion three condensation temperatures. The purpose of this was later generalized to spin-F bosons without [9] or with paper is to explore this rich landscape of transitions in a spin-independent [10] interactions. These results indicate bosonic spinor system and to elucidate the role of atomic that without additional constraints, bosonic statistics fa- interactions. vors ferromagnetism. The present work is to the best of our knowledge In atomic quantum gases with F > 1/2, this type the first comprehensive measurement of thermodynamic of ferromagnetism competes with spin-exchange interac- properties of spinor condensates with conserved magne- tions, which may favor other spin orders such as spin- tization. Previous experimental works exploring finite nematics [6]. Spin-exchange collisions can redistribute temperatures in spinor gases mostly studied spin dynam- populations among the Zeeman states [11–13], but are ics in thermal gases [23–26], or demonstrated cooling of arXiv:1702.08265v2 [cond-mat.quant-gas] 26 May 2017 also invariant under spin rotations. The allowed redistri- a majority Zeeman component by selective evaporation bution processes are therefore those preserving the total of the minority components [27, 28]. The realization of spin, such as 2 × (mF = 0) ↔ (mF = +1) + (mF = −1). dipolar spinor gases with free magnetization [15] was lim- For an isolated system driven to equilibrium only by bi- ited to the study of spin-polarized condensed phases in nary collisions (in contrast with solid-state magnetic ma- equilibrium due to dipolar relaxation. More recently, terials [14]), and where magnetic dipole-dipole interac- a gas of spin excitations in a spin-polarized (mz ≈ 1) tions are negligible (in contrast with dipolar atoms [15]), ferromagnetic Bose-Einstein condensate was observed to the longitudinal magnetization mz is then a conserved equilibrate and even condense at sufficiently low temper- quantity. This conservation law has deep consequences atures [29]. 2

q q q q 1 +1 =8.9 kHz =8.9 kHz =8.9 kHz =2.8 Hz h h h h 0 mz =0.06 mz =0.25 mz =0.63 mz =0.57 Cooling ramp 0.8 -1 a. b. c. d. 0.6

0.4

Optical Density 0.2

0 10 20 30 Trap Depth V0 [µK]

Decreasing temperature FIG. 2. Evolution of peak optical density with trap depth for a particular evaporation trajectory with q/h ≈ 69 Hz and mz ≈ 0.3 at the highest temperature. For these parameters, the mF = +1 component condenses ﬁrst (at a temperature 0.50.5 Tc,1 ≈ 1.8 µK), followed by the mF = 0 component (at a

