physics0203089/OH.tex
1: \documentstyle[preprint,aps,graphicx]{revtex}     
2: %\documentclass[preprint,aps,graphicx]{revtex4}      
3:       
4: \def\btt#1{{\tt$\backslash$#1}}      
5:       
6: \begin{document}      
7: \title{Collisional dynamics of ultracold OH molecules in an electrostatic field.}      
8: \author{ Alexandr V. Avdeenkov and John L. Bohn}      
9: \address{JILA and Department of Physics, University of Colorado, Boulder, CO}      
10: \date{\today}      
11: \maketitle      
12:       
13: \begin{abstract}      
14: Ultracold collisions of polar OH molecules are considered in the presence of  
15: an electrostatic field.  The field exerts a strong influence on both elastic  
16: and state-changing inelastic collision rate constants, leading to   
17: clear experimental signatures that should help disentangle the theory  
18: of cold molecule collisions.  Based on the collision  
19: rates we discuss the prospects for evaporative cooling of electrostatically   
20: trapped OH.  We also find that  the scattering properties  
21: at ultralow temperatures prove to be remarkably independent of the details  
22: of the short-range interaction, owing to avoided crossings  
23: in the long-range adiabatic potential curves.  The behavior of the    
24: scattering rate constants is qualitatively understood in terms of  
25: a novel set of long-range states of the [OH]$_2$ dimer.  
26: \end{abstract}      
27:       
28:       
29: \pacs{34.20.Cf, 34.50.-s, 05.30.Fk}      
30:       
31: \narrowtext      
32:       
33: \section{Introduction}      
34: %\subsection{Background}      
35: Polar molecules bring something entirely new to the field of  
36: ultracold physics.  As compared to the neutral atoms that have  
37: been studied experimentally in the past, polar molecules possess  
38: extremely strong, anisotropic interactions.  It has been speculated  
39: that dipolar interactions will lead to new properties in  
40: Bose-Einstein condensates \cite{Santos,You,Goral} or degenerate  
41: Fermi gases \cite{Shlyapnikov}.  It has also been suggested that   
42: polar molecules in optical lattices may be useful in implementing quantum   
43: logic elements \cite{DeMille}.  
44: On the experimental side, cold polar molecules may be produced in  
45: several ways, including photoassociation of two distinct alkali species  
46: \cite{Shaffer,Schloder},  
47: buffer-gas cooling \cite{Doyle,Egorov},   
48: or Stark slowing \cite{Meijer1,Meijer2,Meijer3}.  
49:   
50: Regardless of the method or production, collisions of molecules are  
51: of paramount importance in describing the properties of the gas.   
52: Collisions should also be interesting in their own right, as detailed  
53: probes of intermolecular interactions.  Several features of 
54: the collisional dynamics of ground-state polar molecules, based on     
55: a simplified ``toy'' model, were discussed in~\cite{bohnpolar}. This model     
56: accounted for the interplay between dipole-dipole interactions, an external    
57: electric field, and states of different parities.  The dipole-dipole    
58: interaction, which    
59: scales with intermolecular separation $R$ as $1/R^3$, renders cold    
60: molecule collisions completely different from cold atom collisions.    
61: This is because a $1/R^3$ interaction is characterized by energy-independent    
62: low energy cross sections in {\it all} partial waves, not just in s-waves    
63: \cite{landau,Shakeshaft,Deb}.  The relatively strong, long-range interactions imply  
64: that  molecules electrostatically trapped in weak-field-seeking states     
65: are generally susceptible to state-changing collisions that can      
66: rapidly deplete the trapped gas.  The rates are in general far larger    
67: than those of magnetic dipolar transitions in stretched-state  
68: alkali atoms, owing largely to the    
69: relative strength of electric, as opposed to magnetic, dipolar    
70: interactions \cite{bohnpolar}.    
71:     
72: In this paper we address ultracold polar molecule collisions in a    
73: more realistic model, considering in detail the OH radical.  This    
74: choice is motivated by the attractiveness of this molecule    
75: for Stark slowing from a supersonic jet \cite{Meijer3,Ye}.    
76: In particular, it has a $^2\Pi$ ground state with a small $\Lambda$-doublet  
77: splitting, making it easily manipulated by modest-sized  
78: electric fields.  
79: A full treatment of cold collisions is somewhat hindered by    
80: the fact that the OH-OH potential energy surface (PES) is     
81: poorly known, although it is known to be very deep and strongly anisotropic    
82: \cite{kuhn,harding}.  It is not even known, for example, whether    
83: OH molecules may suffer chemical reactions at ultralow temperatures.    
84: As a point of reference, it was recently suggested that the reaction   
85: F+H$_2$ $\rightarrow$  HF + H may proceed at appreciable rates at   
86: ultralow temperatures,  in spite of having a chemical barrier   
87: height of 700 K \cite{Bala}.    
88:     
89: However, long-range dipole-dipole    
90: forces strongly dominate the scattering of OH molecules in their    
91: weak-field-seeking states.  In this paper we will show that this arises    
92: from strong avoided crossings in the long-range adiabatic potential    
93: curves, which prevent the molecules from approaching close enough to  
94: one another for exchange potentials to become important.    
95: In this regard cooling and electrostatic trapping of     
96: OH molecules can provide a wealth of information on the long-range    
97: OH-OH interaction.  Thus it appears possible to understand   
98: a class of ultracold    
99: OH-OH collisions without detailed knowledge of the short range PES.    
100: This strategy would be an important stepping stone toward understanding    
101: the full problem of ultracold OH collisions.  Strong-field seekers,    
102: by contrast, will in addition experience the short-range     
103: interaction.  The complete problem of exploring collisions of ultracold    
104: polar molecules might therefore most efficiently proceed by a two-step analysis,    
105: thus simplifying this very complicated problem.    
106:     
107: Accordingly we focus in this paper on the first step, namely, collisions of    
108: weak-field-seeking states.  After some discussion of the relevant    
109: properties of OH molecules in Sec. II and their interactions in Sec. III,     
110: we move on in Sec. IV to illustrate some prominent energy- and    
111: field-dependent features in elastic and inelastic cross sections.    
112: Mapping these features in experiments should help in unraveling     
113: the long-range part of this puzzle.  We also present a simplified    
114: model of the long-range interaction, to help illustrate the    
115: basic physics behind the behavior of the cross sections.  It will  
116: turn out that a new class of long-range bound states of the [OH]$_2$  
117: dimer play a significant role in ultracold collisions of this  
118: molecule.  
119:     
120: \section{OH molecule}      
121:     
122: The OH molecule has a fairly complicated internal structure,     
123: incorporating rotation, parity, electronic spin, and nuclear spin    
124: degrees of freedom, which are  further confounded in    
125: the presence of an electric field.  We therefore begin by     
126: describing the structure of this molecule and the simplifications    
127: we impose to render our model tractable.    
128:     
129: %\subsection{General considerations}      
130:     
131: Molecules cooled to sub-Kelvin temperatures by Stark slowing will be    
132: assumed to be in their electronic $^{2}\Pi$ ground state,      
133: and $\upsilon =0 $ vibrational ground state.  In this state OH is an   
134: almost pure Hund's case (a) molecule, and has a dipole moment of   
135: 1.68 D \cite{kuhn}.    
