Is the fine-structure constant not actually constant? Clarifying the...

8/08/11 9:49 PM

Is the fine-structure constant not actually constant?Clarifying the concept of entanglementTrapped atomic ions and precision sensing

Volume 49, Number 2, Mar–Apr 2012

36 EditorialWarm welcome toSynchrotron funding

37 President’s ColumnWhere are the legislativechecks and balances?

39 News & CommentDecadal plan for AustralianPhysicsMatthew Flinders Medal toKen FreemanActive galactic nuclei tomeasure cosmic distancesBen Eggleton awardedWalter Boas Medal

43 Does electromagnetism’s strength varyacross the Universe?Michael Murphy and astrophysicists at the University of NSWand Swinburne University have found compelling evidencethat the !ne structure constant – and thus the strength ofelectromagnetism – varies across the Universe

49 Cutting entanglementGordon Troup and colleagues at Monash clarify the conceptof entanglement in Special Relativity and QuantumMechanics

53 Precision sensingMichael Biercuk describes a new technique using trappedatomic ions to detect extremely small forces, with asensitivity over 1000 times better than previous techniques

59 ObituariesAngas Hurst (1923–2011) by Alan Carey, Peter Bouwknegtand Max LoheJohn Prescott (1924–2011) by Laurence Campbell and NigelSpooner

61 Product NewsA review of new products from Lastek, Warsash Scienti!c,Coherent Scienti!c and Agilent Technologies

Inside Back CoverPhysics conferences in Australia for 2012

CONTENTS

Is the fine-structure constant not actually constant?Clarifying the concept of entanglementTrapped atomic ions and precision sensing

Volume 49, Number 2, Mar–Apr 2012

CoverObservations of the northern sky fromHawaii and observations of the southernsky from Chile have now been combinedto provide evidence that the !ne-structureconstant varies across the Universe – seethe article by Michael Murphy on page 43[credit: Julian Berengut, University of NSW].

AUSTRALIAN PHYSICS 35MAR–APR 2012 | 49(2)

Australian Institute of PhysicsPromoting the role of physics in research, education, industry and the community

AIP website: www.aip.org.au

AIP ExecutivePresident Dr Marc Duldig

[email protected] President Dr Robert Robinson [email protected] Dr Andrew Greentree [email protected] Dr Judith Pollard

[email protected] Dr John Humble

[email protected] Past President A/Prof Brian James

[email protected] Projects O!cersProf Warwick Couch

[email protected] Olivia Samardzic

[email protected]

AIP ACT BranchChair Dr Anna Wilson

[email protected] Joe Hope

[email protected]

AIP NSW BranchChair Dr Graeme Melville [email protected] Dr Frederick Osman

[email protected]

AIP QLD BranchChair Dr Joel Corney

[email protected] Dr Till Weinhold [email protected]

AIP SA BranchChair Dr Scott Foster

[email protected] Dr Laurence Campbell laurence.campbell@!inders.edu.au

AIP TAS BranchChair Dr Elizabeth Chelkowska

[email protected] Dr Stephen Newbury [email protected]

AIP VIC BranchChair Dr Andrew Stevenson [email protected] Dr Mark Boland [email protected]

AIP WA BranchChair A/Prof Marjan Zadnik [email protected] Dr Andrea Biondo [email protected]

PrintingPinnacle Print Group288 Dundas Street, Thornbury VIC 3071www.pinnacleprintgroup.com.au


Fundamental? Constants?To many physics undergraduates, the fundamental constantsof Nature must seem little more than a rather long list ofobscure numbers at the back of their textbooks. Annoyingly,they have to be committed to memory, not quickly re-derived in a side-calculation as physicists would prefer todo. And they’re (irrational) real numbers, not nice, easy-to-remember, round ones – how many digits of Planck’sconstant can you rattle o! without googling ‘h’?

But these inconveniences go much deeper, right to theheart of our current understanding of physics, in fact.Richard Feynman said it best (as usual) about " = e2/!c,the #ne-structure constant of electromagnetism:

“It has been a mystery ever since it was disco"ered… andall good theoretical physicists put this number up on theirwall and worry about it. Immediately you would like toknow where this number for a coupling comes #om: is itrelated to pi or perhaps to the base of natural logarithms?Nobody knows. It’s one of the greatest damn mysteries ofphysics… We know what kind of a dance to do experimentally

to measure this number very accurately, but we don’t knowwhat kind of dance to do on the computer to make thisnumber come out, without putting it in secretly!” [1].

