Centilitre-scale vacuum chamber for compact ultracold quantum technologies
Oliver S. Burrow, Paul F. Osborn, Edward Boughton, Francesco Mirando, David P. Burt, Paul F. Griffin, Aidan S. Arnold, Erling Riis
CCentilitre-scale vacuum chamber for compact ultracold quantum technologies
Oliver S. Burrow, Paul F. Osborn, Edward Boughton, Francesco Mirando, David P. Burt, Paul F. Griffin, Aidan S. Arnold, ∗ and Erling Riis Department of Physics, SUPA, University of Strathclyde, Glasgow, G4 0NG, United Kingdom TMD Technologies Ltd, Swallowfield Way, Hayes UB3 1DQ, United Kingdom Kelvin Nanotechnology Ltd., 70 Oakfield Ave, Glasgow, G12 8LS, United Kingdom Kelvin Nanotechnology, 70 Oakfield Ave, Glasgow, G12 8LS, United Kingdom (Dated: January 21, 2021)Compact ultra-high vacuum systems are key enabling components for cold atom technologies,facilitating extremely accurate sensing applications. There has been important progress towardsa truly portable compact vacuum system, however size, weight and power consumption can beprohibitively large, optical access may be limited, and active pumping is often required. Here, wepresent a centilitre-scale ceramic vacuum chamber with He-impermeable viewports and an integrateddiffractive optic, enabling robust laser cooling with light from a single polarization-maintainingfibre. A cold atom demonstrator based on the vacuum cell delivers 10 laser-cooled Rb atoms persecond, using minimal electrical power. With continuous Rb gas emission active pumping yieldsa 10 − mbar equilibrium pressure, and passive pumping stabilises to 3 × − mbar, with a 17 daytime constant. A passively-pumped vacuum cell, with no Rb dispensing, has currently kept a similarpressure for a year. The passive-pumping vacuum lifetime is several years, estimated from short-term He throughput, with many foreseeable improvements. This technology enables wide-rangingmobilization of ultracold quantum metrology. I. INTRODUCTION
The ability to laser cool atoms yields orders of magni-tude longer interrogation times than with room tempera-ture atoms for equivalent volume devices, enabling mea-surements with unprecedented accuracy [1–3]. Whilstthe most precise of these instruments are room-sized ap-paratus, a generation of compact quantum technologiesare being developed [4–9] to take ultracold atomic accu-racy out of the lab and into real-world applications. Thekey to achieving this goal is to reduce the size, weightand power (SWaP) of the device’s individual components,whilst increasing simplicity and resilience.Quantum cold atom sensors use magneto-optical traps(MOTs) [10–12] comprising laser-cooling sub-systems of:lasers, magnetic coils, optics, and ultra-high vacuum.Development of compact laser systems is a subject ofon-going research, using diode laser and telecommunica-tions industry technology for robust miniaturisation [13–17]. Magnetic coils optimised for low power consumptioncan be designed and fabricated [18], and the optics forlaser cooling can be simplified to a single beam illuminat-ing a pyramidal or planar optic [19–21]. However, whilekey progress has been made in developing miniaturisedvacuum systems for ‘hot’ ions [22, 23] and laser-cooledatoms [24–28], a chamber should ideally be devoid of anychallenging bulky components or appendages, with inte-grated pump and atom source, enabling ultra-high vac-uum in an apparatus with a truly compact form factor.Custom components have been required to improve onthe vacuum system SWaP, leading to vacuum seal chal-lenges, particularly when including the necessary optical ∗ [email protected] access. Ideally the system would be passively pumped[24, 29], to eradicate vacuum power requirements. Theresulting finite vacuum lifetime could be maximised bycareful choice of materials to minimise outgassing and thepermeation of non-pumpable noble gases [25, 30]. Pas-sive pumping also means that the undesirable volumeand magnetic field of an ion pump can be removed, ame-liorating Zeeman systematic shifts on precision atomicmeasurements.Here, we