136: Spin-orbit coupling involving the lone electronic spin     
137: splits the ground state into $^{2}\Pi_{3/2}$ and $^{2}\Pi_{1/2}$ components,  
138: of which $^{2}\Pi_{3/2}$ is lower in energy and is therefore the    
139: state of greatest interest in ultracold collisions.    
140: In our model we take into account just the lowest rotational      
141: level of the corresponding ground state, $J=3/2$.     
142: The energy of the first rotationally excited state with $|J=5/2,\Omega=3/2>$     
143: is about 84K higher in energy \cite{coxon} and we will neglect this and  
144: higher-lying states     
145: in our scattering calculations.   Such states will, however, contribute  
146: rotational Feshbach resonances in realistic collisions.  
147:      
148: The isotopomer $^{16}$O$^1$H that we consider here has a nuclear spin      
149: of $I=1/2$, which with a half- integer rotational quantum      
150: number  defines the OH molecule as a boson. Thus  we should take into      
151: account hyperfine structure  to ensure the proper Bose symmetry.    
152: We will see below that the inclusion of hyperfine structure is  
153: also important in determining details of collision properties.  
154: The calculations in an electric field also require  knowing the Stark splitting     
155: for OH molecules.   Thus the Hamiltonian for the OH molecule in a field is      
156: \begin{equation}      
157: \label{hamone}      
158: H^{OH}=H_{rot}+H_{fs}+H_{hfs}+H_{field}    
159: \end{equation}      
160:     
161: Wave functions for the spatial degrees of freedom of the molecule    
162: are constructed in the usual way.  Namely, in the zero-electric-field    
163: limit, eigenstates of parity $\varepsilon$ ($= \pm$) are given by the Hund's case (a)    
164: representation:    
165: \begin{equation}      
166: \label{wfs}      
167: |J M_{J} \Omega  \varepsilon>=\frac{1}{\sqrt{2}} \bigl(      
168: |J M_{J} \Omega > |\Lambda \Sigma> + \varepsilon |J M_{J} -\Omega > |-\Lambda -\Sigma>      
169: \bigr),      
170: \end {equation}      
171: where the rotational part is given by a Wigner function:      
172: \begin{equation}      
173: |J M_{J} \Omega > =(\frac{2J+1}{8\pi^{2}})^{1/2} D^{J}_{M_{J} \Omega}(\theta,\phi,\kappa)  ,    
174: \end{equation}      
175: and $\Omega=|\Sigma+ \Lambda|$ is the projection of the total  electronic    
176: angular momentum on the molecular axis.    
177: The total spin of the molecule, ${\bf F}= {\bf J} +{\bf I}$, with laboratory    
178: projection $M_{F}$, is then constructed by    
179: \begin{equation}      
180: \label{onestate}      
181: |(JI)FM_{F}\Omega \varepsilon>=|\Lambda> |S \Sigma>      
182: \sum_{M_{J},M_{I}} |J M_{J} \Omega  \varepsilon> |IM_{I}><FM_{F}|JM_{J}I_{M_{I}}>,      
183: \end{equation}      
184:     
185: The matrix elements for the Hamiltonian (\ref{hamone}) in this basis    
186: can be found elsewhere~\cite{mizushima}. In compact form these matrix      
187: elements are    
188: \begin{eqnarray}      
189: \label{matrix1}      
190: <(JI)F \Omega M_{F} \varepsilon| H^{OH}|(J'I')F' \Omega' M_{F'} \varepsilon'>=\Bigl(      
191: \delta_{\Omega,3/2} \delta_{\Omega',3/2} E_{3/2,3/2}+      
192: \\      
193: \nonumber      
194: \delta_{\Omega,1/2} \delta_{\Omega',1/2} E_{1/2,1/2}+      
195: (\delta_{\Omega,3/2} \delta_{\Omega',1/2}+      
196: \delta_{\Omega,1/2} \delta_{\Omega',3/2} E_{3/2,1/2}      
197: \Bigl) \times      
198: \\      
199: \nonumber      
200: \delta_{J,J'} \delta_{F,F'} \delta_{\varepsilon,\varepsilon'}-      
201: \mu {\cal E} \frac{1}{2}(1+(-1)^{J+J'}\varepsilon \varepsilon')(-1)^{F+F'+M_{F}+I-\Omega+1}      
202: \times      
203: \\      
204: \nonumber      
205: ([J][J'][F][F'])^{1/2}      
206: \left( \begin{array}{ccc}      
207:                   J & 1 & J' \\      
208:                   -\Omega & 0 & \Omega'      
209:                   \end{array} \right)      
210: \left( \begin{array}{ccc}      
211:                   F' & 1 & F \\      
212:                   -M_{F'} & 0 & M_{F}      
213:                   \end{array} \right)      
214: \left\{ \begin{array}{ccc}      
215:                   1 & J & J' \\      
216:                   I & F & F'      
217:                   \end{array} \right\}.    
218: \end{eqnarray}      
219: In this expression $\mu$ is the molecular dipole moment,     
220: ${\cal E}$ is the strength of the electric field, and      
221: $E_{\Omega,\Omega^{\prime}}$ are matrix elements for      
222: the fine structure   
223: $H_{rot} +H_{fs}$ which  can be found in~\cite{mizushima,miller}.    
224: These values depend on the rotational constant, spin-orbit coupling constant,      
225: hyperfine coupling constant, and $\Lambda$-doublet parameters of OH.    
226: All of these constants can be found in~\cite{coxon}.     
227:   
228: Equation (\ref{matrix1}) shows that, strictly speaking, the only  
229: good quantum number for the OH molecule is the projection of its  
230: angular momentum on the laboratory axis, $M_F$.  However, for our  
231: present purposes it suffices to treat the quantum numbers as  
232: ``almost good.''  For example, in view of the fact that OH is nearly  
233: a purely Hund's case (a) molecule, the coupling between $\Omega = 1/2$  
234: and $\Omega = 3/2$ states is fairly weak. We account for this interaction  
235: perturbatively, by replacing the values $E_{3/2}$ and $E_{1/2}$  
236: by the eigenvalues of the $2 \times 2$ matrix  
237: \begin{equation}  
238: \left( \begin{array}{cc}  
239:   E_{3/2} & E_{3/2,1/2} \\  
240:   E_{3/2,1/2} & E_{1/2} \\  
241: \end{array} \right),  
242: \end{equation}   
243: keeping all other quantum numbers constant.  
244:   
245: Likewise, different values of the molecular spin $J$ are mixed in a field,  
246: but this mixing is small in laboratory strength fields.  The total spin  
247: $F$ and the parity are far more strongly mixed.  Accordingly,  
248: in practice we transform the molecular state to a field-dressed basis  
249: for performing scattering calculations:  
250: \begin{eqnarray}     
251: \label{ebasis}     
252: |(\tilde{J}I)\tilde{F}M_{F} \Omega \tilde{\varepsilon};{\cal E} > \equiv  
253: \sum_{JF\varepsilon} \alpha(JF\varepsilon) |(JI)FM_{F}\Omega \varepsilon>,     
254: \end{eqnarray}     
255: where $\alpha(JF\varepsilon)$ stands for eigenfunctions of the Hamiltonian~(\ref{hamone})    
256: determined numerically at each value of the field.  We will continue    
257: to refer to molecular states by the quantum numbers $J$, $F$, and $\varepsilon$,   
258: with the understanding that they are only  approximately good in a field,  
259: and that (\ref{ebasis}) is the appropriate molecular state.  
