v1 15 Dec 2003

M on.N ot.R .A stron.Soc.000,000{000 (0000) P rinted 4 January 2014 (M N LA TEX style le v1.4) O n the m agnetic accretor G K P erseiin outburst C ...
Author: Vanessa Parrish
5 downloads 0 Views 161KB Size
M on.N ot.R .A stron.Soc.000,000{000 (0000)

P rinted 4 January 2014

(M N LA TEX style le v1.4)

O n the m agnetic accretor G K P erseiin outburst C oelH ellier,Sean H arm erand A .P.Beardm ore Schoolof C hem istry and P hysics, K eele U niversity,Sta ordshire, ST 5 5B G

A B ST R A C T

R X T E m ade 5 X -ray observations ofthe m agnetic accretor G K Per during its 1996 outburst,recording a countrate often tim es the quiescentlevel.T he 351-sspin pulse show sa deep,nearly sinusoidalm odulation,in contrastto the weaker,double-hum ped pro le ofquiescence.T he spectrum show sabsorption increased by two ordersofm agnitude overquiescence.W e explain these di erencesin term softhe changing accretion geom etry as the outbursting disc forces the m agnetosphere inwards,and discuss the 5000-sX -ray Q PO s seen during G K Per’s outbursts. K ey w ords: accretion,accretion discs { stars:individual: G K Per { novae,cataclysm ic variables { binaries:close { X -rays:stars.

IN T R O D U C T IO N

T he dw arf-nova outbursts of cataclysm ic variables are thoughtto be caused by accretion-disc instabilities (e.g.O saki1996;Lasota 2001).In principle,such outbursts can still occur w hen the innerdisc istruncated by the m agnetic eld ofthew hitedw arf,asin theinterm ediate polarsubclass(e.g. A ngelini& Verbunt1989).G K Perisan exem plarofthis:its long,2-d orbitand short,351-s,w hite-dw arfspin period lead to a large disc surrounding a relatively sm allm agnetosphere. Its outbursts, w hich last for 50 d and recur every 3 y (Sim on 2002),can thus be m odelled using a disc-instability code w ith theinnerdisc m issing (K im ,W heeler& M ineshige 1992;Y ietal.1992). O utbursts in som e other interm ediate polars m ay also be the result of disc instabilities, for exam ple in X Y A ri (H ellier,M ukai& B eardm ore 1997),Y Y D ra (Szkody etal. 2002),H T C am (Ishioka etal.2002),and possibly EX H ya (H ellieretal.2000).Itislikely,though,thatshort-lived,low am plitude ‘ ares’seen in T V C oland V 1223 Sgr are caused by som ething else, such as m ass-transfer events (H ellier & B uckley 1993). O f the above system s, X -ray observations in outburst have been obtained for G K Per, X Y A ri, Y Y D ra and EX H ya. O utburst observations of G K Per include EX O SAT coverage of its 1983 outburst (W atson,K ing & O sborne 1985), G inga coverage of its 1989 outburst (Ishida etal.1992),R X T E coverage ofits 1996 outburst,and,m ost recently, C handra, X M M -N ew ton and R X T E coverage of the 2002 outburst (M auche 2003). W e report here on the R X T E observations ofthe 1996 outburst.In particularw e addressthe issue ofw hy the 351-s pulsation is strong and single-peaked in outburst (W atson etal. 1985) but m uch w eaker and double-peaked in quiescence (N orton,W atson & K ing 1988;Ishida etal.1992). c 0000 R A S

8

100

9

80

10

X-ray counts/sec

1

60 Visual magnitude

arXiv:astro-ph/0312380v1 15 Dec 2003

4 January 2014

11 40 12 20

13 X-ray quies

14

2450100

2450200

0

Julian Date

F igure 1. T he visual lightcurve of the 1996 outburst (circles), com piled by the A AV SO (M attei2003).A lso show n are the tim es and count rates of the 5 R X T E observations (squares); the line m arks the expected R X T E count rate in quiescence, predicted from the quiescent G inga count rate.

2

O B SE R V A T IO N S A N D R E SU LT S

R X T E m ade ve observations of the 1996 outburst of G K Per,lasting 9 h each,spaced over 41 d (see Fig. 1). W e m ake use ofthe PC A data extracted from the top xenon layer of PC U s 0, 1 & 2 in the energy range 2{15 keV ,and w ith the background estim ated using pcabackest v2.1e.

