Neutron Limit on the Strongly-Coupled Chameleon Field K. Li,1,2 M. Arif,3 D. G. Cory,4,5,6,7 R. Haun,8 B. Heacock,9,10 M. G. Huber,3,∗ J. Nsofini,11,6 D. A. Pushin,6,11,† P. Saggu,4 D. Sarenac,6,11 C. B. Shahi,8 V. Skavysh,9 W. M. Snow,1,2,‡ and A. R. Young9,10,§ (The INDEX Collaboration) 1Department of Physics, Indiana University, Bloomington, Indiana 47408, USA 2Center for Exploration of Energy and Matter, Indiana University, Bloomington, IN 47408, USA 3National Institute of Standards and Technology, Gaithersburg, MD 20899, USA 4Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada 5Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L 2Y5, Canada 6 6Institute for Quantum Computing, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada 1 7Canadian Institute for Advanced Research, Toronto, Ontario M5G 1Z8, Canada 0 8Department of Physics, Tulane University, New Orleans, LA 70118, USA 2 9Department of Physics, North Carolina State University, Raleigh, NC 27695, USA n 10Triangle Universities Nuclear Laboratory, Durham, North Carolina 27708, USA a 11Department of Physics and Astronomy, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada J (Dated: January 27, 2016) 6 The physicalorigin of thedark energythat causes theaccelerated expansion rateof theuniverse 2 is one of the major open questions of cosmology. One set of theories postulates the existence of a self-interactingscalarfieldfordarkenergycouplingtomatter. Inthechameleondarkenergytheory, ] O thiscouplinginducesascreeningmechanismsuchthatthefieldamplitudeisnonzeroinemptyspace butisgreatly suppressedinregionsofterrestrialmatterdensity. Howevermeasurementsperformed C under appropriate vacuum conditions can enable the chameleon field to appear in the apparatus, h. where it can be subjected to laboratory experiments. Here we report the most stringent upper p boundonthefreeneutron-chameleoncouplinginthestrongly-coupledlimitofthechameleontheory - using neutron interferometric techniques. Our experiment sought the chameleon field through the o relative phase shift it would inducealong one of the neutron paths inside a perfect crystal neutron r interferometer. Theamplitudeofthechameleonfieldwasactivelymodulatedbyvaryingthemillibar t s pressures inside a dual-chamberaluminum cell. Wereport a 95% confidence level upperbound on a theneutron-chameleoncouplingβ rangingfrom β<4.7×106 foraRatra-Peeblesindexofn=1in [ thenonlinearscalarfieldpotentialtoβ<2.4×107 forn=6,oneorderofmagnitudemoresensitive 1 thanthemostrecentfreeneutronlimitforintermediaten. Similarexperimentscanexplorethefull v parameter range for chameleon dark energy in the foreseeable future. 7 9 8 INTRODUCTION and dynamical screening mechanism to explain why it 6 hasnotbeenobservedinpreviousprecisiongravitational 0 measurements. . 1 The discovery of the accelerated expansion of the uni- 0 Inthispaperwespecificallyaddressaparticularexam- verse[1,2]incombinationwithothercosmologicalobser- 6 ple of such a screened scalar field called the chameleon vations implies that a component of the universe called 1 field [8–11]. The chameleon field has a nonlinear poten- darkenergyconstitutesabout70%ofthe energydensity v: tial of the form V(φ)=Λ4+ Λ4+n [7] with n the Ratra- of the universe. This original work has been confirmed φn i Peebles index and Λ=2.4 meV based on the acceleration X by more sensitive observations [3–6]. It is known that rate of the universe expansion. Additionally, it has an ar anecgoanttivriebpurteiosnsutroe itnheEvinasctueuinm’sefineelrdgyeqdueantsioitnys,aacntsdlsikinecae extra term that couples to matter field A(φ) = MβPLφ, whereβisadimensionlesscouplingtomatterandM is pressuregravitatesitcancausethe expansionoftheuni- PL thereducedPlanckmass. Sotheeffectivepotentialtakes verse to accelerate. There is no consensus on the na- theformV (φ)=V(φ)+A(φ)ρ. Theappearanceofthe ture and physical origin of dark energy, and most of the eff local matter density ρ in the effective potential makes proposed research in this area consists of astronomical observations to more precisely characterize its effects on the effective mass m2eff ∼ n(n + 1)Λ−14++nn(nMβρPL)nn++12 theuniverse’sexpansionrateandothercosmologicaland of the chameleon field density-dependent, allowing the astronomical observables. However there is an interest- chameleon to evade many of the existing experimental ing subset of ideas for the origin of dark energy which tests of gravity. This also causes the chameleon field to can be addressed in laboratory experiments. One set of be highly suppressed in the presence of even the mod- such