260:     
261: Figure 1 shows the Stark energies computed using all the ingredients  
262: described above. In zero field the energy levels are primarily determined     
263: by the $\lambda$-doublet splitting between opposite parity states,    
264: whose value is $\Delta=0.056cm^{-1}$.  The alternative parity states,  
265: with $\varepsilon=-1$ (f states) and $\varepsilon=+1$ (e states) are  
266: shown in Fig. 1(a) and Fig. 1(b), respectively.  These states are further split    
267: into hyperfine components with total spin $F=1$ and $F=2$.  The Stark shift    
268: is quadratic for fields below the critical field     
269: ${\cal E}_{0}\equiv{\Delta}/{2\mu}$ ($\approx 1000 V/cm$ for OH).    
270: For fields larger than ${\cal E}_{0}$ states with different parity are entirely      
271: mixed and the Stark effect transforms from quadratic to linear.     
272: In this case the molecular states are roughly equal linear combinations  
273: of the zero-field $\varepsilon=-$ and $\varepsilon=+$ states (compare  
274: Eqn. \ref{wfs}) \cite{schreel}:  
275: \begin{eqnarray}      
276: |J M_{J} \Omega  \varepsilon=\pm 1> = \{ \begin{array}{cc}      
277:                                         | J M_{J} \mp \Omega > &  M_{J} < 0 \\      
278:                                         | J M_{J} \pm \Omega > &  M_{J} > 0      
279:                                            \end{array}      
280: \end{eqnarray}      
281:      
282: \section{OH-OH Interaction}      
283: %\subsection{Hamiltonian}      
284: We will consider diatom-diatom scattering as two interacting      
285: rigid rotors in their ground rotational states.     
286: The complete Hamiltonian for the collision process can then be written     
287: \begin{equation}      
288: H=T_{1}+T_{2}+H^{OH}_{1}+H^{OH}_{2}+V_{s}+V_{\mu \mu}+V_{qq}+V_{disp},   
289: \end{equation}      
290: where $T_{i}$ and $H^{OH}_{i}$ are the translational kinetic energy     
291: and internal motion of molecule $i$,  including the electric field    
292: as in Eqn. (\ref{hamone}). $V_{s}$ is the short- range exchange      
293: interaction, $V_{\mu \mu}+V_{qq}+V_{disp}$ are the dipole- dipole,      
294: quadrupole- quadrupole and dispersion long- range interactions      
295: respectively.  Explicit expression for the dipole-dipole ($\propto 1/R^3$)  
296: and quadrupole-quadrupole ($\propto 1/R^5$) interactions are given  
297: in Ref. \cite{avoird}.  Matrix elements for the the dipole-quadrupole   
298: interaction vanish for rigid rotor molecules in identical states
299: \cite{kuhn}, hence will not  be considered here.  
300:     
301: The anisotropic potential between two interacting rigid-rotor molecules    
302: is conveniently recast into a standard set of angular functions~\cite{avoird}:      
303: \begin{eqnarray}      
304: \label{potential}      
305: V_{s}+V_{\mu \mu}+V_{qq}+V_{disp} \equiv     
306: V(\omega_{A},\omega_{B},\omega,R)=\sum_{\Lambda} V_{\Lambda} A_{\Lambda}(\omega_{A},\omega_{B},\omega),      
307: \end{eqnarray}      
308: where $\Lambda \equiv (L_{A},K_{A},L_{B},K_{B},L)$ and the angular      
309: functions are defined as:      
310: \begin{eqnarray}      
311: A_{\Lambda}(\omega_{A},\omega_{B},\omega)= \sum_{M_{A},M_{B},M}      
312: \left( \begin{array}{ccc}      
313:                   L_{A} & L_{B} & L \\      
314:                   M_{A} & M_{B} & M      
315:                   \end{array} \right)      
316: D^{L_{A}}_{M_{A},K_{A}}(\omega_{A})     
317: D^{L_{B}}_{M_{B},K_{B}}(\omega_{B})     
318: C^{L}_{M}(\omega),      
319: \end{eqnarray}      
320: where $\omega_{A,B}=(\theta_{A,B}, \phi_{A,B})$ are the polar angles   
321: of molecules A and B with respect to the lab-fixed quantization axis, and     
322: ${\bf R}=(R,\omega)$ is the  
323: vector between the center of mass of the molecules in the      
324: laboratory fixed coordinate frame.  The indices $K_A$ and $K_B$    
325: denote the dependence of the interaction on the orientation of    
326: the molecules about their own axes; in what follows we will ignore    
327: this dependence, setting $K_A=K_B=0$.  For the long-range part    
328: of the interaction this approximates the quadrupole moment of OH    
329: as cylindrically symmetric.      
330:     
331: The exchange potential $V_s$ is very complicated, consisting of    
332: four singlet and four triplet surfaces \cite{harding}, and is moreover  
333: poorly characterized.  The most complete treatment of this surface to    
334: date computes the lowest-energy potential for each value of    
335: internuclear separation $R$ \cite{kuhn}.  This potential finds    
336: an extremely deep minimum at $R=2.7$ a.u. corresponding to chemically    
337: bound hydrogen peroxide, and a second, shallower minimum at $R=6$ a.u.   
338: due to hydrogen bonding forces.    However, in cold    
339: collisions, the scattering cross sections are so sensitive to    
340: details of the short-range interaction that knowing the complete interaction    
341: probably would not help anyway.  More importantly, as we will see below,   
342: collisions of the weak-field-seeking states are strongly dominated    
343: by the long-range dipole-dipole interaction.  Therefore, we will use    
344: at small $R$ simply the hydrogen-bonding part of the potential    
345: surface (see Fig. 13 of Ref. \cite{kuhn}), and we will treat this    
346: part of the interaction as if it were isotropic.  
347: Finally, we will assert that the spin states of the OH molecules    
348: are in their stretched states, so that ordinary spin-exchange processes  
349: will  not play a role in these collisions.    
350:     
351: We express the Hamiltonian in a basis of projection of total      
352: angular momentum,      
353: \begin{eqnarray}      
354: \label{wave}      
355: {\cal M}= M_{F_{1}}+M_{F_{2}}+M_{l},      
356: \\      
357: M_{F_{i}}=M_{J_{i}}+M_{I_{i}},      
358: \end{eqnarray}      
359: where $M_{F_{i}}, M_{J_{i}}, M_{I_{i}}$ are the projections of the      
360: full molecule spin, rotational motion and nuclear spin on the      
361: laboratory axis respectively for each molecule. $M_{l}$ is the      
362: projection of the partial wave quantum number on the laboratory      
363: axis. In this basis the wave function for two molecules  is     
364: described  as:     
365: \begin{eqnarray}      
366: \label{basis}      
367: \Psi^{\cal M}=\sum_{1,2,l,M_{l}} \left\{|1> \otimes |2> \otimes |l M_{l}> \right\}^{\cal M} \times \psi^{{\cal M},1,2}(R),      
368: \end{eqnarray}      
369: where $\left\{...\right\}^{\cal M}$ is angular momentum  part of      
370: this wave function and $|i>$ is the wave function for each      
371: molecule as described by Eq.(\ref{onestate}).      