2

C .H ellier,S.H arm er & A .P.Beardm ore

150

Obs 1 (1996 Feb 27)

100

Counts/sec

50 0 Obs 5 (1996 Mar 08)

150 100 50 0

0

2000

4000

6000

8000 Time (secs)

10000

12000

14000

F igure 2. T w o sam ples of the 2{15-keV X -ray lightcurve of G K Per during outburst, in 16-s bins. E rrors are typically 3 counts s T im e zero corresponds to JD (T D B ) 2450140.9392 (top panel) and 2450181.6800 (low er panel).

T hem ean 2{15-keV ux in the5 observationsw as50 8 10 11 ergss 1 cm 2 ; for com parison, Ishida etal. (1992) report that in G inga observations ofthe 1989 outburst the 2{20-keV ux w as 48 17 10 11 w hile the quiescent ux is m uch low er at 4.0 0.4 10 11 ergs 1 cm 2 . W e show in Fig. 2 sections of lightcurve from the 1st and 5th observations.T he 5th observation show s the singlepeaked, quasi-sinusoidal m odulation that is typical of interm ediate polars. In contrast, the rst observation show s a shallow er, at-topped m odulation that is sim ilar to the quiescent pulse recorded by Ishida etal.(1992),despite the count rate being already w ellabove the quiescent level. For further com parison w e plot the folded pulse proles ofall5 observations in Fig.3,and their 6{10/2{5-keV hardness ratios in Fig. 4. N ote that observations 3 and 4, nearest the peak ofthe outburst,have the highest hardness ratios.T hiscan be seen from Fig.4 w here these observations are displaced upw ards from equalspacing.Spectralanalysis con rm s that allthe outburst spectra are heavily absorbed, requiring at least tw o partial-covering absorbers ofdensities 3 1023 cm 2 and 2 1024 cm 2 .T his w as also reported by Ishida etal. (1992) during the 1989 outburst, and contrasts w ith typical quiescent absorption of only 1 1022 cm 2 . Interm ediate polar spectra typically show increased hardness in the m inim a of the spin pulses, ow ing to increased absorption,and this is seen in observations 2 to 5. In contrast,the rst observation show s a m ore com plex and shallow er spectralchange. W e have used the last 4 observations,w hich span 35 d, to m easure the pulsation period (the rstw asom itted ow ing

1

.

to the di erent pulse shape). From these w e nd a period of351.335 0.002 s,w here the error m akes no allow ance for possible changes in pulse pro le.T his result is in line w ith M auche’s (2003) re-assessm ent of G K Per’s period change over that originally proposed by Patterson (1991).

3

T H E 5000-S Q P O S

In the EX O SAT observations ofthe 1983 outburst W atson etal.(1985) reported m odulations of the X -ray ux w hich had a strong am plitude ofup to a factor 2{3,and a quasiperiodic tim escale of 5000 s. H ellier & Livio (1994) proposed that they w ere caused by bulges in the inner disc,orbiting w ith the localK eplerian tim escale of5000 s,as m ight arise from an over ow ing stream reim pacting thedisc.T hese bulges w ould periodically obscure the line ofsight to the X ray em ission,causing absorption dips.T he dips are deeper at low er energies,in keeping w ith this idea. M orales-R ueda, Still & R oche (1999) analysed optical spectroscopy ofthe1996 outburstand found thesam e 5000-s Q PO s in the em ission-line pro les.T hey proposed an alternative m odelin w hich blobsatthe innerdisc edge orbitw ith a period ofeither 320 or 380 s.T hese w ould cause enhanced ow to the accretion curtains and hence enhanced absorption w henevera blob lined up w ith the m agnetic dipole,thus producing a m odulation at the 5000-s beat period betw een the blob orbits and the 351-s spin period. W hile this idea adequately explains the optical data analysed by M orales-R ueda etal. (1999), it is less able to explain the X -ray behaviour.Forexam ple,the extra absorption w ould not occur w hen the blob-fed curtain w as on the c 0000 R A S,M N R A S 000,000{000

G K Per in outburst

3

300 6

Ratio

5

Counts/sec

200

4

3

100

2

0 0.0

0.5 1.0 Spin phase

1.5

0.0

0.5 1.0 Spin phase

1.5

F igure 3. T he spin-pulse pro les from the 1st (bottom ) to 5th (top) X -ray observations. W e have added 50 counts s{1 successively to the upper 4 pro les. P hoton-noise error bars are typically 0.5 counts s{1 .T he data are folded on a period of351.335 s w ith phase zero corresponding to JD (T D B ) 2450168.14105.