present a cubic ultra-high vacuum chamber,with 32 mm sides, and an integrated diffraction gratingfor magneto-optical trapping of atoms with a single in-put laser beam: a grating MOT (gMOT [31–33]) vacuumcell (Fig. 1 a). With active pumping the cell pressure is10 − mbar, measured via the MOT loading curves, andmaintained indefinitely – ideally suited for fast operationas an ultracold atom source for metrology experiments.With purely passive pumping, the pressure rises to anasymptote of 3 × − mbar, even with a relvatively im-pure atom dispenser continuously running. A second vac-uum cell has maintained a similar pressure with no activepumping or Rb dispensing for a year.Weighing in at 252 g the vacuum assembly (Fig. 1 (a))has been used in a portable demonstrator (Fig. 1 (b)),laser cooling 10 atoms at 10 Hz repetition rate, with20 mW of gMOT optical power delivered by fibre. Elec-trical power of ≈
10 W is sufficient for Rb dispensers, ionpump and anti-Helmholtz coils. A printed circuit board,with a USB 5 V input, can run the whole demonstratorwith the required voltages for the dispensers, coils andion pump, and a USB battery pack then provides sev-eral hours of use. The vacuum cell’s 25 × . × . volume could be substantially reduced if the redundantion pump is removed. This affordable vacuum cell withno active pumping, ≈ − mbar pressures and projected a r X i v : . [ phy s i c s . a t o m - ph ] J a n M illi o n s o f a t o m s FibreQuarter-wave plateCoils Ion pumpPhotodiodeIrisLenses
MOT beam deliveryDetection
Lenses (a) (b)(c)
Pinch off32mmNEG pumpRb sourceMOT N eq � FIG. 1. The portable vacuum chamber assembly (a), including beam delivery, detection and magnetic systems. A photographof the complete demonstrator chamber, with all delivery optics and electronics is depicted in (b), with the essential passively-pumped vacuum footprint for future devices highlighted in red. An experimental Rb MOT loading curve is shown in (c). helium-limited lifetime of several years will be enhancedin future via a new generation of ‘cleaner’ atomic reser-voirs [34–36].
II. CELL CONSTRUCTION
To realise a portable magneto-optical trap, we devel-oped a novel vacuum enclosure to fulfill the necessary re-quirements (Fig. 1): good optical access, near ultra-highvacuum (UHV) pressures ≈ − mbar, and a control-lable source of Rb vapour. Our system is based aroundthe gMOT architecture, which uses a single input beamand a single 2 × in-vacuo micro-fabricated diffrac-tive optic (patterned with e.g. 1D gratings) to cool ≈ atoms [31], with 3 µ K temperature demonstrated [32] foratom populations similar to those shown here.To achieve the required vacuum, components must bechosen which have low outgassing rates and helium per-meability. Materials and bonding methods used mustalso withstand a high temperature vacuum bake in thecell construction and subsequent outgassing. The gMOTvacuum cell comprises an OFE-copper pump stem assem-bly bonded to a sintered alumina cubic chamber with ahigh temperature metallic braze. The pump stem assem-bly incorporates a Ti cathode sputter-ion pump and anon-evaporable getter pump (NEG) [37]. The stem as-sembly can be attached to a vacuum system for high tem- perature vacuum bake via a standard DN40CF flange,and subsequent removal by a pinch-off technique cold-welding the OFE-copper tube closed under applicationof mechanical pressure.The cubic chamber has 32 mm side dimensions, andhouses the optical grating. A glass-ceramic optical view-port is oriented facing the grating to provide access forthe trapping laser, with two further windows on adjacentsides for probe laser access and observation (Fig. 1 (a),(b)). The windows are bonded using a proprietary glass-bonding technique. An OFE-copper port adjacent to thepump stem allows a SAES Getters rubidium dispenser tobe positioned in the alumina cube, with electrical contactto the dispenser made between the port and the pumpstem. Vacuum bake is performed ≥ ◦ C for 100 hours,during which the Rb dispenser and NEG are activated.