372:       
373: Because the target and the projectile are identical bosons, we must take   
374: into account the symmetry of the wave function under exchange. The    
375: properly symmetrized wave function is then    
376: \begin{eqnarray}      
377: \label{swave}      
378: \left\{|1> \otimes |2> \otimes |l M_{l}> \right\}^{s}=\frac{\left\{|1> \otimes |2> \otimes |l M_{l}> \right\}      
379: +(-1)^{l} \left\{|2> \otimes |1> \otimes |l M_{l}> \right\}}{\sqrt{2(1+\delta_{12})}}      
380: \end{eqnarray}      
381:       
382: Using the expansion of the intermolecular      
383: potential~(\ref{potential}), the wave function~(\ref{wave}), and      
384: taking into account the Wigner-Eckart theorem, we can present the      
385: reduced angular matrix element as      
386: \begin{eqnarray}      
387: \label{matrix2}      
388: <12lM_{l}||| A_{\Lambda}|||1'2'l'M_{l'}>=      
389: \\      
390: \nonumber      
391: (-1)^{L_{A}+L_{B}+J_{1}+J_{1}'+J_{2}+J_{2}'+M_{F_{1}}' + M_{F_{2}}'-\Omega_{1}'-\Omega_{2}'+M_{l}-1}      
392: \frac{(1+\varepsilon_{1}\varepsilon_{1}'(-1)^{L_{A}})}{2}      
393: \frac{(1+\varepsilon_{2}\varepsilon_{2}'(-1)^{L_{B}})}{2}      
394: \times      
395: \\      
396: \nonumber      
397: ([l][l'][J_{1}][J_{1}'][J_{2}][J_{2}'][F_{1}][F_{1}'][F_{2}][F_{2}'])^{1/2}      
398: \left( \begin{array}{ccc}      
399:                   L_{A} & L_{B} & L \\      
400:                   M_{F_{1}}-M_{F_{1}'} &  M_{F_{2}}-M_{F_{2}'} &  M_{l}-M_{l'}      
401:                   \end{array} \right)      
402: \times      
403: \\      
404: \nonumber      
405: \left( \begin{array}{ccc}      
406:                   J_{1}' & L_{A} & J_{1} \\      
407:                   \Omega_{1}' &  0 &  -\Omega_{1}      
408:                   \end{array} \right)      
409: \left( \begin{array}{ccc}      
410:                   J_{2}' & L_{B} & J_{2} \\      
411:                   \Omega_{2}' &  0 &  -\Omega_{2}      
412:                   \end{array} \right)      
413: \left( \begin{array}{ccc}      
414:                   L_{A} & F_{1} & F_{1}' \\      
415:                   M_{F_{1}}-M_{F_{1}'} & -M_{F_{1}}&  M_{F_{1}'}      
416:                   \end{array} \right)      
417: \times      
418: \\      
419: \nonumber      
420: \left( \begin{array}{ccc}      
421:                   L_{B} & F_{2} & F_{2}' \\      
422:                   M_{F_{2}}-M_{F_{2}'} & -M_{F_{2}}&  M_{F_{2}'}      
423:                   \end{array} \right)      
424: \left( \begin{array}{ccc}      
425:                   l' & L & l \\      
426:                   M_{l'} & M_{l}-M_{l'} &  -M_{l}      
427:                   \end{array} \right)      
428: \left( \begin{array}{ccc}      
429:                   l' & L & l \\      
430:                   0 & 0 &  0      
431:                   \end{array} \right)      
432: \times      
433: \\      
434: \nonumber      
435:  \left\{ \begin{array}{ccc}      
436:                   L_{A} & F_{1} & F_{1} \\      
437:                   I & J_{1'} &  J_{1}      
438:                   \end{array} \right\}      
439:  \left\{ \begin{array}{ccc}      
440:                   L_{B} & F_{2} & F_{2} \\      
441:                   I & J_{2'} &  J_{2}      
442:                   \end{array} \right\},      
443: \end{eqnarray}      
444: where $I=1/2$ is each molecule's nuclear spin.     
445:       
446: The reduced matrix elements of the angular functions   
447: $A_{\Lambda}$ between symmetrized      
448: basis states~(\ref{swave}) are      
449: \begin{eqnarray}      
450: <12lM_{l}||| A_{\Lambda}^{s}|||1'2'l'M_{l'}>=      
451: \frac{<12lM_{l}||| A_{\Lambda}^{s}|||1'2'l'M_{l'}>+      
452: (-1)^{l}<21lM_{l}||| A_{\Lambda}^{s}|||1'2'l'M_{l'}>}      
453: {\sqrt{(1+\delta_{1,2})(1+\delta_{1',2'})}}      
454: \\      
455: \nonumber      
456: \times      
457: \frac{1+(-1)^{l+l'}}{2}      
458: \end{eqnarray}      
459: In practice, before each scattering calculation we transform the    
460: Hamiltonian matrix from this basis into the field-dressed basis    
461: defined by (\ref{ebasis}).    
462: We solve the coupled- channel equations using a logarithmic  
463: derivative propagator method \cite{Johnson} to      
464: determine scattering matrices. Using these matrices we      
465: calculate total state-to-state cross sections and rate constants.     
466:       
467:       
468: \section{Results}      
469: This paper considers the scattering problem for OH molecules      
470: in an electrostatic field for cold and ultracold temperatures.     
471: We are interested in particular in the highest energy weak-field-seeking   
472: state of the ground rotational state,    
473: $|(J,I)FM_{F},\Omega  \varepsilon>=|(3/2,1/2)22,3/2,->$.  This state is   
474: indicated by the heavy solid line in Figure 1.  
475: Since the quantum numbers $J$, $I$, and $\Omega$ are the same for    
476: all the scattering processes we will consider, we will refer to this    
477: state by the shorthand notation $|FM_F, \varepsilon \rangle$ $= |22,-> $.    
478:     
479:  The main novel feature of OH- OH scattering, as    
480: compared to atoms or nonpolar molecules, is the presence of    
481: the long-range dipole-dipole interaction and its dependence on    
482: the electrostatic  field.   Because these interactions strongly    
483: mix different partial waves, it is essential that we include    
484: more than one value of $l$.  However, in the interest of emphasizing  
485: the basic underlying physics, we have included only    
486: the s- and d- partial waves.  Sample calculations show that
487: higher partial waves change the results only slightly at the
488: energies considered.  In this case, given the initial    
489: state with $M_{F1} = M_{F2} = 2$, the only allowed values of    
490: the total projection are ${\cal M} = 2,3,4,5,6$.  Among these    
491: channels only the one with ${\cal M}=4$ contains a contribution    
492: from s-wave scattering, and so will deserve special attention     
493: in what follows.  In this case the total number of scattering channels  
494: for all allowed values of ${\cal M}$, is 208.    
495:      
496: \subsection{Prospects for evaporative cooling}    
497:   
498: One of the goals of the present work is to revisit the conclusions    
499: of Ref. \cite{bohnpolar} concerning the effectiveness of evaporative    
500: cooling for electrostatically trapped molecules.  To this end    
501: Figure 2 plots the elastic and state-changing inelastic rate constants versus    
502: field strength for two different collision energies, 100 $\mu K$ and 1$\mu K$.      
503: Here ``elastic'' refers to collisions that do not change the internal    
504: state of either molecule, while ``inel'' denotes those collisions    
505: in which one or both molecules are converted into any other   
506: states.  These transitions are typically exothermic, leading to trap  
507: heating.  Not all these collisions produce untrapped states, however.  
508: We find that the main contributions to the $K_{inel}$ are given    
509: by processes in which  quantum numbers $F$ and/or $M_{F}$ are changed by one.   