$at is, within current physics theories, an expressionfor any observable quantity inevitably includes at leastone fundamental constant, the value of which is knownonly from experiment. $e constants cannot be derivedwithin the theories; hence the title ‘fundamental’ andthe annoying need to commit them to memory. Similarly,the theories say nothing about their constancy – only ex-periments can establish that or rule it out. So far, some ofthe most staggeringly precise laboratory measurementsever made have not revealed any variability (e.g. [2]).

And why so many fundamental constants to ‘worryabout’? Probably because our current theories don’tdescribe the most fundamental physics, but merely a setof approximate physical laws. Much like the Newtonianconcept of gravity is fundamentally incorrect and Einstein’sgeneral relativity is better, probably our current conceptof all physical laws is fundamentally incorrect. $e aim,

Electromagnetism:Does Its Strength VaryAcross the Universe?Michael Murphy

The fundamental constants are central to our theories of physics, yet thosetheories offer no understanding of the constants, no way to calculate them. If theconstants were found to vary, we might learn what they depend on and perhapsglimpse a more fundamental theory, perhaps ‘the’ theory. For a decade we’vefound that !, the fine-structure constant, which characterises electromagnetism’sstrength, is slightly smaller in distant galaxies, at least those appearing in thenorthern sky, than on Earth. But now, with measurements over both hemispheres,we find evidence for a dipole-like variation in ! across the sky. The fine-structureconstant ! could be the first fundamental constant that isn’t.

of course, is to !nd a better theory, maybe ‘the correctone’, maybe one without seemingly arbitrary fundamentalconstants.

"e search for ‘variable constants’ is therefore a basictest of physics beyond the Standard Model. But beyond-Standard theories currently o#er little to guide whereand when in the Universe any variability is strongest, oreasiest to spot. We therefore need to test the variabilityof fundamental constants in as wide a variety of placesand times in the Universe as possible.

!, quasars and the Many Multiplet method"e time variability of $ has, of course, been scrutinisedin highly controlled, Earth-bound laboratory experiments."e degree to which frequencies of electromagnetic tran-sitions depend on $ varies from transition to transition,ion to ion. By comparing the ticking rates of single-ionoptical atomic clocks based on Al and Hg over 10months, the relative rate of change in $ was recentlylimited to just a few parts in 1017 per year [2]. Impressive,certainly, but what if $ changes, say, non-linearly withtime and you want to know its value in a far-%ung galaxy10 billion light-years away? Without a theory of varying

$, that 10-month laboratory experiment in our littlecorner of the Universe can’t say much about the laws ofphysics across the entire Universe and throughout its 14 billion year history.

&uasars, and the odd 10-m diameter telescope, makelooking back 10 billion years routine. &uasars are super-massive (~109 solar mass) black holes at the centre ofgalaxies. Friction and gravitational energy from anaccretion disc of in-falling gas and dust means quasarsoutshine all the stars in their host galaxies, radiating~1040 W of light which, even 10 billion light-years away,appears as a relatively bright, star-like continuum of ra-diation in our sky – see Fig. 1.

While quasars are obviously interesting in themselves,ripe with extreme physics and complex interactions withtheir host galaxies, their brightness, distance, compactnessand spectral simplicity (no narrow features) also maketheir lines-of-sight to Earth perfect probes of the narrowabsorption lines from intervening gas. Fig. 1 shows the(simulated) spectrum of a quasar and labels the absorptionlines. Most arise from the Lyman $ transition of neutralhydrogen in the pervasive intergalactic medium, causinga ‘Lyman $ forest’ bluewards of (at lower redshi's than)

AUSTRALIAN PHYSICS 44 49(2) | MAR–APR 2012

Fig. 1. The quasar absorption line experiment. The top panel shows quasar light intersecting a galaxy on its way to Earth.The bottom panel shows a synthetic (but realistic) quasar spectrum with important emission lines (blue labels) from thequasar and absorption lines (black & green labels) from the intervening galaxy. The relative separations between the metallines are sensitive to the "ne-structure constant !.

the quasar’s Lyman ! emission line. But the line of sightto many quasars also passes near enough to a distantgalaxy to produce a deep, damped Lyman ! line, a Lymanionisation edge, molecular hydrogen bands (sometimes)and, most importantly for us, narrow metallic absorptionlines.

It’s these metal lines from intervening gas cloudstowards background quasars that provide the best (optical)probe of cosmological variations in !. "eir sensitivityto ! is shown in Fig. 2. "is #gure is the essence of theso-called ‘Many Multiplet’ method: like a barcode, therelative wavelength separations of transitions fromdi$erent multiplets, and di$erent ions, encode the valueof !. "us, by comparing the pattern of separations seenin a quasar spectrum with laboratory standards, anydi$erence in ! between a distant gas cloud and the labo-ratory can be measured.