F igure 4.T he 6{10/2{5-keV hardness ratios,folded on the spin cycle,from the 5 R X T E observations.W e have added 0.5 successively to the upper 4 pro les.P hoton-noise error bars are alw ays less than 0.1 (typically 0.05). T he phase zero is the sam e as in Fig.3.

far side of the w hite dw arf,as it w ould be for half of each spin cycle.Indeed,one m ightexpectthe extra accretion ow of a blob-fed curtain to result in enhanced ux for these spin phases.T he observations,though,show that the extra absorption of the Q PO occurs throughout the spin cycle, reducing the ux at allspin phases (see W atson etal.1985 and H ellier & Livio 1994).For this reason w e prefer m odels in w hich the structures causing the dips are circling at the 5000-s quasi-periodicity. M ore recently,W arner & W oudt (2002) have proposed a new understanding of the Q PO s and dw arf-nova oscillations (D N O s) seen in cataclysm ic variables.T hey associate D N O s w ith a m agnetospheric rotation period and suggest thatQ PO sare caused by slow -m oving prograde w avesatthe inner edge of the disc.W arner,W oudt & Pretorius (2003) show that a relation P Q P O =P spin 15 ts m any observations in cataclysm ic variables and X -ray binaries; G K Per obeys this relation w ith P Q P O =P spin 5000=351 14. G iven this,w e retain H ellier & Livio’s (1994) proposal thatG K Per’sQ PO sare dipping behaviourcaused by bulges m oving at 5000 s, but now regard W arner & W oudt’s

(2002) explanation of the bulges as slow , prograde travelling w aves as the m ost prom ising. M orales-R ueda etal.(1999)observed thatthe em issionline Q PO s w ere predom inantly blueshifted, and suggested that this argues against a m odelin w hich structure m oves at 5000 s. H ow ever, it is plausible that the w aves have a leading-edge/trailing-edge asym m etry in em issivity, w hich w ould explain the M orales-R ueda etal.result. T he relevance to this paper is that opticalobservations by M orales-R ueda etal.(1999) and N ogam i,K ato & B aba (2002) show that the Q PO s w ere present during the 1996 outburstdiscussed here.(W e have attem pted to detectthem in the R X T E data, but the search w as inconclusive since the data are broken up by R X T E ’s orbiton the very sim ilar tim escale of6000 s.) T heir existence show s thatG K Per has a su ciently high inclination that,at least during outburst, bulges ofm aterialat the inner disc edge are capable ofobscuring the line ofsight to the w hite dw arfand reducing the ux by factors of 2{3.T his is im portant for the discussion on the origin ofthe spin pulse.

c 0000 R A S,M N R A S 000,000{000

4

C .H ellier,S.H arm er & A .P.Beardm ore

F igure 5. A n illustration of the change in accretion geom etry during outburst,show ing the m agnetosphere shrinking m arkedly as the disc pushes inw ards. T he dotted lines show the 50{73 allow ed inclination range. T he arrow s show how the line-of-sight through the accretion regions can becom e m ore tangentialduring outburst.

4

C O M P A R ISO N W IT H X Y A R I: H ID IN G T H E LO W E R P O LE ?