ideas postulate that dark energy is due to a scalar estly low matter density environment present in most field φ which adopts a nonzero value in the vacuum of terrestrial lab experiments where the effective mass is outer space. For this scalar field to evolve into the dark extremely heavy, thus further escaping detection. As a energy seen today, one must postulate a self-interaction result, the chameleon field, along with a range of simi- 2 lar theories, has yet to be ruled out by experiment. A tron tests involving an apparatus used to test the weak very extensive review of the chameleon field within the equivalence principle for free neutrons [27] and a Lloyd’s broadercontextofmodifiedgravitytheorieshasappeared mirrortypeofneutroninterferometer[28]havealsobeen recently [13]. proposed. A recent review of calculations performed to This paper presents the result of a recent experiment search for dark energy of various types using neutrons, using a perfect crystal neutron interferometer to place laboratoryexperiments to searchfor Casimir forces,and a limit on the chameleon field’s coupling to matter β gravitational inverse square law violations has recently anddoessoinaparticularlydirect,transparentmanner. appeared [29]. We actively modulate the amplitude of the chameleon Calculations which showed that atoms can feel an un- scalar field in a gas cell in one arm of our interferome- screened chameleon field [20] have encouraged experi- ter and exploit the unique ability of neutrons to coher- ments to search for chameleons using atom interferom- ently penetrate the cell wall and access the phase shift etry. The first result of an atom interferometry experi- from the neutron-chameleon coupling. This research ac- ment conducted to search for chameleons has appeared tivity brings together several physics subfields (gravita- very recently [21]. This experiment looked for a phase tionalphysics,atomicphysics,condensedmatterphysics) shift in a cesium atom interferometer operated in ultra- andintheneutroncaseemployscentralizeduserfacilities highvacuumnearasphericalmasswhichcanbeasource constructedmainlyformaterialssciencestudies,thereby ofachameleonfield. Alreadytheconstraintsonthe cou- involving an uncommon diversity of scientific research pling β from this atom interferometry experiment are techniques and environments in the quest to experimen- quite strong. The prospects for further improvement in tally address what is perhaps the most exciting issue in the atom interferometry experiments are very encourag- cosmology. ing [22, 23]. In the regime of small matter coupling β the best lab- As the chameleon fields must couple directly to mass oratory constraints on chameleons over a wide range of to be relevant for the observed universe expansion ac- n come from laboratory tests of the inverse square law celeration,other possible chameleonprobes such as pho- ofgravitywith sensitivity atthe 100microndark energy tons possess a more model-dependent coupling to these scale[14]. Experimentaltestsofthegravitationalinverse dark energy scalar fields. Strong experimental con- squarelawoperatingoverotherdistanceregimes,suchas straints on chameleon-photon couplings already exist. those designed originally to measure the Casimir inter- Examples of photon-based searches for chameleons in- action, can also constrain β. The sensitivity of a force clude the CHASE (the GammeV CHameleon Afterglow sensor specially designed to search for chameleons has SEarch)experiment[16],theAxionDarkMattereXperi- recently been analyzed [15]. ment(ADMX) [17], asearchfor chameleonparticlescre- In the regime of large β techniques which employ two atedviaphoton-chameleonoscillationswithinamagnetic largemassesstarttobecomeinsensitivebecauseboththe field[18], and the CAST CernAxionSolarTelescope ex- sourceandsinkofthechameleonfieldemanateonlyfrom periment [19]. athinregionofthesurfaceoftheobjectsduetothenon- Mostof the chameleonexperiments performed to date linearchameleonself-shielding,anditisthereforeprefer- using neutrons and atoms have sought the chameleon able in this regime to employ test particles whose pres- field by passing the probe close to a dense mass in- encedoesnotsuppressthe chameleonfield. Neutronand side a high vacuum environment and searching for the atom interferometry can be used to searchfor the possi- phase shift from the chameleon-matter coupling β in a bleexistenceofchameleonandrelatedscalarfieldsinthis chameleon field gradient. The chameleon field profile strongly-coupledchameleonregimesincetheseprobesdo is obtained by solving the appropriate nonlinear Klein- not locally suppress the chameleon field. Gordonequationforφusingtheboundaryconditionsset Slow neutrons can be used to perform sensitive by the experimental apparatus. A special feature of the searches for dark energy scalar fields [24]. The fact that experiment reported in this work is that we periodically neutron experiments are a sensitive method to probe introduce a nonzero matter density into the experimen- chameleon fields was pointed out by [25] in the context tal chamber to actively suppresses the chameleon field of an analysis of measurements on the quantum states in the measurement. To perform this type of search we of bouncing ultra-cold neutrons. The disturbance of the exploit the fact that, unlike atoms, neutrons are able to chameleon field near the surface of the flat neutron mir- pass through matter at densities (gas pressures of a few ror employed in these measurements modifies the neu- mbarsuffice)forwhichthechameleonfieldisgreatlysup- tron bound state energies and wave functions as well pressed. We were able to realize an experiment in which as the relative phase of coherent superpositions of the thechameleonfieldintheapparatusseenbytheneutrons neutron gravitational bound states. Recent experiments is repeatedly “turned off” by the addition of a small gas conductedinthissystem[26]havebeenusedtoconstrain pressureintheapparatusand“turnedon”byevacuation chameleon fields, and elaborations of this method are in of the chamber. Moreover, using a neutron interferome- principle capable ofmuchgreatersensitivity. Other neu- ter provides another advantage in that we can keep the 3 pressure difference between the two arms of interferome- in the quantum amplitude of a spin-1/2 particle rotated ter constant. This provides a direct measurementof any by 2π, (3) the most precise determinations of neutron- background that is independent of gas pressure, such as nucleusscatteringamplitudes,(4)sensitivetestsofquan- neutron scattering off the cell walls etc. tum entanglement predictions such as the Bell inequal- ities and the Greenberger-Horne-Zeilinger inequalities, and (5) subtle effects in neutron optics, most recently NEUTRON INTERFEROMETRY the successful manipulation of the orbital angular mo- mentum quantum number of a neutron beam [30]. The experiment was performed at the National Insti- Our experiment searches for the neutron phase shift tute of Standards and Technology’s (NIST) Center for between the two coherent paths of the interferome- NeutronResearch(NCNR)locatedinGaithersburg,MD. ter arising from the coupling of the neutron to the AttheNCNRfreeneutronsaregeneratedusinga20MW chameleon field. The neutron phase shift Φ due to cham researchreactorwhichfeedsovertwodozenindividualin- the chameleon scalar field is strumentsthatareprimarilytailoredformaterialscience β m2φ(x) applications. Here we used monochromatic 11.1 meV Φ =− dx (3) neutrons and interferometric techniques similar to that cham Z MPL ¯h2k of a Mach-Zehnder interferometer for light optics [36]. wheremisthemassoftheneutron,φ(x)isthechameleon Theperfectcrystalneutroninterferometerusedinthis field, k is the neutron wave vector, and the integration experiment consistsof three crystalblades on a common is performed over the neutron’s trajectory. We use the crystalbase;aschematicofwhichisshowninFig.1. The signconventionthat defines the neutronphase suchthat monolithic silicon base below the blades ensures proper positive potentials give negative phases. By measuring arcsecondalignmentbetweenthelatticeplanesofeachof Φ , we can then limit β for a given Ratra-Peebles cham the threeblades. Thefirstbladeservesto spatiallysepa- index n. ratetheneutron’swavefunctionψe−iΦ intotwocoherent The calculation of the chameleon phase shift requires paths (A and B). In order for the two paths interfere, a integratingover the neutron trajectoryof the chameleon centralcrystalbladeactsasalossymirroranddirectsthe field profile φ(x) inside the cell as a function of the gas paths back together onto the third blade. Neutrons exit density. Previous calculations have shown [32] that φ(x) the interferometer along either one of two paths labeled is a rapidly varying function of the density of the gas. traditionallyas‘O’and‘H’andaredetectedusinghighly When the gas density is low the chameleon field φ(x) efficient 3He-filled proportional counters. It should be develops a nonzero amplitude for distances sufficiently noted that there is only one neutronat a time inside the far from the walls of the vacuum chamber. As the gas interferometerandthusitisaelegantexampleofmacro- density is raised into a critical regime, which depends scopicselfinterference. Differencesinphase∆Φbetween on β and the geometry of cell, the chameleon field φ(x) the paths A andB modulates the intensities recordedby is suppressed and tends toward zero. We can therefore the detectors as activelymodulatethe chameleonfieldintheexperiment. I =A +Bcos[ξ(δ)+∆Φ] (1) The first experiment to use perfect crystalneutronin- O O I =A −Bcos[ξ(δ)+∆Φ] (2) terferometrytosearchforchameleonscalarfieldswasre- H H cently performed at Institut Laue-Langevin (ILL) [35]. In order to determine ∆Φ and the other fit parameters In this experiment neutrons also passed through a vac- (AO,H and B) one could vary the cosine term in a con- uum chambermounted in the perfect crystalinterferom- trollable way (ξ(δ)). This is done by the adding what is eter. The effect of the chameleon field was sought both calleda ‘phaseflag’inside the interferometer. The phase by varying the relative separation of the neutron paths flag used here is a 1.5 mm thick × 50 mm wide piece fromthevacuumcellwallsbytranslatingthecellrelative of optically flat quartz and is illustrated in Fig. 1b. By totheincidentbeamandalsobyvaryingthepressurein- rotating the phase flag an angle δ a phase shift of ξ(δ) sidethecell. Sincetheformerislimitedbytheuniformity is caused due to the effective path length difference be- of a machined cell, in our experiment we employed the tweenpathsAandB.Rotatingδ by±2.5degreescreates pressure variation method. an interferogram like the one shown in Fig 2. The perfect crystal neutron interferometery technique employed in this work has been used to conduct a num- EXPERIMENTAL DESIGN AND SETUP ber of textbook experiments in gravitation, neutron op- tics,andquantumentanglement[31]. Theseexperiments A two-chamber vacuum cell is placed in the perfect include, but are not limited to, (1) the first demonstra- crystal neutron interferometer (see Fig. 1b) with inter- tionthatthegravitationalfieldaffectsneutronwavefunc- nal pressures controlled at different values. The vacuum tions asexpectedinnon-relativisticquantummechanics, cell as well as the two neutron beams are both symmet- (2) clear demonstrations of the fascinating minus sign ric about the center walls of vacuum cell, so those two 4 The helium gas pressures in the cell in either configu- ration are low enough that the equation of state of the helium gas is well-described by the ideal gas law. The gas density and resulting neutron phase shift from the neutron optical potential of the helium gas is then pro- portional to the gas pressure. At these low gas densities the neutron phase shift from the helium gas is a few or- dersofmagnitudesmallerthantheultimatesensitivityof FIG. 1: Left is a 3D schematic of the neutron our experiment to phase shifts from the chameleon field, interferometer seen in profile with the two coherent so in practice Φgas can be neglected. Furthermore, even beam paths; right shows the top view of the ifthephaseshiftfromthegaswaslarger,ouractivecon- two-chamber gas cell for the experiment, which fits trol of the pressure difference between the two sides of around the central blade of the interferometer crystal. the cell in the two configurationswould cause this phase shift difference (Φ −Φ )−(Φ −Φ ) gas,1A gas,1B gas,2A gas,2B 0.45 to cancel to high accuracy. The phase shift difference from the two cell walls in the two different configura- 0.40 tions, (Φ − Φ ) − (Φ − Φ ) is also cell,1A cell,1B cell,2A cell,2B 0.35 ) negligible: the only physically plausible mechanism that H +I 0.30 mightcauseadifference,namelysomeabsolute-pressure- O (I 0.25 dependent change of the phase shift from the neutron /O optical potential of the aluminum cell walls, is known to I 0.20 benegligiblefrompreviousmeasurementsatmuchhigher 0.15 Conf. 1 gaspressuresattheNCNRoftheneutron-heliumoptical Conf. 2 0.10 potential. Undertheseconditions,∆Φ isacleanand cham -2 -1 0 1 2 direct measurement of the chameleon phase shift. Ta- d [deg] ble I shows the two sets of pressure configurations that are used in the experiment. The “setpoint” in the table FIG. 2: A typical pair of O-beam interferograms, refers to a low enough pressure at which the chameleon normalized to the total sum of counts, corresponding to fieldmayproduceanextraphaseshift. Thechoiceofgas the two run configurations in the experiment. pressures used in this experiment are also low enough Uncertainties are purely statistical. that, for the chameleon coupling strengths to which we aresensitive,thechameleonfieldseestheheliumgasused in the cell as a homogeneous medium based on previous neutron beams should feel exactly the same chameleon analysis of this issue [32]. field if the pressure inside is same. However the neutron In the experiment we choose to control the absolute interferometer works in such way that it detects the dif- pressure (the low pressure side in Fig. 1) and the differ- ference between the two neutron beams so we need to entialpressureacrosstwochamberstoeliminate possible controlthe pressureofthe twochambersatdifferentlev- systematiceffectsasdiscussedabove. Theabsolutepres- els. First (Conf. 1), the pressure in the right chamber sures in Conf. 2 will deviate slightly from the expected is kept low so that the chameleon field can develop a 1.33 mbar and 2.67 mbar shown in Table I because the nonzerovalue. Meanwhile,the leftchamberisfilledwith absolute pressure gauge used in this experiment has a gas at a higher pressure so that the chameleon field is maximummeasuringrangeof0-0.133mbar. Soinstead highly suppressed. Then (Conf. 2), the gas pressure in of measuring the pressures in Conf. 2 directly, the dif- each chamber is raised by the same amount so that the ferential pressure gauge, which can measure up to 1.33 chameleonfieldgetssuppressedinbothchambers. Sothe mbar,is usedto achievethe desiredpressuresinConf. 2. four phase shift neutron picks up during the experiment We measured the associated pressure uncertainty to be are, smaller than 0.01 mbar. This gives a negligible system- atic uncertainty since the chameleon field amplitude is Φ =Φ +Φ +Φ (4) 1,A cham,1A cell,1A gas,1A closetozeroatsuchhighpressureanditisonlylogarith- Φ1,B =Φcham,1B+Φcell,1B+Φgas,1B (5) mically sensitive to the pressure in this regime. Φ =Φ +Φ +Φ (6) Fig. 3 shows the schematic of the gas handling system 2,A cham,2A cell,2A gas,2A Φ =Φ +Φ +Φ (7) (GHS)usedinthisexperiment. Thevacuumcellismade 2,B cham,2B cell,2B gas,2B of aluminum alloy 7075. Two CF flanges are machined Ideally, the phase shift due to chameleon in Φ , Φ , on the two ends of the cell to accommodate aluminum 1,B 2,A Φ shouldbeclosetozerosowedefinethephaseshiftof gasket seals with negligible outgassing. A wall of thick- 2,B chameleontobe∆Φ =(Φ −Φ )−(Φ −Φ ). ness3mmseparatesthecellintotwochamberswhichcan cham 1,A 1,B 2,A 2,B 5 be filled by gas at different pressures. The GHS employs metal seals (CF 1-1/3”) and ultra-high vacuum (UHV) RunCycle Configurations Pressure(High) Pressure(Low) compatible components which are helium leak tight as Conf. 1 0.67 mbar Setpoint Run1 verified by measurements using a helium leak detector. Conf. 2 0.79 mbar 0.133 mbar All the vacuumtubes andthe mechanicalbellows havea Conf. 1 1.33 mbar Setpoint high-conductance path to the cell (tubing diameters are Run2 Conf. 2 2.67 mbar 1.33 mbar greater than 2 cm) so the pressure gradient inside the gas handling system is minimized. Two absolute pres- TABLE I: Pressure configurations used for two different sure capacitance transducers and one differential capaci- measurement runs. tance transducer areplacedclose to the vacuum cell and used to monitor the absolute pressure and differential pressure across the vacuum cell. Two types of feedback loops controlthe pressures inside the cells. One controls CALCULATION OF THE CHAMELEON FIELD PHASE SHIFT IN THE CELL the differential pressure across the chamber using a mo- torized edge-welded stainless UHV bellows according to readings from the differential pressure gauge. The other Tocalculatethechameleonscalarfieldinsideavacuum feedbackcontrolemploysanabsolutepressuregaugeand cell, one could solve the Klein-Gorden equation, a vacuum compatible sensitive mass flow controller to control the pressure in the low pressure chamber. All ∂V −nΛ4 βρ valves are controlled by air-actuated switches to reduce ∆φ= eff = + (8) ∂φ φn+1 M possible electromagnetic or vibrational noise in the in- PI terferometer environment. Both absolute pressure and Unfortunately there is no known analytical solution differential pressure are controlled with fractional fluc- to this non-linear second order partial differential equa- tuations below 1%. Fig. 5 shows the pressure stability tion so we had to use a finite difference method to ob- data. tain a numerical solution of the 3D field profile inside HP LP HP: High Pressure chamber the experiential cell chamber. We built a 3D mesh with PG LP: Low Pressure chamber roughly 500×500×500 nodes in total and solved the PG: Pressure Gauge non-linearPoissonequationiterativelyusingtheformula DFPG DFPG: Differential Pressure Gauge below (Eqn. 9, where i,j,k denotes the node index in PV PV PV: Pneumatic valve each dimension and l denotes the iteration number). To PG TP: TurbomolecularPump accelerate the convergence of the calculation we further MB PV MB: Motor and Bellow exploitedtheGauss-SeidelmethodshowninEqn.10. To (cid:0) get a precise solution, the number of grid points must PV MFC MFC: Mass Flow Controller be fairly large because the chameleonfield growsrapidly He TP close to the walls. The gradientof the chameleon field is close to infinity at such places, which inevitably causes FIG. 3: Schematic diagram of the gas handling system instabilityintheiterationmethod. Toaddressthisprob- which maintains a constant pressure in one chamber lem, we implemented an uneven grid with more grid and a controlled pressure difference between the two points where the field changes dramatically and fewer chambers. grid points where the field does not change much. The explicit formula is lengthy but is similar to Eqn. 10. φ(l) +φ(l) −2φ(l+1) φ(l) +φ(l) −2φ(l+1) φ(l) +φ(l) −2φ(l+1) −nΛ4 βρ i+1,j,k i−1,j,k i,j,k i,j+1,k i,j−1,k i,j,k i,j,k+1 i,j,k−1 i,j,k + + = + (9) ∆h2x ∆h2y ∆h2z φi(,lj),nk+1 MPI ( nΛ4 − βρ )+ φi(+l)1,j,k+φi(−l+11,)j,k + φi(l,j)+1,k+φi(l,j+−11),k + φi(l,j),k+1+φi(l,j+,k1)−1 φ(l+1) = φ(il,j)n,k+1 MPI ∆h2x ∆h2y ∆h2z (10) i,j,k 2 + 2 + 2 ∆h2 ∆h2 ∆h2 x y z Having solved for the chameleon field inside vacuum cell, the extra phase shift picked up by neutrons due to 6 chameleonfieldis computedaccordingtoEqn.3by inte- torsreduceslightlytheamplitudeofthelineintegralused gratingthechameleonfieldovertheneutronpathlength. tocomputethechameleonphaseΦ (showninFig.4) cham Fig. 4 shows the calculated phase shift caused by the andthereforereducesthecalculatedupperlimitofβ pro- chameleon field with different Ratra-Peebles model pa- portionally. These scaling corrections are also negligible rameters n and β. This result agrees well a previous compared to the uncertainty in our upper limits for β. calculation in the literature [32]. systematic correction uncertainty -2 10 Helium nuclear scattering 0.002 rad/bar 2.0E-6 rad 4 Pressure gauge accuracy 0.3% FS 1.2E-4 rad d] -32 Vacuum cell misalignment 1◦ rotation 0.0005Φcham ra 10 1 mm translation 0.02Φcham | [m 4 Neutron beam divergence 1.5◦ 0.006Φcham a 2 ch 10-4 TABLE II: Estimates of systematic uncertainties. |F 4 n=1, b =10✆ n=1, b =10☎ 2 n=6, b =10✆ n=6, b =10☎ 10-5 Before fitting the measured phase shift to the calcu- -6 -5 -4 -3 -2 -1 lated phase shift due to chameleon field, the sequence of 10 10 10 10 10 10 Pressure [ ] phaseshiftdifferences isfilteredusinga time-seriesanal- (cid:0)✁✂✄ ysis algorithm designed to remove slow zero-point drifts FIG. 4: The calculated phase shift caused by chameleon in data sequences like ours which oscillate between two field versus gas pressure for various values of the states with equal measurement times [37]. To do this Ratra-Peebles index n and β. one takes a weighted average from neighboring points from the original sequence, y = p c u , where u i k=0 k i+k i is the original data sequence andPy is the combined se- i quence. A covariance matrix must be used to properly estimate the uncertainties after the correlations induced DATA ACQUISITION AND ANALYSIS by the weighting algorithm. The weights c satisfy the k equation, Environmental factors are known to cause the phase of a neutron interferometer to drift. To isolate the chameleon phase from environmental phase drifts, we 1 1 1 1 ··· 1 c 0 0 switched between Conf. 1 and Conf. 2 after taking each 0 1 2 3 ··· p c 0 1 interferogram. The contrast (or B/A from Eqn. 1) of tfehreonmeeutterronwainstearrfoeurenndce37p%attaenrndwisitchotnOhseiscteenlltinwtihtheionttheerr- 0... 1... 4... 9... ·.·.·. p...2 c...2 =0... (11) Al cells used inside the interferometer (the empty inter- 0 1 2p−1 3p−1 ··· pp−1c 0 p−1 ferometer contrast is 85%). Since the chameleon field 1 0 1 0 ··· 0 c 1 p is a function of pressure as shown in Fig. 4, we also varied the pressure setpoint in Table I to look for any which is designed so that a zero point drift in the signal pressure dependence of the phase shift. Two sequences withapolynomialtime dependence uptoorderp willbe ofmeasurements weretakenduring two adjacentreactor cancelled by combining each p+1 items in the sequence cycles. Theonlydifferencebetweenthetworunsisinthe while a true signal correlated with the difference in the pressure configurations used. In the first run, nine pres- two configurations is kept unchanged. For comparison suresetpointswerescannedintherangefrom3.33×10−4 we present both the filtered mean and unfiltered