III. VACUUM CHARACTERISATION
To facilitate characterisation a testing rig was con-structed which receives the vacuum cell, mechanicallycentring and aligning it to both the trapping laser beamand magnetic field. The rig has a pair of anti-Helmholtztype coils providing a quadrupole magnetic field with18 G/cm axial gradient, and three pairs of Helmholtzcoils to cancel background magnetic fields. The rig’sdelivery beam optic assembly receives a polarisation-maintaining optical fibre, expands and collimates thebeam to a 1 /e radius of 15 mm and circularly polarisesthe light. The fibre delivers 780 nm light to the beamdelivery system, red-detuned 12 MHz from the F = 2 → F (cid:48) = 3 Rb D2 transition, with a repumper sideband of ≈
1% power [21] provided by a 6 .
58 GHz electro-opticalmodulator.Compared to the atomic fluorescence, the light scat-tered in the cell can be high, due to diffracted light fromthe grating hitting the cell walls. A clean signal fromthe atoms is detected by a photodiode in a spatial fil-tering lens system [27], minimising light not originatingfrom the atom trapping point (Fig. 1 (a)). Loading curvedata (Fig. 1(c)) was taken at 1 ms resolution with a 2 sMOT cycle time. The MOT was turned off and on usingthe quadrupole magnetic field, with coil switching time < . N at time t is: N (cid:48) ( t ) = α P Rb − ( β P Rb + γ P bk ) N ( t ) , (1)where MOT loading is proportional to the rubidiumvapour pressure P Rb , and the coefficient α is specific tothe experimental apparatus and parameters. Cold atomsexit the MOT at two rates: due to collisions with theRb vapor or other non-rubidium background gases atpressures P Rb and P bk , with corresponding loss coeffi-cients { β, γ } = { . , . }× mbar − s − [38]. The non-rubidium gases are assumed to be dominated by H , withthe cross-sections of other species varying by at most afactor of 2. Two- and three-body rubidium collisions arealso assumed to be negligible.A brief delay, of a few ms, was observed between ini-tiating laser cooling and measuring the fluorescence sig-nal from trapped atoms. As the imaging system here isonly sensitive to atoms at the MOT position, this is dueto the time taken for the atoms to cool and trap fromthe beam overlap region and then congregate at the trapcentre. A simulation of forces in our atom trapping vol-ume and the imaging region, yielded comparable values (a few ms) to the observed trapping time. This effect wasempirically modelled by replacing the constant α , with α ( t ) = α (1 − e − t/δ ) , where δ is the characteristic time foratoms entering the trap volume to arrive at the imaginglocation.Using this delay time and the following relations forthe measurable quantities, the MOT’s equilibrium atomnumber and lifetime are: N eq = α P Rb τ, τ − = β P Rb + γ P bk , (2)respectively, with solution N ( t ) = N eq (cid:18) δ e − t/δ − τ e − t/τ τ − δ (cid:19) + N (3)where N is a removable residual fluorescent backgroundsignal. This recovers the result from [39] when δ → N eq and τ are determined, however there are three unknowns inEq. 2: P Rb , P bk and the experiment-specific α . Whilstthe procedure laid out in Ref. [39] can be followed tomeasure α , and hence P Rb and P bk , we found that thequantity αP Rb = N eq /τ actually varied with τ . This canbe clearly seen in Fig. 2, where over