510: In particular, the process $|22,- \rangle + |22,- \rangle$
511: $\rightarrow$ $|22,->+|21,->$  generally makes the  
512: largest contribution to $K_{inel}$, especially at high electric field.  
513:   
514: At low field the rates are nearly independent of field, but begin to evolve    
515: when the field approximately exceeds the critical field ${\cal E}_0=\Delta/2\mu$    
516: where the Stark effect changes from quadratic to linear.  Above    
517: this field the rate constants exhibit oscillations as a function    
518: of field.  These oscillations provide an experimentally variable  
519: signature of resonant collisions, meaning that mapping this field  
520: dependence  should help in untangling details of the long-range  
521: OH-OH interaction.  This is similar to the ability of magnetic-field  
522: Feshbach resonances in the alkali atoms to yield detailed scattering  
523: parameters \cite{Roberts,Loftus}.  
524:     
525: Following the example of ultracold atoms, we expect that evaporative    
526: cooling can proceed when the ratio of elastic to  inelastic collisions,    
527: $K_{el}/K_{inel} \gg 1$.  Figure 2    
528: shows that this is hardly ever the case for large field values    
529: ${\cal E} > {\cal E}_0$, except, perhaps, at very special field    
530: values where $K_{inel}$ is at a minimum of its oscillation.     
531: Since the losses are dominated by exothermic processes,    
532: the ratio $K_{el}/K_{inel}$ in the threshold scattering limit    
533: scales as the ratio $k_i / k_f$ of the incident and final channel    
534: wave numbers, as can be seen from the Born approximation.  
535: Thus at high electric fields, where the Stark splitting  
536: is large (hence $k_f$ is large), the ratio may become more    
537: favorable.  In our calculations this apparently happens for    
538: fields above $10^{4}$ V/cm.    
539:     
540: For fields below ${\cal E}_0 \approx 1000$ V/cm, a favorable ratio    
541: of $K_{el}/K_{inel}$ is only somewhat more likely.  For fields this low,   
542: however, the maximum depth of an electrostatic trap is $\approx 8$mK,    
543: as given by the magnitude of the Stark shift (Fig. 1).  The temperature of    
544: the trapped gas must therefore be well below this temperature.  In the  
545: example of a 100 $\mu$K gas, (Fig. 2a), the ratio $K_{el}/K_{inel}$  
546: may indeed be favorable.    
547: However, if the gas is cooled further, say to 1 $\mu$K (Fig. 2b),    
548: this ratio becomes less favorable again.  This is because of the Wigner    
549: threshold laws: the exothermic rate $K_{inel}$ is energy-independent    
550: at low energy, while the elastic scattering rate plummets to zero    
551: as the square root of collision energy.  Thus, in general, evaporative  
552: cooling seems to be viable only over an extremely limited range of temperature    
553: and field  
554: range for the OH molecule, if at all.  We therefore reiterate the message of    
555: Ref. \cite{bohnpolar}, and recommend that cold OH molecules    
556: be trapped by a far-off-resonance optical dipole trap, in their lowest  
557: energy $|F|M_F| , \varepsilon \rangle = |11,+ \rangle$ states.  
558:   
559: At this point we emphasize an essential difference between evaporative  
560: cooling of electrostatically trapped polar molecules and of magnetically  
561: trapped paramagnetic molecules.  For polar molecules the transition   
562: from weak- to strong-field seeking states is {\it always} exothermic,  
563: even in zero applied field.  This is because the lower  
564: member of a $\Lambda$-doublet is always strong-field-seeking (e.g. Fig. 1).  
565: For paramagnetic molecules, by contrast, weak- and strong-field  
566: seeking states can be nearly degenerate at low magnetic field values  
567: (e.g., $^{17}$O$_2$ as discussed in Refs. \cite{Avdeenkov,Volpi}.  
568: In the present case, OH is also paramagnetic and hence could in principle  
569: be magnetically trapped.  For example, the low-energy states with  
570: $|FM_F \varepsilon \rangle$ $= |11, + \rangle$  
571: might be suitable candidates.  The influence of electric dipole  
572: interactions on evaporative cooling of magnetically trapped OH has  
573: yet to be explored.  
574:   
575:   
576:     
577: \subsection{Analysis of the long-range interaction}      
578:     
579: The general behavior of the rate constants in Fig. 2 can be    
580: explained qualitatively by simplifying our model even further,    
581: to a case that contains only the essential ingredients:     
582: the dipole-dipole interaction, the $\Lambda$-doublet, and an electric    
583: field \cite{bohnpolar}.  Roughly speaking, the electric field has two effects    
584: on the molecules: 1) it mixes molecular states of opposite    
585: parity, thus creating induced dipole moments; and 2) the    
586: resulting dipole-dipole interaction strongly couples     
587: scattering channels with different partial waves, leading    
588: to long-range couplings between two molecules.   
589:    
590: As a starting point in this analysis, 
591: Figure 3 breaks down the elastic and inelastic rates into their    
592: contributions from different values of the total projection     
593: of angular momentum ${\cal M}$.  This is done for the rates    
594: calculated at an energy $E = 100$ $\mu$K, from Figure 2(a).      
595: In both elastic and inelastic scattering, the rates are dominated    
596: by the contribution from ${\cal M}=4$, which, it will be recalled,    
597: is the only value of ${\cal M}$ that incorporates s partial waves  
598: in the present model.  
599: We will accordingly consider only this case in what follows.    
600:     
601: The model used to obtain the results in Figs 2 and 3 consists    
602: of 32 channels for the block of the Hamiltonian matrix with    
603: ${\cal M}=4$.  To simplify the analysis of this block even further,    
604: we focus on the sub-Hamiltonian with fixed quantum numbers    
605: $F = M_F = 2$ for each molecule.  This reduces the effective    
606: Hamiltonian to six channels:  there are three non-degenerate    
607: thresholds $E_{\varepsilon_1 \varepsilon_2}$ corresponding     
608: to different possible values of the field-dressed parity quantum number  
609: $\varepsilon_i = \pm$.    
610: For each of these three thresholds there are two channels, corresponding   
611: at large $R$ to s- and d- partial waves.    
612:     
613: The simplified six-channel Hamiltonian matrix then consists of     
614: $3 \times 3$ blocks $\hat{V}^{ll^{\prime}}$ parameterized     
615: by partial wave quantum numbers $l$, $l^{\prime}$:    
616: \begin{eqnarray}    
617: \label{model}    
618: \hat{H} =   
619: \left( \begin{array}{cc}    
620:                   \hat{V}_{diag}^{00} &   \hat{V}_{\cal E}^{02} \\    
621:                   \hat{V}_{\cal E}^{20} & \hat{V}_{diag}^{22}+\hat{V}_{\cal E}^{22}    
622:                   \end{array} \right)    
623: \end{eqnarray}    
624: Here the diagonal components $\hat{V}_{diag}^{ll^{\prime}}$ include    
625: the parity thresholds and the centrifugal interactions,    
626: \begin{eqnarray}    
627: \label{modeldiag}    
628: \hat{V}_{diag}^{ll}=    
629: \left( \begin{array}{ccc}    
630:                   E_{--}+{\hbar^2 l(l+1) \over 2m R^2} & 0 & 0  \\    
631:                   0 & E_{-+}+{\hbar^2 l(l+1) \over 2m R^2} & 0 \\    
632:                   0 &  0 & E_{++}+{\hbar^2 l(l+1) \over 2m R^2}    
633:                   \end{array} \right)    
634: \end{eqnarray}    
635: where the electric-field-dependent thresholds are given by    
636: $E_{\varepsilon_{1}\varepsilon_{2}}    
637: =E_{+}+E_{-}-({\varepsilon_{1}+\varepsilon_{2}}) \Delta    
638: \sqrt{1+k^{2}}/2$, in terms of the dimensionless parameter   
639: \begin{equation}  
640: \label{kdef}  
641: k \equiv {2 \mu{\cal E} \over \Delta}  
642: \end{equation}  
643: that relates the electric    
644: field strength ${\cal E}$ to the zero-field lambda-doublet splitting   
645: $\Delta = E_- - E_+$.    