It’s worth noting that the set of transitions observed inany given quasar spectrum, from any given absorptionsystem, varies considerably. Looking back at Fig. 2, thismeans that the pattern of wavelength shi%s from the labo-ratory standards will be di$erent from one absorber to an-other; each one has a di$erent barcode depending onwhich transitions are observed. From the point of view ofminimising systematic e$ects, this is very much a goodthing: most instrumental systematics one can imagine tendto a$ect all quasar spectra in the same general ways, sodi$erent barcodes will be a$ected in di$erent ways bysuch systematics. "at is, the diversity of transitions observedin di$erent absorption systems helps to average over and/orexpose simple systematic errors.

A stubborn result: smaller α in quasarabsorbers"e Many Multiplet method of analysing quasar spectrafor the variability of ! was proposed and #rst demonstrated[3, 4] in 1999 by John Webb, Victor Flambaum andVladimir Dzuba at the University of New South Wales.At that time, the #rst largish samples of high-quality,high-resolution spectra were coming from the 10-metrediameter Keck Telescope in Hawaii. Even with a sampleof 30 absorption systems, there was an indication that !was smaller in the absorption clouds, on average, than inthe laboratory by about 10 parts per million, with a sta-tistical signi#cance of about 3& [4]. Only the redder Feand Mg lines could be used in those spectra – those onthe right of Fig. 2 – and, consequently, the absorptionredshi%s were all below 1.8, placing the transitions inthe visible observing window.

Many 3& results go away, usually quite quickly. Notthrough lack of attention, this result didn’t… and stillhasn’t. Its #rst test was at higher redshi%s where, as you’llnote in Fig. 2, an entirely di$erent pattern of line shi%swas expected. But we saw the same result [5, 6]: a smaller! in the absorption clouds by about 7 parts per millionat redshi%s 1.8–3.3, the same as the (re-analysed) lowerredshi% results. Combined, they constituted 4& evidenceof ! varying on cosmological scales.

A larger sample added in 2002, this time coveringboth low and high redshi%s, gave the same result as theprevious two datasets [7]. By 2004 we had published143 measurements of ! in distant galaxies with the con-sistent result that ! was about 6 parts per million smallerthan the current laboratory value [8]. "e formal statisticalsigni#cance of the evidence for varying ! was now above5&.

So the signal had not disappeared with added data.Indeed, it had become clearer, as one expects from a realvariation in !… or, of course, a well-de#ned systematicerror. For an observational (rather than experimental)result like ours, this is, and remains, the most di'cultquestion to answer: can systematic errors explain it?


Fig. 2. Sensitivity to α variation of metallic transitions seenin quasar absorption systems. Here Δα/α is the relativedeviation in α from the current laboratory value, while ‘I’denotes a neutral species, and ‘II’, ‘III’ denote singly- anddoubly-ionised species. Note that the redder Fe and Mgtransitions have a very di#erent pattern of shifts to the bluertransitions; low- and high-redshift absorption systemsrespond very di#erently to systematic e#ects.

“The fine-structureconstant ! could be the firstfundamental constant thatisn’t.”

One by one, we analysed and ruled-out or limited thee!ect of many possible astrophysical and instrumentalsystematic errors [7, 9]. None could explain what we’dfound.

However, all the indications for a varying " had so farcome from just one telescope, the Keck Telescope inHawaii. No matter how good the Keck is, no-one shouldtrust a fundamental result like this from just a single in-strument. Enter the VLT – the Very Large Telescope –in Chile.

New result: an α dipole?In 2004 we began the long task of compiling a similarlylarge sample of high-quality, high-resolution quasarspectra from the VLT. #e previous Keck data were gen-erously donated to us (see the Acknowledgments) already‘reduced’ from their raw form taken at the telescope intoanalysis-ready, science-grade quasar spectra. #e VLTspectra had to be reduced from scratch. We also discoveredproblems with the way VLT spectra were wavelengthcalibrated – the crucial calibration step when lookingfor shi$s between absorption lines – and these neededto be overcome [10]. But with the hard work of PhDstudent Julian King at UNSW, we began measuring " in153 absorption systems in 2008, over the same redshi$

range, using the same Many Multiplet method as before.In my mind, there were two likely possibilities: #e

VLT would show no variation in ", or it would give thesame, smaller value of " in the absorbers as the Keckdata. Neither was true. Nor was it something in between.In fact, the VLT spectra showed the opposite variation in" to the Keck spectra! #at is, on average, the VLTvalues of " were larger in the absorbers than the currentlaboratory value [11, 12].