the course of a weekwe study the MOT fill parameter behaviour ((a) N eq vs τ and (b) N eq /τ vs τ ) with constant Rb dispenser current[42] and only passive pumping.This behaviour was inconsistent with our expectationthat P Rb remained constant. Using low-intensity 780 nmabsorption spectroscopy of the MOT vapor cell [43], as-suming a uniform Rb distribution, we independently con-firmed a constant P Rb = 2 × − mbar. With constant P Rb , and varying αP Rb we have therefore found evi-dence justifying an improved MOT loading model withpressure-dependent α . Moreover, by considering the in-creasing importance of background collisions which occuron timescales comparable to the MOT characteristic cap-ture time τ C [44], one can obtain an excellent fit to thedata using a Poisson distribution and consider only atomssurviving the cooling process relative to the pressure-dependent collisional loss rate [45]. In particular Fig. 2(a) and (b) are fit with the curves: N eq = α P Rb τ e − τ C /τ , NP Rb τ = α (cid:48) e − τ C γ P bk , (4)respectively, where α (cid:48) = α e − τ C β P Rb ≈ α at low Rb pres-sures [46].Utilising the equilibrium MOT atom number N eq andtime constant τ to determine pressure via Eq. 2, we haveexplored the behaviour of the vacuum chamber under avariety of conditions. Following the characterisation of N eq - τ evolution of the cell with the ion pump off anddispensers on in (Fig. 2 (a), (b)), we could determine thebackground pressure P Rb rise over the course of the week. . . . . . . . (a) . . . . . . . . (b) (c) N ( m illi on s o f a t o m s ) Lifetime � (ms) P (10 mbar) -7 bk P ( m ba r) - Time (days) T = 16.7 days P = 2.8 10 mbar eq -6 eq V b k � ( m ba r s ) - - FIG. 2. The evolution of MOT atom number N eq vs. lifetime τ in one week of continuous loading curve measurements, withonly passive pumping and constant Rb gas dispensing (a).The ratio N eq /τ is proportional to the trap loading rate αP Rb (b). The pressure rises when the ion pump is turned off, andthe rate atoms are loaded into the trap decreases when theatom loss due to collisions becomes comparable with the fittedMOT capture time t C = 16 ms used via Eq. 4 for the theorycurves in (a) and (b). From these results and Eq. 2 one caninfer the background gas pressure against time (c) during theweek of passive pumping and Rb dispensing. Fitting this curve with a function P ( t ) = P + ( P eq − P )(1 − e − ( t − t ) /T V ) , yields the parameters for asymptotic equilibrium pres-sure P eq = 3 × − mbar and rise time constant T V =17 days.This observation of passive pumping reaching equilib-rium is supported by a pressure measurement in a simi-lar cell of 1 . × − mbar after 7 months with no activepumping during which no Rb was dispensed. We believethe pressure rise in Fig. 2 (c) is largely due to non-Rbemission from the dispensers, as the pressure ceases torise dramatically when the dispensers are off for severalhours, and the pressure load from the dispensers due toRb is small – P Rb = 2 × − mbar at equilibrium withpassive-only or active pumping.Noble gases are of particular concern when trying toeliminate active pumping from a vacuum system. Pas- P r e ss u r e ( − m ba r) I onpu m po ff Load i ng c u r v e s m ea s u r ed I onpu m pon C e ll m o v ed t op r e ss u r e v e ss e l m i nu t e s r H e C e