646: The simplified field-dependent dipole-dipole interaction term     
647: $\hat{V}_{{\cal E}}^{ll^{\prime}}$  is readily parameterized in the   
648: field-dressed  basis as   
649: \begin{eqnarray}    
650: \label{dipole}    
651: \hat{V}_{\cal E}^{ll^{\prime}}=    
652: \left( \begin{array}{ccc}    
653:                   k^{2} & \sqrt{2}k & 1  \\    
654:                   \sqrt{2}k & 1-k^{2}& -\sqrt{2}k \\    
655:                   1 &  -\sqrt{2}k & k ^{2}    
656:                   \end{array} \right)    
657: \frac{C^{ll^{\prime}}}{(1+k^{2})R^3},    
658: \end{eqnarray}    
659: whose coefficient $C^{ll^{\prime}}$, which is independent of both $R$  
660: and the electric field, is given by  
661: \begin{eqnarray}   
662: C^{ll^{\prime}} =     
663: -\mu^{2}   
664: ([l][l'])^{1/2}   
665: \left( \begin{array}{ccc}      
666:                   l' & 2 & l \\      
667:                   0 & 0 &  0      
668:                   \end{array} \right)^{2}   
669: \frac{\Omega^{2} M_{F}^{2}(J(J+1)+F(F+1)-I(I+1))}{2(J(J+1)F(F+1))^{2}}     
670: \end{eqnarray}   
671:  Notice that the    
672: dipole-dipole interaction vanishes for s-waves, so that    
673: $\hat{V}_{\cal E}^{00}=0$.    
674:     
675: Within this simplified model we will refer to the scattering    
676: channels by the parities $\varepsilon_1$ and $\varepsilon_2$   
677: of the two molecules, along with the partial wave     
678: quantum number $l$.  Thus the incident channel for weak-field-seekers will    
679: be denoted $|\varepsilon_1 \varepsilon_2, l \rangle$ $=|--,0\rangle$.    
680: Recall that all other quantum numbers ($J$, $I$, $\Omega$, $F$, $M_F$)   
681: are assumed to have fixed values for each molecule.  
682:     
683: The explicit field-dependence in the coupling matrix (\ref{dipole})    
684: explains qualitatively the behavior of our ultracold weak-field-seeking    
685: molecules, which have incident quantum number $\varepsilon=-$.    
686: For zero electric field ($k=0$), there is no direct dipole-dipole coupling between  
687: identical molecules.  There is, however, an off-diagonal coupling to  
688: different channels with opposite parity,  
689: as can be seen in the form of the Hamiltonian (\ref{model}).    
690: This interaction    
691: brings in the dipole-dipole coupling in second order, contributing    
692: an effective dispersion-like potential $C_6^{eff}/R^6$, with a coefficient    
693: \begin{equation}    
694: \label{C6}    
695: C_6^{eff} = \frac{(C^{20})^{2}}{2\Delta}    
696: \propto \frac{{\mu}^{4}}{\Delta}   
697: \end{equation}  
698: for both s- and d- partial waves.   
699: For the OH molecule this effective coefficient is $\approx 4 \times 10^4$    
700: atomic units, far larger than for the alkali    
701: atoms that are familiarly trapped.  Thus, even in zero external field    
702: the effective interaction strength of polar molecules is quite large.    
703: This may imply the breakdown of the contact-potential approximation to describing    
704: Bose-Einstein condensates of polar molecules, even when their dipoles  
705: are not aligned by an external field \cite{Santos,You,Goral}.  We note that the     
706: quadrupole-quadrupole interaction is relatively unimportant, becoming  
707: larger than this effective dispersion interaction only when   
708: $R > \approx 3 \times 10^{5}$ a.u.   
709:     
710: When the field is switched on, the s-wave channels undergo a qualitative   
711: change.  Now the incident channel $|--,0 \rangle$ sees a direct coupling   
712: to its d-wave counterpart $|--,2\rangle$, via the matrix element    
713: $V^{l=0,l'=2} \propto \frac{k^{2}}{1+k^{2}} \frac{\mu^{2}}{R^{3}}$.    
714: This perturbation  generates an far stronger effective long-range potential   
715: of the form $C_4^{eff}/R^4$, with   
716: \begin{eqnarray}    
717: \label{slong}    
718: C_4^{eff}=  -(\frac{k^{2}}{1+k^{2}})^{2} \frac{\mu^{4}2m}{l(l+1)},   
719: \end{eqnarray}    
720: where $l=2$.   Thus the electric field is able to completely alter  
721: the character of the intermolecular interaction.  
722:     
723: For d-wave collisions, the dipole-dipole coupling is direct, but not in    
724: the limit of zero field, where the molecules are in parity eigenstates.    
725: At low fields ($k \ll 1$, where the Stark effect is quadratic),    
726: the diagonal coupling $V_{\cal E}^{22} \propto k^2/(1+k^2)$, is small.  
727: In this limit the molecules are nearly in parity eigenstates,    
728: hence do not ``know'' that they have dipole moments.  At larger fields  
729: this interaction grows in scale, thus ``activating'' the dipoles.     
730: This is why the rate constants shown in Figs 2,3 begin to evolve     
731: at fields near ${\cal E}={\cal E}_0$.  It is also why the contributions    
732: from all angular momentum projections except the one with ${\cal M}=0$   
733: contribute only weakly to scattering at low field.  The    
734: channels with ${\cal M} \ne 4$ are all of d-wave character, hence  
735: obey a threshold law $\sigma \propto  E^{2}$ at low fields,  then  
736: evolve to a $\sigma \propto {\rm const}$ threshold law at larger fields.   
737:     
738: \subsection{Large-field oscillations and long-range states of the    
739: [OH]$_2$ dimer}    
740:     
741: At fields larger than the critical field ${\cal E}_0$, the rate    
742: constants in Fig.2 exhibit oscillations with field.  Significantly, these occur    
743: only when the projection of total angular momentum ${\cal M}=4$,    
744: which is the only case in which s- and d- partial waves are mixed (Fig. 3).    
745: To understand this oscillating behavior of cross sections we     
746: show in Fig. 4(a) the adiabatic potential curves in the simplified 6-channel     
747: model~(\ref{model}).  In the case shown the field is ${\cal E}=10^4$ V/cm.    
748: In this figure a strong avoided crossing can be seen at $R \approx 60$ a.u.,    
749: corresponding to the crossing of the attractive $|\varepsilon_1  
750: \varepsilon_2,l \rangle=$  
751: $|--,0\rangle$ channel    
752: with the repulsive $|-+,2\rangle$ channel.  The strong dipole-dipole    
753: interaction between these different partial waves creates the     
754: adiabatic potential shown as a heavy black line and labeled $U_0$.  