00 temperature Tc,2 ≈ 560 nK). No condensate was detected in OD 1 0 +1 1 0 +1 1 0 +1 1 0 +1 the m = −1 component. The curves for m = +1 and m = − − − − F F F 0 have been shifted vertically by 0.2 and 0.1 for clarity. The FIG. 1. Illustration of stepwise Bose-Einstein condensation error bars denote statistical uncertainties at a 66 % confidence in antiferromagnetic spin 1 gases. Each column is formed level. The solid lines indicate the piece-wise linear fits used by juxtaposing absorption images of spin distributions with to determine the critical trap depths. monotonically decreasing temperature T from top to bottom. The quadratic Zeeman energy q and low-T magnetization mz are indicated at the top of each column. a. Only m = 0 F time on which the gas returns to thermal equilibrium. condenses (mF = 0, ±1 are the three Zeeman states). b. (c.) As a result, the kinetic equilibrium state for the quan- For low (high) magnetizations, mF = 0 (mF = +1) condenses first followed by mF = +1 (mF = 0). d. For small q and high tum gases studied in this work is still determined by mz, mF = +1 condenses first followed by mF = −1, while a magnetization-conserving Hamiltonian. Furthermore, mF = 0 does not condense. the ODT is tight enough such that a condensate forms in the so-called single-mode regime [32], where the spatial shape of the condensate wavefunction is independent of Our experiments are performed with ultracold 23Na the Zeeman state. In the following, we characterize our atoms confined in a crossed optical dipole trap (ODT). data for a given value of q by an evaporation “trajec- The longitudinal magnetization mz = (N+1 − N−1)/N tory” (N, T, mz)V0 , taking four experimental realizations acts as an external control parameter independent of the for each point in the trajectory. externally applied magnetic field B. Here, NmF is the Absorption images as shown in Fig.1 are recorded after reduced population in Zeeman state mF and N the to- 3 ms of expansion in an applied magnetic field gradient tal atom number. We vary mz between unmagnetized [31]. We perform a fit to a bimodal distribution for each (mz ≈ 0) and fully magnetized samples (mz ≈ 1) using component to extract the temperature, the populations a preparation sequence performed far above Tc [30, 31]. NmF , and the condensed fraction fc,mF per component An applied magnetic field B shifts the single-atom en- [31]. We found that low condensed fractions < 5% are 2 ergy by ∆EmF = pmF + q(mF − 1). The conservation difficult to detect with the fit algorithm due to a com- of magnetization makes the linear Zeeman effect ∝ p ir- bination of low signal-to-noise ratio and the complexity relevant in the equilibrium state. The quadratic Zeeman of fitting the three Zeeman components simultaneously. energy (QZE), which lowers the energy of mF = 0 with The signature of BEC, the appearance of a dense, narrow respect to mF = ±1, is the relevant term, and is given peak near the center of the atomic distribution, can in- 2 2 by q = αqB with αq/h ≈ 277 Hz/G for sodium atoms. stead be tracked by monitoring the peak optical density (OD) taken as a proxy for the condensed fraction [33]. The depth V0 of the ODT determines the tempera- This procedure avoids relying on bimodal fits or other ture T and total atom number N for a given V0. We indirect analyses with uncontrolled systematic biases. find that the magnetization mz also varies with V0 (by up to 15 %), a byproduct of evaporative cooling. Once Fig.2 shows such a measurement for a particular evap- a condensate forms in one of the Zeeman components, oration trajectory. The peak OD increases sharply when evaporation tends to eliminate preferentially atoms in the Bose-Einstein condensation is reached, demonstrating in other Zeeman states. The evaporative cooling dynamics this particular example a two-step condensation where is very slow compared to the microscopic thermalization mF = +1 condenses first, followed by mF = 0. For a 3

mF = +1 mF = 0 mF = 1 Summary − a. b. c. d. q/h = 8.9 kHz 0.8 Normal

0.4

0 1 BEC OD c,id 0

T/T e. f. g. h. q/h = 69 Hz 0.8

0.4

Reduced temperature i. j. k. l. q/h = 2.8 Hz 0.8

0.4

0 0 0.5 1 0 0.5 1 0 0.5 1 0 0.2 0.4 0.6 0.8 1 Magnetization mz

FIG. 3. Thermodynamic phase diagram of an antiferromagnetic spin F = 1 Bose gas. The peak optical density of each Zeeman component is reported for the entire set of data at each value of the QZE – q/h = 8.9 kHz (a-c), 69 Hz (e-g) and q = 2.8 Hz (i-k). The temperature is normalized to Tc,id, the critical temperature of a single-component ideal Bose gas with the same number of atoms. The grayed areas indicate the absence of data in the corresponding regions. The right column (d,h,l) shows the measured critical temperatures of the mF = +1, 0, −1 Zeeman components (red, gray, and blue markers, respectively). The solid (dashed) lines are the predictions of a Hartree-Fock (HF) model with spin-independent interactions (ideal gas theory). The dotted line in k shows the expected Tc,2 where mF = 0 condenses according to the HF model.