ll r ep l a c ed i n r i g D i s pen s e r s a c t i v a t ed FIG. 3. Pressure measured in the cell before and after anexposure to 1 bar of He for 5 minutes (9-14 minute region). sive pump mechanisms do not pump noble gases, and assuch their throughput into the vacuum cell must be min-imised. Helium is of particular concern as the smallestmono-atomic gas readily leaks through any channels, andcould permeate directly through the walls and windowsof a poorly designed vacuum cell. Amongst its many ex-citing properties, graphene coating also offers prospectsto prevent He permeability [47]. The partial pressure ofhelium in Earth’s atmosphere is 5 × − mbar, meaningthere is a strong pressure differential between this andthe desired 1 × − mbar overall pressure in the vacuumcell.We studied accelerated He permeation, by first placingthe cell in equilibrium with 6 . . . × − mbar,whereas the maximum expected pressure rise establishedin Fig. 2 would be negligible (and over-estimated as thedispensers are initially off). Attributing this pressurerise solely to the 5 minutes at 1 bar of helium, the timescale for a 10 − mbar pressure rise with atmospheric He(5 × − mbar i.e. 5 ppm in pressure) would thereforealready be 8 years. IV. DISCUSSION
We have created a centilitre volume vacuum cell, withintegrated grating-MOT optics, which can be used as arobust cold atom source. With active ion pumping thecell is expected to function for several years at 10 − mbar,based on the specified total pumping capability. Withpassive-only pumping the cell’s pressure increases to ahigher equilibrium pressure of 3 × − mbar, with anexponential time constant of T V = 17 days, but still hasan expected lifetime of several years.The passive-pumping pressure rise is largely due to theRb dispensers, which at above 500 ◦ C are likely to beoutgassing other species significantly. Moreover, the dis-penser is currently the main contributor to the system’selectrical power. A different rubidium source would ame-liorate the situation, e.g. lower temperature methods in-clude capturing Rb in graphite, for either thermal [34] orelectronic release [35]. Microfabrication-based beam op-tions are also now available [36]. Rubidium metal couldin principle be used, as it has 2 × − mbar room tem-perature vapor pressure, however it would have to bedistilled to a cold point in the chamber, or released froma small ampoule, after the cell fabrication and bake pro-cess.Accelerated He permeation tests already indicate thepassively-pumped cell can withstand eight years of at-mospheric helium leakage, and finding the main channelfor He permeation in future tests could extend this life- time significantly. Whilst the cell does contain a non-evaporable getter pump, optimising the quantity and ac-tivation of this remains a fertile avenue of investigation,with potential for a large improvement.Our vacuum device will aid the mobilization of quan-tum technologies, across a diverse range of applicationsthat require the accuracy that only cold atoms, ions andmolecules provide. Details on experimental data and con-ditions are available in the Dataset [48]. ACKNOWLEDGMENTS
We are grateful for support through InnovateUK grant103880: ‘gMOT - Magneto optical trap system for coldatom technologies’ (EP/R020086/1), as well as Quan-tum Technology Hub EPSRC grants EP/T001046/1 andEP/M013294/1. We thank Alan Bregazzi and JamesMcGilligan for guidance in the design of the imaging sys-tem and for useful discussions. [1] I. Dutta, D. Savoie, B. Fang, B. Venon, C. G. Alzar,R. Geiger, and A. Landragin, Continuous cold-atom in-ertial sensor with 1 nrad/s rotation stability, Phys. Rev.Lett. , 183003 (2016).[2] W. F. McGrew, X. Zhang, R. J. Fasano, S. A. Sch¨affer,K. Beloy, D. Nicolodi, R. C. Brown, N. Hinkley, G. Mi-lani, M. Schioppo, T. H. Yoon, and A. D. Ludlow, Atomicclock performance enabling geodesy below the centimetrelevel, Nature , 87 (2018).[3] C. Overstreet, P. Asenbaum, T. Kovachy, R. Notermans,J. M. Hogan, and M. A. Kasevich, Effective inertial framein an atom interferometric test of the equivalence princi-ple, Phys. Rev. Lett. , 183604 (2018).[4] Y. Bidel, N. Zahzam, C. Blanchard, A. Bonnin,M. Cadoret, A. Bresson, D. Rouxel, and M. Lequentrec-Lalancette, Absolute marine gravimetry with matter-wave interferometry, Nat. Commun. , 627 (2018).[5] L. Liu, D.-S. L ˜A ¼ , W.-B. Chen, T. Li, Q.-Z. Qu, B. Wang,L. Li, W. Ren, Z.-R. Dong, J.-B. Zhao, W.-B. Xia,X. Zhao, J.-W. Ji, M.-F. Ye, Y.-G. Sun, Y.-Y. Yao,D. Song, Z.-G. Liang, S.-J. Hu, D.-H. Yu, X. Hou, W. Shi,H.-G. Zang, J.-F. Xiang, X.-K. Peng, and Y.-Z. Wang, In-orbit operation of an atomic clock based on laser-cooled Rb atoms, Nat. Commun. , 2760 (2018).[6] D. Becker, M. D. Lachmann, S. T. Seidel, H. Ahlers,A. N. Dinkelaker, J. Grosse, O. Hellmig, H. M ˜A ¼ ntinga,V. Schkolnik, T. Wendrich, A. Wenzlawski, B. Weps,R. Corgier, T. Franz, N. Gaaloul, W. Herr, D. L ˜A ¼ dtke,M. Popp, S. Amri, H. Duncker, M. Erbe, A. Kohfeldt,A. Kubelka-Lange, C. Braxmaier, E. Charron, W. Ert-mer, M. Krutzik, C. L ˜A ¤ mmerzahl, A. Peters, W. P.Schleich, K. Sengstock, R. Walser, A. Wicht, P. Wind-passinger, and E. M. Rasel, Space-borne Bose-Einsteincondensation for precision interferometry, Nature ,391 (2018). [7] J. Grotti, S. Koller, S. Vogt, S. H¨afner, U. Sterr, C. Lis-dat, H. Denker, C. Voigt, L. Timmen, A. Rolland, F. N.Baynes, H. S. Margolis, M. Zampaolo, P. Thoumany,M. Pizzocaro, B. Rauf, F. Bregolin, A. Tampellini,P. Barbieri, M. Zucco, G. A. Costanzo, C. Clivati,F. Levi, and D. Calonico, Geodesy and metrology with atransportable optical clock, Nat. Phys. , 437 (2018).[8] D. C. Aveline, J. R. Williams, E. R. Elliott, C. Dutenhof-fer, J. R. Kellogg, J. M. Kohel, N. E. Lay, K. Oudrhiri,R. F. Shotwell, N. Yu, and R. J. Thompson, Observa-tion of Bose-Einstein condensates in an Earth-orbitingresearch lab, Nature , 193 (2020).[9] M. Takamoto, I. Ushijima, N. Ohmae, T. Yahagi,K. Kokado, H. Shinkai, and H. Katori, Test of generalrelativity by a pair of transportable optical lattice clocks,Nat. Photonics , 411 (2020).[10] E. L. Raab, M. Prentiss, A. Cable, S. Chu, and D. E.Pritchard, Trapping of neutral sodium atoms with radi-ation pressure, Phys. Rev. Lett. , 2631 (1987).[11] C. Monroe, W. Swann, H. Robinson, and C. Wieman,Very cold trapped atoms in a vapor cell, Phys. Rev. Lett. , 1571 (1990).[12] J. F. Barry, D. J. McCarron, E. B. Norrgard, M. H. Stei-necker, and D. DeMille, Magneto-optical trapping of adiatomic molecule, Nature , 286 (2014).[13] F. Lienhart, S. Boussen, O. Carraz, N. Zahzam, Y. Bidel,and A. Bresson, Compact and robust laser system forrubidium laser cooling based on the frequency doublingof a fiber bench at 1560 nm, Appl. Phys. B , 177 (2007).