755:     
756: This potential curve supports bound states of the [OH]$_2$ dimer.    
757: These bound states are of purely long-range character, similar to     
758: the long-range states of the alkali dimers \cite{Movre} that have been used  
759: in photoassociation spectroscopy studies of ultracold collisions    
760: \cite{Jones,Stwalley}.  Moreover, in the case of the [OH]$_2$    
761: states the shape of the potential $U_0$, hence the energies of the bound states,    
762: are strongly subject to the strength of the applied electric field.    
763: The curve in Fig. 4(a) in fact possesses no bound states in zero    
764: field, but five by the time the field reaches $10^4$ V/cm.      
765: More realistic adiabatic potentials are of course more elaborate,    
766: as shown in Fig. 4(b) for the more     
767: complete Hamiltonian that includes hyperfine levels.  Nevertheless,    
768: in this figure, too, can be seen adiabatic potential wells    
769: that will support bound states.    
770:     
771: The significance of these curves is twofold: the crossing is    
772: very adiabatic, implying that coupling to lower-energy channels    
773: is weak, and that therefore the cross sections depend only weakly    
774: on details of the short-range potentials.  This we have indeed     
775: verified by altering the short-range potential in the full calculation.    
776:     
777: Additionally, as the field strength grows and the potential becomes    
778: deeper, new bound states are added to the potential, causing    
779: scattering resonances to appear.  This is the cause of    
780: the oscillations observed in the rate constants in Fig. 2.  To    
781: illustrate this, we reproduce in Fig. 5 the complete elastic    
782: scattering rate constant (solid line), along with the    
783: same quantity as computed in the simple six-channel model    
784: (dashed line).  The qualitative behavior is nearly the same, namely,  
785: oscillations appear at fields above ${\cal E}_0$.  Moreover,  
786: the arrows in the figure indicate the values of the field for which    
787: a bound state of $U_0$ coincides with the scattering threshold.  
788: These fields correspond fairly well    
789: to the peaks, although they are somewhat offset by coupling to    
790: other channels.  Nevertheless, this simple picture  
791: clearly identifies the origin of these oscillations with the  
792: existence of long-range bound states.   
793:     
794: These resonant states are not Feshbach resonances, since there is    
795: no excitation of internal states of the molecules; nor are they    
796: shape resonances in the usual sense, since there is no barrier    
797: through which the wave function must tunnel.  Instead, they are the  
798: direct result of altering the interaction potential to place a  
799: bound state exactly at threshold \cite{landau}.    
800: Probing these states through direct scattering of weak-field-seeking    
801: states should reveal details about the long-range OH-OH interaction,    
802: making possible a comparison of theory and experiment without    
803: the need to fully understand the short-range [OH]$_2$ potential energy  
804: surface.    
805:     
806:     
807:       
808: \section{Conclusion}      
809: In this paper we theoretically investigated ultracold collisions of  
810: ground state polar diatomic molecules in an electrostatic field,       
811: taking OH molecules as a prototype.  Focusing on weak-field-seeking  
812: states, we have strengthened the suppositions in Ref. \cite{bohnpolar}  
813: that long-range dipolar interactions drive inelastic scattering processes  
814: that are generally unfavorable for evaporative cooling of this species.  
815: However, at electric fields above a characteristic value ${\cal E}_0$,  
816: oscillations occur in both elastic and inelastic collision rates,  
817: implying that a regime may be found where the ratio  
818: $K_{elastic}/K_{inelastic}$ is favorable for cooling.  
819: Even though evaporative cooling may be difficult, the inelastic rates  
820: may nevertheless prove useful diagnostic tools for cold collisions  
821: of these molecules.  The Stark slowing technique  provides  
822: a means of launching a bunch of molecules toward a stationary trapped  
823: target, i.e., of making a real scattering experiment \cite{Meijer3,Ye}.  
824:   
825: For actual trapping and cooling purposes, for instance as a means of  
826: producing molecular Bose-Einstein condensates or degenerate Fermi gases, it  
827: seems likely that the molecules must be trapped in their strong-field  
828: seeking states.  Collisions of these species will present their own  
829: difficulties, since they will depend strongly on the short-range  
830: part of the potential energy surface.  However, the scattering length  
831: for OH-OH scattering may be determined by photoassociation spectroscopy  
832: to the long-range bound states we have described above.  This will be  
833: analogous to the determination of alkali scattering lengths   
834: \cite{Jones,Stwalley}, except that microwave, rather than optical, photons  
835: will be used.  The detailed properties of the long-range [OH]$_2$ states,   
836: and prospects for using  them in this way, therefore deserve further attention.  
837:       
838: %{\bf \Large Acknowledgment}      
839: This work was supported by the National Science Foundation. We acknowledge
840: useful discussions with J. Hutson and G. Shlyapnikov.
841:       
842: \begin{references}      
843:   
844:  \bibitem{Santos} L. Santos, G.V. Shlyapnikov, P. Zoller and M. Levenstein,     
845:  Phys. Rev. Lett. {\bf 85}, 1791 (2000).;\\ Stefano Giovanazzi et.al.,  J.Phys.B: At.Mol.Opt.Phys. {\bf 34}, 4757 (2001)     
846:       
847: \bibitem{You} S. Yi and L. You, Phys. Rev. A {\bf 63}, 053607 (2001).  
848:   
849: \bibitem{Goral} K. G\'{o}ral, K. Rz\c{a}\.{z}ewski, and T. Pfau, Phys.  
850: Rev. A {\bf 61}, 051601 (2000); J.-P. Martikainen, M. Mackie, and  
851: K.-A. Suominen, Phys. Rev. A {\bf 64}, 037601 (2001).  
852:   
853: \bibitem{Shlyapnikov} M. A. Baranov, M. S. Mar'enko, V. S. Rychkov, and  
854: G. V. Shlyapnikov, cond-mat 0109437 (2001).  
855:   
856: \bibitem{DeMille} D. DeMille, quant-ph/0109083 (2001).  
857:   
858: \bibitem{Shaffer} J. P. Shaffer, W. Chalupczak, and N. P. Bigelow,  
859: Phys. Rev. Lett. {\bf 82}, 1124 (1999).  
860:   
861: \bibitem{Schloder} U. Schl\"{o}der, C. Silber, and Z. Zimmerman, Appl. Phys.  
862: B {\bf 73}, 801 (2001).  
863:   
864: \bibitem{Doyle} J. D. Weinstein, R. deCarvalho, T. Guillet, B.  
865: Friedrich, and J. M. Doyle, Nature {\bf 395}, 148 (1998).   
866:   
867: \bibitem{Egorov} D. Egorov,  
868: J. D. Weinstein, D. Patterson, B. Friedrich, and J. M. Doyle,  
869: Phys. Rev. A {\bf 63}, 030501 (2001).  
870:       
871: \bibitem{Meijer1} H. L. Bethlem, G. Berden, A. J. van Roij, F. M. H.  
872: Crompvoets, and G. Meijer, Phys. Rev. Lett. {\bf 84}, 5744 (2000).      
873:       
874: \bibitem{Meijer2} H. L. Bethlem, G. Berden, F. M. H. Crompvoets,      
875: R. T. Jongma, A. J. A. van Roij, and G. Meijer, Nature {\bf 406},      
876: 491 (2000).      