Your immediate reaction is probably similar to mine:the VLT results simply disagree with the Keck results,neither are correct, and both are the product of (probablydi!erent) systematic e!ects. Well, we had not found asimple systematic error that explained the Keck resultsand it is similarly di%cult to explain the VLT results bya systematic error. Explaining both away probably requirestwo di!erent systematic errors. But, I’m sure you say,this is still the most likely answer. Maybe.

#ere’s another possibility. We projected the resultsonto the sky, as in Fig. 3. #e Keck and VLT see di!erentskies on average – Keck is at +20° latitude and the VLT isat –25°. Our Keck quasars were predominantly northernones, our VLT quasars predominantly southern, but manyin both samples were equatorial. In Fig. 3 you notice a ten-dency for larger blue squares – greater positive deviations


Fig. 3. Distribution of Δα/α values from Keck (circles), VLT(squares) and both telescopes (triangles) across the sky. Therelative deviation from the laboratory value is colour-coded,while the point size scales with the signi#cance of thatdeviation. The grey band indicates the Milky Way disk,behind which few bright quasars are known. The red blobnear the Galactic centre (grey bulge) is the direction and 1σcontour of a dipole #tted to all Δα/α measurements, whilethe green and blue blobs are the dipole directions #tted tothe Keck and VLT data independently. The cross marked ‘A’and the blue dashed curve indicate the Keck + VLT dipole’santi-pole and equator, respectively. Weighted mean valueswere used for quasar lines of sight with two or moreabsorption systems.

Fig. 4. Binned values of Δα/α versus the angle from the best-#tting dipole direction. The bottom panel separates theKeck (pink) from the VLT (blue) values, while the top panelcombines them. The red line indicates the best-#tting dipolemodel, while the blue dashed line shows its 1σ error.

in ! from the laboratory value – to appear near the bottomright of the plot, and larger pink circles to appear near thetop le", indicating larger negative deviations. In the middleyou notice a mix of smaller symbols, ie. less signi#cant de-viations from the laboratory value. $e data seem to suggestvariation in ! across the sky!

Modelling ! as a dipole on the sky (ignoring anyredshi" dependence for now), we can show in Fig. 4 thedeviation in ! as a function of angle from the best-#ttingdipole direction. $e dipole model does seem to describethe data very well. Indeed, compared to a monopolemodel (ie. constant o%set in ! in all directions), thedipolar variation in ! is signi#cant at 4&, as assessed byboth analytical means and by a bootstrap technique (ie.randomly reassigning ! measurements to di%erent quasarsight-lines) [11, 12].

$e dipole direction is shown as a red 1& error blob inFig. 3. Because it’s deep in the southern hemisphere, youmay still suspect that the apparent dipolar variation in !is driven by systematic e%ects between the two di%erenttelescopes, Keck and VLT, even though they agree forequatorial regions. But the blue and green blobs in Fig. 3give signi#cant pause for thought: when dipoles are#tted to the Keck and VLT data independently, theypoint in the same direction on the sky. If the dipolar

variation from the combined results was driven by sys-tematic errors between Keck and VLT, you would notexpect this. We estimate such close alignment wouldhappen 6% of the time by chance alone, given the distri-bution of quasar sight-lines across the sky [11, 12].

We also #nd similarly-close alignment between dipoles#tted independently to low- and high-redshi" subsamplesof the combined Keck + VLT results, with a 2% chanceof occurring by chance alone. $e joint probability ofboth these chance alignments happening is di'cult toestimate, but a bootstrap technique suggests that it’sabout 0.1% [11, 12]. All in all, the current data leave uswith the strange picture of the Universe illustrated inFig. 5.

Conclusions and future testsA"er a decade of #nding ! to be ~6 parts per milliontimes smaller than on Earth in distant galaxies, predom-inantly in the northern sky, we now #nd the oppositeresult on the other side of the sky. But, rather than con-tradicting each other, these two results show a remarkableconsistency: there is 4& evidence for a dipole-like variationin ! across the sky; the dipoles from the two telescopescoincide; the low- and high-redshi" dipoles also coincide.And we still have not found systematic errors that can


Fig. 5. Illustration of the ! dipoleand the spectral line shifts (seealso Fig. 2) which are observed, onaverage, over many quasar lines ofsight from two di"erenttelescopes, the Keck and the VLT.[credit: Julian Berengut, UNSW]

“The aim, ofcourse, is tofind a bettertheory, maybe‘the correctone’, maybeone withoutseeminglyarbitraryfundamentalconstants.”

explain the results from either telescope, let alone thecombined results.