mean mbar to 2.67×10−3 mbar. In the second run, we chose in Fig. 6. The good agreementbetween the mean values three pressure setpoints that span over a wider range, andstatisticaluncertaintiesofthe filteredandunfiltered 3.33 × 10−4 mbar, 3.33 × 10−3 mbar and 2.00 × 10−2 phase shift data shows that any possible effects of inter- mbar. The measuredphase shift alongwith the pressure ferometer phase drift are negligible in our measurement. and differential pressure stability is shown in Fig.5. We use the filtered data to extract our limit. Systematic uncertainties in our measurement are neg- Toestablishanupperlimitofβ atthe95%confidence ligible compared to the statistical uncertainty. Table II level,thesquareoftheweightedresidualsissummedover lists the major systematic uncertainties and corrections. all measurements (χ2(β)) The first two items could lead to a nonzero phase shift even in the absence of a chameleon field but both are [ζ(β) −∆Φ ]2 much smaller than the statistical uncertainty (typically χ2(β)= i i (12) 0.0025 rad) in the chameleon phase. The last three fac- Xi σi2 7 a) mbar] 00..666666661150 Z0χ2(βlim)p12(χ2′)dχ2′ =0.95 (13) [ P 0.666605 where p (χ2) is the χ2 distribution with 12 degrees of 12 -2 freedom,correspondingtothe12pressuresetpoints. The 10 ] calculated limit is shown in Table III and the excluded r ba 10-3 areaisshowninFig. 7asafunctionoftheRatra-Peebles m [ index n. P 0.2 a) ] 0.1 0.03 d a r 0.02 [ m 0.0 ] ha ad 0.01 c r [ -0.1 m 0.00 a h c -0.01 -0.2 =1.0X107 Unfiltered mean 0 2 4 6 8 10 -0.02 ✢ 6 Filtered mean =5.0X10 Days ✢ 6 b) -0.03 =2.0X10 ✢ 2 3 4 5 6 7 8 9 2 3 4 5 ] 1.3333 0.001 ar Setpoint Pressure [mbar] b m 1.3332 [ b) P 0.008 Unfiltered mean 1.3331 Filtered mean -2 7 10 =1.0X10 P [mbar] 100-.32 [rad]ham 00..000004 ✢✢✢==52..00XX110066 c d] 0.1 -0.004 a r [ m 0.0 a -0.008 h c 2 4 6 8 2 4 6 8 2 -0.1 0.0001 0.001 0.01 Setpoint Pressure [mbar] -0.2 FIG. 6: a) and b) shows measured phase in two runs 0 10 20 30 compared to calculated phase with different values of β Days (n=1). FIG. 5: a) and b) shows the pressure difference between the two cells (∆P), the Conf. 1 pressure in the low pressure chamber (P), and the measured chameleon n 1 2 3 4 5 6 phase shift in the first and second runs (∆Φcham). βlimit×106 4.7 8.2 12.7 17.9 20.4 23.8 TABLE III: The calculated upper limit on β with 95% confidence level. where ζ(β) is the expected chameleon phase shift and i ∆Φ is the measured phase shift with uncertainty σ for i i the ith pressure setpoint. To estimate β, χ2(β) is then minimizedwithrespecttoβ foragivenRatra-Peeblesin- CONCLUSION dex n. However,the typical computation of a fit param- eter confidence interval is not valid in this case, because We have conducted a search for chameleon dark en- for our measurements this function reaches its minimum ergy fields using neutron interferometry. We realized an at β = 0 for all values of n due to the constraint β > 0. experiment in which the chameleon field is periodically Tofindthe95%confidenceinterval,χ2(β)wassolvedfor varied in magnitude with no change in the experimen- the value of χ2 that gives: tal geometry. Our upper bound of β < 4.7×106 for a 8 tal refutation of a plausible dark energy theory. Scalar field candidates for dark energy which employ other screening methods, such as symmetrons, might also be constrained by this and other experiments with further analysis. ACKNOWLEDGEMENTS We acknowledge the support of the National Institute of Standards and Technology, US Department of Com- FIG. 7: The excluded regions in (β, n) parameter space merce, in providing the neutron facilities used in this work. This work was supported by NSF grants NSF compared to other experiments. From bottom to top: torsion pendulum experiment [12]; atom interferometer PHY-1205342, PHY-1068712, PHY-1307426 and DOE experiment [21]; this work; ILL neutron interferometer awardDE-FG02-97ER41042. Financialsupportprovided by the NSERC CREATE and DISCOVERY programs, experiment [35]. Other experimental constraints in the regime of large β are less stringent. CERC, and the NIST Quantum Information Program are gratefully acknowledged. K. Li and W. M. Snow acknowledge the support of the Indiana University Cen- Ratra-Peeblesindexofn=1toβ <2.4×107forn=6is terforSpacetimeSymmetriesandtheIndianaUniversity Faculty Research Support Program. the most sensitive direct constraint on the free neutron- chameleon coupling in the strong coupling regime of the theory. It is more sensitive than a recent neutron inter- ferometerexperimentattheILL[35]byaboutafactorof 5forn=1toafactorof30atn=4,anditcutsintopart ∗ Corresponding author: [email protected] of the projected sensitivity of a proposedexperiment us- † Corresponding author: [email protected] ing an optimized force sensor [15]. The constraints from ‡ Corresponding author: [email protected] this work on the chameleon are consistent with but less § Corresponding author: [email protected] stringent than a very recently-published atom interfer- [1] A. G. Riess et al.,Astron. J. 116, 1009 (1998). ometry experiment using cesium atoms [21]. Under the [2] S. Perlmutter et al.,Astrophys.J. 517, 565 (1999). [3] E. Komatsu et al. Astrophys.J. Suppl.192, 18 (2011). assumption that chameleon dark energy obeys the grav- [4] D. Larson et al. Astrophys.J. Suppl.192,16 (2011). itational equivalence principle and that there are no es- [5] N. Suzukiet al. Astrophys.J. 746, 85 (2012). sentialdifferencesbetweentheresponseofaneutronand [6] A.G.Sanchezet al.Mon.Not.R.Astron.Soc.425,415 a cesium atom to the chameleon field, the atom inter- (2012). ferometer constraintsare about two ordersof magnitude [7] B. Ratra and P. J. E. Peebles, Phys. Rev. D 37, 3406 more stringent at present. (1988). [8] J.KhouryandA.Weltman,Phys.Rev.Lett.93,171104 This neutron interferometer experiment can be im- (2004). proved by (a) using an interferometer crystal with a [9] J. Khoury and A. Weltman, Phys. Rev. D 69, 044026 larger path length, (b) improving the contrast of the in- (2004). terferencesignalintheinterferometer,(c)optimizingfur- [10] P. Brax, C. van de Bruck, A.-C. Davis, J. Khoury, and therthe pressurerangeofthemeasurements,(d) operat- A. Weltman, Phys.Rev. D 70,123518 (2004). ing the interferometer on a more intense monochromatic [11] S. S. Gubser and J. Khoury, Phys. Rev. D 70, 104001 neutron beam, and (e) varying the neutron coupling to (2004). [12] A. Upadhye,Phys. Rev.D 86, 102003 (2012). the chameleon both by changing the cell geometry and [13] A. Joyce, B. Jain, J. Khoury, and M. Trodden, Phys. also by varying the gas pressure. With these improve- Rep. 568, 1 (2015). ments, the statistical sensitivity of this measurement to [14] E. Adelberger, B. R. Heckel, S. A. Hoedl, C. Hoyle, D. the coupling β can be improved by at least two more Kapner et al.,Phys. Rev.Lett. 98, 131104 (2007). orders of magnitude with negligible systematic effects, [15] A. Almasi, P. Brax, D. Iannuzzi, and R. I. P. Sedmik, which could then surpass the existing atom interferome- Phys. Rev.D 91, 102002 (2015). ter limits at larger n. [16] A.S. Chou et al.,Phys. Rev.Lett. 102, 030402 (2009). [17] G. Rybkaet al.,Phys. Rev.Lett. 105, 051801 (2010). A experimental lower bound on β > 50 for n = 1 al- [18] J.H. Steffen et al.,Phys. Rev.Lett. 105, 261803 (2010). ready exists from gravitational inverse square law tests. [19] V. Anastassopoulos et al. (CAST Collaboration), Phys. By improving the atom interferometry limits by another Lett. B 749, 172 (2015). few orders of magnitude, laboratory experiments will ei- [20] C. Burrage, E. J. Copeland, and E. A. Hinds, J. Cos. ther discover chameleons or provide the first experimen- Astro. Phys042 (2015). 9 [21] P. Hamilton, M. Jaffe, P. Haslinger, Q. Simmons, H. [31] H. Rauch and S. A. Werner, Neutron Interferometry: Muller, and J. Khoury,Science 349, 849 (2015). Lessons in Experimental Quantum Mechanics, Wave- [22] S. Schlogel, S. Clesse, and A. Fuzfa, arXiv:1507.03081 Particle Duality, and Entanglement, second edition, Ox- (2015). ford UniversityPress (2015). [23] C. Burrage and E. J. Copeland, Contemporary Physics [32] P. Brax, G. Pignol, and D. Roulier, Phys. Rev. D 88, DOI:10.1080/00107514.2015.1060058 (2015). 083004 (2013). [24] G. Pignol, Int.J. Mod. Phys. A 30, 1530048 (2015). [33] T.C.Black,P.R.Huffman,D.L.Jacobson,W.M.Snow, [25] P. Brax and G. Pignol, Phys. Rev. Lett. 107, 111301 K. Schoen, M. Arif, H. Kaiser, S. K. Lamoreaux, and S. (2011). A. Werner,Phys. Rev.Lett. 90, 192502 (2003). [26] T. Jenkeet al.,Phys. Rev.Lett.112, 151105 (2014). [34] K. Schoen, D. L. Jacobson, M. Arif, P. R. Huffman, T. [27] Yu.N. Pokotolovskii, JETP Lett. 96, 751 (2012). C.Black,W.M.Snow,S.K.Lamoreaux,H.Kaiser,and [28] Yu.N. Pokotolovskii, JETP 116, 609 (2013). S. A. Werner,Phys. Rev.C67, 044005 (2003). [29] P.BraxandA.-C.Davis,Phys.Rev.D91,063503(2015). [35] H. Lemmel et al., Phys.Lett. B 743, 310 (2015). [30] C. W.Clark, R.Barankov, M.G. Huber,M. Arif, D.G. [36] M. Arif, M. S. Dewey, G. L. Greene and W. M. Snow, Cory, and D. A.Pushin, Nature525, 504 (2015). Journal of Research of NIST98, 135 (1993). [37] E. Swanson and S. Slamminger, Meas. Sci. Technol. 21, 115104 (2010).