given evaporation trajectory, we identify the critical trap quential condensation is always observed for mz 6= 0, but depth V0,c where condensation is reached by a piece-wise the overall behavior changes drastically with q. linear fit to the data, taking the intercept point as the We first discuss the cases with largest QZE, q/h ≈ experimentally determined V (see Fig.2). We inter- 0,c 8.9 kHz (Fig.3 a-d) and q/h ≈ 69 Hz (Fig.3 e-h). For polate numerically the atom number, magnetization and q/h ≈ 8.9 kHz and highly magnetized samples, the ma- temperature to obtain the critical values Nc, Tc, mz,c jority component mF = +1 condenses first at a critical from V0,c. temperature Tc,1, followed by the mF = 0 component at a lower temperature T . For low magnetizations, the Fig.3 summarizes the results of this work. We show c,2 condensation sequence is reversed. For q/h ≈ 69 Hz, we the peak optical density for each Zeeman component and observe only one sequence, a two-step condensation with each value of q in a (T − m ) plane (Fig.3 a-c, e-g and i- z m = +1 first and m = 0 second. k). In this plot, all data taken at a given QZE q are F F binned with respect to magnetization and temperature. This behavior can be understood qualitatively from The domains where condensation occurs appear in light ideal gas theory, taking the QZE and the conservation colors. For convenience, the temperature is scaled to of magnetization into account [19]. For ideal gases, BEC the critical temperature of a single-component ideal gas occurs when the chemical potential µ equals the energy 1/3 kBTc,id = ~ω[N/ζ(3)] , with ω the geometric average of the lowest single-particle state [34]. The same criterion of the trap frequencies and ζ the Riemann zeta func- holds for a spin 1 gas with µ0 = µ and µ±1 = µ±λ, where tion [34]. The same plot also shows the measured critical λ is a Lagrange multiplier enforcing the conservation of temperatures (Fig.3 d, h, l)[35]. The phenomenon of se- mz. For mz = 0 (λ = 0) and q > 0, the QZE lowers the 4 energy of mF = 0, which is therefore the first component below Tc,2. While coexisting mF = 0 and mF = +1 to condense when µ = −q. For mz > 0, λ is positive and components form a M⊥ phase with hSˆx + iSˆyi= 6 0, coex- increases with mz. The energetic advantage of mF = 0 isting mF = ±1 components correspond to a phase with is in balance with the statistical trend favoring the most hSˆx + iSˆyi = 0 but where the spin-rotational symmetry populated component mF = +1. Eventually, this trend around z is broken by a non-zero spin-quadrupole ten- ∗ takes over at a “critical” value mz (where λ = q). For sor (“quasi-spin nematic” phase -qSN). At T = 0 and in ∗ mz > mz, the mF = +1 component condenses first. the single-mode regime, the M⊥- qSN transition occurs p 2 Coexisting mF = 0 and mF = ±1 components with at a critical magnetization mz,c = 1 − [1 − (q/Us)] , a well-defined phase relation correspond to a non-zero with Us ≤ q the spin-dependent interaction energy [37]. ˆ ˆ transverse spin hSx + iSyi 6= 0 (“transverse magne- When q > Us, there is no phase transition and only the tized” phase – M⊥). For large q, the condensate is re- M⊥ phase is present. This explains the qualitative differ- duced to an effective two-component system mF = 0, +1 ence between the data for q/h = 2.8 Hz and the other two ∗ with mF = −1 mostly spectator. The case mz = mz values. We estimate Us/h . 50 Hz and mz,crit ≈ 0.3 for (µ0 = µ+1) realizes the Siggia-Ruckenstein (S-R) sce- a BEC without thermal fraction [30]. This agrees well nario, where condensation and ferromagnetic behavior with the lowest temperature measurements reported in appear simultaneously. Away from that point, the S-R Fig.3j-k. picture breaks down (µ0 6= µ+1) and sequential conden- In the experimental data in Fig.3 i-l, the region of the sation takes place. phase diagram occupied by the M⊥ phase shrinks with Figure3 d-h show the critical temperatures and com- increasing temperature. In fact, we find that mF = −1 pare them to ideal gas theory. Although the general condenses at Tc,2 for all parameters we have explored, trends in the theory are the same as in the experi- with mF = 0 condensing at a third, lower critical temper- ment, we observe a systematic shift of Tc,1 and Tc,2 to- ature (except for mz ≈ 0, where all components appear wards lower temperatures, and an experimental “critical” to condense together within the accuracy of our mea- ∗ mz ∼ 0.3 larger than the ideal gas prediction. The be- surement). Finally, the dashed line in Fig.3k shows Tc,2 havior for q/h ≈ 69 Hz (Fig.3 e-h) is qualitatively similar predicted by the HF model with spin-independent inter- ∗ to the largest q case, but with a small mz that cannot actions. Although the model incorrectly predicts that be resolved experimentally (the ideal gas theory predicts mF = 0 should condense below Tc,2, the predicted tran- ≈ 0.002). sition closely matches the observed boundary between Repulsive interactions between the atoms can be ex- single-component mF = +1 BEC and qSN mF = ±1 pected to lower the critical temperatures as in single- BEC. This indicates that the transition line itself (but component gases [36], with an enhanced shift of Tc,2 due not the magnetic order below it) is determined by the to the presence of a condensate. We use a simplified ver- thermal component alone. sion of Hartree-Fock (HF) theory to make quantitative In conclusion, we have studied the finite-T phase dia- predictions [22]. Our self-consistent calculations include gram of a spin-1 Bose gas with antiferromagnetic inter- the trap potential in a semi-classical approximation, and actions. For condensates in the single-mode regime, we treat the interactions as spin-independent. These ap- observed a sequence of transitions, two for high QZE and proximations are valid only above Tc,2, where at most three for low QZE, with the lower two leading to differ- one component condenses [31]. As a result, the HF model ent magnetic orders. We have found that a simplified HF cannot make any prediction for the low-temperature be- model reproduces the trends observed in the variations havior below Tc,2. The results of the HF calculations, of the critical temperatures Tc,1 and Tc,2 with magne- performed for atom numbers and trap frequencies match- tization and QZE. A more complete theoretical analysis ing the experimental values [31], are shown in Figure3. accounting for all experimental features –in particular the The HF model qualitatively accounts for the experimen- harmonic trap, which is crucial to stabilize an antiferro- tal data, explaining in particular the strong downwards magnetic condensate in a single spatial mode [32]– and ∗ shift of Tc,2 for all q and the shift of mz to higher val- elucidating the exact nature of the low-temperature tran- ues for q/h ≈ 8.9 kHz. The residual discrepancy around sitions for low QZE remains open. A natural extension of 7 − 8 % could be partially explained by finite-size and this work would be to study the critical properties of the ∗ trap anharmonicity effects not included in the Hartree- observed finite-T transitions, in particular near mz = mz Fock calculation [31]. and between the M⊥ and qSN phases at very low q. Two- At the lowest field we studied, q/h ≈ 2.8 Hz (Fig.3 i-l), dimensional systems provide another intriguing direction we observe a change in the nature of Tc,2. For high values to explore. Several Berezinskii-Kosterlitz-Thouless tran- of mz, Tc,2 corresponds to condensation into mF = −1 sitions mediated either by vortices or spin textures have while mF = 0 remains uncondensed. This phenomenon been predicted [38, 39]. We expect that such topologi- is incompatible with ideal gas theory [16, 19] and with cal features will further enrich the already complex phase our HF model with spin-independent interactions. It cor- diagram observed in three dimensions. responds to a change of the magnetic ordering appearing We acknowledge stimulating discussions with B. 5