Like G K Per,X Y A rishow sa low -am plitude,non-sinusoidal pulse in quiescence. T his m ost likely results from m inor asym m etries betw een tw o accreting poles, so that the disappearance of one as the w hite dw arf spins is not entirely balanced by the appearance ofthe other. In outburst, X Y A ri’s pulse changed to a very deep, quasi-sinusoidalm odulation. T he m inim a w ere not accom panied by increased hardness suggesting that absorption played no role. N oting that X Y A ri w as deeply eclipsing and thus at a very high inclination, H ellier etal. (1997), suggested that the inner disc had pushed inw ards during outburst,cutting o the line ofsight to the low er accreting pole.T husw e saw only the upperpole,and thispassed from the visible face to the far side and back as the w hite dw arf rotated,producing a near-totalm odulation. G K Per’s pulse also changes from low -am plitude and non-sinusoidalin quiescence to deep and sinusoidalin outburst (Fig. 3). It is thus natural to ask w hether the sam e e ect is occurring.A n im m ediate counter is that,unlike in X Y A ri,G K Per’s outburst pulse is energy dependent and caused prim arily by an increase in absorption at spin m inim um .T hus the m odelcannot apply straightforw ardly. N evertheless,letusconsiderw hetherthesystem param eters are com patible w ith hiding the low er pole in outburst. W e assum e a 1-M w hite dw arf (see M orales-R ueda etal. 2002),w hich,w ith a 351-s spin pulse,produces a m agnetosphere of 10R w d ,assum ing thatitcorotatesw ith a K eplerian innerdisc.T here are argum ents for a m uch larger m agnetosphere, based on spectra and m odelling the outbursts (e.g.K im etal.1992),but the presence ofpulsed X -rays in quiescence im plies that accretion is not centrifugally prohibited, so w e place the inner disc edge at the corotation radius. A linearscaling w ith opticaland X -ray brightnessw ould suggest that the accretion rate rises by a factor 10 during outburst.H ow ever,thisisnotin accord w ith disc-instability m odels ofthe outburst,and K im etal.(1992) and Y ietal. (1992) argue that the accretion rate rises by a factor 100{ 1000,w ith the X -raysbeing supressed by opacity.T he standard scaling of disc radius w ith accretion rate (r/ m_ 2=7 ) then im plies that the inner disc radius decreases by a factor

of betw een 4 and 7 (w e show an illustrative reduction by a factor 4 in Fig. 5). T he system inclination is w ithin the range 50{73 (M orales-R ueda etal.2002). T heaboverangesarenottightenough to tellusw hether the bottom ofthe w hite dw arfis hidden by the disc in outburst (particularly given the uncertainty in the disc thickness),but allow both possibilities.W e thus turn to considering the observations. T he high am plitude ofthe 5000-s Q PO s (reductions in X -ray ux by factors up to 2{3) im ply that bulges near the inner-disc edge obscure the upper accretion pole, and are thus of a height above the disc plane com parable to the w hite-dw arfradius.T his is reinforced by the observation of very high absorption (1024 cm 2 ) at all tim es during outburst.G iven the inclination range of50{73 ,and the likely innerdisc radius of< 3R w d ,such m aterialw ould likely hide the low er pole continually. N ote,how ever,that w e don’t see episodes ofnear-zero ux each spin cycle,w hich w e do in X Y A riw hen the upper pole sw ings round to the hidden face of the w hite dw arf. T his im plies either that the low er pole is visible at phases w hen the upper pole disappears, or that accretion regions at the upper pole are visible at allspin phases.T he latter is,at rst sight,im plausible for the above inclination range. T he sm allm agnetosphere during outburst im plies accretion regions far from the poles (> 35 m agnetic co-latitude for rm ag < 3R w d ), w hich, added to any dipole o set from the spin axis,w ill likely exceed the 90 i polecap region that is alw ays visible. So can w e reconcile thesecon icting indicators? W esuggest that, in outburst, the accretion ow overw helm s the m agnetosphere su ciently that accretion ow s to the poles from allazim uths (w hereas,in quiescence,each pole w ould be fed from a restricted azim uthalrange),and thus falls at allm agnetic longitudes.H ence accretion at the upper pole w ould alw ays be visible.T here is evidence from eclipse tim ings that exactly this occurs at the peak ofthe outburst in X Y A ri(H ellier etal.1997).Ifthe above is correct,it suggests that the angle betw een the spin and m agnetic axes is relatively sm all. c 0000 R A S,M N R A S 000,000{000