I have not mentioned other constraints on variationsin ! from work by other groups using the Many Multipletmethod, or from radio and millimetre-wave absorptionlines, or from meteoritic analysis, or from the Oklonatural "ssion reactor, etc. While none of these othermethods has yielded evidence for a varying !, they are allconsistent with the ! dipole from our work [13].

Dedicated observing programs on both Keck and theVLT are now aimed directly at refuting or con"rmingour results. And we are starting to target the anti-pole ofthe dipole with quasars high in the northern sky andnear our Galaxy’s anti-centre (see Fig. 3): the signal islargest there and even a modest sample could rule it out(or con"rm it!). Crucially, we are trying to do this withtwo telescopes to make absolutely sure that systematicerrors are not responsible for whatever we "nd.

What if all these future astronomical measurementscon"rm that ! varies on cosmological scales? It would stillbe important to test with a di#erent technique altogether.And laboratory atomic clock experiments may be the keytest. We see tentative evidence for the amplitude of the !dipole increasing radially outwards from Earth, supportinga ‘simple’ picture of an ‘! gradient’ in the Universe [11,12]. If such a gradient exists in our Solar System, it may bedetected as an annual modulation in the relative ticking

rates of atomic clocks at the 10-20 level [13]. While this isthree orders of magnitude below current sensitivity, thelast decade has already seen a 1000-fold improvement.With such rapid progress, a laboratory test may see avarying ! in the next decade or two.

Acknowledgments$is article reports work done in collaboration with theauthors of references [3–13] below, especially JulianKing, John Webb and Victor Flambaum (University ofNSW). Keck spectra were kindly contributed by ChrisChurchill (New Mexico State Univ.), Jason Prochaska(Univ. California, Santa Cruz), Arthur Wolfe (Uni. Cal-ifornia, San Diego) and Wallace Sargent (California Inst.Tech.). VLT spectra are publicly available on the EuropeanSouthern Observatory Archive. I thank the AustralianResearch Council for a QEII Fellowship (DP0877998)and a Discovery Project grant (DP110100866).

References[1] R. P. Feynman, ‘QED: $e Strange $eory of Light and

Matter’ (Princeton Univ. Press, NJ, 1985).[2] T. Rosenband et al., Science 319, 1808 (2008).[3] V. A. Dzuba, V. V. Flambaum and J. K. Webb, Phys. Rev.

Lett. 82, 888 (1999).[4] J. K. Webb et al., Phys. Rev. Lett. 82, 884 (1999).[5] M. T. Murphy et al., Mon. Not. Roy. Astron. Soc. 327,

1208 (2001).[6] J. K. Webb et al., Phys. Rev. Lett. 87, 091301 (2001).[7] M. T. Murphy, J. K. Webb and V. V. Flambaum, Mon.

Not. Roy. Astron. Soc. 345, 609 (2003).[8] M. T. Murphy et al., Lecture Notes Phys. 648, 131 (2004).[9] M. T. Murphy et al., Mon. Not. Roy. Astron. Soc. 327,

1223 (2001).[10] M. T. Murphy et al., Mon. Not. Roy. Astron. Soc. 378,

221 (2007).[11] J. K. Webb et al., Phys. Rev. Lett. 107, 191101 (2011).[12] J. A. King et al., Mon. Not. Roy. Astron. Soc. accepted

(2012), arXiv: 1202.4758[13] J. C. Berengut and V. V. Flambaum, Europhys. Lett.

accepted (2012) arXiv:1008.3957.


BIOMichael Murphy is an Associate Professor at Swinburne University of Technology’s Centre forAstrophysics and Supercomputing. He received his PhD in physics at the University of NewSouth Wales and spent !ve years at the University of Cambridge (UK) as a Research Fellow. Hereturned to Australia in 2007 as a lecturer at Swinburne and began an ARC QEII Fellowshipthere in 2008. He studies quasar absorption lines and the distant galaxies that cause them.

“After a decade of finding !to be ~6 parts per milliontimes smaller than on Earthin distant galaxies,predominantly in thenorthern sky, we now findthe opposite result on theother side of the sky.”

Is the fine-structure constant not actually constant? Clarifying the...

Documents

Transcript of Is the fine-structure constant not actually constant? Clarifying the...