Evrard, L. De Sarlo, E. Witkowska, J. Beugnon, L. de condensate with free magnetization,” Phys. Rev. Lett. Forges de Parny, A. Ran¸conand T. Roscilde. This work 108, 045307 (2012). has been supported by ERC (Synergy grant UQUAM). [16] T. Isoshima, T. Ohmi, and K. Machida, “Double phase TZ acknowledges funding from the Hamburg Center for transitions in magnetized spinor Bose-Einstein condensation,” Journal of the Physical Society of Japan 69, 3864– Ultrafast Imaging, and KJG from the European Union’s 3869 (2000). Horizon 2020 research and innovation programme un- [17] W. Zhang, S. Yi, and L. You, “Bose-Einstein condensa- der the Marie Sk lodowska-Curie grant agreement No. tion of trapped interacting spin-1 atoms,” Phys. Rev. A 701894. 70, 043611 (2004). [18] Kao, Y.-M. and Jiang, T. F., “Transition temperatures of the trapped ideal spinor Bose gas,” Eur. Phys. J. D 40, 263–269 (2006). [19] G. Lang and E. Witkowska, “Thermodynamics of a spin- 1 Bose gas with fixed magnetization,” Phys. Rev. A 90, ∗ Current address: Department of Physics, University of Basel, Klingelbergstrasse 82, 4056 Basel, Switzerland 043609 (2014). [1] D. Vollhardt and P. Wölfle, The Superfluid Phases of [20] S. Uchino, M. Kobayashi, and M. Ueda, “Bogoliubov Helium 3 (Taylor and Francis, London, 1990). theory and Lee-Huang-Yang corrections in spin-1 and [2] Michael R. Norman, “The challenge of unconventional spin-2 Bose-Einstein condensates in the presence of the superconductivity,” Science 332, 196–200 (2011). quadratic Zeeman effect,” Phys. Rev. A 81, 063632 [3] T.-L. Ho, “Spinor Bose condensates in optical traps,” (2010). Phys. Rev. Lett. 81, 742–745 (1998). [21] N. T. Phuc, Y. Kawaguchi, and M. Ueda, “Effects of [4] T. Ohmi and K. Machida, “Bose-Einstein condensation thermal and quantum fluctuations on the phase diagram 87 with internal degrees of freedom in alkali atom gases,” of a spin-1 Rb Bose-Einstein condensate,” Phys. Rev. Journal of the Physical Society of Japan 67, 1822–1825 A 84, 043645 (2011). (1998). [22] Y. Kawaguchi, N. T. Phuc, and P. B. Blakie, “Finite- [5] J. Stenger, S. Inouye, D. Stamper-Kurn, H.-J. Mies- temperature phase diagram of a spin-1 Bose gas,” Phys. ner, A. Chikkatur, and W. Ketterle, “Spin domains Rev. A 85, 053611 (2012). in ground-state Bose-Einstein condensates,” Nature 396, [23] H. K. Pechkis, J. P. Wrubel, A. Schwettmann, P. F. Grif- 345–348 (1998). fin, R. Barnett, E. Tiesinga, and P. D. Lett, “Spinor dy- [6] D. M. Stamper-Kurn and M. Ueda, “Spinor Bose gases: namics in an antiferromagnetic spin-1 thermal Bose gas,” Symmetries, magnetism, and quantum dynamics,” Rev. Phys. Rev. Lett. 111, 025301 (2013). Mod. Phys. 85, 1191–1244 (2013). [24] X. He, B. Zhu, X. Li, F. Wang, Z.-F. Xu, and D. Wang, 87 [7] Eric D. Siggia and Andrei E. Ruckenstein, “Bose conden- “Coherent spin-mixing dynamics in thermal Rb spin-1 sation in spin-polarized atomic Hydrogen,” Phys. Rev. and spin-2 gases,” Phys. Rev. A 91, 033635 (2015). Lett. 44, 1423–1426 (1980). [25] M. Erhard, H. Schmaljohann, J. Kronjäger,K. Bongs, [8] Si-Cong Ji, Jin-Yi Zhang, Long Zhang, Zhi-Dong Du, and K. Sengstock, “Bose-Einstein condensation at con- Wei Zheng, You-Jin Deng, Hui Zhai, Shuai Chen, and stant temperature,” Phys. Rev. A 70, 031602 (2004). Jian-Wei Pan, “Experimental determination of the finite- [26] B. Naylor, M. Brewczyk, M. Gajda, O. Gorceix, temperature phase diagram of a spin-orbit coupled bose E. Maréchal, L. Vernac, and B. Laburthe-Tolra, “Com- gas,” Nat Phys 10, 314–320 (2014). petition between Bose-Einstein condensation and spin [9] K. Yamada, “Thermal properties of the system of mag- dynamics,” Phys. Rev. Lett. 117, 185302 (2016). netic bosons : Bose-Einstein ferromagnetism,” Progress [27] R. Olf, F. Fang, G. E. Marti, A. MacRae, and D. M. of Theoretical Physics 67, 443 (1982). Stamper-Kurn, “Thermometry and cooling of a Bose gas [10] Eli Eisenberg and Elliott H. Lieb, “Polarization of inter- to 0.02 times the condensation temperature,” Nat. Phys. acting bosons with spin,” Phys. Rev. Lett. 89, 220403 11, 720–723 (2015). (2002). [28] B. Naylor, E. Maréchal, J. Huckans, O. Gorceix, P. Pedri, [11] M.-S. Chang, C. D. Hamley, M. D. Barrett, J. A. Sauer, L. Vernac, and B. Laburthe-Tolra, “Cooling of a Bose- K. M. Fortier, W. Zhang, L. You, and M. S. Chapman, Einstein condensate by spin distillation,” Phys. Rev. “Observation of spinor dynamics in optically trapped Lett. 115, 243002 (2015). 87Rb Bose-Einstein condensates,” Phys. Rev. Lett. 92, [29] F. Fang, R. Olf, S. Wu, H. Kadau, and D. M. Stamper- 140403 (2004). Kurn, “Condensing magnons in a degenerate ferromag- [12] H. Schmaljohann, M. Erhard, J. Kronjäger,M. Kottke, netic spinor Bose gas,” Phys. Rev. Lett. 116, 095301 S. van Staa, L. Cacciapuoti, J. J. Arlt, K. Bongs, and (2016). K. Sengstock, “Dynamics of F = 2 spinor Bose-Einstein [30] D. Jacob, L. Shao, V. Corre, T. Zibold, L. De Sarlo, condensates,” Phys. Rev. Lett. 92, 040402 (2004). E. Mimoun, J. Dalibard, and F. Gerbier, “Phase diagram [13] T. Kuwamoto, K. Araki, T. Eno, and T. Hirano, “Mag- of spin-1 antiferromagnetic Bose-Einstein condensates,” netic field dependence of the dynamics of 87Rb spin-2 Phys. Rev. A 86, 061601 (2012). Bose-Einstein condensates,” Phys. Rev. A 69, 063604 [31] (See Supplemental Material at [URL will be inserted by (2004). publisher]). [14] L.-P. Levy, Magnetism and Superconductivity (Springer, [32] S. Yi, O.¨ E. Müstecaplıo˘glu,C. P. Sun, and L. You, Berlin, 2000). “Single-mode approximation in a spinor-1 atomic con- [15] B. Pasquiou, E. Maréchal, L. Vernac, O. Gorceix, and densate,” Phys. Rev. A 66, 011601 (2002). B. Laburthe-Tolra, “Thermodynamics of a Bose-Einstein [33] S. Trotzky, L. Pollet, F. Gerbier, U. Schnorrberger, 6

I. Bloch, M. Troyer, N. V. Prokof’ev, and B. V. Svis- densate fraction and critical temperature of a trapped tunov, “Suppression of the critical temperature for super- interacting Bose gas,” Phys. Rev. A 54, R4633–R4636 fluidity near the Mott transition,” Nat. Phys. 6, 998–1004 (1996). (2010). [37] W. Zhang, S. Yi, and L. You, “Mean field ground state [34] F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S. Stringari, of a spin-1 condensate in a magnetic field,” New Journal “Theory of Bose-Einstein condensation in trapped of Physics 5, 77 (2003). gases,” Rev. Mod. Phys. 71, 463–512 (1999). [38] Subroto Mukerjee, Cenke Xu, and J. E. Moore, “Topo- [35] In one case, mF = 0 when mz ≈ 0.3 and q/h = 2.8 Hz, logical defects and the superfluid transition of the s = 1 the lowest temperature images do show a condensed com- spinor condensate in two dimensions,” Phys. Rev. Lett. ponent but the critical temperature could not be ex- 97, 120406 (2006). tracted reliably from the fitting procedure due to sparse [39] A. J. A. James and A. Lamacraft, “Phase diagram of sampling. This particular point is not reported in Fig.3l. two-dimensional polar condensates in a magnetic field,” [36] S. Giorgini, L. P. Pitaevskii, and S. Stringari, “Con- Phys. Rev. Lett. 106, 140402 (2011).