G K Per in outburst 5

T H E C H A N G E IN T H E SP IN P U L SE

T he fact that the pulse pro le in G K Per is largely an absorption dip, and not the near-total, energy-independent m odulation seen in X Y A ri in outburst, im plies that sim ply hiding the low er pole does not explain the change in pulse pro le betw een quiescence and outburst.W e are thus left to explain the fact that (1) in outburst G K Per show s a ‘typical’pulsation w ith a quasi-sinusoidalpro le resulting from a broad absorption dip,and (2) the absorption dip is m uch reduced in quiescence leaving a m ore-com plex,low eram plitude m odulation. W e suggest that the change results straightforw ardly from a com bination ofthree factors.First,the increase in accretion rate by a factor 100{1000 w illincrease the colum n density oftheaccretion curtains.Second,them arked shrinking oftheaccretion curtainsin outburst(seeFig.5)w illforce theincreased ow through a m uch sm allercircum ference and thusa m uch sm allerarea,furtherincreasing thecolum n density.T hird,as illustrated in Fig.5,the change in geom etry can easily result in the line-of-sight to the accretion regions taking a m ore grazing path through the accretion curtains, thusincreasing the line-of-sight colum n yetfurther.Such effects can explain w hy w e observe a phase-varying colum n as high as 2 1024 cm 2 in outburst,w hereas the quiescent spin pulse is m odelled by phase-varying absorption of only 6 1021 cm 2 (Ishida etal.1992). T hus,in outburst,the e ectofintense absorption dom inates the pulse pro le, w hereas in quiescence the m uch w eaker absorption, from tenuous accretion curtains m uch further from the w hite dw arf,is less signi cant than other factors,such as an asym m etry betw een the tw o poles.

6

C O N C L U SIO N S

(1) X -ray Q PO s are caused by bulges at the inner disc edge,travelling ata 5000-speriod.T hey likely correspond to the travelling w aves discussed by W arner & W oudt (2002). T he bulges are at a height above the disc plane com parable to the w hite-dw arfradius. (2) W e suggest that the X -ray Q PO s are not seen in quiescence ow ing to the inner disc being m uch further out, so that the bulges do not obscure the w hite dw arf. (3)D uring outburst,accretion occursfrom allazim uths, form ing a com plete accretion ring at the poles. T he low er pole is likely hidden. (4) T he pulse pro le changes in outburst to becom e dom inated by absorption;thisresultsfrom thegreateraccretion ow ,the sm aller accretion-curtain area,and the change in how the line-of-sight passes through the curtains. (5) M easurem ent ofthe 351-s spin period con rm s the period change reported by M auche (2003).

A C K N O W LE D G M E N T S W e thank Janet M attei and the m any observers w ho contribute to the A AV SO for com piling and supplying visual observations ofG K Per’s outburst. c 0000 R A S,M N R A S 000,000{000

5

R EFER EN C ES A ngeliniL.,V erbunt F.,1989,M N R A S,238,697 H ellier C .,B uckley D .A .H .,1993,M N R A S,265,766 H ellier C .,Livio M .,1994,A pJ,424,L57 H ellierC .,K em p J.,N aylorT .,B ateson F.M .,JonesA .,O verbeek D .,Stubbings R .,M ukaiK .,2000,M N R A S,313,703 H ellier C .,M ukaiK .,B eardm ore A .P.,1997,M N R A S,292,397 Ishida M .etal.,1992,M N R A S,254,647 Ishioka R .etal.,2002,PA SJ,54,581 K im ,S.-W .,W heeler J.C .,M ineshige S.,1992,A pJ,384,269 Lasota J.-P.,2001,N ew A R ,45,449 M auche C .W .,2003,astro-ph/0301582 M atteiJ.A .,2003, O bservations from the A AV SO International D atabase,private com m unication M orales-R ueda L.,StillM .D .,R oche P.,1999,M N R A S,306,753 M orales-R ueda L.,StillM .D .,R oche P.,W ood J.H .,Lockley J.J., 2002,M N R A S,329,597 N ogam iD .,K ato T .,B aba H .,2002,PA SJ,54,987 N orton A .J.,W atson M .G .,K ing A .R .,1988,M N R A S,231,783 O sakiY .,1996,PA SP,108,39 Patterson J.,1991,PA SP,103,1149 Sim on V .,2002,A & A ,382,910 Szkody P.etal.,2002,A J,123,413 W arner B .,W oudt P.A .,2002,M N R A S,335,84 W arner B ., W oudt P.A ., P retorius M .L., 2003, M N R A S, 344, 1193 W atson M .G .,K ing A .R .,O sborne J.,1985,M N R A S,212,917 Y iI., K im S.-W .,V ishniac E .T .,W heeler J.C .,1992, A pJ,391 L25