[14] S. S. San´e, S. Bennetts, J. E. Debs, C. C. N. Kuhn, G. D.McDonald, P. A. Altin, J. D. Close, and N. P. Robins,11W narrow linewidth laser source at 780 nm for lasercooling and manipulation of rubidium, Opt. Express ,8915 (2012). [15] T. L´ev`eque, L. Antoni-Micollier, B. Faure, andJ. Berthon, A laser setup for rubidium cooling dedicatedto space applications, Appl. Phys. B , 997 (2014).[16] J. Pahl, A. N. Dinkelaker, C. Grzeschik, J. Kluge,M. Schiemangk, A. Wicht, A. Peters, and M. Krutzik,Compact and robust diode laser system technology fordual-species ultracold atom experiments with rubidiumand potassium in microgravity, Appl. Opt. , 5456(2019).[17] E. D. Gaetano, S. Watson, E. McBrearty, M. Sorel,and D. J. Paul, Sub-megahertz linewidth 780.24nm dis-tributed feedback laser for Rb applications, Opt. Lett. , 3529 (2020).[18] Private communication: Nathan Welch and MarkFromhold, University of Nottingham.[19] K. I. Lee, J. A. Kim, H. R. Noh, and W. Jhe, Single-beamatom trap in a pyramidal and conical hollow mirror, Opt.Lett. , 1177 (1996).[20] M. Vangeleyn, P. F. Griffin, E. Riis, and A. S. Arnold,Single-laser, one beam, tetrahedral magneto-optical trap,Opt. Express , 13601 (2009).[21] M. Vangeleyn, P. F. Griffin, E. Riis, and A. S. Arnold,Laser cooling with a single laser beam and a planardiffractor, Opt. Lett. , 3453 (2010).[22] Y.-Y. Jau, H. Partner, P. D. D. Schwindt, J. D. Prestage,J. R. Kellogg, and N. Yu, Low-power, miniature Ybion clock using an ultra-small vacuum package, AppliedPhysics Letters , 253518 (2012).[23] P. D. D. Schwindt, Y.-Y. Jau, H. Partner, A. Casias,A. R. Wagner, M. Moorman, R. P. Manginell, J. R. Kel-logg, and J. D. Prestage, A highly miniaturized vacuumpackage for a trapped ion atomic clock, Review of Scien-tific Instruments , 053112 (2016).[24] D. R. Scherer, D. B. Fenner, and J. M. Hensley, Charac-terization of alkali metal dispensers and non-evaporablegetter pumps in ultrahigh vacuum systems for coldatomic sensors, J. Vac. Sci. Tech. A , 061602 (2012).[25] J. A. Rushton, M. Aldous, and M. D. Himsworth, Con-tributed review: The feasibility of a fully miniatur-ized magneto-optical trap for portable ultracold quantumtechnology, Rev. Sci. Instrum. , 121501 (2014).[26] A. Basu and L. F. Vel´asquez-Garc´ıa, An electrostaticion pump with nanostructured si field emission electronsource and ti particle collectors for supporting an ultra-high vacuum in miniaturized atom interferometry sys-tems, Journal of Micromechanics and Microengineering , 124003 (2016).[27] J. P. McGilligan, K. R. Moore, A. Dellis, G. D. Mar-tinez, E. de Clercq, P. F. Griffin, A. S. Arnold, E. Riis,R. Boudot, and J. Kitching, Laser cooling in a chip-scaleplatform, Appl. Phys. Lett. , 054001 (2020).[28] B. J. Little, G. W. Hoth, J. Christensen, C. Walker,D. J. D. Smet, J. Lee, and P. D. D. Schwindt, A passivelypumped vacuum package sustaining cold atoms for morethan 200 days, arXiv:2101.01051 (2021).[29] R. Boudot, J. P. McGilligan, K. R. Moore, V. Maurice,G. D. Martinez, A. Hansen, E. de Clercq, and J. Kitch-ing, Enhanced observation time of magneto-optical trapsusing micro-machined non-evaporable getter pumps, Sci.Rep. , 16590 (2020).[30] A. T. Dellis, V. Shah, E. A. Donley, S. Knappe, andJ. Kitching, Low helium permeation cells for atomic mi-crosystems technology, Opt. Lett. , 2775 (2016). [31] C. C. Nshii, M. Vangeleyn, J. P. Cotter, P. F. Griffin,E. A. Hinds, C. N. Ironside, P. See, A. G. Sinclair, E. Riis,and A. S. Arnold, A surface-patterned chip as a strongsource of ultracold atoms for quantum technologies, Nat.Nanotech. , 321 (2013).[32] J. P. McGilligan, P. F. Griffin, R. Elvin, S. J. Ingleby,E. Riis, and A. S. Arnold, Grating chips for quantumtechnologies, Sci. Rep. , 384 (2017).[33] R. Elvin, G. W. Hoth, M. Wright, B. Lewis, J. P. McGilli-gan, A. S. Arnold, P. F. Griffin, and E. Riis, Cold-atomclock based on a diffractive optic, Opt. Express , 38359(2019).[34] R. N. Kohn, M. S. Bigelow, M. Spanjers, B. K. Stuhl,B. L. Kasch, S. E. Olson, E. A. Imhof, D. A. Hostutler,and M. B. Squires, Clean, robust alkali sources by inter-calation within highly oriented pyrolytic graphite, Rev.Sci. Instrum. , 035108 (2020).[35] J. P. McGilligan, K. R. Moore, S. Kang, R. Mott, A. Mis,C. Roper, E. A. Donley, and J. Kitching, Dynamic char-acterization of an alkali-ion battery as a source for laser-cooled atoms, Phys. Rev. Appl. , 044038 (2020).[36] C. Li, X. Chai, B. Wei, J. Yang, A. Daruwalla, F. Ayazi,and C. Raman, Cascaded collimator for atomic beamstraveling in planar silicon devices, Nat. Commun. ,1831 (2019).[37] The ion pump has nominal speed 0 . , 033420 (2012).[39] R. W. G. Moore, L. A. Lee, E. A. Findlay, L. Torralbo-Campo, G. D. Bruce, and D. Cassettari, Measurementof vacuum pressure with a magneto-optical trap: Apressure-rise method, Rev. Sci. Instrum. , 093108(2015).[40] J. Scherschligt, J. A. Fedchak, Z. Ahmed, D. S.Barker, K. Douglass, S. Eckel, E. Hanson, J. Hendricks,N. Klimov, T. Purdy, J. Ricker, R. Singh, and J. Stone,Review article: Quantum-based vacuum metrology at theNational Institute of Standards and Technology, J. Vac.Sci. Tech. A , 040801 (2018).[41] Similar results are also obtained by fitting an exponentialrise ( N eq − N )(1 − e − t/τ ) + ( N + N ), omitting data atearly times for τ , in conjunction with the known offset N to determine N eq .[42] We used a lower current for the dispensers than the datashown in Fig. 1(a) to keep P Rb low and have minimalerror on P bk in Eq. 2.[43] J. Keaveney, C. S. Adams, and I. G. Hughes, ElecSus:Extension to arbitrary geometry magneto-optics, Comp.Phys. Comm. , 311 (2018).[44] This capture time is in reasonable agreement with a sim-ple Doppler cooling model [49] that has been enhancedusing stochastic processes.[45] Alternatively, constant αP Rb can be more simply recon-ciled if one allows a τ offset for the minimum lifetime τ C = 16 ms at which MOTs are obtained. A linear fit inFig. 2(a) gives αP Rb = N eq / ( τ − τ C ).[46] We note that the pressure decay coefficient P = ( τ C γ ) − links τ C and γ – enabling a clear determination of one pa-rameter provided independent measurement of the other. [47] P. Z. Sun, Q. Yang, W. J. Kuang, Y. V. Stebunov, W. Q.Xiong, J. Yu, R. R. Nair, M. I. Katsnelson, S. J. Yuan,I. V. Grigorieva, M. Lozada-Hidalgo, F. C. Wang, andA. K. Geim, Limits on gas impermeability of graphene,Nature , 229 (2020).[48] Dataset DOI TBA. [49] J. P. McGilligan, P. F. Griffin, E. Riis, and A. S. Arnold,Phase-space properties of magneto-optical traps utilis-ing micro-fabricated gratings, Optics Express23