877:   
878: \bibitem{Meijer3} H. L. Bethlem, F. M. H. Crompvoets, R. T. Jongma,  
879: S. Y. T. van der Meerakker, and G. Meijer, to appear in Phys. Rev. A.  
880:      
881: \bibitem{bohnpolar} John L. Bohn, Phys.Rev.A {\bf 63}, 052714(2001)  
882:  
883: \bibitem{landau} L.D. Landau and E.M. Lifshitz, Quantum Mechanics: Non- Relativistic Theory   
884: (Nauka, Moscow, 1989, 4th ed.)    
885:   
886: \bibitem{Shakeshaft} R. Shakeshaft, J. Phys. B {\bf 5}, L115 (1972).  
887:   
888: \bibitem{Deb} B. Deb and L. You, Phys. Rev. A {\bf 64}, 022717 (2001).  
889:   
890: \bibitem{Ye} J. Ye, private communication.  
891:      
892: \bibitem{coxon} J.A. Coxon,  Can.J.Phys {\bf 58}, 933 (1980).;\\     
893: Thomas D. Varberg and Kenneth M. Evenson, J.Mol.Spec. {\bf 157}, 55 (1993).     
894:      
895: \bibitem{kuhn} Bernd Kuhn  et.al., J.Chem.Phys. {\bf 111} 2565 (1998).  
896:       
897: \bibitem{harding} Lawrence B. Harding, J.Chem.Phys. {\bf 95} 8653 (1991).      
898:   
899: \bibitem{Bala} N. Balakrishnan and A. Dalgarno, Chem. Phys. Lett.    
900: {\bf 341}, 652 (2001).    
901:   
902: \bibitem{Johnson} B. R. Johnson, J. Comput. Phys. {\bf 13}, 445 (1973).  
903:     
904: \bibitem{mizushima}  M.Mizushima, The Theory of Rotating Diatomic Molecules (Wiley, New York, 1975);\\     
905: Helene Lefebvre- Brion, Robert W. Field, Perturbations in the Spectra of Diatomic Molecules      
906: (Academic Press Inc.(London) Ltd., 1986);\\     
907: Karl F. Freed, J.Chem.Phys. {\bf 45} 4214 (1966).     
908:       
909: \bibitem{miller} Steven M. Miller and David C. Clary, J.Chem.Phys. {\bf 98} 1843 (1993).      
910:       
911: \bibitem{schreel} K.Schreel and J.J.Meulen, J.Phys.Chem.A {\bf 101}, 7639 (1997).      
912:      
913: \bibitem{avoird} Ad van der Avoird, Topics in Curr. Chem, {\bf 93} 1 (1980)      
914:   
915: \bibitem{Roberts} J. L. Roberts, J. P. Burke, S. L. Cornish, N. R. Claussen,   
916: E. A. Donley, and C. E. Wieman, Phys. Rev. A {\bf 64}, 024702 (2001).  
917:   
918: \bibitem{Loftus} T. Loftus, C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin,  
919: submitted to Phys. Rev. Lett.  
920:   
921: \bibitem{Avdeenkov} Alexandr V. Avdeenkov and John L. Bohn, Phys.Rev.A {\bf 64}, 052703(2001)      
922:       
923: \bibitem{Volpi} A. Volpi and J. L. Bohn, to appear in Phys. Rev. A.  
924:   
925: \bibitem{Movre} M. Movre and G. Pichler, J. Phys. B {\bf 10}, 2631 (1977);  
926: W.C. Stwalley, Y.-H. Uang, and G. Pichler, Phys. Rev. Lett.  
927: {\bf 41}, 1164 (1978).  
928:   
929: \bibitem{Jones} K. M. Jones, P. S. Julienne, P. D. Lett, W. D. Phillips,   
930: E. Tiesinga, and C. J. Williams, Europhys. Lett. {\bf 35}, 85 (1996).  
931:   
932: \bibitem{Stwalley} J. P. Burke, Jr. {\it et al}, Phys. Rev. A {\bf 60},  
933: 4417 (1999); C. J. Williams {\it et al.}, Phys. Rev. A {\bf 60}, 4427 (1999).  
934:        
935: \end{references}      
936:      
937: \newpage     
938: \begin{figure}     
939: \label{stark}     
940: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig1a.ps}}     
941: \centerline{\includegraphics[width=.50\linewidth,height=0.60\linewidth,angle=-90]{fig1b.ps}}     
942: \caption{     
943: The Stark effect in ground state OH molecules, taking into account hyperfine splitting.  
944: (a) shows the states that have odd parity $\varepsilon=-$ in zero electric field   
945: (f states), whereas (b) shows those of even parity (e states).  
946: The weak-field-seeking state with quantum numbers $F=M_F=2$,  the subject 
947: of this paper, is indicated by the heavy solid line.    Note that states with
948: $M_F = \pm |M_F|$ are degenerate in an electric field.
949: }     
950: \end{figure}     
951:      
952: \begin{figure}     
953: \label{rates}     
954: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig2a.ps}}     
955: \centerline{\includegraphics[width=.50\linewidth,height=0.60\linewidth,angle=-90]{fig2b.ps}}     
956: \caption{     
957: Rate constants versus electric field for OH-OH collisions with molecules   
958: initially in their  $|FM_F,\varepsilon \rangle$ $=|22->$ state.  
959: Shown are the collision energies $100\mu K$ (a) and $1\mu K$(b).    
960: Solid lines denote elastic scattering rates, while dashed lines 
961: denote rates for inelastic collisions, in which one or both molecules
962: changes its internal state.
963: These rate constants exhibit characteristic oscillations in field when  
964: the field exceeds a critical field ${\cal E}_0 \approx 1000$ V/cm.  
965: }     
966: \end{figure}     
967:      
968: \begin{figure}     
969: \label{elastic}     
970: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig3a.ps}}     
971: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig3b.ps}}     
972: \caption{     
973: Elastic (a) and inelastic (b) rate constants versus electric field for the      
974: same circumstances as in Figure 2(a).  The rates are separated into
975: contributions from different values of ${\cal M}$, the projection of
976: total angular momentum on the laboratory $z$-axis.
977: }     
978: \end{figure}     
979:      
980:      
981: \begin{figure}     
982: \label{adiabatic}     
983: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig4a.ps}}     
984: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig4b.ps}}     
985: \caption{     
986: Adiabatic potential energy curves.  The curves in (a) correspond to the  
987: simplified six-channel model described in the text, and show a long-range  
988: potential well (labeled $U_0$) that can hold bound states of the [OH]$_2$  
989: dimer.  The curves in (b) are those for the more complete calculation that  
990: includes hyperfine structure.  
991: }     
992: \end{figure}     
993:   
994:   
995: \begin{figure}  
996: \label{bound}  
997: \centerline{\includegraphics[width=0.50\linewidth,height=0.60\linewidth,angle=-90]{fig5.ps}}  
998: \caption{  
999: Elastic rate constants versus field, as in Figure 2.  
1000: The solid line reproduces the elastic rate constant from figure 2.  The  
1001: dashed line is an approximate elastic rate constant based on the simplified  
1002: six-channel model described in the text.  The arrows indicate values of  
1003: the electric field at which bound states of the long-range potential  
1004: $U_0$ (Figure 4) coincide with the scattering threshold.  
1005: }  
1006: \end{figure}  
1007:      
1008: \end{document}   
1009: